#off-topic-tech
1 messages Β· Page 205 of 1
Man, the whole world is patchwork.
I think we got some roman walls left somewhere around in antwerp.
and a old sewer system.
But I think we put the wall back into the ground π€£
like we found while digging, and then we either destroyed it, put dirt over it or moved to bokrijk I forgot.
We had it all here. I once investigated for school who all "invaded our country". Pretty crazy π€£
I can't say country, but lets says the region that now is my country.
costs money we don't have as a country.
Well tourism. If you have a nice aztec pyramid. You can make it into a cash cow.
I ask that most of the time I see tourist anywhere over here... wtf are you doing here?
We got nature. But doesn't everyone?
Especially for older buildings where the government won't let you insulate outer walls so people give up in the end and let the outer walls decay
Because the government logic is that it's better not to touch older buildings and let them decay instead of renovating them....
that's comming.
I heard it on a podcast. EU will do it for every building to get climate neutral.
I wonder how they going to make historical building climate neutral 
We are going to adopt a new law where every new public and non-residential building from 2026 needs to have solar panels and all new residential buildings will need to have solar panels from 2030
How do you insolate a pyramid?
The goal should be neutral houses.
So happy EU can make descision on all these little details 
Anyway, good news though!
I wonder what happens if electrical utility company won't give you permit for solar panels because the substation is overloaded. Are you going to get fined?
the 1.5 degree we are not going to make for climate change.
But the doom scenarios are also off the table.
So we didn't get the milestone from the climate top of Paris, but we averted the worse case scenarios.
I think we are aiming for 2-3 degree now, but that will have severe consequences.
Are you sure it's the worst case scenario? 
Did you know that hail*, grows exponational with rise of earths temperature?
So, we all going to have solar panels right?
And we know that hail is getting bigger and bigger with the rise of temperature.
And what we feel today, are the effects caused 30 years ago.
So in a few decades, our solar panels will be blasted with hail and we going to need money to repair them.
Solar panels I am going install are tested on 35mm hail. I don't think we ever had hail larger than 10mm
Like I said.
Hail is GROWING.
The higher our global temperature, the bigger hail is. What you see now, the size of yuor hail, is not the future.
I highly doubt it's going to grow from 10mm to more than 35mm in a short amount of time. Manufacturers will have enough time to adapt and create thicker glass
Depending on the location that is
Hail larger than 2 cm in diameter is classed as severe, and once hailstones reach 5 cm or more, they are considered very large, and at that size, they can smash car windscreens, puncture roofs, flatten crops, and even threaten aircraft safety or result in injuries or - in rare cases - fatalities.
that's 20mm to 50mm.
But don't believe science, I am not going to discuss science π€·ββοΈ
R.I.P. to people living in areas where hail can reach 5 cm
But I found it interesting how silly we are as humans.
Doing stuff and not looking ahead in the future. Just patchworking problems one by one. Creating new problems.
Like going full electric cars, and politicians thinking, we did a good job.
Without keeping other options opening and probably causing different issues in the future.
You had a car company pretty far with hydrogen.
probably the whole world π€£
Near-surface hailstones <4βcm are found to decrease in frequency by an average of 25%, whereas the largest stones are found to increase by 15β75% depending on the greenhouse gas emissions pathway.
This is from the paper itself. (www.nature.com/articles/s41612-024-00728-9)
I highly doubt every place on earth is going to have 5 cm hail though. Don't know much about weather but there are probably conditions which need to happen in order to have such a huge hail

You do know what **__climate __**change is right? π€£
It's not becausse it never rains in, let's say a desert somewhere, that it wont be snowing there eventually if climate shifts.
The world has been frozen over and dawned at a few points in it's history.
also, if climate becomes more "extreme" and "sporadic/unpredictable".
I wonder if it ever will have an impact on our air traffic.
like that we need to protect airplanes for even MORE extreme weather conditions.
They said Croatia is going to become a desert by 2100 but I think it's a very short period and the more realistic prediction would be by 2150 or 2200
Yeah, and if the ice keeps melting we going to lose land.
we shouldn't have pissed off earth π
I mean, a place with dirt can become a desert but I higly doubt it would happen in a very short period of 100 years
isn't las vegas troubled by desertification? and they see it pretty clearly if I am not mistaken.
Isn't whole Nevada desert?
βVegas has done great things such as ripping out the grass, but weβve lost 20% of the flow of the Colorado River since 2000 and another 10% loss by 2050 is completely possible,β said Brad Udall, a water and climate scientist at Colorado State University whose research has focused on the stresses facing the river.
i mean iv nerver seen hail large enough to damage anything
iv seen a shitload of ice damage but nerver hail
lucky you.
I had. Was a crazy period.
Every car in Antwerp, in the city, full of dents.
Like I was walking around and all I saw was damage, damage, and car damage.
If you had any glass in your roof, gone. Happened to friends of my parents.
and that was almost 15-20 years ago.
let me see if I can find a picture on the internet.
what even is hail lol
like it feels like angry snow
its frozen water instead of freezing water basically
Basically ice balls and you hope they don't hit you in the head 
Yeah, snow doesn't have the density.
This is just very dense ice at high speed.
So, you are right, it's pissed off snow π€£
no ik theres hail around here just its usually so small it just feels like a more agresive rain drop
like you wouldent know its hailing unless its super windy and its hitting you in the face
and another thing i dont get is why does hail only show up in hot areas when its literally ice
you will feel hail at certain conditions π€£
let me see if hail ever killed someone.
Look at this. Holy Moly.
The World Meteorological Organization reported that the highest mortality associated with a hailstorm happened near Moradabad, India, on April 30, 1888. The deadly storm killed 246 people with pieces of hail as large as βgoose eggs, oranges and cricket balls.β
definitely more dangerous in the hsort therm than what im use to
The largest one looks like around 35mm
When the sky starts to fire bullets at us π€£
Snow is tiny ice crystals forming inside the clouds.
Hail is more like raindrops freezing on their way down
and somehow clamping together, to get bigger in size.
Instead of forming crystals, it's just forming random ice
So it's less controlled and faster giving you clumps
And higher speed cuz bigger mass
ofc.
We have less snow and ice. We have a lot of tree damage over the years.
That's just branches heavy with snow and strong winds moving them to a point where the roots give in
The destruction doesn't come from the ice, the ice just provides one component
Not from Cologne and didn't exist in the 90s
Well, the ice adds to weight.
whens the last time you drove a full size tractor on a snow bank without it digging into the snow even a CM
The 2000 isn't there for no reason
thats how strong that ice is
If ice builds up on one side of the tree a little more, and then with some nice wind.
Except snow doesn't compact that strongly on a tree branch
It's just heavy is all
But if it's really cold? Wont the snow freeze together on the branches over time?
And if it sticks to the branches?
Also, compacted snow is technically different from ice.
It's the difference between a quartz crystal and a handful of sand
Other way.
Cold snow is light and fluffy.
Wet snow is very heavy.
Mhm, that snow looks frozen on those trees. Not snowy. But can be my eyes.
If it's really cold, what's there to freeze?
snow usually falls off and dosent stick unless its so cold it turns to ice on contact basically instantly
And you get frozen stuff in surfaces with either subcooled water, or with snow melting and refreezing.
It needs to melt a little before it can compact
Well it can happen under certain conditions.
Both which happen near 0C
but it's also hard to see whats under the snow of the trees in the pictures I see π€£
thats all ice
but I would suspect snow/ice does add to the weight of a tree.
And if that tree falls over + extra weight ...
oh yea a lot
Basically rain with little under 0C temperatures causing stuff like that.
Yeah, I think so too. The other trees (background) are totally fine.
i dont remember how much exactly but isent it like atleast a 3X increase in weight or so on the average tree?
like 3.5X on average or so
I happen to live in Finland...
I have lot of experience on what kind of weather gives what kind of ice, snow etc. π€£
i do to but i dont really pay attention cause all i really care about on a day to day is "is there too much snow to go out or not"
Although... maybe I am wrong.
Can someone be crushed by the weight of snow when they hit by an avalanche? Like literally, the body crushed by the weight.
crushed idk but there gonna suffocate for sure
Maxeek keeps switching his picks.
But like I said, to get lot of snow into trees etc, it needs to be wet when falling.
Cold snowfall is very fluffy and only accumulates on flat surfaces without wind.
Yeah, but I wonder. If snow keeps adding to weight. Or balances out.
i mena at the end of the day it depends how much
if you get a boulder sized snow ball the density of ice and it absolutely decks you rolling down a hill at you know pretty darn fast speeds do you think you would survive?
Snow that has had time to pile up can be very heavy and solid, as the weight of the upper layers compresses the lower ones, also heating them to nearer to water, so it starts to turn more to ice form.
youd probably survive but you wouldent walk away from it and probably will die soonero r later
Or if there are suitable temperature cycles etc.
Is the image only not loading for me? I put so much effort into it π€£
Didn't load
working for me
That was the second post that worked.
But that's what I was wondering.
Even more, because of that compression making the lower layers more dense.
As the fluffier snow compacts and turns partly to ice.
Yeah, I just pulled numbers out of my ass. To put into a picture what I head in my head π€£
So it would go from 10 to 25 to 60 or something,
yea kinda
usually the top is hard the middle is soft and the bottom is bedrock basically
Ok, sorry for the derailing. I recently saw YT video of someone being rescued under an avalanche
But basically for snow to stick to tree branches etc. it needs to be wet for the starting layers.
The cold fluffy snow just falls off if there aren't enough flat area to start.
And same during nearing to spring, any snow on roofs etc. gets heavier as it starts to get wet from ambient moisture, which can break roofs that handled the same snow during the winter, without any new snowfall.
and the flooding from the snow melting
Fuck, I wanted to know if fallen trees still had snow sticking to them.
But if you look at the bark it has snow on top, so it meant it was still snowing so the snow build up after the tree fell. π
Well Baldur says it falls off.
when it dry and fluffy yea
And you need wind for the tree to fall? Wouldn't wind blow snow off.
but the wet stuff is still gonna build up
Cold snow, not wet snow.
Oh missed the wet snow. Ok. So two types of snow. Makes sense.
More, sliding scale in a way.
I understand you.
You can only make snowballs with wet snow for example.
If it is too cold, you cannot make them.
In the middle you can kind of make them when your hands heat the snow and make it partially melt.
That's how I make snowballs.
Basically you cannot make snowmen below -10C (IIRC) at all for example.
yea dry snow is literally just sand
a very obnouxious powder that you sink in effortlesly
And -15 - -20C and below any new snow is just powder.
Interesting. It never snows enough here to build up that kind of knowledge.
We got a lot of fallen leaves. And if not that, we got grass. Or dead grass π
basically flour
Basically with proper wet snow, you don't need to heat it at all to make a snowball, it just sticks together naturally without any extra heating, you just add the shape.
And when you throw it at wall or like, most of it just sticks.
As kind of reverse crater.
let me pull up some videos from my gopro lol
i remember one winter i said fuck it and tried to get through the soft snow anyway
it was fun but its a dang good example of what you DONT want to wake up to on a week day
Worst is when there is thin hard top layer, that doesn't bear your weight.
Just full fluff is easy compared to that.
use snow boots/paddles whatever they are called
Very cumbersome. Skis are much nicer.
thats... nervermind
Ah found it π€£
snowshoes, like I said, cumbersome.
snowshoes, not boots right.
cant find what i was looking for but i found this
This is the dry crap you dont want to wake up to on a week day
its literally just powder with a insane amount of resistance
And seems few mm. thick part-ice layer on top, then powder, then more solid older layer below.
Right on top of the ground.
that was first big snow storm of the year if i recall so in that case no
it was a bit over a foot of all soft
hadent hardened at all yet
but usually yes
is your snow scooter struggling? (don't have audio on atm)
Yes, snow can be too light for snowscooters, and they sink in.
Backend only in some cases, and you don't get any traction, or front too in more severe ones.
In easier cases you get off it, and run it forward/backwards slowly or in bursts while next to it.
While fighting with the same snow you might be waist deep in.
In worse cases it needs towing or like.
a snow scooter?
thats one iv nerver heard before π
Oh you have normal one?
thats not something oyu can just go and buy you need to mod those
you can get snowmobiles but yea there expensive and not worht it if you dont have trails around
its just a old fourtrax atv from the 80's
Sorry, english you know π€£
there nice for snow because there light and small compared to the new ones
so you can fit in a lot of places and if you get stuck you can lift them completely in the air with 2 people if your desperate
but yes to answer your question i was flat out the entire time in that video so yes the engine was holding on for dear life
the motor has no cooling?
so only airflow as cooling
yep natural airflow
and environment
no radiator, fans anything
Air cooled engines are quite common
yea at that size you really dont need it
They're a rather simple and reliable design
Not trying to lecture you lol
why not?
But yes
On most lower-end or utility focused atvs they're often used
im curious
Because of the power. But maybe I overestimated it π
I though people would go for more HP, turbo, I don;'t know. Stuff like a car.
But I know very little of it as you can see π
Oh you can, but only for serious competative racing
And it adds a lot of weight
So only in very specific scenarios
And essentially never practical
a turbo wont help a engine like this much unless its carburator is set wrong
mine is too low so it gains power in winter when the air is more dense
so winter is the turbo
Winter means cooler, denser air
but if i adjusted it like you should it wont make a diffrence
So you can create more power per stroke
Cool stuff!
satisfactory paintjob
Truth is power dosent matter much in snow
as long as you have enough power to spin the tires you dont need more
in 99% of cases your choke point will be traction and ground clearance
any vehicle with a motor where you're outside or on it scares the crap outta me
i see way too many stupid people on those kinds of things
How many years did it take Microsoft to implement a separate icon for when cable is disconnected instead of any problem with internet?
That cable disconnected was there before internet detection?
IIRC
todays was first time I saw cable disconnect instead of globe with crossed circle
Modded ATV in the pic?
Because actual snowmobile has single long tread:
Blue Origin Literally NAIL The Landing Of Their New Rocket:
https://www.youtube.com/watch?v=XAYYWjvXgaM
It's been 10 years since Blue Origin demonstrated landing and reuse of their New Shepard rocket, now it's time for New Glenn to show off this capabilty. On only the Second flight of their orbital class rocket they successfully landed it on a drone ship, taking a few more steps ferociously.
Follow me on Twitter for more updates:
https://twitter....
will be interesting how that works long term
reading a GN review that ends with this note:
With the 9070 XT, we think the $600 MSRP is the minimum that AMD had to do to get some positive groundswell and attention. And it achieved that: there has been a lot of enthusiasm over the last few days, in no small part thanks to NVIDIA continually f***ing everything up. We wish AMD would have come in slightly cheaper on the 9070 XT to really kick in the doors to the mass market, even though thatβd hurt short-term margin, just to make sure it can generate some momentum and keep developers integrating its own technologies.
π
Yeah, AMD wasted very good moment to try to get more market share, by pricing 9000-series too close to Nvidia, at time when Nvidia was fucking stuff up.
And now the RDNA 1/2 driver stuff.
But of course lowering the prices would possibly have caused the profits to drop, even if revenue would have gone way up.
And of course would have affected profit margin anyways.
i hate corpos : I
And I am not a fan of how adrenalin "tunes" GPU.
Default is a mystery. Undervolt, just undervolts, no underclock option. Then overclock, overclock vram. Ok.
Then you have quiet. Which ... barely does anything. Then you got balanced. Which ... I don't know is the same as default? And you got rage.
And if you put it on costum and disable most OC, you get different behavior than default? Why is this so convoluted.
Also, I think from their perspective it would have been pointless to lower prices until now because for several months they were selling everything they had even when it was priced above that msrp
Yup.
They said they sold more cards than anything in recent memory, but the whole time nvidia was outselling them over 5 to 1
they seem unwilling to produce enough gpu's to maintain their market share, even as low as it is, let alone radically grow it to push back towards the 60/40 days.
GN said nvidia owns 90% of GPU market π
Free market at work 
I like they mention that we shouldn't give AMD/Intel pity purchases.
But no-one should buy AMD GPUs as they have:
A: bad drivers
B: No DLSS
C: No CUDA
/s
Where that A is mostly just a myth that just keeps living.
And to most people C doesn't matter at all.
I know you are passionate about AMD, but you can't blame it all on Nvidia. AMD is (partly) responsible for the situation hey are in.
Nah, I just hate Nvidia.
AMD is nowhere perfect, Nvidia is just horrible in comparison.
If it's either by bad descision, bad reqruitment, bad engineers, bad QA, bad software, cheaping out on products, bad costumer service, lack in features, ....
But they did something to push a lot of people to Nvidia.
But those three are the normal things that people use to make people they recommend GPUs to always go for Nvidia.
Mindshare is big thing, and Nvidia has been great at spreading that.
Leading to point where large part of consumer space only consider Nvidia, not because of merit but just because of the baggage in minds.
The one that wouldn't surprise me, that holds a bit of truth is, bad drivers π
DLSS/CUDA I can live with. Opinion ofc.
Yes, used to be significantly behind 10+ years ago.
But it just keeps living even today.
Spread by people who for most part haven't had any experience with AMD GPUs since ATi times or like.
I don't know, I am thinking last few days if I should keep AMD or go Nvidia. The price of 5070 is doable. If it's the GPU, then AMD fucked up imo. Or XFX.
But somebody fucked up. And Nivida would also mean I can excludee AMD drivers being culprit or fucking over my new card.
Yeah, but broken/unstable from factory cards happen on both sides, with about same frequency AFAIK.
True, in my situation, it just gives me more safety I think.
And is suprisingly high, up to few percent of production.
Like I said, I can slap in 600 euro again. If drivers are fucking my card, in 2 years I am in same situation.
For the factory OC models.
How high?
0.5-2% for factory OC models AFAIK.
both AMD and Nvidia? π€
and mine isn't even OC model lol.
Yeah, no. I don't think I shall give AMD my money again. (actually, no clue tbh, the CPU is fine).
Yes. Part of why Nvidia has clamped down hard on OC models IIRC.
Partners setting card clocks higher than the base model.
it's literally models with OC in the name right?
Depends on card manufacturer if they add that OC or not.
a lot do.
And for some, the highest OC model doesn't contain OC in the name.
That doesn't help me at all π€·ββοΈ
Yup.
Basically need to compare card spec to AMD spec.
As naming isn't reliable and there can be multiple levels of OC over stock.
Yeah, I figured that out.
Where the higher level cards are much more likely to end up unstable over few years of use.
But it's just again... shit practices.
Or be unstable straight from box in some specific corner of turbo-table.
Like they put all this information on the box to make it easy for consumers to pick up a GPU.
Yup, basically the card manufacturer scene hasn't caught on to the current per-die level of turbo table variability.
but in reality, you got to compare full spec with manufacturer if you want to know the truth.
a GPU can say OC or not OC. Doesn't mean it's OC out of the box π€·ββοΈ
And then you have manufacturers going off the specs.
And do the binning and final card testing without enough coverage.
And then you have the motherboard manufacturers fucking with the GPU in unimaginable ways.
So, everything is just doing shit to everything and then the consumer can figure it all out if it goes wrong.
That's basically how I see it π€£
Oh, and all at the top you have also windows doing it's shenigans. Installing drivers automatically etc, turning stuff on/off to "optimize".
Also, logitech got hacked.
And I feel like it's only getting worse. So even with security improving over the year. It's seems breaches are gaining.
Maybe it's time to not try to solve this. But just make sure there isn't anything valuable to get.
Every website/tech company, needs to quickly stop data hoarding private information about users and start respecting privacy. So even if hacked, at most they get a name and mail adress, but not a phone number, street, age, gender, ...
Proactive protection is just cost center and unneeded.
And data is profit center.
(CEO think)
Then goverments should slap with fines for every data breach for neglectence of protection and/or hoarding of unneeded data.
I sure hope that graph is cumulative (it probably isn't)
i mean yea mine is a OC that i havent touched and the vram is starting to show signs of noping out so yep its starting to have its tole
that statistic would not surprise me in the slightest
especially with enviromental diffrences making the overclocks no longer stable even
And the very limited final testing that seems to be done based on the various GN GPU factory tour snippets of the test stations.
Basically just testing that the cards work at few max load scenarios when new without crashing fully.
yea just a short furmark burn in usually right?
Man alive
I'm having a wild morning today
sometimes with an extra step
That last was single 3DMark run and some FurMark.
I went to go listen to my discover weekly on spotify and it is entirely AI generated
not surprised
For Yeston
yea i watched it just forgot what specifically but i knew it was furmark and something else once
Don't remember which specific 3DMark test.
How to insult older gen Z and gen alpha without them knowing 
Younger gen Z might have used those phones and will understand
Yeh
This doesn't make sense to me.
Wouldn't older genz have used em and not younger genz?
Oh, sorry
I mean, older gen z
When you think one thing and write another 
At least they didn't have to deal with rotary phone and older CRT TV you had to slap in order to fix the image
I learned how to use a rotary phone but never actually used one.
As for crt, ours worked without being slapped
yea iv never heard of slapping a crt before but yea the only one i have works without it
AFAIK only needed for failing ones, not for fully working ones ever.
Was it the one from the 80's or older? Newer CRT's didn't have issues
I was born in 2000, idk when that CRT was made
I had an issue with a wooden sony trinitron which was probably made in 70's or 80's. Had no issues with newer CRT's though
Probably cuz this
#off-topic-tech message
Not sure tbh. Heard a lot of people doing that back in the days
Unless TV's were unreliable in general
Age failure, and there were also (semi-hidden) potentiometers for tuning the sync circuitry.
So that happened when the sync circuitry doesn't actually sync.
The real fix was tuning, but that hit causes enough jostling in the electronics for it to sync for a time.
Later that tuning was automated and not with pots/trimmers.
As aging and electronics temperature etc. caused changes.
This. Hitting the TV only worked for 10 minutes and you had to do it again lol
Oh I remember the times of hitting my TV π€£
common gotta work for that lazyness a little atleast
I don't own a TV anymore..
Stuff like this, first with some of them in the back, sometimes all are inside.
And lot more for full set:
All moved to software and/or automation later.
Some of the very old TV:s had most important ones at the front, as they needed constant adjustment to keep that main sync, as the TVs temperature changed etc.
chrome hates me and decided to duplicate, no, sorry, triplicate my tab groups...
i bought a 9070 xt because im afraid that the gpu prices over the next month and year might get fucked by the ram issue
was waiting on a 5080 super, but thats as good as canceled
Never forget what they took from us! https://youtu.be/sp_OISQDtAM
Start learning cyber security with TryHackMe: https://tryhackme.com/salem Use my code "SALEM25" to get 25% off on annual subscription.
This Dell Vostro is just 1 example of the type of laptops that used to come into the repair shop I worked in where I would geek out over how cool it was. I've always liked these bulky laptops due to their build ...
i had the very first gen dell vostro, was a miniature beast until the class action suit about shitty gpu soldering happened
separate panels to access parts, and a swappable battery
I got pissed one month ago when I noticed new laptops don't have cmos battery socket like on desktop motherboards anymore
Hasn't been normal ever.
Mostly same as older desktops where the battery is some kind of nonstandard thing with wired connector.
Some have them soldered.
You have to basically buy a cmos battery with 2 pin connector and that's going to cost at least β¬10 minimum while just buying a battery costs few euros
The laptop in the video has that standard cmos battery socket
Basically desktop MBs standardizing to 2032 cell battery holders at some point is the outlier.
Some have, some had, but in minority.
Makes repair more difficult. When the cmos battery died, I was naive without even checking my laptop and thought I could replace it in 5 minutes....
Boy, was I wrong when I opened the laptop cover
Ok, it seems it was a bug in UE5 itself. Deleted some config files and it booted up. I was already confused how my GPU crashed while my game was still open π€£
Ofc the error message makes sense, it's a multi billion dollar company 
and I could open another project without a problem. Leading me to believe that particular project was botched.
I think windows update killed my parents windows 10 and I will have to reinstall windows again....
Ok, it managed to reboot and reinstall updates somehow so I don't have to do anything
Almost thought windows installation was toast
Hope Microsoft isn't trying to kill older laptops which don't support W11 with updates even though they said there would be no updates for W10....
Correction: there will be security updates though
really sad that that's not AI
i checked my 4090's connection on my last pc cleaning, and it was fine, but a post on reddit made me paranoid that i didn't check the PSU side of the connection
also i completely forgot that my case actually has spots for 2x 2.5" drives if i wanted to. thought they had ditched them for my case
new GN video on RAM π€£
Yeeep
Bought and OBD2 ecu flasher, a chinese 10β¬ copy of 150β¬ product. But now I have problem as it came with drivers on CD and DVD and none of my PC:s have drive anymore π
Theres a sticker on the product that says "use only our drivers or the device will be damaged" .. hmm
So one using copied chip design, and masquerading as that copied chip, but with incompatibilities with the official drivers.
As the copy isn't 100% accurate
Yeah, my plan was not use it to reflash working ECUs. Just use the standalone microSD logging with bikes, as in offroad use its possible get water into that thing. Even though I've used plastic bag and duct tape protection
And been worried bricking my original one
I bought USB DVD drive just for things like that when I got rid of internal drives.
oh, now that I read some forums, the original driver may have some built in protection to detect china copies and will brick them
Yeah, at some point that was added, basically inputting commands that specific clones cannot deal with, but the real deal can.
Because those clones were way too wide-spread, even making into official supply chains.
For the specific USB <-> RS-232 converter chip.
And new enough drivers are now in Windows Update too.
That cause some of the bad copies to brick with the mode-set or like commands that they cannot deal with.
Is this normal behavior for memory clockspeed for GPU?
13xx MHz to 22xx MHz rise and drops, sometimes to 9xx MHz
After some reddit reading, seems to suggest it's normal.
Yeah, when at desktop and nothing GPU heavy is running.
Well... UE5 is running and the game on the taskbar.
This is when sampling interval is set to 0.25
This is my iGPU **without ** 6950 XT in the case SAME situation (game, UE5, discord). Interval set to 0.25ms.
But maybe the two work differently π€·ββοΈ
iGPU memory is system memory, system memory does not downclock
Yeah, main RAM doesn't have capacity for speed changes for power saving.
GPU VRAM does.
Thanks
Technically it does, but it's some jank with overclock profiles that nobody uses (it's not very good) π
Not frequency change even then.
Just either full power save sleep, or full active.
And that is on by default on AM5, because the memory training retaining needs that.
Testing atm to make sure I can't provoke hard reboot without 6950 XT, for a moment I was wondering if it had somethign to do with the clock speed of my GPU* memory, since it seemed to fluactuate heavily.
There is speciifcally a feature to train two different memory profiles (e.g. 4800mt/s JEDEC and 6000 OC) and to swap between them for power savings after memory training (like in-OS)
but you can only specify settings which fit into an EXPO profile to do that, which excludes most of them
it's also quite minimal power difference and affects performance
Second question, I can't disable memory on 6950 XT in BIOS or something right? (I did look around but didn't find anything).
Maybe in the GPU bios but i don't think so, and that's not designed to be user accessible.
Yeah, wondering if problem would persist if I disabled GPU memory. If problem dissapears, then I am 99% sure it's GPU. That was my reasoning.
It cannot work without that memory.
at least without some of it π
That fucking sucks.
I'm fairly sure it's demons in the power delivery stuff. VRM etc
Yeah yeah ... two PSU's. Both problems.
No GPU in case. System stable (for now... will test 24 hours).
Only thing I am not sure off is, either motherboard or GPU.
The PSU provides 12v power to the graphics card. The card then has a bunch of power circuitry to drop that to several low voltages (around 1v) and regulate them. That takes like half of the graphics card area, it's a big job. It fails much more often than other parts.
Also the only part of the GPU that can make sounds, outside the fans.
Well, the rest can, once.
You mean like that.
yeah
Yeah, I think so too.
But somewhere in the back of my mind is this little voice, "maybe it's the motherboard" π€£
it can also feed back to make PSU act weird in rare instances because it's fed by 12v from the PSU, and if the demand suddenly changes then the PSU "feels" that
Which shouldn't be able to affect that VRM side of the GPU.
Sure enough to bet on it? I don't think so.
mobo does supply some power too via pci-e slot. But i would say graphics card pretty much 100%
100%! I talked about this with Baldur. I though I was crazy.
you could ask buildzoid if he wants to tear apart a demon gpu and take a look at the power regulation π
But he was like "nah, you're not fully crazy yet" π€£
maybe I can get my money back that way ahaha
That reminds me, there was a problem on a motherboard VRM not too long ago where for some batches of mobo's, Asus literally attached a capacitor backwards
and it caused havoc
I wrote down some timings, normally I am able to crash system in 10 minutes to 26 minutes.
I took out 6950 XT. Booted up everything that provokes crash, we are sitting at 35 minutes now.
Don't tell me this. Please π
Yeah, someone on assembly line didn't check which way the capacitors were mounted on the reel when installing the reels.
And they were random.
Same per reel, but could be either reel to reel.
PS; a second reason I still have my motherboard in sight is this.
I found out that a lot of things weren;t put to auto on the motherboard bios.
As if someone fucked around with it for a bit, and then send it back to vendor. And that vendor sold it back to me.
Just a theory.
And the optical checking wasn't set up to check for that on the finished boards either.
π§
Just that the cap was there, not that it was right way around.
How does it even work.
that should not happen
Which of the two? Settings set to values (not auto) or vendor selling crap back to me? π€£
The mobo not being on defaults
it should also have been sold as used/refurb
both are legal requirements here at least
I was wondering about that. In EU we have the right to send goods back within 30 days without any reason.
What happens to those goods?
If it's open, they have to list it as B stock
You had updated the BIOS at that point, which resets everything to BIOS defaults.
Just that AsRock wasn't setting those settings to Auto, but to specific settings.
usually with a discount, and making sure that it still works fine.
Nah, I don't think this was the case.
I reloaded "default user" multiple times.
I updated the bios multiple times.
And after a year or two we found out shit was set to values and not auto.
The only fix was, for me to go over everything I can find and check if I can turn it to auto.
So it's fully possible the motherboard was shipped with values set.
Point was that AsRock had programmed the BIOS to have the default to be "FCLK 2000MHz", and not "FCLK Auto", etc.
Well, that doesn't give me more confidence that mobo isn't culprit π€£
Image, I get new GPU, it goes good for a year. Then starts hard rebooting again because faulty motherboard fucked my GPU. I would lose my shit I think.
Well, I might do a coin toss eventually.
btw, Adrenaline can also set BIOS settings.
FCLK 2000 is within spec
AMD actually has auto fclk set to objectively bad settings sometimes, probably for lack of testing
so they are just forcing it to not be dumb
And this is how I know.
It might just always say that
They wont be saying that for nothing. AMD Adrenaline, has access to bios to set things. π€·ββοΈ
The spec is for FCLK to be 1733MHz, no matter the RAM speed.
The unofficial recommendation is for that change depending on RAM speed.
AsRock on that specific MB, with couple of BIOS versions had FCLK always at 2000MHz, no matter what.
For AM5.
let me check. One second.
1733 is default with 5200mt/s RAM, but the spec is anything up to 2000 AFAIK. I had that confirmed by the AMD employee who wrote the settings
It seemss to be set to 1,796.8 MHz (auto).
your base clock is bad
It's auto π€·ββοΈ
99.82 bclk tho
That 2000MHz is in line with the unofficial recommended settings.
That 6000MT/s RAM is the sweet spot with specific ratio.
Again difference between the official spec, and the semi-official "almost every CPU should be capable of this" recommendation.
That's just a coincidence, it's guaranteed up to 2000
for example if you run the 2dpc 3600mt/s spec, it uses 1800 fclk, not 1733
Where do I find base clock?
Not in any official material.
And not warrantiable for not hitting that.
Stable.
Not sure why it isn't 1733MHz though. The FLCK. bclk seems to be what Aeryn said.
@twin dew How do you know FLCK is recommended 1733MHz? I cannot find it on AMD 7800X3D documentation or techpowerup.
Asus was using 1733MHz FCLK with FCLK: Auto, RAM: Auto, no matter the RAM speed.
And then switching to the ratio when RAM was manually set, either directly or via memory profile, up to 2000MHz.
And it was part of the original launch material IIRC.
Asus is your reference?
No, that original AM5 7000-series launch material.
i believe otherwise, but hard to dig up the right evidence at this point π
I had to do few rounds of back and forth in emails to get my first 7800X3D hit by the VSOC thing get warrantied before it died.
Couldn't do 1833MHz FCLK stable anymore with IGP on, and that max was going down.
That is part of why i confirmed it with bill
1900MHz wouldn't post, 1866 MHz was very unstable, 1833MHz was unstable, 1800MHz still worked for most part.
At that time, but it was going down somewhat fast.
if 1900MHz woulnd't post, what did you do? Because you can't access bios to unfuck it?
Asus defaulted to that 1733MHz on CMOS clear.
Ah nice.
Weird thing is, whenever I do a CMOS clear, everything stays auto π€·ββοΈ
MSI was 2000, and they didn't do anything anywhere to my knowledge which violated spec on OOTB (out-of-the-box) settings
So if what you said was true, and Assrock set values other than auto, wouldn't they reappear with a cmos clear 
the only thing that reappears after a cmos clear is "auto driver installer: enabled".
But I was able to argue that as the CPU had run with that Auto VSOC feeding too much voltage, and the max FCLK was going down at significant rate, to get the CPU replaced before it would have died within warranty period anyways.
Because originally it could do that 2000MHz FCLK and more with IGP on.
And still could with IGP off.
I hope my CPU is healthy. I did check both RAM sticks and they are out of the clear.
I was actually happy, with how the prices are going for ram π€£
IIRC your original CPU had run with that VSOC problematic BIOS with non-auto RAM speeds, so had hit that VSOC.
And did have FCLK stability issue as kind of confirmed.
fml, sorting by price doesn't even work.
look at the prices, they aren't sorted π
699 is also between 703 and 748 ... ofc makes sense.
Ooooooooh. It's adding the cost for delivery. Confusing.
automatic OC's are a shitshow. AM5 wasn't the first and won't be the last
if you know what you are doing, should manual OC. If not, don't OC IMO
Honestly, as a noob. I have no clue if my motherboard is set to default or not. I am so lost with all these shenanigans.
Not Auto-OC.
But MB makers automatically raising VSOC when RAM frequency was raised manually or with XMP/EXPO.
That is automatic overclocking, setting stuff to OC speeds (or using profiles to do so) without controlling the relevant variables. If you don't do that then the mobo will automatically do it for you, and they do a very bad job of it.
No motherboard actually overvolted SOC out-of-the-box, or set to spec speeds, or if you set SOC yourself. It had to be explicitly set to overclock speeds + auto SOC. That's automatic OC.
I though I didn't OC. But it turned out, it was OC'd. And I never overclocked values myself.
And that's what I mean. You slap in a board. You slap in a GPU and CPU.
You start playing games without fucking around in BIOS. System reboots. You start investigating.
And eventually find out BIOS isn't default as you think it is.
People, who don't overclock, shouldn't be dealing with broken hardware as the result of OC. That's just not fair.
And if Assrock, AMD, or anyone else want to auto overclock, the waranty should be WAY longer on the hardware because it's their responsibility.
Maybe we need 3 types of warranty.
OC warranty: you fucked with it, no money back.
AUTO-OC warranty: 5 years or something, AMD/Assrock fucked up, they pay for it.
Normal warranty: 2-3 years.
We're not on the same page here.
AMD mandates that out-of-the-box settings are within spec.
When i say auto OC, i mean that somebody went into the BIOS and e.g. pressed one button which says "make memory go OC speeds" and then the motherboard automatically set 30 different things related to doing that. It requires explicit user intervention, but the user isn't controlling the settings themselves. They are using like a one-click "easy OC" button. Most commonly these are XMP/DOCP/EXPO, but there are a few other names.
That stuff is awful π
Which also "auto" OC stuff. Even the default profile is overclocking.
Like if I set a costum profile and turn all OC off. It's different than default π€―
So default is overclocking.
And I set VDIMM, and RAM frequency manually.
Which then also changed VSOC without prompt while that was still on Auto.
Basically table in BIOS, that changes what VSOC is on Auto, depending on set RAM frequency.
Which has been a thing for ages.
But wait, I though auto meant "default". In bios?
Yeah, you have to methodically and carefully control lots of stuff like that when overclocking. When not overclocking though, it doesn't hurt you as it will be at/below spec.
For example the spec defines SOC voltage of 1.05v with 5200mt/s RAM, the fastest supported speed.
If you set RAM to 6000 or 6400 and have auto SOC, board will make up its own thing and you're at high risk. It can kill the CPU instantly.
because there is no default option in bios.
Auto is Auto, in most cases means single default, in some other it changes depending on other stuff.
=_=
And I did catch it in less than 15 minutes, as next I booted back into OS, checked the various voltages and saw that VSOC had been raised to insane levels, went back to BIOS and lowered it manually.
But too late.
How fucking obnoxious can you make a bios.
They shouldn't use the word auto and just use default.
and when something can change depending on other stuff, call it auto.
So the user can decide to lock in values. Or let bios "auto" pick values.
Yeah that would be nice π
I'd rather have it fail because i didn't set something, instead of having potentially lethal settings applied because i didn't set something. You need a ridiculous level of paranoia to tune anything in BIOS these days.
And that's ... why there are prebuilts π€£
unfortunately
most prebuilds are automatically overclocked and poorly tested
you have yet to grasp the true extent of the shitshow
With the added benefit that if your prebuilt has issues, and whoever prebuilt it is a good vendor, you can bring it back and get a new one.
Reason why lot of prebuilds come with RAM on Auto, and XMP not enabled.
Marketed with the XMP RAM speed, but not with it enabled.
Yeah, that's good at least
particularly if you have a local store
i don't, but i live in the roof of an amazon distribution center so it's not so bad
Yeah, I know prebuilds have drawbacks. But I am not sure if I could go back in time if I rather spend a few more on prebuild or have my current situation.
Because that was the main argument I heard two years ago, prebuilds are way to expensive. Prebuilds are garbage etc.
I do not tolerate unexplained instability, but one of the most powerful diagnostic tools is swapping parts and you need to be around a bunch of computers to do it easily. Like there are 3 graphics cards in my house atm.
People forget the pay for people to build the computer, and for warranty.
I did tell you what my local shop said right? I asked if they could stress test my GPU.
No mobo π
Motherfuckers.
bad shop
Yeah. Like asking for a spare tyre at a mechanic or some other basic shit like oil, and they say they don't have it.
In that case, what you can technically do, is buy a new GPU and run it, and you can return it within X days. It will work or it won't.
Honestly. That was my idea.
ideally you plan to keep it if it works π
But that's how I got so paranoid with my motherboard.
My paranoia;
If the GPU works for a few weeks without any hard reboots, I think I am out of the clear. But if there is something else in my system, gradually degrading my GPU, in a year or two.
I am back at square one.
And no, I cannot refund it that easily π
I could refund it and say "hey the GPU is broken" ... but in reality it would be my fault and not vendor
The vendor doesn't know that ofc.
Random shit is factored into margins
you've done more than most to ensure that you're not accidentally damaging hardware
That is true. Baldur did help a lot on that department too.
I think that will be my new GPU. Sapphire Pulse Radeon RX 9070 XT
I read some reviews saying it's very quiet, and good cooling.
In order to protect your card, the SAPPHIRE cards have fuse protection built into the circuit of the external PCI-E power connector to keep the components safe.
π€£
Kind of a lie.
Protects the PCB when a VRM power phase goes and shorts.
So makes any GPU repair much easier.
Or the PowerColor Hellhound. But reviewer warns of slightly higher V temperatures. "Slightly".
9-11 degrees isn't slightly to me. And isn't 88 getting near to point of "Oh fuck?"
Although the testing is done on BIOS 25.4.1 with reports that 25.5.1 reduce temperatures.
90C for VRAM is still perfectly fine AFAIK.
110C is AFAIK the limit for most such chips.
spiral for the win!
https://www.youtube.com/watch?v=QtxVdC7pBQM
the most interesting part will be to know how that pigtail got discovered/introuced to the process,
cause it's some alchemy level kind of shapes affects results
How does a jet engine not melt? Sponsored by KiwiCo - Use code VERITASIUM to get 50% off your first monthly KiwiCo Crate! https://www.kiwico.com/VERITASIUM
Want to start your own crystal garden? https://www.kiwico.com/crystalchemistry
Our friends at Science Alert are doing a sweepstakes where you can win a 5-day space themed vacation for two. ...
RDNA3 Hellhound made me hate that name forever
fucking bullshit lighting
I am looking at GPU models and brands. Investigating the brands itself too... they all look ... I don't know off.
And what do you mean with bullshit lightning, the RGB?
Basic aerodynamics?
If you mean that the turbine blades have backwards tail at the outer edge?
For the model I am looking at, it has a switch. Bless em π I really like that. Giving the user options.
Other than that. It features some weird fan stuff.
middle fan spins opposite way.
Man, the image quality is so bad on this website lol.
no no. the pigtail in casting mold
I never see memory temps over 60c or so
on my card
helps oc and longevity, some vendors are just too cheap to cool stuff
the temps are ofc under full load
Ye ofc
let me see if I can find more specifics about the test conditions.
i cannot get the memory on my 5090 to 90c because i cannot turn the fan below 30%, and that's like 70c lol
I am not sure what test they using for full load to test VRAM temperatures on GPU.
Kinda weird what you say.
Because the testers can't get it below 70. π€£
Must be a 9070 XT thing.
yeah they mostly settle for good enough on midrange cards
Might explain price a bit π But honestly, I can live with average for average price. But that's opinion/subjective.
AMD also for most part often uses older tech memory at higher voltages to get higher frequency for that tech, compared to Nvidia jumping to next VRAM tech instead.
gddr6x was hot π π
6 not so much, but hotter than 7
Atm i have gpu temp 43c, memory 52c, ~3100mhz core, 2833 mem (2176 GB/s). Low fans. But it is not at 100% load, it's CPU bound WoW with LSFG adaptive on
That 4080 Super is GDDR6X at 1438MHz.
9070 XT is GDDR6 at 2518MHz.
According to TechPowerUp.
23 Gbps effective vs. 20.1 Gbps effective (per pin?)
6x is pam4
I forgot what 7 is
I might be getting mixed up on which ran super hot also. Maybe it was 5x
What gigabyte did for those mem temps btw, they have a big copper plate for the GPU core, and pads for the memory to contact it too.
A lot of cards with high mem temps have seperate memory cooling, or little cooling for memory at all
it doesn't pull that much power overall and isn't that power dense either, so if you attach it to cooling it typically runs cold.
Not that big a deal though. Just don't get one that runs at 110c π
and does not have 12v2x6 for the love of god
Same with thermal putty for that Gigabyte card.
sapphire makes some 12v2x6 9070 / 9070xts
and someone posts a melted one every week or two
Ah, sorry, that Nvidia thing has thermal chamber under both.
vapor chamber yeah i think
What was the name for the giant heatpipe thing.
That Gigabyte has the classic thing with separate GPU die contact patch, and then surround part for the memory, with thermal putty, and that surround part contacts the heatpipes.
sapphire melts? Wait whaat?
Usually it is vapor chamber, they spread heat very well for heatpipes
As in some Sapphire cards have the new power connector.
And those have had the same melting problems as with Nvidia cards, just in pretty small numbers because the lower max power.
The pulse has 2x 8-pins (6+2)
and nobody who follows tech will buy the 12v2x6 variant when there is a pci-e 8-pin version next to it
Why do they say 6 + 2? π€
that's just because some pci-e connectors are 6+2 so they can plug into 6 or 8
Well, most PSUs have that type AFAIK.
There were some with actual 8-pin at some point, but then those went back away and everyone returned to 6+2.
Haha, I remember this. The first time I though I broke my PCIE π€£
2 years ago I posted something like "Euh, my 8 pin turned into a 6 and 2 pin, I broke it."
yea i hate 6+2 connectors
I don't get it how to put them back together π€£
I fiddled with it, but whatever I do, I can't connect them together again.
They don't connect together, the 6 just holds the 2 into the plug.
In some PSUs they have been type where you can separate the parts by cracking some plastic, but never reattach.
In most they are never actually connected.
With the Seasonic Focus, they should never have been connected in any way, except by tiny amount of friction.
Just those two small stubs on the +2 part that go under the lip of the 6-pin part to keep them in while both are inserted.
You have to slide them but its such a weak system that it nerver holds
and every time you go to plug them in they seperate
so you cant plug it in cause the slots dont line up so its to big to fit
and its just a nerver ending pain
and it also looks absolutely terrible if you only plug in the 6
Shitty design.
No no no, they fit together.
Because I have two of them. One cable I accidently seperate. And the other ones are still attached but losely.
So there is a way to attach them back together.
and that's from seasonic focus.
the other cable from seasonic focus, yeah it seperated and can't reattach. But there should be a system/logic in it.
Ah, that is older Seasonic cable type.
Just clipped together with that 2-pin side going over bump on the 6pin side for the whole length.
I do agree with the rest, kinda poopy design π
4 trillion dollars and they can't be arsed to have a decent cable connector
right about now would be a hilarious time for china to break out their bootleg specialists and dump a bunch of 50 series knockoffs with 8 pin connectors
They need a new (or at least expanded) PCB to do it properly
what can be done with no additional hardware on the board is some cable as thick as a couple of fingers just wired right up to the 12v π
Some partner card PCBs already have the pads for 8-pins, but Nvidia didn't let them to be used in the end.
Btw. Hungriest game i've ever seen (Satisfactory)
I'm testing with desynced fclk to push the available memory bandwidth up higher, at the cost of a slight bit of latency, and it's loving it
It's also made 2ccx perform better than 1ccx
My total OC gain is literally ~+50% here, in game FPS. Just locking to the vcache CCD it's +41%, but the interconnect OC makes 2CCX outperform 1CCX and that does +50%.
OC is not dead β¨
I think it's me, but in general, the connections to the motherboard all feel cheap.
I pulled off plastic caps from my motherboard while trying to pull cable.
I think all QA goes into the testing the actual hardware performance (heat, voltages, etc) but they cheap out on production quality of board itself.
It's really expensive to make those more quality, and they land in a weird spot
They're designed to be user-serviceable but those are the same connection standards that are used internally on many many other devices
And for something that will only be connected/disconnected once or twice throughout its entire lifespan it would drive prices infeasibly high to do a higher-quality connection
Yeah, I am probably spoiled or something. But if the connectors would be metal or something it would add a lot.
But that's probably costly.
It is mainly the USB3.0 dual-front panel connector that has problems with the plastic surround coming off with the cable connector.
Like when I pull out a cable, I am sometimes afraid of bending the board, they are so rough to pull out. It's weird.
There must be a solution.
Yeahh
I mean
there are solutions
But it being hard to disconnect means that it'll stay connected reliably without having to build some sort of latching retention mechanism like they use in automotive
Take the jack of a guitar for example, not the best example, but if you look at it, you pull/plug it in with ease. Even ethernet jacks or HDMI connects easier.
Yep
and connecting to the motherboard isn't much of an issue. It's mostly disconnecting stuff that worries me.
But again
Those are all user-facing connections and orders of magnitude more expensive than internal PC headers
I don't think I follow the distinguish between user facing and motherboard connections, which are to me also user facing.
Who else is going to connect the GPU to the motherboard? The user.
The OEM
In what reality?
Self-built is very small market compared to prebuilt.
What are you talking about?
OEM/Prebuilt pcs outnumber DIY builds many times over
So, if I buy 9070 XT, I need to ask AMD to connect it?
Sorry, clarification:
Said OEM, meant system integrator
And it is the large OEM companies that decide the connector specs.
Ah, ok.
Still, they know there is self built market.
And it doesn't seen that small to me anymore.
Gaming been on the rise. Building PC's too. Or maybe I just got more involved in it and it looks that way π
They do
But they are competing on cost
Would you pay 20-50 dollars extra for exact same mobo with a slightly nicer to use atx 24 pin?
And more for your power supply?
yes. If I don't pull of my caps. Or pull apart my motherboard.
And also then have mobo/psu that are not compatible with any existing devices
And I can just click and pull out cables with ease.
100% I would pay an extra 50 for that.
Per component?
Because especially for your mobo and psu it really could be that much
Now I am fiddling with ... what do you call it again... tweezers? ... inside the case π
What on earth do you need tweezers for?
Oh
Pliers
Those are not tweezers.
So, I would rather have a more efficient connecting/disconnecting system for motherboards, yes. And more quality. It would also help with people not connecting shit properly if they did a redesign.
Now they add colors, if you see yelllow, it means you didn't plug it in fully etc.
Yeah, it might
I do understand your argument here but my personal preference would be opposite of that
I am so sorry I don't know the full english dictionary.
it's not the matter of not knowing english. Tweezers are very different tools.
Check out my Reaction Supercut of Linus Tech Tips' Linux Daily Driver Challenge:
https://www.youtube.com/watch?v=up2Za7luucU
I don't know much about making computers, but I sure learned a lot when The Verge tried to make one. Tells you a lot about the credibility of these "tech reporters", which, in case you didn't already know, are run by Vox...
I'm sure your language has different words for tweezers or pliers
Very different?
It has to legs. You clamp shit with it.
One is used in construction. The other for euhm... beaty and stuff?
Way different force amounts.
yeah, but I am not that off the mark.
i wouldn't call it clamping either. clamps are meant to tighten down on something and clamp against it
It's not that I said hammer. π€£
Tweezers don't give the leverage that pliers can with the hinge
It's the difference between a carpentry hammer and a ballpeen hammer. Similar form factors for very different functions
I fully understand the added costs though. But I just thinkj it could add so much benefits.
And like I said, HDMI connection feels amazing. It just plugs in and out.
They even add some idiot proof stuff to the HDMI. It's a neat design.
I know, user facing, needs to be idiot proof. But one can dream π€£
I do understand why you want that, and I do think it's good to make a hobby more accessible
But I don't want everyone in the hobby to pay for the inexperience of others that are trying to do a technical project but aren't familiar with the connections/standards used in the space
and then you got display connectors. Which I like less. Because of the lock again.
Inexperience? Rofl.
Half of them burned down their GPU's 
Might have been preventable if they had better connection system.
and if it was metal connectors, it wouldn't melt either.
Not possible to have full metal connectors?
Need insulators, most which melt at low temperatures.
Unless you go to ceramics or like, which in turn are brittle.
Not fully, but atleast something like HDMI.
Oh
So nothing plastic that enters the "female."
The metal part is the male that goes into the female.
There is just supporting ground metal around the insulator, that then has the pins in it.
The black plastic stays OUT of the GPU.
That's not true
Pcie connectors, the plastic goes inside the female.
Look at the inside of the ground wrap
See the black plastic inside the ground.
That then has gold pins in it.
It's very impractical to develop a multi-pin connector at small scale without plastic support
Yeah, I know, but you seen how much plastic the 6+2 has?
Also, plastic on PCIE is outside, there it's inside.
That isn't really impactful to the reliability of the connector though.
Look at how much plastic it is! The different is glaring to me.
Plastic marked with red outlines:
All I know is, USB, HDMI, display port, guitar jack, everything that has metal shell. Connects easy. And disconnects easy.
All the PCIE cables with plastic shell, and mostly plastic, are glued to the board.
Because higher gauge wire-to-connectors basically require something like that for insulation.
Or you have major short circuit risks.
I never had to use my pliers to disconnect my usb C 
Again, that's partly because PCIe is designed to be very resistant to accidental disconnect
And one large unified ground like that wouldn't make nearly as much sense for a power-only connector
Fuck. Nothing we can do about it? Or engineers can't come up with solution?
I think that most enginners aren't looking for a solution because they don't believe one is needed
The issue with 24pin ATX power is pin-creep over time, and cases where the MB has connector at the smaller end of the permitted tolerance, and the PSU connector is at the larger end of the tolerance.
Wait untill they meet me.
I'll yap their ears off π€£
Even this thing looks easier to work with.
And is pretty expensive implementation.
And very easy to break at the edge of the PCB.
Also not high-current.
Also, why trhe fuck wouldn't engineers think there is a problem?
We had a year or two with these pictures:
I know they blame it on us. "Not connecting properly".
Rated for half of the power of an 8 pin, to be precise
But in the end, it's a design fault π€·ββοΈ
Maybe some are fake. But it's pretty brutal.
As I said, 12v 2x6 is a bad connector design(without temperature sensing and balancing)
8 pin is quite a bit more debatable on that front
First pic is 8 though.
Yeah, I saw that
So 5 pins of +12V for 66W total, 1.1A per pin?, and 4 pins of 3.3V for the rest. (For PCIe)
And for comparison, 12v 2x6 is 8.3A per pin
iv had to do it for rgb and fan headers since they tend to be hard to reach and pull on but usb-c is a first
and my hands get sweaty, and it's such small pieces to grab π€£
and you can't just "yoink" the cable to pull it out because then you pull out the wires itself probably.
i would suggest getting the smaller ellectrician ones however
cause those will tear plastic to shreds
I found that out.
When trying to disconnect the new PSU (which I got a refund from) from my GPU.
yea those really love denting stuff
But hey, I got my money back π€£ The cables connectors are a tiny bit scratched.
but with those 6+2 you pull out the 6, and the 2 remains stuck and stuff haha
get one of the pointy ones
maybe I should use tweezers π€£
you can be a bit more precise with them
you know... it could work if there large enough i bet
but i doubt it would be a good experience
On the other hand, I am getting better in it.
Disconnected so much hardware and reconnected the past few days.
Removed GPU again today and it went pretty "ok" to disconnect PCIE. Just had to wiggle wiggle wiggle a bit.
and make sure I don't rip GPU out of motherboard.
Is it better to take off GPU and then disconnect PCIE?
i had the hardest time plugging in the 24 pin more than anything
i disconect 12VHWPR first personally
Yeah, I am not sure. Maybe I am doing it wrong π€£
I should rewatch the verge guide. they know for sure what to do first.
NZXT really dosent want you having an outwards facing 24 pin (Also yes i plugged in the cable all the way after)
honestly if i could i would just leave the cover off but its not removable
no?
bold of you to assume i go places
the only place to go on the east coast is the sea
I will never understand that sort of mentality
the last thing youl find is valleys
?
The "nowhere to go, nothing to do" mentality
Nah, but maybe you were like "Oh we have something like the grand canyon called x" π
i live in a small town thats falling apart split in half by a collapsed bridge
and this is NA so walking in the pouring rain probably getting yelled at is the best you get when you cant drive
Yeahh
I grew up in a small rural town that was too small to even have a stoplight, so I have a hard time understanding how when people say that they don't have anywhere to go
i mean i could go get food if i walked and actually cared enough and also had money but i only have cash and everything takes card now
I shouldn't have asked π
i prefere it than living in the city but having to go elsewhere for everything you need makes days quite slow or uneventfull
random landscape from canada rendered in UE5 π
It's east of stein valley park.
Ignore the fog, quickly put it together as practice. I can reduce the fog. It has to do with my camera.
The fog is thick I see
Well it is canada π€£
id say thick fog is pretty acurate for canada yea
i had this one screenshot of the windows sign in screen that summed it up pretty well let me see if i can find it again
Yeah, I don't have a weather system yet π€£
"feels cold" is bold for -20Β°C
"humid" is funny - what humidity?
the one all the dads across the country seem to magically see π€£
I saw people post some screenshots of the weather where they lived, and I noticed that most apps now have "feels like X" noted.
What does that even mean. It's 20 degree, but it feels like 15 ... WHAT?
wind makes temps FEEL very diffrent
Yeah, but I was fine with just 20 degree and windy.
for example rn
10C sunny without wind feels a lot warmer than 15C thick clouds strong winds
feel like only really matters in sub zero
you dont really notice it since in the heat the wind feels nice
but a cold wind in a cold air does not in fact feel nice
but whatever is measuring that 15C is also impacted by clouds and/or wind, no?
If I have a thermostat and I blow cold wind on it, wont it lower temperature?
yeah, 35C with wind is a lot more pleasant than 32C with no wind whatsoever
probably not much unless its pretty strong and consistent
not if the wind is the same temperature as the air around the thermometer (which it usually is)
but we speaking equipment of weather stations π
if you blow on a thermometer when it's cold outside, its temp will actually go up because your breath is warmer than the environment
If the clouds block the sun. It must also impact the weather stations. And so does wind afaik.
We place these things around the country as a weather station.
but the "chilling" effect you experience from wind is only partly physics and primarily just biology
your body maintains a film of air around the skin which insulates you against the environment.
if there's wind, that film of air is blown away and there is no more isolation
Makes sense.
isent that why your nose is always cold right?
but on a thermometer, if there is no localized heat difference, it will measure the same with and without wind
since your breathing does the same thing
oh, that's why hair raises on the arm when you get a chill.
to enlarge the film of air. Now I remember.
but if there is a localized heat difference (like a CPU producing heat), then introducing airflow will cause that difference to equalize
yup, exactly
where is this taken at?
Ok, that's a good explenation.
bleekveld.
sunlight introduces heat
wind introduces airflow
both of which are intended effects that you intend to monitor with such a station
it's why outside temps are always given as shadow temps - to eliminate the inconsistent heat input from the sun
No, I understand now why the "feels like X" has value.
honestly, I never look at the weather so I never noticed it π
I just go outside and feel, go back inside to adjust if needed
If I go on a long daytrip, I do check weather. Especially hiking.
thats crazy
i have a downright annoying amount of specialized pliers & tweezers
i think my next ones will be refurbed hemostats
actually, check out the tool at 3:45
Adam's stand on storage drawers is that they're where tools go to get lost. To keep his essential workshop tools easily accessible, he built these custom stands as a solution for organizing all of his hand tools. Adam explains the design of these stands and shows off some of the more esoteric tools they keep within reach.
Shot and Edited by Jo...
orthoscopic stapler
ok dumb question time
with how 6000MT/s is the "sweet spot" for the X3D chips, is 9000 going to be the next "sweet spot" up?
seems we almost double with each new generation so DDR6 will probably be 10K+ yea
That's probably less likely than you'd think
we're assuming that we'll have any access to DDR6
rn DDR5 and 4 is apparently a rarity
6000 is just the "this always works for automatic overclocking" point. It's not the fastest.
It's likely that next gen will see the best performance from DDR5-9000+
oh i know it's not the fastest
just the current timing sweet spot
wondering if +50% is the next tier up for that, or if it has to jump all the way up to double
There's no one particular point
DDR5 9000 is done with the memory controller between current and next gen
and closer to 10,000 even on intel's CPU which supports CUDIMM DDR5 (as next gen imc should)
as for what bandwidth scaling looks like, that depends mostly on the details of the new interconnect that they're using
8000's with one-click auto OC should be very achievable
raphael / granite ridge are happy at 7600 with ancient IMC, the reason they do 6000 is just because of interconnect bottlenecks, the 7600 is not faster than 6000.
6000 is the highest frequency for 1:1 operation on AM5 before it becomes too unstable
Or rather, before instability becomes too likely to be worth risking
For automatic overclocks on a 2022 BIOS, yes.
Yeah, xmp/expo operation.
If you're going manual, it doesn't really matter much what speed your kit's profile has, so I assumed auto-OC
Btw, since you mentioned 2022 BIOS specifically, has that changed by now?
Work on AGESA has raised what speeds for FCLK, MCLK and UCLK are attainable somewhat easily.
Sooo... are 6400 profiles low enough risk now to be recommendable for 1:1?
For the "set and forget" crowd
no, but uclk=memclk doesn't really matter
not nearly as much as uclk=fclk. uclk=fclk is what people were talking about with the "1:1" stuff on zen 1, 2 and 3, and it kinda got co-opted wrongly here.
so you can one-click anything up to 7600 (or 8000 on some boards)
and manual ocers would use 6200-6600 or 7800-8400 depending on the memory config and the cpu sample (they can be good at one or the other or both)
Point was mainly what RAM speed is possible get stable with just setting the RAM profile on, on 99.9+ of the CPUs.
Actually stable, and not near stable.
That's 7600
On 1:2, but what is the same on 1:1?
Because that 7600 needs more VDIMM.
Technically higher freq needs more VDDIO, but we're talking like needing 1.3 VDDIO instead of 1.15 here when it's standard for EXPO to do 1.45 anyway.
3800:1900:1900 is actually generally easier to hit and less stressful on the CPU than 3000:3000:2000, while also being faster. For example 1.05 vsoc will work every time, instead of 1.2 - 1.3 that is pushed for 3000:2000:2000 expo.
But it wasn't even possible at launch, because the microcode to support it didn't exist yet.
it is just a little complicated and AMD doesn't wanna update their whole marketing sheets and techtuber presentation for relatively small improvements
people already get way too confused with the info out there as it is
And as a plus, that would mean no need to manually switch to 1:1, when on default it should go to 1:2 when over 6000 MT/s.
Then the question becomes:
If 3800:1900:1900 outperforms 3000:3000:2000, where's the threshold before which 3:3:2 performs better?
cuz if 3200:1600:1600 is outperformed, no reason to recommend 6400 kits over 6000 ones
And if a 7200cl34 kit is the same price as a 6000cl30 one, would that be a valid pick?
oh hi Aeryn
btw it's kinda weird how bclk oc behaves nonlinearly with 9000 max boost
it's like reliably /2 or something
so pm is doing some weird calculations based on base frequency
likely is kind of calibrating each start over ocl
R.I.P. Vibe coders https://youtube.com/shorts/vTIdcI5l-jw
has anyone come across steam showing games as still running? after checking task manager they are not yet the resources on the GPU is still being taken up? (happening on 3 games now)
Long, long, loooong ago.
Sometimes after closing a game played through steam itll still say its launched and im playing for up to ten minutes. I have no clue why this is a thing. Although it is more common on my intel/nvdia Laptop then on my amd desktop
kinda reminds me of how SF will refuse to sync the save, so you can still play, but your ghost is just sitting there
it's a pain