off-topic-tech

[jagged snow]

Project Digits!

If you're stupid into home automation and want to run a powerful llm locally, it's actually the perfect solution for that

www.nvidia.com/en-eu/project-digits/

The marketing around it is dumb

But it is a nifty product

ofc it is 3 grand us

[soft bloom]

https://www.youtube.com/watch?v=o1sN1lB76EA
looks like it can fit on a really tiny PCs. nice
i should try it

[mental oriole]

But wont that be slow af?

[soft bloom]

yeah, but if that's at least possible, it means 2 things:

• it can run on arm
• it can run on weak hardware

aka your home PC might be just enough to use it for "ask-and-forget" kind of interaction

[mental oriole]

So could other models hosted locally.

Anyways.

[soft bloom]

but they don't have nearly as good ratio of power

[mental oriole]

Sure, but point still valid.

[soft bloom]

and this headline might be just enough for some investors to turn away from most open model back to the greedy players.

[mental oriole]

You'd want at least 20t/s to be usable imo.

[soft bloom]

this vid shows 30-50ish on a single amd pro w7700 (yeah i also don't know wtf that is)

good news is that you don't need that tower of macminis

[mental oriole]

We'll have to see how it develops, its nice to see another competitor.

[night girder]

And "open-source".

[soft bloom]

I couldn't sit through it.
Does it mention the map of plate sunking/rising?
If you look at it, to an eye, it actually looks like sinking is a very localized and slow phenomenal while there are massive and fast oceanic ridges that form new surface

[dire igloo]

the map of what?
and about the "it actually looks like" part: https://youtube.com/shorts/_WFLiX3HI7o

he watches the Neal Adams video and adresses the claims while also showcasing the holes in the theory

biggest hole of them all:
where's all the extra shit coming from? you got the laws of thermodynamics, where's all the expansion mass coming from? where did the ocean water come from?

[soft bloom]

Minimal effort to answer the question: mass is energy, sun is source of energy, there you have it

But honestly I don't think it's fair to claim unanswered questions as holes while not doing the same to other theories. And remember that not every possible or plausible answer is the answer

Though, so far, I know about nuclear reactions going on inside Earth, so they work the opposite way

[dire igloo]

okay, holy fuck, did you do the maths on that?

E=mc²

[twin dew]

And what the fuck would cause that energy to be converted into matter deep inside the earth?

But yeah, even creating one gram of matter would need so much light...

[dire igloo]

even if you managed to convert that energy do mass, you'd need HEAPS of energy

[twin dew]

Even at 100% conversion ratio.

And Nuclear Fission is where matter splits and releases energy, not the other way around.

[dire igloo]

the expansion would also deccelerate

[twin dew]

After the fission, you have tiny bit less mass.

[dire igloo]

not accelerate

because the mass is proportional to the volume while the energy absorption is proportional to the surface area

[soft bloom]

Eh, how is that for or against anything? Sorry, I don't think I know how change of speed is ralted

[soft bloom]

Fair

[dire igloo]

Proof by contradiction.
The claim of an acceleratingly expanding earth can not be physically possible if the only source of expanding mass is absorption of the sun's radiated energy

[soft bloom]

I couldn't find today the excat one I remember, but essentialy, there are maps of age of ocean floor, easy to find.
And there are also for speed of their movement. I found some less detailed ones, but there has been one looking like those for age, just for speed

[dire igloo]

besides all that, earth itself is radiating off heat - and you still haven't explained the creation of new H2O

[soft bloom]

Who claimed acceleration of expansion?

[dire igloo]

had you watched Milo's video, you'd know about continental crust (the old af shit that's hardly moving) and oceanic crust (conveyor belt)

[twin dew]

Earths crust is being recycled all the time.
Not being expanded in mass...

[dire igloo]

oceanic crust arises somewhere in the ocean, moves towards continental crust, dips below it, melts up, repeat

[soft bloom]

If we still play this game: h20 is o3 + hydrogen

[dire igloo]

so there was a surplus of ozone and hydrogen that slowly gets converted into water?

except the much smaller earth had a much weaker gravitational pull so it couldn't have maintained a sufficiently dense atmosphere

[soft bloom]

So what makes it go under, melt and sink? I just honestly never got this part.
As well as how it forms anew on top.
Regardless of the more global theory.

[dire igloo]

plus, why is it only converting slowly as the earth expands?

[twin dew]

https://en.wikipedia.org/wiki/Subduction

https://en.wikipedia.org/wiki/Plate_tectonics

[dire igloo]

on example of where it forms anew is at the mid-atlantic ridge

[soft bloom]

Ok, but how is it pro/against/related the existence of maps that are like on most articles on plate tectonics?

[soft bloom]

Yes. Clear from maps

[soft bloom]

Wdym?

[twin dew]

The rate of new water creation would need to match the rate of earths expansion, to keep the water level same.

[soft bloom]

Better find out how ozone got here in the first place, and how it's levels are replenishing (considering that arctic hole restores - they are)

[dire igloo]

because oceanic crust is constantly being recycled in a conveyor-like fashion, it's obviously much younger than continental crust

[soft bloom]

But it isn't the same?..

[twin dew]

Because there is ratio that normal oxygen and ozone try to keep in upper atmosphere.
The holes were caused by chemicals changing that ratio and changing ozone to normal oxygen faster.

[dire igloo]

one of the very key "gotcha" points of expanding earth believers is that the earth's surface material is much younger underneath the oceans and moving towards the landmasses

[soft bloom]

Actually, it's much rarer to see map that tales ages for both. i was talking purely about ocean

[dire igloo]

and that's sufficiently explained by the oceanic plate conveyor belt

[twin dew]

Only happens in upper atmosphere because that needs external energy to trigger.

Or we can do that ourselves with electricity etc.

O3 isn't stable, O2 is.

When there isn't external trigger to convert 3x O2 to 2x O3.
Which then reduces with time back to just O2.

Way up oxygen is mostly O1.

(Single oxygen atoms, looking to pair with anything)

[dire igloo]

you have maps that tell the age of either, you can just combine them. you just need different means of measurement to obtain the data, so it's better to not mix the two

[soft bloom]

Yes, but I wasn't talking about that.
Basically, I remember that "by looking* at the map of speed of ocean floor relative to regions of subduction and .... Forming, it is "easy" to conclude that the rate of growth (aka formation) is higher than rate of sinking.
But that would need that map to check and see. Although age of crust should have relation with that as well.

[dire igloo]

for O3+2H → O2+H20 to work, you'd need a lot of H2 and a lot of O3 at the beginning of earth's existence.
If what you said was true, it'd mean that it's only converting over time, so why didn't it just all react almost instantly

[soft bloom]

Btw tht claim about anthropic chemicals is challenged by idea that hydrogen escapes earth into atmosphere and combines with o3

[dire igloo]

this is soooo super handwavy, I sure as fuck hope you don't base any actual educated opinion on that

ahem, LOOKS LIKE
https://youtube.com/shorts/_WFLiX3HI7o

[soft bloom]

Idk like why you pay so much attention to what started as "minimal effort" answer

[night girder]

That's this channel buddy.

[dire igloo]

also, let's assume that earth did have some means of converting sunlight into matter - what does it turn that energy into?

[night girder]

Say one thing wrong, or one misunderstandment and you have content for days,

[twin dew]

Extraordinary claims, with absolutely no backing?
When the normal explanation is usually seen as covering about everything already?

[soft bloom]

I need contextanager

[twin dew]

When the extraordinary claims seem so bogus even on surface level.

And there doesn't seem to be anything underneath either.

And seem like mostly a conspiracy theorist flat earth cult redux.

[dire igloo]

why does it pick silicon and carbon, what about some of the heavier metals?

[soft bloom]

I can't get enough of bot blocking scientific links

www.ngdc.noaa.gov/mgg/image/crustalimages.html

[dire igloo]

"minimal effort answer" ain't cutting it if you're spreading pseudoscientific mumbojumbo acting like you're proposing valid scientific theories

[night girder]

Aah, just saw this video. Too funny. "Petrified something" 🤣

I want to play this drinking game.

[soft bloom]

Yeah but I wasn't spreadind anything. Just asked questions and shared opinion clearly outlining the "looks like"

[night girder]

But I knew that giants were real.

Here 4 giants got stuck into a mountain.

Allegedly.

[dire igloo]

"I'm just asking questions" doesn't count.
My dude, you are actively engaging and perpetuating easily debunkable pseudoscientific bullshit

you're proposing this shit as an actually valid scientific theory

I can get the appeal of making a ludicrous assumption and playing "What If...?" but this ain't the way to do it

[night girder]

you know what you two should do? For every argument you make. Have some sources to back it up. Even wikipedia is fine 😉

[soft bloom]

I thought chat agreed that discussion needs a simplified version we share within the chat

[dire igloo]

already provided one, the milo rossi video

[night girder]

and else, maybe don't discuss anything since it always keeps boiling down to:

No proof.

[dire igloo]

only upon request if the provided source was deemed to complex to be evaluated

also, just use simple wikipedia

[dire igloo]

Hitchens's Razor:
what can be assorted without evidence can be dismissed without evidence

[night girder]

Atleast Baldur has sources to back up whatever he is saying.

[dire igloo]

??? #off-topic-tech message

[soft bloom]

Wait, what did i propose? The sunlight to matter? No, it was the intend even in the slightest. The growing earth? I asked if they mentioned my eyeballed observation

[night girder]

No offense to you fireworker. Just didn't see that. I saw Baldur 😊

[dire igloo]

YOU were the one that brought up expanding earth?!?!

[night girder]

But to me you two seems to get a lot into: "Give me proof" - "No proof given, but more statements made." - "Give me proof about that statement!" - "More talking without sources" ...

[dire igloo]

#off-topic-tech message

[soft bloom]

So my simple question is: why is there much more younger crust than older?
According to the conveyor belt, i would expect it to be close to 50/50

[dire igloo]

I'm repeating the talking points of the video (since they said they didn't watch it fully) and adding some of my own holes in the theory

[dire igloo]

why would you expect it to be 50/50?

have you ever looked at the distribution of landmass across earth's hemispheres? far from 50/50 too

why would anyone suspect a 50/50 distribution?

[soft bloom]

It's not about hemispheres

[dire igloo]

also: had you read Baldur's wikipedia article, you'd have seen it mention that the conveyor belt also takes off a bit of continental crust

[dire igloo]

it's merely an example of why expecting a 50/50 distribution is nonsensical

[soft bloom]

For a real conveyour belt, imagine assiging age as what time that line of belt spent being on top.
In that case, half of top lines are below middle age (exact half of max age, middle of the belt), and half above middle age

Ofcourse that's simplification that doesn't account for stretch and compression

[night girder]

How do you know there is more "young" crust than older based on this image? Image is from url you provided.

[dire igloo]

ding ding ding

[soft bloom]

I hvae more questions after you sced this. What does "takes of a bit" mean?

[dire igloo]

you measure by volume, not by distance or surface area

[soft bloom]

Green is middle, there's color bar at the bottom left

[dire igloo]

if it crumples up towards the continental plate, that's gonna look like a slowdown without necessarily being one

[night girder]

yes, I know.

How do you know there is more young than old? You calculated it by color?

Or just gut feeling by eye?

red is young btw. blue is old.

[dire igloo]

as long as the same amount of MASS is melted as is expulsed, it's a loop

[soft bloom]

"by eye", you can fit pretty much all blue areas into red ones (very young) without moving the latitude much

[soft bloom]

But, you see, this won't cut it for the surface to stay the same.
I brought this up because i thought maybe that was what fueled the idea of expanding earth. Because when I looked at these maps, i had a question why it looks like that, aka why it doesn't satisfy this intuitive conclusion.

[dire igloo]

https://tenor.com/view/shot-shot-glass-shut-the-fuck-up-cheers-tasty-gif-12762582

[dire igloo]

have you tried "maybe my intuition is wrong"?

[soft bloom]

No, it's job of scientists to explain how the world works. They proposed they theory

[dire igloo]

it's a perfectly valid question to ask, but it in no way justifies bringing up pseudoscientific bullshit

[dire igloo]

exactly. so go ask the scientists how the world works

[night girder]

Young Oceanic Crust (less than 10 million years old): Approximately 20-25% of the oceanic crust is considered young.
Old Oceanic Crust (more than 10 million years old): Approximately 75-80% of the oceanic crust is considered old.

Source: you don't want to know.

Young Crust (Oceanic Crust): Approximately 60% of the Earth's crust is oceanic, which is relatively young.
Old Crust (Continental Crust): The remaining 40% is continental crust, which includes both young and very old sections.

So, ocedanic crust vs continental is 60-40. Split oceanic further down to new and old and yo uget 20-25% vs 75-80%.

[dire igloo]

https://tenor.com/view/armstrong-made-it-up-gif-25653819

[night girder]

No. AI

Now I am curious where that answer comes from. What data it was trained on to give that answer.

But what should* this proof anyway?

[soft bloom]

You mean saying that maybe it was behind conspiracy (as in making people believe, not in the meaning of someone actually hiding anything)

[soft bloom]

I will also ask them if they have done math to check if the map even alligns with their claim, and by how much.

[dire igloo]

I meant that it's perfectly fine to ask that question and that there is zero reason to bring up expanding earth or other pseudoscientific stuff

[soft bloom]

If thta was true, it would put under question why we have so fucking large proportion of a very young crust (the maps are usually ranging in 200-300my).

[night girder]

That's explained in the answer.

While oceanic crust is constantly being created and destroyed, continental crust tends to be more stable and can contain much older rocks.

And I think fireworker or Baldur pointed that out earlier in the conversation?

[dire igloo]

i think it's about relative amounts, not relative age

basically: why is there more of one than the other

[night girder]

Can you clarify?

Why is there more oceanic crust than continental? Or why is there more younger than old?

[soft bloom]

But isn't it interesting where scientific curiosity turns into conspiracy?
I mean, you won't find much followers of idea that water is actually an alien poop, but when flat earth gains followers because it's "baked" by everyday observation and "intuitive" conclusions that most people are simply not educated enough to understand accurately.

[dire igloo]

that was my attempt of clarifying someone else's point

[night girder]

Oh 🤣

[soft bloom]

Well, the map i linked only has are and age. Somewhere there is a map for speed.
You should be able find a map for thickness as well. Maybe gravity anomalies are related as well

[dire igloo]

the difference between science and pseudoscience is that one draws conclusions from evidence while the other draws evidence from conclusions

you have a theory, you look for evidence why it is correct

[night girder]

Reasons:

• https://en.wikipedia.org/wiki/Seafloor_spreading
• https://en.wikipedia.org/wiki/Subduction (posted by Baldur)

[soft bloom]

I am not talking bout continental crust at all. Purely about oceanic one.

[charred relic]

lol silly globe earhters...round

how silly...

[dire igloo]

sounds correct enough on a surface level, but then you get to the actual scientific part:
you gather evidence on why it wouldn't be correct.
You collect so much evidence that your theory is the only viable conclusion

[night girder]

well, seafloor spreading is about oceanic. and is one possible* explanation why there is more young than old.

[dire igloo]

the flat earth theory has supporters all around the globe

[charred relic]

it's true!

[soft bloom]

That's really not how either of them work. Sadly.
And that sentence perfectly fits the pseudoscience science driven by confirmation bias. But even more it describes the worst things in how do i name it, official science? Numerous "scientific" studies and research that are not even attempted to be replicated, that "found evidence" etc. that's what is behind those scams with bogus research without reviews.

Without even trying to disprove your findings you are destined to make same errors as flatearthers

[dire igloo]

practiced science isn't free of flaws either. scientists need to eat and sleep, their money has to come from somewhere or someone

go look into anti-intellectualism

the underlying motivations are different - pseudoscience aims to plant the seed of mistrust

[night girder]

Is flatearth pseudoscience?

And if yes, what is there* to gain in planting that seed of mistrust I wonder.

[dire igloo]

and about that last part: one scientist can, maybe even an entire team. but across millions of people on earth, eventually you'll reduce the amount of random mistakes
do you remember what happened to the proposed room temp superconductor LK99?

[dire igloo]

for the most prominent flatearth spreaders: attention from a very committed community
for most believers: perpetuating a planted seed of mistrust

[soft bloom]

Thing is that I don't distinguish between official, governments private, pseudo, alternative etc. I shouldn't care.
There's just a question of having evidence, its quality.
Maybe a bit of goals, aka what theory is attempted to get proven (like, would you even listen to someone trying to prove that some race is somehow superior?..).

You shouldn't let authority or dogmas influence your ... Towards new evidence, or even ideas.
In your mind, you can make conclusions towards anything, but actually contributing takes effort.

[dire igloo]

I recommend you watch Milo Rossi's lecture on the pseudescience pipeline

[soft bloom]

In some countries, you have to not trust officials.

[night girder]

No, I cannot make conclussions towards anything into my mind. You need facts. Proof. Data to make good conclussions.

[soft bloom]

Yeah but I did

[night girder]

Science provides data. So we can make conclussions on it.

[soft bloom]

There was no success for most of them, and then... Someone found something similar? I don't remember resolve

[edgy hazel]

Jesus fucking christ

[dire igloo]

you're doing a bad job of making a good point.

arguments of authority on their own shouldn't hold much value and they usually don't
it also shouldn't matter who came up with a theory or evidence, however it should absolutely matter HOW it was gathered - look up the chocolate diet. Especially in fields with lots of variables or uncertainty, there are ways to manipulate experiments and data to make them produce the result you want.
however, this should not be justification to just treat every single theory proposed by anyone as a valid scientific theory - a good theory has to be independently provable/confirmable and attempt to explain observations we make. a theory that fails either of the two criteria is inherently worthless.

[soft bloom]

Okay, prepare for something mind bending.
Do you know who discovered volcanism on our Moon?

[dire igloo]

a group of German scientists followed their instructions to the letter while eliminating impurities the best they could
turns out that the Korean group of researchers had a contaminated sample that would've shown to not be superconducting after some more testing - and the German group tested LK-99 to actually be more of an insulator than a conductor

[dire igloo]

https://tenor.com/view/trump-china-gif-284632875632280687

[soft bloom]

Ok so no good end sadly.
Remember the insulator part, but not the contamination.
Surprising how great promise made so many teams try this.

[dire igloo]

ding ding ding

that's the idea behind science

you think you have a revolutionary idea or breakthrough?
make it accessible to as many people as you can.
make it as easy as possible for your work to be verified.
but if you turn out to be wrong, admit as such

peer review holds immense power

[dire igloo]

ironically, the contamination is what made the sample test the way it did

[soft bloom]

I broke chatgpt trying to get a short summary of history, hehe

I was talking about Kozyrev

He has his name associated with quite a bogus metaphysics topics... But the way he's idea that moon has volcanism was treated in ussr is quite sad.

[soft bloom]

Dam, i think the term "mainstream science" got to me from reading what chatgpt writes sometimes...

[manic cipher]

Any information on these CPUs?

https://pcpartpicker.com/products/cpu/#m=21&s=13,61&F=63&f=78&k=35&sort=price&R=5,4

[jagged snow]

Those are probably bogus

I'm not at all sure what's going on there

Either way, don't pay for skylake unless you get it very cheap and used

[soft bloom]

speaking of Siberia heating up
https://www.youtube.com/watch?v=bYnLCpYoqas

[glossy glacier]

Old Intel HEDT CPUs.
Got beaten by AMDs Threadripper, now they are discontinued

[manic cipher]

I see.

The motherboard that I have is a X299 Micro made byu EVGA and it only supports Intel Processors in a LGA 2066 socket.

[glossy glacier]

If you don't need the additional memory channels or PCIe lanes anything modern will run cycles around these CPUs

[manic cipher]

I use a 7740X Processor currently.

[glossy glacier]

Get any AM5 or Whatever's the current intel socket system and you'll almost double or more your CPU performance

[charred relic]

So many GB of games installed last night...

[pure karma]

get a new one and get a actually decent cpu instead of staying on a platform thats already been dead for a while now

[languid gulch]

for the price, yea

[charred relic]

I have a real nice 2.5" SSD I'll sell ya, whopping 90GB

You will never need storage again

[manic cipher]

I already have a 500GB SSD that uses NVME.2 and I have a 1 TB HDD.

[wanton orchid]

$300 💀

[wanton orchid]

you can get 9800x for less than $100 on ebay or something (technically new)

[wanton orchid]

even when talking about memory channels it get blasted by good ddr5 dual channel

newer cpu got better cache

[soft bloom]

https://youtu.be/8nIB7e_eds4?si=kxseEDo8O8FUuMN9
just beautiful

[dire igloo]

Don't filter by reviews

How reviews work on PCPP:

• you need to make a build on the site and have it registered as a build
• in your build post, you need to review the part and give it a star rating
• PCPP only counts reviews that came from PCPP user builds

Even compared to other customer reviews, the informational value they hold is incredibly low.

And you can just ignore customer reviews in general.
A bad review could've been user error and most often addresses issues that fall under warranty.
A good review almost always comes from someone who doesn't have the knowledge/experience/skills/equipment to make an educated judgement and more often than not has people give good reviews to bad products because of it - plus they usually don't have a good base for comparison

Basically:
"Wow, this 32 inch full HD curved VA monitor is incredible, much better than the old 40 inch burnt in plasma TV I got handed down. Five stars!"

A shitty product can still look good when compared to an even worse one

[verbal raft]

HUH

[pure karma]

Seems about right since thats in line with what they use to charge once upon a time

[dire igloo]

I dislike these calculations, way too often they're disregarding the entire supply chain and only consider cost of raw material plus labor

[twin dew]

Just cost of the wafer that contains the die probably.

And very hand-wavy calculation of that too, and that headline cost is the lower bound one.

[dire igloo]

Also: quite a misleading headline.
"Could cost as little as" means nothing other than "this is the lowest number we came up with"

[twin dew]

Might not even take yields into effect, just wafer area / per die area to get how many dies per wafer.

[dire igloo]

If I wanted to make a 5090, assuming I somehow got my hands on Nvidia's schematics, I'd have to buy machinery, pay for a place to put it, etc etc

It'd be much more affordable to just buy one from an AIB

[tribal kraken]

In 2,5 hours we see how the pricing matches to demand.. Looks like the number of 5090 cards are low

But if script bots beat me to buying it, just game with 4090 until get one of the shelf. There was Reddit post from Microcenter stock numbers, about 1:10 ratio of 5090 to 5080. And total of 233pcs. 5090 cards all Microcenters in USA...

[dire igloo]

With almost a third of the MC cards being in Tustin, CA

[charred relic]

i love microcenter

too bad my closest one is a 3 hour round trip assuming traffic is reasonable

[night girder]

Security researchers from the Georgia Institute of Technology and the German Ruhr-Universität Bochum have found new vulnerabilities in Apple chips. This makes so-called 'side-channel' attacks possible, with which sensitive data can be stolen.

Herg, you already read this news?

• predictors.fail/files/SLAP.pdf
• predictors.fail/files/FLOP.pdf

Using indirection, we
show out-of-bounds addresses being read, the wrong function
being executed, and finally sensitive login-protected data be-
ing exfiltrated across websites in Safari and Chrome. Finally,
we identify that the LVP can be disabled via DIT.

Bettter patch up those Apple machines Apple response:

"We want to thank the researchers for their collaboration as this proof of concept advances our understanding of these types of threats," Apple told BleepingComputer.

"Based on our analysis, we do not believe this issue poses an immediate risk to our users."

[dire igloo]

https://www.bleepingcomputer.com/news/security/new-apple-cpu-side-channel-attack-steals-data-from-browsers/

Sauce

[night girder]

So, it turns out.

[dire igloo]

RUB is one of the better universities in Germany if you wanna get into infosec btw, not surprised to see their name there

[night girder]

social.treehouse.systems/@marcan Hector Martin from Asahi Linux did more investigations.

And it turns out, if you enable right CPU browser security, it's a non issue.

Researchers doing vuln research for Apple hardware, you might want to consider talking to us before publishing. We know how this stuff works and how to reverse engineer chicken bits in a few minutes.

Oof.

Edit: Duh, the paper has a subtle mistake. It shows the array contents as "S", "2S", etc. However, the array pointers are unscaled indices, so it should be "S/4", "2S/4", etc. (because the items are 32-bit ints). So in the ecore case, it exceeds dcache size. Fixing this makes the ecore/pcore behavior identical... so something is still off.

So, pretty interesting stuff going on.

So there you have it. DIT fixes FLOP, and SSBS fixes SLAP. Those are documented, architecturally defined mechanisms to control this behavior. No chicken bits needed.

The CPUs are working as intended. Browsers just need to get their head out of the sand and flip those bits when running untrusted JS.

Fucking JS, I knew it.

[dire igloo]

Hm, I wonder if it's a "stop running our stuff out of spec" type of thing or more of a "we introduced an easy to miss requirement to cover our asses" situation

Could also be a situation similar to the Luca app in Germany:
A COVID contact tracing app where users could manipulate their account data in such a way that the department of health officials would receive macro malware when downloading the app's Excel reports.
Developers said that that wasn't actually a security vulnerability because "everyone knows you shouldn't enable Excel macros"

Basically: Apple fucked up but because there's some common practice that should prevent that, they're out of the waters

[night girder]

developer.arm.com/documentation/101550/0001/AArch64-registers/AArch64-register-descriptions/AArch64-Generic-System-control-register-description/SSBS--Speculative-Store-Bypass-Safe

Seems more like an architecture thing?

SBSS:

Speculative Store Bypass Safe.

Prohibits speculative loads or stores which might practically allow a cache timing side channel.

A cache timing side channel might be exploited where a load or store uses an address that is derived from a register that is being loaded from memory using a load instruction speculatively read from a memory location. If PSTATE.SSBS is enabled, the address derived from the load instruction might be from earlier in the coherence order than the latest store to that memory location with the same virtual address.

0b0

Hardware is not permitted to load or store speculatively, in a manner that could practically give rise to a cache timing side channel, using an address derived from a register value that has been loaded from memory using a load instruction (L) that speculatively reads an entry from earlier in the coherence order from that location being loaded from than the entry generated by the latest store (S) to that location using the same virtual address as L.

0b1

Hardware is permitted to load or store speculatively, in a manner that could practically give rise to a cache timing side channel, using an address derived from a register value that has been loaded from memory using a load instruction (L) that speculatively reads an entry from earlier in the coherence order fro that location being loaded from than the entry generated by the latest store (S) to that location using the same virtual address as L.

And there are systems in place to protect it. But if people flip those off 🤷‍♂️ (as claimed by Hector Martin)

[twin dew]

Basically lot of the speculative execution side channel protections cause lot of performance regression.
So that is designed so that by default it is open to attack, but developer is supposed to mark the code that needs that protection in place.

Like executing untrusted JS from internet.

[night girder]

That's why he points towards browsers/websites and not CPU.

[twin dew]

Default is unsafe but fast, and by setting a flag, you get slow but safe.

Or if you have code that handles secrets, you can set the flag for that code, and external things cannot attack.

So you needed both browser that didn't set the flag for JS execution, and secrets handling code that didn't set the flag for itself.

[night girder]

Again, that's exascxtly what Hector Martin said 🤣

[twin dew]

Yeah, just dumbing it down to few lines.

[night girder]

Why. Just read source. No cause for argument if we just post sources 🤣

[twin dew]

What argument?

[night girder]

Nvm 🤣

I do wonder, why is this a Mac problem?

Is it because of the architecture of the CPU?

[twin dew]

Just someone who selected specific apple chip as target to try to find something.

[night girder]

paper describes L1 and L2 caches being shared. But, isn't that for a lot of CPU's?

[twin dew]

And found that you get speculative execution data leakage if you don't use speculative execution data leakaga protection mode.

[night girder]

Both types of cores have private L1 caches and
shared L2 caches within a cluster of the same core type. The
caches are set-associative: they are partitioned into multiple
cache sets using part of the memory address, and data from
that address can fit into any of the cache ways in that set,
where each way contains a cache line.

Os this specific for Apple silicon?

[twin dew]

Just normal cache behavior AFAIK.

[night girder]

Ok, need to read the papers more. I am not sure why it's pinned on Apple and not more as a global vulnerability for CPU's.

[twin dew]

There isn't anything special but that by default the specific CPU being tested has been designed to not be "safe", but to be fast.
And given that special flag to toggle to safe mode instead.

Most CPUs don't have that kind of thing.

Intentionally.

[night girder]

In the landscape of the decline of
Moore’s Law birthing more exotic microarchitectural op-
timizations, we believe that LAPs may not be an Apple
exclusive, either now or soon. As such, we emphasize
the need for novel hardware and software countermeasures
against LAPs in future work.

That's from SLAP paper. So they acknowledge it.

On the other hand, we note that the DIT bit is not set (and
thus the LVP is enabled) in the macOS kernel, where in Sec-
tion 4.4 we demonstrated the possibility of in-kernel LVP
training and exploitation.

And that;'s how the FLOP paper ends.

[wanton orchid]

it can also be a we cover our ass with features that was actually already used, and claim the browser didnt use them

[dire igloo]

fwiw, it could be a general problem. It's just that the researchers who "found" it did so on Apple devices because that's what they were testing.

Could also be an issue with apple-specific versions of these apps/browsers where the devs didn't turn ln the protective features - while the Android/Windows/Linux counterparts do have them enabled

[dire igloo]

That'd be the case if it was apple releasing the "it's the browsers" statement.
It was an independent researcher tho who confirmed and explained this behavior through their own testing.

It's just a question of why the features were left disabled - did the browser makers screw up because they forgot it or did apple make it super easy to miss? Could also be both fwiw

But it sounds like those settings are on a deeper fundamental architectural level that apple silicon is built on top of.
Stuff you'd find in any ARM-based CPU, not just Apple's

In which case, definitely a browser maker screwup

[dire igloo]

nvm

[wanton orchid]

yea usually it's safe by default, and performance apps can toggle it off

[dire igloo]

Odd

[languid gulch]

so a fun thingy happened. literally no budget, so top of the line everything

[dire igloo]

Why Corsair?

And why strix?

If you have actual unlimited money, you would want to go for the best of the best hardware, no?
Not generic big marketing budget fancy name brand stuff

[languid gulch]

Was brand restricted

One of the usual caveats when you get the 'no budget' option

If I had unrestricted options, then yes, I would have made different choices

[dire igloo]

How is that a usual caveat?
Only things I can think of are sponsored builds and uneducated client

Anyways, it's not top of the line everything, it's as good as it gets within the limitations

[stray badger]

Rich ppl wanting big brand most likely

[languid gulch]

yup, sponsored

and it's a 9800X3D & 4090 OC

and definitely not an uneducated client

[pure karma]

Im not sure what id do if i were to win some kind of giveaway pc

Probably part it out and use the stuff thats worth it

[sharp oasis]

Sell it

Just straight up sell the thing

[dire igloo]

many giveaway TOS have a clause that disallows selling the item within some time after

[languid gulch]

It handily beats my 5600X/6800XT, so I'm hanging onto it

Plus this lets me revamp my current system and hand it off to a family member

[dire igloo]

how'd you win it?

[soft bloom]

So i have been playing with correlation lately and got an idea...

www.aimspress.com/article/id/668fb539ba35de127fcb6dde#:~:text=The type of causality appears,by [CO2] increase.

This is very recent, but somehow was the first result.
Ye, i also was like "wait T causes Co2? No way"
Surely someone must have tried that sooner, right?

I wanted to ask is there anything about lag between CO2 and T, but this paper I mentioned above claims that those samples also support the T causes Co2 relationship.

Pretty damning

[dire igloo]

solubility of gases in water goes down as temperature increases, so yes, increasing temperatures cause a higher CO2 concentration. that is a well known and well documented effect

doesn't mean that that is the only way the causality goes

to add to this: the greenhouse effect is also well-documented

and of course: looking across millions of years of earth climate history neglects one crucial part: human civilization has been around for only a couple thousand years

so maybe the planet will be fine, doesn't mean we as humans will be

[flint stone]

Switched my PC case and CPU cooler, Satisfactory now runs up to 40°C cooler.
Hooray

Pardon me wrecking the convo

[soft bloom]

Speaking of how modern (aka past 50 years) can be manipulated, sadly.
Or rather, how many forms of activities that rely on grants, can be put to work in certain direction.
It's quite easy to notice how the "anthropic co2 is bad for climate" has become dogma, directing research to look for proofs and how to deal with it, without questioning much what are other variables, causes, nature of that relation.
I have heard a couple times that papers that are not inline with this notion are simply not published.
You can kinda say it's a conspiracy, but sadly no more a theory than avenue paying for music band to play certain playlist live (allowing for artistic expression within)

[soft bloom]

I have a feeling you missed reply addressed at some person who tries to deny climate change

[night girder]

40 degrees Celsius is a crazy drop though.

[dire igloo]

"temp causes CO2" is a common climate change denier talking point

[twin dew]

And many have switched to that "it isn't us, it is just the world changing"

From complete denial.

[flint stone]

Ever since I upgraded from a 3 3200g to a 3 3600, while keeping the stock cooler and a shotty Be Quiet 500 something with poor airflow it went to 80°C in satis

Now it hovers at 43°C in a Corsair 5000Daf using a Thermalright Phantom spirit 120 SE

[dire igloo]

or "it's not THAT bad"

[twin dew]

As in just another natural swing.
When it is just off the charts in change speed compared to anything before.

[flint stone]

(Overkill in size and cooling power but it is also planned for an upgrade later on)

[night girder]

You mean ryzen 5 3600?

[flint stone]

Oh yeah my bad

[night girder]

Thermalright, the one you got, is a good CPU cooler (not my opinion, opinion of multiple people in this discord).

[flint stone]

Yeah. Good ones, and so silent too

[night girder]

I am not familiar with the Corsair 5000Daf airflow though.

[dire igloo]

Thermalright first and foremost is a brand

[night girder]

Especially for its price.

[night girder]

Read my friend. Read.

[dire igloo]

ain't that 4000D?

[flint stone]

Paid like 55€ for it which is fine tbh

[dire igloo]

I did, your wording was just poor

[dire igloo]

case or cooler?

[night girder]

I just got unlucky when I ordered one. I got a crooked one 😄

[flint stone]

Nope a 5000D Airflow

[dire igloo]

first is dope, second is too much

[night girder]

No it wasn't. Your english is at that point.

[flint stone]

Cooler. Case was 180€

[night girder]

It's frustrating, not the first time you call me out on something unnecesary.

[dire igloo]

what reason was there to explicitly name the brand?

it's redundant info

[night girder]

Thermalright, the one you got

Ok, let dissect. Thermalright is a brand, "the one you got" refers to the model PD plankton25 bought.

So what I am saying is; Thermalright Phantom spirit 120 SE is a good CPU.

[soft bloom]

Glad you didn't say "looks like" or I would have took a shot.
Ding ding ding

Why not just ask?

[night girder]

Did you watch some reviews about the case? On* what did you decide your decision for it? (just curious)

[dire igloo]

yikes, what country are you in?

[soft bloom]

So it's good strategy to label based on the intersection of talking points?

[dire igloo]

"the one you got" already references the product, that info includes the brand
there is zero reason for you to explicitly bring up Thermalright if you're merely talking about the product

[stray badger]

Yes

[night girder]

Thank you Flare 🙏

[dire igloo]

why bring up conspiracy talking points without proper context?

[night girder]

https://tenor.com/view/oh-f-off-gif-27333127

[dire igloo]

it's the exact issue with that information - it is correct but presenting it without appropriate context makes fertile ground for bullshit

[night girder]

High horse. Get off.

[dire igloo]

"These results contradict the conventional wisdom, according to which the temperature rise is caused by [CO2] increase."

yeah, except you can easily do a greenhouse gas experiment at home and produce evidence for CO2→T

[soft bloom]

Which are?...
That grants have goals?
Or that scientists need money to live?

[twin dew]

Which causes individual papers to be crap.
Not that all research for last 50 years is crap.

[dire igloo]

Experiment setup starts around the 04:00 mark
https://youtu.be/m7o5O5JrMEk

[dire igloo]

that all of science is flawed, that scientific consensus is inherently dubious and that any alternative proposition is to be treated as valid

individual issues even out when looked at on a large enough scale

[soft bloom]

I think you are missing the point.
And historical data.
...
Basically, that's strawman in practice: you don't argue with me, you argue with some dumb climate denier.

[soft bloom]

Is there any basis to that statement?

[dire igloo]

true, to an extent

[soft bloom]

Science is flawed, I thought multiple members of chat agree on that.
Consensus can be manipulated, and it's something what has evidence to it, and could be proven or debunked.
Nobody said that any alternative should be considered valid, especially without evidence.

[flint stone]

Ye, Gamers Nexus

Decision based on spacing, looks, and cooling performance

[dire igloo]

the mathematical principle of averages: random variance disappears over a large enough sample size

[night girder]

If price isn't a concern 😄

[flint stone]

It usually is, I’m a student in last year

But I thought I might as well do it properly

[soft bloom]

But here's the thing: was it 50 years of trying to prove (and proving?) that co2 causes T on small and large periods of time? No. There are works on that, but how many and what are their results.
We shouldn't compare longevity and amount of research that is based on assumption to the quality of research that tests that assumption.
It's the basis of critical thinking and fact checking. Quantity doesn't equal quality.

[dire igloo]

Science is flawed, doesn't mean that anything science ever produced is to be held as flawed.
Consensus can be manipulated, but it's hard and takes a lot of resources - also just because it CAN be done, doesn't mean that it is generally the case
and it's not about validating things without evidence, it's about validating alternatives with lackluster evidence, improper context and neglecting consensus evidence

[dire igloo]

the more indepentent researchers, experiments and projects come to the same conclusion, the more reliable this conclusion is.
So no, in this context, quantity is absolutely a metric of quality.

[soft bloom]

This is no more valid than flatearth claiming Sun is much closer "because math and geometry". Where are actual observations? Experiments?
Oh, wait, right. The razor...
No, razor is sharp. You talk with analogy, i can too: averages work, in orchestra, when many people play the same notes, individual variance disappears. Because they play the same music, orchestrated.

See how easy it is to just yap?

[soft bloom]

Oh boi, you continue on arguing with someone else. Ok

[twin dew]

When that quantity is cross referencing each other by different people with different methods.
And no-one else has been able to point to another direction in any meaninful way.

[dire igloo]

actually, you are the one that's generalizing. Show that issues like publication bias actually reinforce a false/inaccurate consensus

[soft bloom]

Easy experiment. Go on and list at least a dozen research on causation between T and CO2.
And reas results, including caveats.

Being educated is not about simplifying everything to binary answer, but about knowing nuances.

[dire igloo]

then stop acting like you are someone else???
i don't believe in pseudoscientific misinformation, stop bringing it up? stop reinforcing misleading or inaccurate viewpoints?

[night girder]

I do agree that science has flawns, but imo this is the best we have at the moment:

https://en.wikipedia.org/wiki/Scientific_method

[dire igloo]

it feels like you want to combat a black-or-white view but in the process muddy the waters and present everything as the same shade of grey

there are greyscales, but there are definitely very dark and very light greys

[night girder]

If this procedure isn't followed, I have a hard time believing in it.

The image is also missing a part, peer reviews.

[dire igloo]

I've sent you a video. it's in German, but the setup is simple:
one thermometer underneath a glass dome, another directly next to it. shine an IR lamp on the dome to heat it up, let the thermometers settle and write down the temp difference.
Next, flood the dome with CO2 (dude in the video used dry ice), then repeat.
Observation: the air inside the dome gets much hotter when it's CO2

simple proof for the greenhouse effect, you can even reproduce it at home

yet the study you posted claimed that it's not true that higher CO2 levels cause higher temps

[languid gulch]

i've just stopped interacting with kek because i think it's just turned into a misinfo account

[tough owl]

Dead internet theory

[dire igloo]

spread misinfo, get called out on it, "nuh uh, I'm just asking questions", deny any accusations of perpetuating conspiracies, insinuate strawman

[languid gulch]

yea, the "just asking questions" makes me instantly reject it. "wanting to learn more" never seems to be brought up

[night girder]

fair point

Also, I just think there might be a bit of miscommunication because of language.

[languid gulch]

a lot of it is cherry picking & gish gallop

[dire igloo]

also: fallacy fallacy
yourlogicalfallacyis.com/the-fallacy-fallacy

[dire igloo]

not just a bit

another common issue with online discussions is moving further away from each other's position in an effort to move the person only a bit closer

[languid gulch]

also called walking off a cliff

[night girder]

also called: pointless

It is the longest discussion I've read on this channel. Must have been a few days by now.

And after all those days. I am non the wiser tbh.

I did learn a bit about earth crusts, things I didn't know.

[dire igloo]

btw: Hitchens's Razor requires a request for said proof to be made and denied/ignored

you can't just dismiss something because there wasn't evidence attached to it, you should at least have the decency to request it and see if you get it

no mathematician will request you to prove that Pi is the ratio of a circle's circumference and its diameter

[edgy hazel]

https://tenor.com/view/monkey-gif-18890916

[soft bloom]

https://link.springer.com/article/10.1007/s00704-023-04444-x#Tab7
This one is great, especially for me trying to understand nuances of Granger
This one also studies CH4

[soft bloom]

It got revived today.
And I don't think it's as long... Definitely won't compare to hardware stuff. Maybe longer than ai "take cover!"

[night girder]

This is when it started. 17 days ago.

[soft bloom]

mmm, I think that is fairly distant topic, but sure can be grouped into a larger twingle of threads

[edgy hazel]

it's the same circlejerk popping up over and over and over again

[soft bloom]

You had me in the first half, and then you lost me.
Essentialy, you broke the scientific method, jumping from hypothesis to conclusion. Or can you point ot where I say anything that contradicts the "best we have at the moment" or anything regarding that it should completely disregarded, or completely not trusted or any other radical claim you seem to be countering?

[charred willow]

I am only skimming, but this is in that article you linked

[soft bloom]

I will simplify a bit
I don't deny climate change, In fact often I have to describe to people that it's indeed very real and will impact them.
Where I don't allign with consensus is in what causes observations, the scope of changes, how soon they will occur and how large the impact will be.
In that sense, I have much more dramatic view.
And the problem I was pointing that in science is that because "co2 warms atmosphere" has been put as THE climate change, many other observations get discmissed, which is quite dangerous. (aka we would be better of studing the cause of wildfires than how to make cows fart less)
Though, thanks to last year, there' planty of public data for anyone to look at and gasp.

[soft bloom]

I hope you read the abstract and results in full, otherwise that's just cherrypicking results and fueling confirmation bias

[soft bloom]

but also, yes, that's one of the findings

[charred willow]

its not confirmation bias, I just think its interesting that that exact quote is in there, more so sad that this channel somehow has strayed from what it used to be. and a real confirmation that this is not a place I really want to come back to. I personally am a radio astronomy/radio guy and I can make my own understanding of a lot of the results based on how light/blackbody objects work.

I could be wrong here, I am not a climate researcher, but a astronomy one. but wouldn't the warming of the atmosphere cause an overall decrease in surface water in airid locations causing a higher fire risk? it seems to me like that issue is targeted because it is one that we can do direct research in easily, and is publicised because it is one that can be slowed in some ways.

again- not a climate researcher, but that is just my perspective from the knowledge I have on the things that I know about. which could be wrong.

[night girder]

RJGames, I got a question for you. It might be dumb.

But what am I looking at? #off-topic-tech message

[charred willow]

oh, sure.

its a gif from the radio telescope, frequency on the bottom, db (power) on the left

its a reading that we took over a couple days, and the bump about 1.420.4 ghz you see develop is actually the arm of the milky way passing over

and the smaller spikes when it does pass over are likely radio galaxies

you can use the signal that we are, as a reference to the 21cm line (hydrogen) to figure out the doppler shift (and rotational speed) of the milky way

@night girder hopefully that is some explanation, if you have any questions i'd love to answer them

I'm relatively new to radio overall, at least compared to my peers, but its an awesome field of interest at the moment

[night girder]

It's just fascinating to me you can pick that up with radio signals.

[charred willow]

yeah! its a TON of gain, and you have to think, every object emits some radiation, and it happens to be a lot of objects have a lot of hydrogen

so its a relatively strong radio signal. I'll get you pictures of the setup tonight. Its relatively crude (at least for a uni) but we're happy to have it, in addition to a really nice optical telescope.

if you want to learn more, and happen to know python, I'd appreciate any help me and my friend can get on the data processing side

[night girder]

what work do you need to be done in python?

[charred willow]

Well, it’s somewhat convoluted. But as of now we have a good format of data, and we are working to get more ways to display the data in ways that are helpful

So I’m working on displaying the data next to a sky map to help us pinpoint where it is pointed and what is passing over

A lot of it is going to be reworking a lot of the “get it to work” type code, to make it less slow, and more ideas/research on good and useful graphs

Heatmapping the path that it’s on is another goal, so we can do a full sky radio heatmap

Then other things like peaks, Doppler shift calculators

[night girder]

is it on git?

[charred willow]

https://github.com/RESBI/RadioDataProcessor/edit/main/README.md

Yeah.

And you don’t have to code or anything, I’ll try to send more in here.

It’s a REALLY interesting field. If you want to build a radio telescope yourself, there’s many many diy ones under 100$

Can do stuff like observe storms on Jupiter

[night girder]

yeah, just looking at the code atm.

[charred willow]

It might be a bit messy, me and resbi have literally been developing for about a week straight now

[night girder]

it seems you have some unused variables 🤣 Or I cannot find the use of them.

[charred willow]

depends on which script

specifically, I know resbi is completely overhauling the main script

and hasn't implemented the inclusion of the skycharts one quite yet

[night girder]

RadioDataProcessor.py

line 150: chart_config = config[1]

[charred willow]

yep, thats for the skycharts, so theres no config there because implementation isn't done yet.

def very beta as of right now

[night girder]

that's the thing about coding, there is so much context to it 🤣

[charred willow]

😆 I apologize if I didn't give the whole story, wasn't intending not to

[night girder]

No no!

If you work on a project for a week, you know every line of code.
But when you look at it for the first time, it's a big puzzle 😄

[charred willow]

😆 yeah, I know a couple of the unused variables are just so we know what we are planning to use

[night girder]

It makes sense, but it's just figuring out what does what.

[charred willow]

👍 the charts creation has been more mine, the radiodataprocessor.py is mostly resbi's

just because I'm not particularly familiar in mathplotlib, but I do know how to run calls to the tcp server on cartes du ciel

[night girder]

You might want to see if implementing a logger, is quicker than print().

If you want to speed up the script (if speed is an issue).

[charred willow]

speed on radio data processor is actually solid

after resbi multithreaded it, it will do like 1000 files in under 20s

which is good considering for the charts script takes ~2s per image

so if I did 1000 images from that one, wouldn't really be a limit in the radiodataprocessor.py

[night girder]

Ok, yeah. Just saying it could speed things a tiny bit up.

[charred willow]

👍

do you see anything major that I could speed up in the skychart script?

[night girder]

Actually, I could test it out 🤣

[charred willow]

thats by far the biggest drag script

would you like some data to test it on?

[night girder]

I do need a data set though.

[night girder]

Yeah, it's not going to be for this evening. But might be fun tomorrow 😄

[charred willow]

gotcha 😆 . the skychart script is not implemented

almost sure it will be by tomorrow

i appreciate you peeking over it letter, this is one of my first bigger projects, and I'm really happy with where its at.

[soft bloom]

i was referring to that if that's all you take from it simply because that alligns with what you want to find - that is confirmation bias.
Or if someone was to make such headline form that article, ignoring other finding, and the T -> CO2 and T -> CH4 causations. Because in fact conclusion is more complicated.

[charred willow]

its definitely still in the ADHD hyperfocus stage, which helps it get through 😆

I'm considering rewriting the skychart script into either F90 or C++ which are closer (or in c++ case) my home languages

[night girder]

I love to mess around in code 🤓

[charred willow]

just to see if the cartes du ciel app is the limit, or something weird within the python implementation of calling into it

[charred willow]

to be clear this was in response to letterdief.

[night girder]

If I understand a little bit of the code, I need to run the RadioDataProcessor.py with -p "test1". Optional I can give it threads argument. And I can also it with a config_file.

And if I look at it right, there is no way to give it a directory. Only a prefix.

Ah whatever, I will find out tomorrow 😄

[pure karma]

guess im aparently doing robotics tomorrow?

just got a email saying im eligible for a robotics class tommorow

i mean i doubt much will be done but guess its worth showing up

[soft bloom]

I was referring to the common experience that its better to do something and reiterate with new experience, than to organize everything without doing the core

[soft bloom]

Which is one of the things I am pointing at in this discussion.

[soft bloom]

eh, where i the vid?

[dire igloo]

#off-topic-tech message

[soft bloom]

Oh. sorry. I thought there was a second one

[dire igloo]

nope, that is the one

[dire igloo]

it's one of the things you're doing in this discussion (e.g. strawman)

[dire igloo]

where do I get one of those?

[manic cipher]

I don't know. I saw that image on another discord.

[dire igloo]

Dangit

I want one

[soft bloom]

let's put an analogy:
I am not shitting on the game of "science"
(and scientific method is actually a cool thing)
I am saying that there are bugs
and that some levels should get fixed, because I don't like (I mean, in this analogy, there's no other sensible motivation) the artdirection of one of the chapters

Did Baldur talk atoday about consesus? I think I can't find the message to reply...

[manic cipher]

Look on Ebay

[dire igloo]

Ehhhh, rough comparison but I think I'm getting your drift.
You're just doing quite a poor job of clarifying that you're trying to have the holes closed instead of ripping them open to tear apart the entire game

[soft bloom]

anyway:
from today's search, in the top results, I didn't find any paper saying something like "consensus is that CO2 is primary cause for temperature rise, not the other way around, for all observable period", only references to that idea, with proves of the opposite for at least distant (200+ years) past
That last paper Is indeed confirming that there appears causation in that direction, but also in the opposite.
So, I feel like that (quote from second line) exists only in media

[dire igloo]

If CO2 isn't the cause of the rapid temp increase, then what is?
You're comparing geological activity versus human intervention

That's the biggest fallacy to fall into when analyzing Earth's history.
The impact that humans had and still have is immense - and it's only getting stronger

Maybe if earth was left alone without humans, then temp fluctuations would cause CO2 variations to follow.
That is in no way contradictory to the claim that humans adding greenhouse gases in record speed is not the cause of the current temp increase

The point that climate scientists in public media are trying to bring across isn't how the world has worked historically, it's how we as a species have to take care of our home if we want to keep living in it

[soft bloom]

Hmm, today, I would look at plots for water vapor in atmosphere.

oh come on!
the .gov domain should totaly be allowed

which weirdly reminded me of this page
www.ncei.noaa.gov/access/global-ocean-heat-content/

huh, I think today it's broken and doesn't show the color-to-label matching. big sad

[soft bloom]

but really, more observations should be made

[dire igloo]

Eh, anyways, bed time

[soft bloom]

the header has quite a detailed plot-rich article

in short words: heat content of ocean layers has increased, a whole lot, since 1955

the discussion section is quite interesting, but i will refraing from sharing short quotes

You know what is lacking from research?
How does plastic garbage (and it's smaller forms, aka microplastics, that can be carried in other means, not just garbage disposal) impact heat conductivity of the ocean?
Does it make it so that ocean's surface can't store the heat from atmosphere?
Or would it speed up evaporation?
I don't know, but even chatgpt doesn't dare to halucinate - so probably very little research done.
And that's sad, considering that water vapor is a very potent greenhouse gas, and it's even in the same 'green'/ecology direction of science.

[charred relic]

chewed through that one pretty good

According to Al Gore I should be underwater by now...

Seems his calculations were... incorrect

[manic cipher]

How much money did he make from his "calculations"?

[charred relic]

Ah I'm sure his calculus on that angle was flawless...

[manic cipher]

So he basically made a lot of money predicting something that did not materialize.

[charred relic]

Nooo surely he would never

Probably just an innocent mistake in the math, decimal point in the wrong place or some such.

You know that's how spinnach became so linked to iron... an early study misplaced a decimal point. Turns out it is indeed great for iron just not nearly as great as they thought at first.

[languid gulch]

at some point you'd just be eating scrap metal

[willow pike]

https://twitter.com/IanCutress/status/1885088050587328710 jfc

𝐷𝑟. 𝐼𝑎𝑛 𝐶𝑢𝑡𝑟𝑒𝑠𝑠 (@IanCutress) on X

➡️Based on industry feedback, plan to leverage Falcon Shores as an internal test chip only without bringing
it to market. !!

X · Jan 30, 2025, 10:09 PM

clearwater forest delayed

that's Intel's big e-core xeon

the big ticket 18A bet the company on it chip

supposed to be this year

now they say 2026

call it; flatlined

time of death, 31 Jan 2025

[twin dew]

That first part would suggest it is chip design fault, not process fault.
Just bad product.

[willow pike]

falcon shores was always cursed

everything intel is cursed

also they insist 18a is good, but they always big up their process

they bigged up 10nm

[dire igloo]

two questions about that:

• I've only ever seen ocean plastic discussed in the context of its impact on ocean life, idk if it has anything to do with atmospheric temperature and couldn't tell you if there's even a mere theoretical link. Why would you think there is a connection?
• I agree that water vapor is a potent greenhouse gas, what I don't know is how big its influence is. Primary school teaches about the water cycle, so from what I know, atmospheric water vapor will leave the atmosphere within a few days - maybe there is some analysis of atmospheric water ppm over time

[twin dew]

And many of the additional factors have only come into play as things boosting the change that humans started.
Making stopping the warming much harder.

[dire igloo]

Not addressing this to you, KEKw, just stating smth I've seen as a general trend:
I noticed more and more people mixing different topics of environmental protection, primarily pollution and climate change.
idk how often I've seen someone ask how not littering helps fight climate change or how less plastic in rivers reduces GHGs in the atmosphere

[dire igloo]

Henry constant being a prime example of this

[twin dew]

And all the extra stuff now starting to be emitted from thawing tundras permafrosts, bottom of sea, etc.

[edgy hazel]

Yeah no. This channel has become horrible and has nothing to do anymore with its original purpose of being a tech channel. Wish you all well. Be better.

[twin dew]

Bye 😭

[night girder]

Shit, bye herg 👋

[verbal raft]

bye Herg R.I.P.

yeah, sure buddy

[willow pike]

seems legit

[verbal raft]

oh hey

another f"cktard

[verbal raft]

quiz which of these will fall over?

[dire igloo]

The grey one in the blue frame

[wanton orchid]

these vapors are quickly becoming clouds
which reflect sun energy outward

less vapor produced = less cloud = more vapor in atmosphere

keep in mind to always make sure not to do shortcuts

[verbal raft]

nope

https://www.youtube.com/watch?v=LLSHiOfHHE8

lmao

[verbal raft]

https://www.youtube.com/watch?v=iibW1CAHggw

[verbal raft]

anyone remember this?
https://www.youtube.com/watch?v=xUT4d5IVY0A

[languid gulch]

i love when corporate delusions get shattered by reality

and for the moment, AMD has better CPUs for gaming & general tasks, Intel really only has the upper hand in some specific applications

[soft bloom]

What is this based on? Perhaps a paper that studies causes?

[languid gulch]

not to mention nobody will trust intel for the next couple of years until they can prove that both the 13th/14th gen microcode fixes are actually fixes, and that there's not some equally horrific issue with the 200 series

[wanton orchid]

no reasonable people will ever buy the 200

they will at least wait 300 or something

[languid gulch]

i wanna see how the next X3Ds with more cores will compare to 200 series

[soft bloom]

I mean, if there's a flywheel and you time your efforts to speed it up correctly you will contribute to it's energy
it's not up for discussion
but how correct it is to say that you are the one responsible for it spinning, even if there's some feedback mechanism that causes industrial machine to pump more energy into it if it spins faster?
What i'm trying to say is... If we want to understand it, it's not about finding 1 cause, it's about finding The cause and all the other causes.
For example, yesterday, in that paper from NOAA, I encountered phrase that, essentially, even if humanity stopped existing, and no more heat was trasnfered to ocean whatsoever, it's current (as of 2010) capacity is enough to heat atmosphere 36C more. Aka, perhaps, we need much more drastic measures.

[soft bloom]

yeah, but it's a conveyor. So what matter's a lot is how much of it is at any given moment.

[dire igloo]

wha... what? conveyor?
if you mean what I think you mean, that's referring to oceanic plate, the ocean floor. not stuff floating near the surface.
Ocean currents are a WHOLE different thing

or you mean the water cycle that as far as I know doesn't act as a conveyor

[soft bloom]

referring but only as a conveyor, not as causal link

[dire igloo]

when water evaporates, it takes barely anything with it, that's the only reason why distillation works

[soft bloom]

bruh i wrote long msg with a link and now its gone 😦

I just remembered that ocean playes its role in co2 circulation

[dire igloo]

ctrl+z

[dire igloo]

oh, absolutely, see Henry constant above
I just questioned the role of ocean plastic

[glossy glacier]

Gaming a bit worse than the single chiplset versions, MT almost on-par with non X3D.
Same as last gen with a small boost in MT tasks because the X3Ds don't quite loose as much clockspees now

[twin dew]

And by using manual core pinning tools like project lasso, the last differences in games should disappear and turn into better performance from the higher clocks on the X3D CCD in the multidies (from segmentation)

[soft bloom]

uh, I was just shitting on some search results that pop up that don't link any actual research, instead using words of some "expert" and deriving conclusions without any basis.

[dire igloo]

Words of experts can be used as a reference, it's not inherently bad.
A scientist talking about their research has no less value than them writing it down and publishing it. It's just usually shortened for the sake of audience retention.
The issue with appeal to authority is when you introduce an expert statement as truth not because the expert is talking about their research, but merely because they're an expert.

Basically: ask a famous physicist what he thinks of any given theory outside his field and their answer will be meh. Ask them about their field tho and their testimony holds more value now.
Even better if you can just link their written research for further reading

Which btw is a mistake I've seen happen a few times now:
People mixing sources for their work with material for further reading.
Nobody can go through every single paper and article there is on a given topic, so at some point you have to cut your reading short or exclude information from your research.
However, anything you reference directly should've been read for the full context

[night girder]

GN doing Gods work 🙂

[night girder]

fosdom immich live stream starting in 6 minutes for those interested: https://www.youtube.com/live/DjyQ4b7r-XE

[night girder]

https://www.youtube.com/watch?v=8utpbbdj0LQ

[pure karma]

well well well the cycle continues

3rd satisfactory sprint 3rd hardware failure

[soft bloom]

is there any evidence that DeepSeek actually stole anythign from OpenAI?

[pure karma]

or not?...

huh im so confused

frametime is peaking without the cpu or gpu peaking or being busy

it better not be XMP again

[night girder]

All I read today is that VS is going to investigate if Deepseek illegally bought nvidia gpu's from Singapore.

[soft bloom]

Visual studio?

[night girder]

Ah sorry, VS is dutch acronym (verenigde staten)

verenigde staten = united states (US).

[soft bloom]

I saw couple vids explaining how its funny that openai claims they were used, but it looks like nobody even questions if the claim is true or not.

Which would be a good way to plant idea that DeepSeek is somehow not deserving the praise, that actual cost is still very high etc, even if everyone is ok with them stealing from OpenAI

[night girder]

This was my first reaction: #off-topic-tech message

Companies are suing openAI for using their data without permission.
OpenAI investigating Deepseek, because they think Deepseek used openAI data without permission.

Is it me, or does this sound hypocritical?

[soft bloom]

It would, but I really think that the real problem is not in them doing that or not, but in people accepting thta they did so simply from accusation. Because of the things outlined above

[night girder]

People just don't care about it. It's like two gangsters blaming eachother for a crime they both commited.

If openAI paid every single ~~website ~~ source (website, videos, books, chats, whatever) they used to train their model on, that's a different case.

[soft bloom]

Wrong example

[night girder]

It's just my opinion, it cannot be right or wrong. It's personal 😄

You can disagree with it though. I don't mind.

Anyway, that's how I look at it atm.

[soft bloom]

It could be that deepseek didn't steal.anything from openai. In that case openai just uses it's resources to discredit competitor who outran them by so many parameters. Meaning that you don't need billions of budget or government regulations to do quality LLM product. Which will decrease need for Nvidia gpu's, costly training and usage.

And if they are actually to blame for any stealing, which would be much less destructive than openai robbing.millions of websites from views, it's still an impressive display of.how much less effort LLMs need compared to a giant bill OpenAI pays and requests.

[soft bloom]

I mean, until there's any proof, I feel no need to even question if OpenAI deserves being robbed. So far they are just expensive, and try to regulate their competition on government level, whole already selling their tech to military.

[night girder]

if Deepseek is not trained on openAI, but trained on the internet for example, it's still a moral grey zone no? They did the same as openAI. However I think about, the way they trained AI rubs me the wrong way.

Or do you believe deepseek paid every source? The people who put in work writing wikipedia's etc.

Ok, let's not even think about paying them.

Let's just start with asking permission to use the data.

There is also news like this: www.wiz.io/blog/wiz-research-uncovers-exposed-deepseek-database-leak

And: www.garanteprivacy.it/home/docweb/-/docweb-display/docweb/10096856#english

TLDR:

The Authority also asked the companies what kind of information is used to train the artificial intelligence system and, in case personal data are collected through web scraping activities, to clarify how registered and non-registered service users have been or are being informed about the processing of their data.

[soft bloom]

Well, yes, but openai definitely won't do shit about protecting those they robbed themselves.
So i see it just as their battle for attention.
They try to smear DeepSeek's reputation and stay on top. Somehow people believe them...

[night girder]

What people?

[soft bloom]

On internet, who make videos

[languid gulch]

suing each other would provide some hilarious evidence in discovery

i bet it's magically settled

[night girder]

Yeah 🤣

[languid gulch]

btw i had to order a new desk for the new pc 🤣 it's 35kg & i don't trust my current one to handle it

10mm plywood is fine for light stuff, not for massive stuff

[pure karma]

yea that sounds about right

[soft bloom]

oh, so Alibaba too now
https://www.youtube.com/shorts/bAtuUeRWV3w

[manic cipher]

Does anyone here know of any data recovery services that can unlock SATA locked hard drives?

[twin dew]

Password locked with unknown password?
Or broken and unresponsive?

Or accessible but some SMART functions cannot be accessed like secure erase because the utility says the drive is locked?

[manic cipher]

Locked with a forgotten password.

[verbal raft]

OOF

look at FX

[twin dew]

AMD FX series, good integer, bad floating point.

And like you see, didn't pan out in long run.

[verbal raft]

this is how i used to play ARK evolved

[charred relic]

I keep meaning to try that game...

[willow pike]

not even that good at integer cause the modules had one front end to fight over, the L3 cache had such high latency it was the same as RAM, and it was a long pipeline P4-like design that never ran near the target frequency

though yes floating point was somehow even worse

that jump to zen 1 is nothing short of miraculous

[twin dew]

Lets skimp on the design costs by using "AI"!
Not doing full manual layout etc. but trying to cut costs by letting programmatic layout handle lot of the fine details...
Leading to various problem with max frequency etc.

[willow pike]

it's hard to design a worse cpu

[charred relic]

I have a Pentium-D around here somewhere. Nothing like shoving two notoriously toasty cores onto one package

[willow pike]

HEY YOU WANNA KNOW WHY I USE AMD USERNAMES IT'S THIS SHIT

pentium suck my D

god I hate that thing

[charred relic]

I pulled it out of a bit of audio recording hardware GC was trashing, neighbor worked there

[soft bloom]

how come I have file modified in the future?
aka it's recorded that way

[charred relic]

some huge rack mount thing.. only thing wrong was the power supply

[soft bloom]

what is this nonsence ls -la output?..

[charred relic]

That and they had pretty terrible cooling for that cpu

Time travel

[soft bloom]

indeed

but also PyCharm broke, even though it doesn't interact with that file

[mental oriole]

Change os clock modify change back time

Or you're using a library that fucks with time

[soft bloom]

it;s like if i have had gremlins in my pc sometimes sabotaging

[twin dew]

And AFAIK there are calls that allow you to set whatever.

[charred relic]

I can see scenarios... like somebody on a timezone ahead of you sends a file saved recently

[soft bloom]

or maybe just sun radiation hitting hard?...

[charred relic]

it would have a timestamp from the "future"

[soft bloom]

thankfully not yet

[soft bloom]

back to where?

[soft bloom]

those are timezone aware datetimes. so wouldn't work

[charred relic]

Alternatively do not play with the time machine... it's not safe

[mental oriole]

enters paradox

[charred relic]

caves into a singularity

[mental oriole]

Love when that happens

[soft bloom]

the problem i have is that that file is written by a proces that has been inactive for a while
and i don't know if process caused such weird datetime, or something else, which resulted in prcoess to halt
but that's .gz file, so if i end the process I might loose it's contents
shit

[charred relic]

it's a mystery wrapped in an enigma!

[soft bloom]

yep, its fucked up

[twin dew]

And how much in future are those times?
2 hours?

Could point to timezone problem.

That UTC+2 timezone getting applied twice.

Once by the program and second time by OS.

Or something similar.

[soft bloom]

more like 1 hour. THough i am not sure when it stopped writing

and ls -l shouldn't put year inthe time column

[twin dew]

It does, when the time isn't in last year before current time IIRC.
And when it is in future, it isn't.

[soft bloom]

ohhhhh
ok i might now what happened
i started the thing after windows
but before connect to internet
so while it was running it updated the clock...
thoguh still not sure why would that break PyCharm and holt the process

[twin dew]

So Windows had set the HW RTC 2 hours into "future" in Linux eyes, and the Linux fixed that once NTP worked.

Or something similar.

[soft bloom]

95% certain that happened

[twin dew]

Please fix your Linux to use local, or Windows to use UTC...

Both are easy to do.

(and don't do both)

[soft bloom]

that's the sad part...

[twin dew]

Ah, you had done both?

So Linux was in local, and Windows in UTC (how they thought about HW RTC stored time)

[soft bloom]

no, not yet. sad that I can't do both...
aka i don't won't any of them to rely on setting of the other

[soft bloom]

no, Windows always loads displaying true UTC time on the clock

[twin dew]

No.
This isn't what time is shown.
Windows by defaults stores local time in RTC.
Most Linux distros by defaut store UTC time in RTC.

So with time zone set to UTC+2 in OS and local time is 14:00.
Windows stores 14:00 in RTC.
Linux stores 12:00 in RTC.

Then if you at 16:00 local, quit from Windows, to switch to Linux.
Windows stores 16:00 in RTC, computer reboots.
Linux loads, sees 16:00 in RTC, and then sets its own clock to be 18:00 in local time.

So clock moves 2 hours into future on switch from Windows to Linux.
And 2 hours into past on switch from Linux to Windows.
Until NTP based clock correction is done.

And you correct that by changing one setting in either:
Linux (RTC contains local time, instead of default UTC)
Windows (RTC contains UTC, instead of default local time)

Which one you change depends if you prefer BIOS/UEFI or timezone non-aware throw away OSes to show UTC or local time.

So if you want BIOS to show local, you change how Linux handles RTC.
If you want BIOS to show UTC, you change how Windows handles RTC.

[charred relic]

Timezones 101

[soft bloom]

I want bios to show UTC, but I realize that if I will make in Windows that next isntallation might break again. or even... if i change in Ubuntu, than next install might miss it. hmmm

[twin dew]

Like with any settings change, it need to be done again after reinstall.

[charred relic]

holy shit...a guitar that use a e-ink type display as the paint/finish on the body

what a time to be alive...

[wanton orchid]

you want utc as rtc
Windows is the boggus one
but keep it like this for compatibility

[wanton orchid]

you can have a .reg file with the key
very easy to set

[twin dew]

And RTC in UTC would be pretty weird to most consumers.

[wanton orchid]

next restart you sync date or boot Linux and it's all OK

[twin dew]

When in BIOS or like.

Which is the stated reason for Windows default.

[wanton orchid]

why ?
it still shows local in os

common consumer don't go in bios

and people can know about utc existing
as almost everything is internationalized online
people can understand GMT +2
means utc is -2

other people will not go to bios

imo this is yet again lazyness
compatibility with old behavior
and yet an other way to easily track on you and be incompatible with Linux systems

rtc utc period

best example is you moving with your laptop from country to country

you want to update timezone
not reset time references

(and bios screen could implement selecting timezone)

[stray badger]

Certainly a... choice

[verbal raft]

Interesting

[pure karma]

[sand saddle]

Played europa universalis the price of power board game. It’s a bit complex and long…
3 players had played once before, for 1 it was the first time.
We were able to finish one age(4 rounds) after 8 hours…

[pure karma]

Sounds… complicated

[soft bloom]

Keep like what?

[soft bloom]

Wonder if there's some programatic way Ubuntu could detect if Windows changed the time

It has access to the drive, could be as easy as reading some journal

[sharp oasis]

@pure karma what's wrong with this photo.

[glossy glacier]

PhysX and Re-BAR not enabled/available?

[sharp oasis]

I mean physx not being able is known cus its amd

Rebar support is one

Its doing something that may basically be impossible

[glossy glacier]

regarding physX, it is?
maybe thats a bug then when you have a nvidia and AMD (i)GPU in the same system, because it was shown as available for my AMD iGPU

[wanton orchid]

if I had to do bios I would simply Splat "UTC" close to the rtc

[wanton orchid]

keeping it not utc

i.e whatever

[wanton orchid]

too complicated for no real reasons, as I said earlier it's easy to set windows rtc in utc

also windows is not supposed to "change" it once set to whatever local the windows is configured to

rtc stands for Real Time
and Real Time standard is UTC

[soft bloom]

I htink it does so when ask it to set time atuomatically to fix displayed clock. otheriwse it shows UTC time

though, yesterday I didn't .... so not that sure

One of the things you don't want to hear from scientists is apology stating that they didn't have the task of analysing multiple evidence as parts of one process

[pure karma]

the default clock being like 700MHZ above normal???

not to mention 4000MHZ boost

this fucker is basically a cpu

[sharp oasis]

Hehehehe

[sharp oasis]

That's cause driver default is set higher then reference default cause of bin

[charred relic]

Glad I went with a good CPU cooler this build. It can run all the cores at 5.5GHz without even touching anything with plenty of temperature overhead.

Now idea how overclockable these chips are

[soft bloom]

https://www.youtube.com/watch?v=l-9ALe3U-Fg
"used mechanic motor to turn the dials"...

[soft bloom]

good example on how common simplification/aproximation simply doesn't reflect a bit more complex reality
but most not only forget these nuances, but also claim them to be accurate representation
https://www.youtube.com/watch?v=DQv0nlGsW-s
in optics, when working with lenses
what's the relation between numerical aperture and f?

[charred willow]

numerical apeture is physically how big the apeture is, on a 16" telescope, it would be 16 inches. but a f value is a ratio of the focal length to the apeture size, if you had a 100mm f2 lens, for example, then 100/2 the apeture is 50mm.

the f ratio effects overal lens light collection, as the larger your focal length, the lower the field of view, which also means a lower light collection, so then you need a larger apeture to compensate

the reason the f notation is used is for cameras so that you can get a real value representing how much light is being collected in reference to your other parameters, shutter speed, gain, etc.

because if you have a 400mm lens with a 200mm apeture, it would be easy to assume that it picks up more light than a 200mm lens with a 100mm apeture, but these are actually the same overall light collection

[soft bloom]

but the 1/2 rule is just an aproximation that doesn't tell the full story
and applying aproximations without understanding its limitations will lead to undesirable outcomes

[soft bloom]

I feel like Destin made a video about nuclear power just as excuse to do silly things on camera

[twin dew]

https://www.youtube.com/watch?v=Nh1FHR9fkJk

[charred relic]

They added some extra frames to your frames so you can game while you game.

[fathom patrol]

Good lord, Windows 11 24h2 is absolute TRASH. I had to roll back to Windows 10 because screw this noise.

Prior to the update of 24h2, my FPS was around 60-70 and then it would just tank to like sub 20s

[twin dew]

All the Nvidia fuckery around PresentMon fork...

So typical Nvidia.

[charred relic]

I caught part of that video a bit ago, was distrated doing other shit for most of it tho

I just run whatever I can get away with. 1080p, 1440p or 4k

i can do pretty much anything in 1440

[soft bloom]

amd drivers keep somewhat breaking. now anytime I try to open AMD it says this.
another windows update fuckery?..

[charred relic]

o.O

[twin dew]

Probably.

[charred relic]

I had Davinci giving me fits because the iGPU was enabled and for some reason it freaked it out and put it in a state where it saw no GPU's

I doubt I'll miss those 2 GPU cores...

[twin dew]

Windows Update doesn't check if newer driver is installed already.
Just if the specific version it has has ever been installed.

So consistently downgrades GPU drivers for Intel and AMD twice a year or so.
Nvidia has some thing that prevents that.

And that specific AMD error only happens when that has happened.

[charred relic]

guess i hadn't had mine long enough to get hit by it

yeah ordered it oct 13th... so not that long

[twin dew]

IIRC Nvidia uses just the VEN and DEV for Windows Update drivers, and also that SUBSYS for their manual drivers, bypassing the issue.
But Intel and AMD don't.

And just use VEN & DEV for both.

As Windows uses more specific driver if possible.
So with that split, WU will install Nvidia drivers too, but they don't get used if there is manual driver installed.

DLSS also stands for Dumb Lame Super Stupid.

[fathom patrol]

Anyone else have FPS issues with the new Windows update?

[twin dew]

If you mean on Windows 11 24H2 specifically, then no idea.
You need to remember that "new Windows update" can mean the major version like that, or the latest monthly patch.

[fathom patrol]

Okay

Yeah I mean specifically 24h2

[twin dew]

But first thing to do would have been to DDU GPU drivers and reinstall, and not completely reinstall OS.

[fathom patrol]

Yeah that may be true but hell with Windows 11 for now

[twin dew]

9 months left, and no major Win11 update coming before that.

[fathom patrol]

Yup I know

I'll deal with it then

[soft bloom]

Trying to remember how you gold me to install windows driver first and then update from amd website.
Aka, how that windows driver got installed...
Although, it's probably already in?
So just download and install

[verbal raft]

i finally know what DLSS stands for

[twin dew]

Yes, just download and install, and don't use the factory reset option.

[twin dew]

Probably better to write as "Dumb, Lame and Super Stupid", instead of that original commenters version.

[charred relic]

How dare you

That as insult to the stupid

What is FSR tho

[soft bloom]

free speed (fps) restoration

[twin dew]

But it is as stupid thing as DLSS is?

[charred relic]

I try to just run without that stuff

haven't really needed it to hit my target resolution and framerate so...

flight simulator will stress it but that is a game that has traditionally been all about making hardware cry

[soft bloom]

probably dumber, but it makes it better

i mean, rtx is smart, but... what's the value in it?

[charred relic]

but you can trace those rays bruh

[soft bloom]

Lumen

[soft bloom]

with the need for special hardware

and its used only in a couple of games to this day

(dozen)

I was that the simpler the additional tech is (to rasterization), the more users will use it

[charred relic]

I remember the first time I bought a game where the hardware was holding me back

[verbal raft]

man,i am so happy i didnt buy the 3050 6GB instead of this RX 6600

[charred relic]

Ultima V on mt 8088, it was sluggish compared to 286 systems

I upgraded from a 2060 to an AMD card because best value and it hit my res/fps sweet spot... in my price range at the time

[tribal kraken]

Motorcycle exhibition over, 5h train trip to home. So much wanted to buy electric dirt bike but it's worth 6x RTX5090 and they only pay 2,5 for my old trade in bike. Availability is also the same. All production is sold and ordering now ships the bike in summer.

[mental oriole]

To continue to use bloatware, install bloatware

[twin dew]

How is intentionally installed software bloatware?
AMD GPU driver installer has three options:
Driver only, limited Radeon Software (no recording etc. just basic options), full Radeon Software.

[willow pike]

no amd driver has ever asked me to sign in to use it

[wanton orchid]

but how would they maintain a subscription model if you didn't sign in ?

[surreal socket]

mine runs at a 120 TDP @sleek prairie

the intel is at 125

[sleek prairie]

what is tdp

[surreal socket]

Thermal Design Power

[sleek prairie]

isn’t that just max heat

idk i just know it’s more common for intel cpu’s to generate more heat while you’re using them

[surreal socket]

Something like that

What is the Thermal Design Power (TDP)? TDP stands for Thermal Design Power, in watts, and refers to the power consumption under the maximum theoretical load. Power consumption is less than TDP under lower loads. The TDP is the maximum power that one should be designing the system for.

[sleek prairie]

yea

[pure karma]

The max power draw under default operating settings

[opaque grove]

i got dis oxy-acetylene cutter kit in a second hand store for $20.56 and i can buy a new similar kit version for $493.41

[languid gulch]

https://www.youtube.com/watch?v=0L1Uyw22UAw

i don't think i've seen anyone else do a long-term analysis like this, so condensed down to 2 simple charts

[verbal raft]

this is so CRINGE

[languid gulch]

that's the kind of shit that makes people really suspicious that you're on their payroll

[twin dew]

How do you expect he even got a card?

You either need to be very big (and even that isn't certain), or asslicker, to get review sample from Nvidia.

[languid gulch]

yup

[twin dew]

And if you are in the asslicker category, anything but pure asslicking will drop you out of the category.
And too critical take from actual reviewer also has very high chance to do that.

No matter how valid take.

[stray badger]

if you have to have secrets for your next gen to beat your last gen, ur kinda screwed

[languid gulch]

the giant pile of caveats that they hide behind is just hilarious

it used to just be a "this card gets higher fps [but don't notice that it's on lowered settings]"

[dire igloo]

remember when Nvidia claimed 4090 level performance for the 5070?

[languid gulch]

i laughed my ass off at everyone who burned $2k on a 4090 when they said that, because if it had been true, my god someone might have firebombed nvidia HQ

[lucid yarrow]

What am4 CPU would pair well with a 4080? Currently have a 5600x but just bought a 4080 and figured it was time for an upgrade but I don't want to change the mobo yet. A 5800x3d would be my first choice but it's unobtainium new and used prices are still bit steep imo

[languid gulch]

5700X3D

[twin dew]

5800X3D is out of production, 5700X3D is available.

[lucid yarrow]

I see how big is the difference between the 5700x3d and the standard 5800x iirc the 5800x is about 80 dollars cheaper

[languid gulch]

5700X3D is like 5% short of the 5800X3D

aka 95% of the performance for like 66% of the cost

[glossy glacier]

well, 95% of the performance that does not benefit from extra cache

[twin dew]

?
That was comparing higher clocked X3D to lower clocked X3D?

Where the only difference is that lower max frequency on 5700X3D compared to 5800X3D.

[glossy glacier]

right, skipped over the second X3D

[twin dew]

And both are much better at most games than 5800X.
To level that 5800X3D still bests 14900K in few games when CPU limited.

Outside of sim etc. genres.

Where the X3Ds really shine, including Satisfactory etc.

5600X to 5800X3D was about 75% increase in Satisfactory performance, little lower for 5700X3D

When someone switched on Reddit.

[willow pike]

x3d is superb

the post in question

[dire igloo]

5800X is hardly an upgrade over your current 5600X.
the X3D is what's important

and for X3D: it depends on your usecase
if you're playing endgame Satisfactory at 1080, wtf did you get a 4080 for?
and if you're playing 4K Cyberpunk with Pathtracing, no need to upgrade CPU

for now, you should just continue with your build as is and only consider an upgrade if you're noticing CPU performance issues

and at that point, depending on usecase and desired performance, an in-place upgrade on AM4 may not be sufficient

[verbal raft]

2X performance ?

wow

[twin dew]

When CPU limited, yes.

[verbal raft]

yeah i know

[twin dew]

Satisfactory absolutely loves that extra L3.

[verbal raft]

i just thought it would be reasonable like +30%

but thats crazy

[twin dew]

When cache trashing, more cache will cause insane performance improvements.

[verbal raft]

how does intel fare?

🤔

[twin dew]

Best I can do:
https://overclock3d.net/reviews/cpu_mainboard/amd-ryzen-7-9800x3d-review/10/

Those bad 79X0X3D results are because Factorio isn't game by default in Game Bar, so it is running on wrong CCD.

And that 14900KS result is just weird and probably error.

[verbal raft]

low quality TAA

[lucid yarrow]

Play at 1440 and yes I play cyberpunk lol also some heavily modded Skyrim so it's gonna be a blast getting over 60fps lol

[twin dew]

But basically this same thing repeats in lot of other cache / ram latency/bandwidth hungry games in the sim side of games.

Just absolute curb stomping of everything else by X3Ds

[pure karma]

I had a 8X performance increase with the drop of full release

To be honest im still impressed by that

[verbal raft]

8

X

8X ? HOW

[twin dew]

That pic is from same computer, same game version, same game, same day.
Just CPU switched.

[twin dew]

Because MagicZ had something broken.

[pure karma]

Yea multicore spreading got huge upgrades

[verbal raft]

the game did once trick me into thinking i had 2X fps

[verbal raft]

by forcing FG with FSR

without letting you turn it off while using it

[verbal raft]

classic

I guess

huh

[pure karma]

Get something better used at that price

[verbal raft]

huh

[verbal raft]

my stratagy is : wait.

has never failed me yet

[lucid yarrow]

The focus has turned from raw raster to ai, frame generation and ray tracing honestly sucks

[swift elm]

Yep, it does, but we're also running out of ways to make the raw raster go faster.

(Hey that rhymes lol)

[lucid yarrow]

Yeah the die can only get so dense with silicon until we have a breakthrough in material the focus will have to keep shifting

[swift elm]

Yep, and we're starting to run into literal quantum junk with how small the process nodes are getting.

[dire igloo]

Is it fair to call the 4090 the Titan Lovelace?

[verbal raft]

Yeah I guess

[languid gulch]

RTX 40Titan

[soft bloom]

https://www.youtube.com/watch?v=NvNyU6ExSrA
another example of how ideas might appear, but finding evidence immediately is just really hard until a new way is discovered

[verbal raft]

https://videocardz.com/newz/amd-instinct-mi400-chiplet-design-to-feature-new-multimedia-io-dies

[stray badger]

i got my 7900 GRE for about $600 including taxes (probably a bit less tbh). no way im buying either for that

[stray badger]

rt is solves many problems that raster is unable to. the problem is nvidias lying

[willow pike]

in interviews they say MFG is only used when your inputframe rate is already 60. watch how they will soon market a 5060 as being able to path trace at 60FPS using an input of 15

[pure karma]

yea frame gen is meant to be used at 60fps+ but its only used, usefull and demonstrated at framerates bellow 60

[stray badger]

wdym useful? below 60fps the results are atrocious

[pure karma]

because why would i use it beyond 60fps its useless

i use it at 20fps or less

bumps it up to 45+

[charred relic]

Stand back I'm about to fire up this GT 1030

[languid gulch]

start up a complete SF save 🤣

[sharp oasis]

@pure karma

Cursed power setup.

[languid gulch]

truly cursed would be if you have pigtailed from one gpu to the other

[charred relic]

I have a RX 6750 XT... not great, not bad

the size of some of the cards are pretty ridiculous

[languid gulch]

yea, installing the 4090 was terrifying just because of how damn heavy it is

[charred relic]

the case cane wutg an anti sag system, installed it even though i don't need it with that card

some extra stability doesn't hurt and it's not noticable

kinda want to swap in some rgb fans on the rad... or put these on the outside and some rgb on the inside and do push/pull

i like the way these fans will all sync together.. still a hot mess but it's more usable than anything else i've had

(assuming I stick inside the Lian Li RGB garden)

[safe trench]

there goes my entire life purpose

expect for farming stuff to be sinked

[soft bloom]

https://www.youtube.com/watch?v=IPBdC0YUKKE
heh, reminds of Hydridic Earth

so, time has passed, and I would summarise the core behind hydridic earth as "there's a shit ton of hydrogen inside Earth, because that's how it should have formed"

[twin dew]

Another thing about the Nvidia 5000 series launch that hasn't been mentioned here directly.
Nvidia launched with tiny initial stock, just as all Chinese factories shut down for a month.
So next restock is probably in march.

[mental oriole]

Seems they always launch with no stock...

[twin dew]

4090 launched with 3+x larger initial stock than 5090.

4080 with way larger than 5080.

Never near large enough to fill the initial demand, but this launch was tiny even compared to previous ones.

[languid gulch]

already seeing reddit posts of people with dead 5080s

[twin dew]

Like GN speculated, this might just have been about having the cards for sale before any Trump tariffs hit import costs.

No matter how tiny amount.

[languid gulch]

i didn't read the details, did they say that the tariffs only apply to brand-new products coming in & all the 5090/5080s are now exempt? or just this initial batch

[twin dew]

Tariffs happen at importation step, so anything already in country doesn't get affected by any changes in them.

No idea what is the status of any new tariffs intended by Trump.

[soft bloom]

DlSS5: only every fourth customer has gpu, the other 3 have AI aproximation of it

[charred relic]

https://tenor.com/view/wavcord-wav-content-triples-triples-discord-meme-gif-47473542699554618

[verbal raft]

https://youtu.be/KEtb0punTHk?si=U7i1AtDyHMkuEfpn

TI is back at making video s

[verbal raft]

That would be crazy

[pure karma]

here we go again

[languid gulch]

looks like DOA out of the box & not just frying the connection

[twin dew]

Or just DOA from handling while installing (ESD)

[languid gulch]

i feel like ESD is largely not an issue anymore

[twin dew]

Worst time for ESD on northern hemisphere.

[twin dew]

No real change over time.

[languid gulch]

even with all the new protections?