#off-topic-tech
1 messages · Page 41 of 1
i still feel like programmers create acronyms as dirty as possible while still officially publishable. because that looks too much like FUCK to be a coincidence
so they could have gone with FACK
clock as in frequency signal.
oh i know
But is there a way to know if it's CPU or Mobo? 😒
or only solution: replacing them.
but it's like the Proceedings of the National Academy of Sciences. PNAS
I'll still do it for bulk storage... it's pretty hard to beat the ability to buy an 8tb drive for under $100
By trying to get CPU RMA replacement?
The price gap is closing though, gen 4 drives are getting very affordable at 4tb
for the moment, but i think by the time these drives start having issues, solid state storage capacity will be big & cheap enough
Ask the CPU seller first.
And if that doesn't work, then I can help you try to get direct RMA with AMD like I did.
yea, my gen 3 4TB drives were like $140 each
I think there's potential for another big breakthrough in data density on hdds, but I honestly don't see them dropping in price too much further than they already have
4/5 years? imagine 16tb drives for that price
At the very least I'd expect 8tb nvme drives to cross below the 200 dollar mark within 4-6 years
at least for HDD, yea. just the base manufacturing of the parts is probably $50
I'd wager it's significantly lower than that, but they have to make margins somewhere still
i wonder if shipping cost is factored in
mhm ... then I don't have a PC. I have to think about this.
Another path is: I was interested in the 7800X3D, wait for a sale and buy it. Then replace CPU.
AM4 came out in september 2016. i'm starting to feel old
over 7 years later & AMD is still putting out new cpus for it
Yeah, that's actually pretty cool
The even more impressive part is that the 5800x3d is still in the top 5 gaming cpus
i think that's proof that fat cache is probably the future
Just for bad code, like most games are.
Oh absolutely, especially considering how cost-effective it is
Not really, SRAM hasn't gone down in size much in last process shrinks.
So large cache takes humongous amounts of die area.
To be fair, games are by far the most complex applications that most people interact with on a regular basis and development times would skyrocket if you expected them to write perfectly clean & optimised code
relative to spending the time to not have crap code, i bet it's cheaper 🤣
It's stacked on top of the die though, that's the entire point of amd's approch
the actual size of the chip doesn't have to grow significantly
And that's only true to an extent
IMO that mindset tends to come from senior devs that have been working at the same software company since they started as a junior working in cobol
Mainly meant the lack of intentional cache/memory management.
To keep the hot data in the CPU cache intentionally by using certain access patterns etc.
Oh yeah, that's entirely valid
Although I expect with something like satisfactory it's simply not feasible to keep all the important stuff in 15mb of l3
google should add to its google images, an ai generation tool where anyone can just plug in phrases to generate images 😄
Just saw an ad from a small local ISP where they mentioned a speed of 1000 Mb/s in the title and then 1000 MB/s in the description
As an ISP, you should know the difference between Mb/s and MB/s....
ISPs yes, but the ad people? very rare.
But their marketing team didn't
My isp accidentally labeled 1 gigabit service as 1GB/s
Afaik they resolved it pretty quickly but it was still funny
There's absolutely no way they were offering 10Gbit service for under 100/month
Where can I find this ISP so I can download all the games that exist on steam servers? 
They're pretty small and actually do provide excellent service now
They used to be terrible but they dumped a massive amount of money into running all of the fiber for the entire couty(except one mainline from a major telecom) and they no offer consistent symmetrical gig to the home for $80/month
Which is a masive improvement considering that before they did that the fastest plan avaliable from any isp was 22mbps
(and it was the same price too)
They only serve a few thousand people, but it's fast & reliable ftth
That's pretty nice. I am currently at 500 Mb symmetrical for that same price.
Once there is more fiber competition here, the pricing should go down.
There are 2 ISP's who want to cover my city with fiber here (I work for one of them lol)
My colleagues from the acquisition are trying to obtain as much as consent from apartment owners as possible so we can cover the apartment buildings (you need at least consent from 50% of the apartment owners)
I hope this is not one of those discoveries where they hype up everything and you later completely forget about the discovery.... https://www.reddit.com/r/Futurology/comments/192tkrg/new_material_found_by_ai_could_reduce_lithium_use/
So we are now at the point that anything ML-based is now AI?
Yes
They even call excel formulas AI these days 
gotta love simplicity
"Wait, you use programming in Excel? Gotta be AI!"
Wish I knew VBA in excel. Mostly, I just extract data from SAP and then use Pivot table to get the info I need
VBA would be very useful in some cases
Careful what you wish for. VBA is a pain in the***
I created a template in Excel VBA that I could paste an email into, It would pull the data into a report and after entering a few fields would send an email. I guess that is AI.
so have you noticed the 'reaction emoji' that discord randomly switches too when you hover of it ... is never actually in the list you can select yourself? they are all different ...
like where is this one ?!?!
Party breaker 
That one looks like it might be from a different server
you know, i wouldnt even call excel formulas ML even
i checked each server, and none of the ones it randomly flips too is on any of them... its so weird
its like discord came up with different versions of the basics... just for the purpose of trolling you when you hover over the lower right corner 😄
Not sure why discord implemented that. Doesn't really make any difference imo
And I just learned about that feature from you lol
right, you cant even say "go ahead and use the random one you flipped too"
I see what you mean now. That is dumb...
glad im not the only one!
I bet nobody even noticed the "feature" (including myself)
haha
ive noticed it for a long time, but i never bothered to "notice" until now when the emojis it was using wernt even in the damned list to choose from
lets hope it doesnt mean they plan on changing the default base emojis to those ugly ones its now flipping randomly too!!!
Speaking of AI, here is a music video with some weird AI-generated video https://www.youtube.com/watch?v=sfa4fjXc_10 
Streaming: https://dora.lnk.to/OneDay
Autor glazbe i teksta: Nemanja Filipović
Autor aranžmana: Nemanja Filipović
Video: Kelly Boesch
Post produkcija: Paolo Gentilini
2024 MISHA REPUBLIC
#MISHA #OneDay #Dora2024
You are mine
But you were always so hard to find
Like true love is
In disguise
And heaven knows I tried
Love takes you
Baby up to...
i mean if you want trippy shit... ai is def the way to go
and with just how much 'content' is out there these days... ai can easily have a heyday melding stuff up
And I thought my dreams were weird 😅
coulda done without the text .... kinda ruined it
Yeah
my dreams are never subtitled 😉
I wonder what would happen if AI created videos from people's dreams
Because nothing in my dreams make any sense lol
everything makes sense in a dream until you realize you are in a dream, then you start poking holes in all the flaws
then you wake up cause its like "for fucks sake already"
My worst moment is when you think you have woken up from the dream and then you realize you are probably still dreaming because you saw something supernatural
and it just keeps going
i usually end up trying to take control of things before i wake up ....
I just scream at monsters and hyperventilate
lol
=p
Theoretical limit is 100 meters
Worst is when some modem arrives with cat5 (not cat5e) cable and technician takes the extra cable and gives it to a customer who has a gigabit modem
I have Asus RT-AC85P
It was the only asus router in the store and decided to pick it up
The discord custom ones look creepy AF. Almost worse than MS teams version.

so when a spec like thunderbolt says it can handle 240W, does that mean it can do it 100% of the time?
& i assume that it can either do 240W or data, not both 240W AND data at the same time
That would seem to defeat the purpose of some of the docks I recall reading about? I thought I recall 140W TB docks that power the laptop and add connectivity through one cable, but I could be misremembering.
that's what i was thinking, feels like it'd be too much for a skinny cable like that
ok if they don't call this thing the Borzoi i'm gonna be really mad
anyone know how to disable a iGPU
starting to get really anoyed of it running as a second GPU even tho theres nothing plugged into it and stutering the living hell out of everything
It shouldn't be stuttering anything
prettymuch SLI microstuters
its being used as if its a second dedicated GPU when its not
point is i want to know if its posible to disable it so it stops doing that...
You're not plugged into the motherboard display output are you?
Then in that case it's not possible that it's behaving as you're describing
The most likely case is that Windows is offloading hardware-accelerated tasks to the igpu instead of the 4070ti while gaming... which can only help
well its not its also spiking cpu idle load like hell
You're blaming irregular CPU load on your igpu?
il send you a picture of the back io because you probably dont believe me...
and im blaming ram on that cpu load but il get to that in a second
thats the back io
now about the stuters im not denying it could be ram instability but its odd that everytime i notice it not only is the IGPU on but my CPU is in the shit
like 50% all core load kinda in the shit on a 13700K
I honestly have no clue what that is intended to mean
yea i see that...
Can you explain with less slang?
to put it in the simplest of terms
every single time that i see games stutering
the IGPU is running for no reasson
my ram looses all stability
and my cpu spikes from like 4 cores at max to all cores at around 50% ussage or more
my theory is that the IGPU is causing extreme stability issues
and i was asking if its posible to disable it
Highly unlikely
and if so how
Correlation is not causation
well it shouldent be running at all so its a good start point?
I just explained why it would be used
but i dont want it to be used because it causes stability issues
But utilization of your gpu shouldn't directly impact CPU usage and certainly can't cause memory instability
When you say stability issues, what do you mean?
mostly stutering and every single thing i have open turning into a slideshow like chrome or firefox or reallly any browser even task manager whenever i had it on my side screen
Earlier you mentioned microstutters, is that just something you feel or are the 1% lows actually dropping?
probably actually droping
some of them are like good 1 to 2 second freezes even sometimes
Probably? Test it objectively
idk i dont have anything to test 1% lows
Rtss will do it, but actual freezes like that are definitely a problem
these two im guessing right?
Yup
guess i just missed them
would benchmark loads work or do i just have to play something until it starts dying again
probably game load cuz i havent seen stuters in stuff like 3D mark
That depends on weather it does it in benchmarks or not
How often does it happen?
If you're getting multi-second freezes the lows don't really matter though
fairly randomly for sprints at a time like right now its perfectly fine
but im testing a fairly low stress game to start off whit
Well, if you want to disable your igpu you should be able to look up instructions easily enough
seems ok whit games right now but as soon as i open a browser it just slows to a creep
or really anything thats ussing ram at a decent rate it seems
Ram or hardware acceleration?
whats the difrence
One is a method of offloading computation for an application to the gpu, one is a system component
ram then
Why do you say that?
you know, i put a tiny dot of silicon right in the groove, and its shut up completely
Almost certainly
I hate this so, so much https://arstechnica.com/tech-policy/2024/01/film-studios-demand-ip-addresses-of-people-who-discussed-piracy-on-reddit/
can't they do something like claim litigation abuse or frivolous lawsuits since this kind of thing has happened numerous times before & has been slapped down every time
"The studios argue that the information requests do not implicate the First Amendment and that the rulings around the two aforementioned subpoenas are not applicable because the new subpoena is only about IP address logs and not other user-identifying information."
"we're saying it's not about identifying them while clearly asking for identifying information"
asshats
either that or the lawyers are geniuses for getting big studios to pay them for useless work
LOL or could you imagine if the courts let it thru, then you'd have god knows how many other court cases cite it as precedent that IP addresses somehow don't count as identifying information, & whatever implications that has for their cases
Somewhere in the MB BIOS.
For my AM5 Asus board: Advanced mode, then Advanced/NB Configuration/IGFX Multi-Monitor.
My ASRock AM4 boards seem to not have it in the ASRock menus at all, but under AMD CBS which will work with any manufacturers AM4 or AM5 board.
On my old Asus Intel 4th gen MB it is under Advanced mode, then Advanced/System Agent Configuration/Graphics Configuration.
GeForce RTX 4090 sold without a GPU A HKEPC media reader fell victim to a scam that resulted in substantial financial loss. The buyer was deceived into purchasing a counterfeit RTX 4090 graphics card on a second-hand market. According to HKEPC’s coverage, a new scam has emerged in the second-hand market, resulting in one of […]
Because it sends some audio to the last two nonexistent speakers?
Separate sound cards (internal or external) are still a thing.
And even Creative Sound Blasters still exist, just that the Audigy branding is just for the crappiest models.
Because they are often software configurable now.
You plug something in and you get popup asking what you connected.
Yes, if you aren't using the side/back speakers in 5.1, the center one is almost useless.
i've been suspicious about product quality on just about everything that comes out of a factory, there's no way in hell i'm buying secondhand unless it's cheap, easily replaceable, & something i can physically verify in front of me before buying
Think I am going to look out for a 7800X3D (just in case my CPU is dying, I got backup)
Who knows. Bit in the dark here.
Just want to get a stable system is all. So we'll just keep throwing money at the build untill it's stable. 
Not much I can do to get RMA I think. It's such a niche case atm.
Also, not sure if this is tech. But holy moly look at this nose:
It's the new X-59, supersonic and silent. Manufactured by Lockhead Martin and NASA.
Silent as in: breaking sound barrier without bang.
👀
scroll up a bit 😛
With bang, but that it would dissapate to low enough volume before getting to ground level from the intended flight level.
yea, sounds like they're expecting it to be more of a soft thud than a crack
sorry 😦 wait what? I don't 👀 anything up I am blind.
I'm making a fire extinguisher robot as sort of an optional school project.
Can't really find a motor driver which could be controlled by raspberry pi 4 and output 18V 40A, it would be fun if the water pump could have variable speed control.
I guess just get a relay and just have on/off?
But that nose, it can spear someone. Also, the angle of the picture, it makes it look like it's going to tip over to the front 😂
Crazy shit
It looks imbalanced. I am 99% it isn't since it's NASA.
The nose is probably really lightweight
Yes I agree, and it's needed to break sound barier without the bang I assume.
I think it looks dope though. Futuristic. But not Cybertruck level futuristic.
Yup
Just me thinking from QA perspective, but do you need the water to be variable speed? Since the goal is to put out fires, I would assume you just want to spray as fast as possible to get the fire out.
Fire-Extinguishers don't have a way to adjust the pressure.
Do I need to? No
Would it be cool? Yeah
Lot of PWM speed controlled motor controllers around?
Fair 😄
I mean the budget is coming out of an university budget, not my wallet
Or any potentiometer controlled one could have the potentiometer replaced by the Pi with little extra work.
I kinda forgot how much we had, but it's a decent bit
We were thinking to control it with a PS4 controller which has pressure buttons
The programmer in the group has already gotten smth to work from my understanding
So find a controller thingie that fits the the motor you are going to use, and we can look more.
On how to control that with Pi.
not sure if this can help golden: www.instructables.com/Auto-tracking-Water-Blaster/
The water inlet is piped into a 12v water valve that is connected to a ¼” ID x 3/8” OD vinyl tube. That in turn is connected to a ¼” barbed tubing to ¾ slip fit PVC connector and glued to a ¾” PVC water cap with a 1/16” hole drilled for the water stream.
That one isn't controlling water pump, it is using mains pressure.
I can't find the controller thingie.
Raspberry pi compatible stuff seems to be quite low powered.
I mean a powerful enough relay is fairly cheap and easy, just no variable speed.
Just putting it out for inspiration. Thinking out of the box 😄
And just toggling that on or off with simple valve.
And then low power turning ones.
I meant find ANY thingie that can vary the speed of the wanted motor.
And then we see how to direct that with the Pi.
Not Pi specific thing, motor specific thing.
You mean ones which have a manual knob? Found plenty of those.
Yes, most can be trivially changed to work with Pi replacing the knob.
Motor just needs enough power, it's DC
That's awesome
I don't know which motor I'll use yet, dad will get me some drill motor(most are 18V, some 24V) and we already have the water pump bought that needs a motor mounted to it.
Can't wait to see the end result 😄
You don't need anything Pi specific.
Just anything that takes voltage or PWM input.
And most of the potentiometer ones are voltage input with 3.3V or 5.0V for the potentiometer.
Post the model first, when you find suitable one.
11th May
Btw, anyone got advise on Prime95? What to do and what not to do?
For what?
Don't run the "real" mode on possibly unstable computer?
Run the Torture Test?
What else?
I though that would be obvious. To make my PC crash so I don't have to wait for hours and hours in GW2.
Need a more reliable way to make my PC crash, before I can start fixing it. That's what I am thinking.
Put the torture test on on defaults as first thing?
Since it's probably the CPU.
the thing is i checked and there wasent a single thing other the The voltage control for it in the bios
its either named a completely random unrelated thing or not a option
E :)
@edgy hazel Finally managed to setup FTB Launcher. Just started using Xnet and it's awesome
image doesn't work the hell?
Oh i see...
Wtf is wrong with discord....
yes. Discord behave!
Might be one of those Image Screen stuff that just does dmg to ur computer idk
Im being carefull tbh XD
asablic wouldn't do that 🙂
Is there an image hosting site that works with discord here?
and if they do, I am going to develop their app. And we all know how that ends.
Ok, there is seriously something wrong with discord
Not sure if I uploaded png before on discord
Just ran Prime95 torture test for 15 minuts, no crash and no errors.
It must has been .Gpej u should have used @visual tree idk
That's normal. For a flacky system, it should hit issues around 10-15 I've read.
But for Prime95 to find errors, it's up to 24 hours.
Can someone try uploading a random png image? Not sure if it's discord tbh
Yep, discord is definitely broken....
That's from their official documentation.
Oh, RIP discord image uploading.
Q) How long should I run the torture test?
A) I recommend running it for somewhere between 6 and 24 hours. The program has been known to fail only after several hours and in some cases several weeks of operation. In most cases though, it will fail within a few minutes on a flaky machine. When overclocking it is entirely feasible to run short 10-15min tests at each increase in clock speed to quickly assess the feasibility of running at those speeds, then run longer tests later.
And I can categorise my machine as flaky at this point 
rofl, discord url is blocked on discord.
is it me, or are the emoticons also broken?
Test 🤔
jacelul shows up as a weird emoticon
I can see everything normally
also jacelul?
Yes
weird, it shows up broken here. I would send a screenshot but ...
It's probably something wrong with discord. Hope they will fix the images and emoticons soon
they already know. check url I posted.
Investigating - We are currently investigating an issue that is preventing embedded media from loading.
Jan 13, 2024 - 06:18 PST
Oh
I think that jacelul emoticon is uploaded on a server that's not reachable for certain people.
Not for me and gs.
Yeah, it's their Media Proxy: "Service responsible for resizing images for thumbnail display".
it's been down for 25 minuts by now. Also, it's been 4 years since this proxy has been down.
last time was december 4th 2020 for 7 minuts.
I just think it's cool that all this data is available for us 🙂
What MB?
we only have two ISP providers dominating the market. Telenet and Proximus.
Because I have not seen single MB with IGP support that didn't have the option, even OEM boards.
gigabyte z790 UD AC
*hides in the corner
Settings, IO Ports, Internal Graphics: Disabled
Below "Initial Display Output" setting, that needs to be PCIe 1 Slot
but thats not in the bios is it?
il look again
And the current Discord issues striking.
So enter the BIOS, then to Settings tab, then IO Ports should be second from top, between Platform Power and Miscellaneous.
And then that Internal Graphics should be second option after you open that menu.
Only for businesses afaik

It is 😅
At least, most of the hard work is done and I am less stressed now lol
Private users can get up to 1 Gbit/s but I don't know about the business clients
Larger clients can get dedicated fiber but that's probably expensive
found it guess i just missed it or probably dident check the setings section
maybe discord shouldn't have done that lay-off.
I wouldn't be surprised if it's because of employees being fired
it has been days for me since discord breaks
Like usual :(
before this issue occured nothing was wrong
everything would be smooth sailing
prob not? they store alot more than just 2trillions messages... in total (Dont take it seriously)
Imagine that 😂
issues should be solved.
Who?
I wonder if discord regrets declining Microsoft offer by now.
Microsoft would have just bought up discord and still add all the features we need to pay for.
Not because they need it (they should be financial healthy) but just because they can.
I don't want Discord to fail with this project. But I always get annoyed with tactics to open my wallet, feels "too agressive". This is highly subjective.
lol 😄
Saw you trying to post discordstatus.com, whitelisted it now. 😄
Thank you 😄
We'll need that more often in the future, right lol
You know it's funny, they said they implemented a fix and are monitoring the result.
Either the fix is still rolling out. Or something is still wonky. Since my latest image is still "loading".
Oh, if I click the image, it shows my screenshot. It just can't embed it in Discored. Let me restart client. Update: restarting client did the trick.
They should make a special channel for you Maxeek 😉
"Maxeek's deep thoughts channel"
Did you order a new GPU?
ah thanks for reminding me.
No no, I needed to look up if there is a place where I can give me old hardware and get money for it.
Recycle Parks = we need to pay to deliver our old hardware. Then they recycle the materials and sell those. So they are double dipping and I don't like that much.
Any electronics store needs to take old electronics for free here, and deliver to right recycling.
Or all the larger waste yards have containers for that too.
That's better system imo.
I don't mind it, otherwise recycled materials would sell at a much higher price and then not be used(defeating the purpose of recycling)
There are small companies in Belgium, they come pick up old electronics and pay you for the value of the old electronics. So I know it's possible.
Then they are probably
- supported by public funds in some way
- upcycling instead of recycling
- reselling at a higher price or to different markets
For example, I pay to recycle my car tires
There are places that take them for free, but they're usually subsidized
Everything 2.0 wasnt possible 🤣
what's the difference between upcycling and recycling?
Reduce, Reuse, Recycle.
Usually we jump straight to the last.
Upcycle is the second
In here if you see something you would like in the wastepile, you cannot take it and try if it still works...
Find a second use for the hardware without stripping it down to raw materials
We technically cannot either, although I have done it in the past and ended up with functional stuff
How is reusing -> not the same as recycling? 😒
Did it in past.
Now the employees at the various places have been taught to prevent that.
Actively.
I always thought recycling was: reusing materials. For example, a plastic bottle. Make it back into basic plastic and make something else. Like plastic fork or someething.
Reusing the item as is.
The same what I though urban mining was doing with hardwre. Get the metals out of it. Smelt them. Sell them. To me this is still recycling. But I might be wrong here.
Reduce: make is so that the item never needs to be manufactured.
Reuse: after the first owner discards the item, find another user for it.
Recycle: Break down the item and reuse the materials if possible.
Ok, I understand.
It just feels weird, that I have to pay to give away something that still has value.
And, wouldn't it motivate more humans to recycle if they get money for it? Money is a powerfull motivator sometimes.
The Finnish system actively reduces/prevents the second step happening.
And jumps direct to third.
To be precise, I am talking mostly big hardware here. Monitors, old PC's, projectors, speakers, ...
We are very big on reusing. Every village has at least one "kringloop" in it's vicinity. That's where you give away stuff that's being sold again.
Inside it looks like this:
a big flea market 😄
Like goodwill in the US
Yes, exactly
its actually a great way to find work clothes (stuff you would otherwise kind of mess up with paint/dirt/grease)
no point paying $40 for a long sleeve shirt you are going to work on the car with ... just grab something from goodwill/flea for $2
i hate a frend who swappend my m and n key
what's really fun is setting someone's keyboard to dvorak
What might be more sinister is something close to QWERTY like AZERTY. Something that's just slightly off
That wouldn't matter if you touch-typed
i'm just glad my mom made me attend typewriting classes when i was in middle school. thought it was too girly of a class to take at the time, but good god it's paid for itself so many times over
naw, switch "C" and "V" on a programmers keyboard, and watch the flames ensue
i did hear about one that's just satan, & it's that apparently there's some kind of (maybe?) cyrillic punctuation that's very similar to ; but different enough that it has its own dedicated code
Oh no
I would cry
I think you might mean Greek Question Mark? U+003B ; SEMICOLON – inherited from ASCII
U+037E ; GREEK QUESTION MARK
woah! reminds me about a story a friend told me many years ago:
a colleague of him made a small USB gadget that did nothing other than every now and then replaced some characters with either the very opposite or similar things
< instead of >
, ;
" '
or switched z with y for a few keystrokes
drove another coding colleague a bit crazy and on a swearing crusade
after a few wasted hours of buy fixing he was told what was going on...
short period of furious anger, but nothing a few beers couldn't fix.
since then it's kinda a running gad if someone made a few typos to 'look at your keyboard cable' or 'your keyboard needs some fixing' and stuff like that
Last week I learned about the alt shortcuts.
For example: press alt -> type in 0220 -> release alt = Ü
press alt -> type in 0196 -> release alt = Ä
A bit of an outdated system (who is going to remember all these codes?), but can be handy for niche symbols.
yea that's it 🤣 couldn't remember which language it applied to
& i've been tempted more than once to get a ridiculous keyboard or one of those programmable keypads just so i could macro special letters
I have an azerty keyboard. But I put it on qwerty. 
So, I can type in most of the symbols. But the weird ones like that "Ü" ... no clue where it is on the keyboard.
I did this because I switched 5 years ago to qwerty suddenly (easier programming). But I didn't want to buy a new keyboard.
You know what I am really looking for? My ideal keyboard. I struggle to find keyboards that are maintenance friendly. All these keybaords have so many places where dirt can really get stuck to making it harder to clean.
i definitely miss having just screws holding things together instead of those stupid easily breakable plastic tab clip things
idk if i could get used to the feel of round keycaps
typewriter style, interesting
i still have my great-grandma's woodstock typewriter. she could do 100wpm on it
Have you ever heard of washable keyboards?
They have drainage holes on the bottom.
Thats logitech K310
Usually those are just for spills.
So most with such holes aren't suitable for immersion cleaning without disassembly.
Look at cleaning instructions... they are literally called washable keyboards.
That one is.
But meant the "They have drainage holes on the bottom." part.
That most with the holes aren't washable.
I want one. But afaik, it's discontinued.
Haha, you clever. You making sure nobody who reads this is putting their keyboard in a bathtub because it has "holes" 😂
but in the end, I want a keyboard with as less crevices as possible. Should make it easier to maintain. Maybe I also will try out a 60% for gaming.
For the rubber dome type I just disassembled the things and washed the case parts and keycaps.
Cannot do with current mechanical without desoldering, as the PCB is under the top case, and the switches are above, with the case sandwitched in between.,
So can just wash the keycaps, and try to clean the case around the switches.
my new keyboard has hot swappable switches finally
I got my keyboard to work today by smashing it on my desk a few times
I really need a new one
It should arrive in like a week
same here.

That's the reason I was looking into ducky DIY keyboards.
You can deconstruct it fully.
same 😄
Switches are still mounted onto the front plate.
So would need to remove the switches from the plate, and their mounting isn't made to take multiple cycles of that.
The mounting clips on the switch bodies are pretty standard, and they are flimsy, meant for single install, not removal intact.
They can be removed intact, but not really intended for it.
So when you unmount them, you might not get all back intact each time.
You mean these two iron clips?
Those are contact pins, not mounting.
the switches themself are made for it.
and you can just buy new switches anyway 🤷♂️
would be weird since it's Keychron "selling" point is to have hot swapable switches. Would be surprised if the switches weren't tested on that.
Center pin, two contact pins, and some small clips on the side.
and like I said, if one breaks, you can replace it. Better than to have to buy a whole new keyboard.
But seems to be cleaner in the pic than I had pictured.
personal preference.
I'll get gateron browns with it tho 🤮
so I'll need to get some cherry blues seperately
If there was 100% size hotswap base with separate multimedia keys (back, forward, play/pause, mute, volume + and -), top-side leds and 1.25/6.25 lower row...

Haven't been able to find, Keychron has south/near-side leds...
And most don't have multimedia keys except via FN-type comboes.
maybe ducky
Only the three volume keys there separately in their full-size models.
Like I said, Keychron has the leds on wrong side of the switches.
Not suitable for keyboard marking lighting, just for key surrounding lighting.
For glow between the keys, not through they key markings.
think on my corsair K70 the leds are at the bottom.
Have 0 issues with it. I can see my keys. Some leds are broken (faulty colors).
Corsair has that type of keycaps, but on most the markings are on top side.
And Corsair uses different bottom row layout.
yeah, transparant.
I never, ever use my media keys. Never. So I will not buy a full keyboard again. No use.
That was some Corsair model, and checked the K70 pics otherwise, that was just first one which I could find with some keys removed.
K70 MK2 SE for the pic one.
those media keys are totally different than mine. I probably have MK1
Different model, just reused the K70 name.
uhu
But all the Corsair keyboards I quickly looked through had north-side leds and markings.
the leds on this corsair died way too quick
will not buy corsair again. Too pricy for the quality I got.
well, the frame is solid. And that's the only thing positive I can say about it.
your resistance to not making your pfp a gate is amazing
my guild is pretty fun now. Every new member is a reference to BG3. "Gale", "Lolth", ... they all joined recently. All BG3 references.
I am surprised Baldur isn't playing BG3 24/7 
or maybe they are. Who knows haha 😄
yea, been seeing a spike in BG3 names in my mmo
Would need to start with BG1 at some point.
Never finished it as I always went to wrong directions and "broke" the plot.
i mean, it's at least mildly appropriate considering it's D&D
Was made to allow for that, but if you hadn't played it once the right way it started to get weird fast.
still applies for the 3rd game 😄
but i've also never understood the need to make character names that have NOTHING to do with what your build is 🤣
I just picked up a cool name. My fighter in BG3 is called Fenix.
in DDO you can reincarnate which gives that character a Past Life of that class or race, which lets you start over at level 1, just more powerful each time, so you'll see people who clearly started off as a barbarian named something like "Hacknslash" who are now cleric healbots
& the storage/inventory situation there is awful, so you'll see characters named "leveltwostorage" running around
ah the classic character bank
My keychrons lights are fine
which one do you have?
k is also north
Which makes it even weirder that the Q-series ones have south side leds.
When K-series (both low-profile and normal profile) have north side leds.
Part of K-series seems to also have south side, at least K10 Pro has.
K10 non-Pro has north side.
And the ones available in ISO, at 100% size are Q6, V6, K5 Pro and K10 Pro
Only K5 Pro of those has north side leds, and is for slim switches and keycaps.
normal k10 can also be iso
I never noticed the leds being north/south or whatever. Does it bother you?
Want keymarking illumination without excessive between keys illumination.
I didn't like the look of the k10 pro
Ok, they didn't list it on their site when selecting only ISO keyboards.
depends on what iso
Talk was about hotswap switch ones.
Corsair K70 Pro Mini Wireless is hot swapable.
Lot different keyboards available that fill all the other ticks but that.
And none of the hotswap ones seem to fill the requirements.
And isn't 100%, doesn't have separate media buttons.
Seems to have the standard bottom row and not the Corsair one...
uhu, mini
.
just saying there is a K70 model that supports it 😛
just not waht you are looking for.
And it's the first corsair, released last year I believe. So who knows for the future.
- Keychron Q6: program your own media buttons.
I'm kind of done with popular tech brand-peripherals tbh. I mayhaps get another logitech mouse if my current one dies but the keychron will definitely be my entry point for custom keyboards
South side leds.
So doesn't work with see through markings.
So the AK35 would fit, if it was available in ISO.
Would need to check.
No
And has south side leds too...
For the current V2.
Original model seems to have had north side.
just turn keyboard 180 degrees and you got north side 😉
Or just use the current non-hotswap keyboards I have that hit everything but that.
Which just makes cleaning harder.
I mean, I want to clean my keyboard monthly.
So the easier it is to clean, the better for me.
one of those battery-powered canned air things is a great way to keep on top of it
takes care of like 90% of the garbage that accumulates in it
https://www.reddit.com/r/SatisfactoryGame/comments/196dswu/does_satisfactory_take_advantage_of_amd_5800x3ds/ OMFG this topic is giving me a headache, especially the answers
how can some people be totally clueless and yet still offer an advice on something is beyond my imagination
"i'm not a doctor, but i play one on tv"
XD
the only game i play that actually uses more than one core that i've noticed is KSP, & that's because vessels within physics distance get their own thread
"unless Satisfactory is in some way special (it isn't to my knowledge) it should behave like any other CPU-bound game."
This answer though. Makes no sense to me. It's not satisfactory being special. It's the CPU.
can each core grab ahold of the full L3 cache?
Because the CPU is different, some type of games (mostly the ones that benefit from cache) have better performance.
exactly, they are totalyl clueless
L3 cache is like a shared storage between CCX
in Ryzen cpus, if there is 1 ccx, there is 1 l3 cache
i can never remember the specific architecture like that
Essentially giving you another L3 cache, 3D V-Cache allows your processor to distribute and store more instructions, meaning that it doesn’t have to take them from RAM as often.
Zen 3 and Zen 4 have 8 core CCX:s, so single CCX per CCD.
And the L3 is shared inside the CCX/CCD.
X3D, afaik, has additional layers of cache stacked on top of CPU.
but also, 5800X3D is still what, top 5 for gaming?
at that level it basically doesn't matter
In theory with the dual-CCD 7900X3D and 7950X3D can have the non-Vcache accessing the V-cache CCDs L3, but in practice that is so slow it isn't done.
let's gooo my cpu is still good

my 5600X still feels completely fine for about 95% of gaming
technicalyl speaking, if they went with intel approach and had a hardware for core assignment it would work
Problem is the latency over the IF
yeah, Infinity Fabric is the current limitation for now
It can be accessed, but the access has so much latency that it isn't useful
Baldur, correct me if I am wrong, but according to that picture.
L3 is accessible by every core.
That isn't for any late AMD CPU.
that's 1 ccd that has all CCX inside
Trying to find a good architecture picture mhh.
How the L2 and L3 are split or shared between cores depends on the architecture.
Almost all current Intel and AMD CPUs have per core private L1 and L2 caches, and shared L3.
Intel Atoms have shared L2 per X cores (4 at last I checked IIRC).
are all 5800X3D CPUs single CCD?
Yes.
all xx800x3d cpus are single ccd up to now
i know there was a big agesa brain fart with 7000 series & whether or not they were 1 or 2 CCD
that's not 7800x3d
it's dual ccds here
so either 7900x or 7950x or 7900/7950x3d with 1 ccd being v-cache stacked
8 core and down only has single CCD.
Then the 12 and 16 core ones have two.
In some cases there is second inactive die in the CPU when delidded, sometimes not.
was that due faulty 7950x dies being used in r7 series?
iirc the agesa issue was because they were using dual CCDs with models like 7600X with single deactivated cores on each CCD, so it was a 2x3 setup, & the agesa would just shut off a CCD & people were suddenly having 3-core systems
their documentation is ok though
it takes some time to understand, it's a complex thing
I do remember something to that, but if that is true, the dual-CCD 7600X would have significantly worse performance than single-CCD 7600X in some things that need cross-core communication...
So I really cannot believe AMD would do that type of thing.
And AFAIK they cannot disable cores in that amount anyways, only in steps of 2.
So 8 core CCD cannot be cut down to 3 cores.
2+4 would be in theory possible.
But if you have only 2 cores functional from 8...
is that even possible?
No, two CCDs max.
yeah
For the IO-die used and the substrate used.
i'll have to hunt it down. the bug was definitely shutting off an entire CCD
Is this one correct?
what i remmeber is what i wrote above, AMD repurposes faulty dual ccd cpus, for example when a 7950x dosn't pass a QC check in certain cores, they can use the good ones in 7600x etc
i read that many months ago so i don't fully remember the exact details
Yes, but one of the CCDs is completely disabled.
Yes.
If you skip the right side that is MB specific at least partly.
Aren't the X3D chips or at least ryzen 5000 X3D chips like the highest bin possible other then server grade.
not necessarily
Iirc the 5800X3ds were failed Milan-x ccds?
Well, at first the 5800X3Ds were server line reject CCDs.
7800x3d is I assume not the same case
"MSI, ASRock & ASUS rolled back the BIOS due to a major flaw that disabled a CCD on a few Ryzen 5 7600(X) CPUs that feature dual CCD layouts."
so 7600X definitely has dual CCD variants
as far as i know, the highest bin possible is on 7950x3d ones
Question is if that explanation is right or not?
those are the top QC pass ones
Highest bin is server grade. Iirc
What is the source for it?
Because might be just speculation originally without official AMD or MB maker word.
So no, no official word, just the word of chi11eddog.
https://www.techpowerup.com/277053/dual-ccd-ryzen-5-5600x-and-ryzen-7-5800x-in-the-wild it even goes back to 2021 5600x dual ccd
For consumers yeah that's probably the highest binnec chip. But for servers iirc the 5800X3d had some of the best vf curves and still didn't pass for server bin.
Which is a thing:
There are, apparently, some 5600X and 5800X built from dual-CCD MCMs, in which an entire CCD, although physically present on the package, is disabled. A 5600X based on a dual-CCD design is essentially a 5900X from which one of the CCDs didn't fully qualify;
But having two CCDs with small parts of each enabled isn't AFAIK.
that i never heard of tbh
that kind of screwup sounds like AMD completely forgot that it was a thing 🤣
if you see their arc. it makes sense, repurposed 7900x > 7600x
repurposed 7950x > 7700x etc
but idk how they manged to do that if it's real
some parts of each ccd being active on 7600x is wild
Which has been the point of the talk, if there have been 7600X:s with two active CCDs with just 3 cores each.
Which isn't possible.
More probably 7900X that had faulty CCD0.
And the 1.0.0.4 AGESA tripped with that config, when the CCD1 was supposed to be the active one.
might just be the cores that aren't up to spec, with the rest of the CCD being fine
aah now it makes more sense , i see.
And worked when CCD1 was disabled and CCD0 was one working.
my current amd setup has a 7600x in it, i was supposed to get a 7950x3d but couldn't find a one in stock back then, i'm quite happy with it
gonna get a 8800x3d when it gets released
i'm not even sure if it's even neccesary for satisfactory only usage though
7600x handles it quite nicely
yea, basically any cpu that's not in a laptop made in the last 3-4 years should really handle SF just fine
unless you're trying to do something insane with the map
you can melt just about any cpu's capabilities with just about any map building game if you try hard enough
Like I have said several times here, my 7800X3D could do 90-100FPS in Taros 29MB save if I had good enough GPU.
eh, i've seen some usage of x3d cpus in satisfactory and i gotta say, the difference between a 7600x/7700x to 7800x3d is not that big compared to my expectations
it's probably has to do how a game data is optimized to be calculated
people with superior game dev knowledge know something about it
my knowledge is none in that part
Send your save in PM and we can do comparison of Game time while at HUB?
meanwhile, my 18 year old MMO has an area that's absolutely murder everyone's fps no matter what because of all the assets because of how badly optimized it is
sadly, atm im on my old ass laptop. My pc is in my main city
we can arrange that once i go back
but my save is not that big, maybe we can use a really big one
to see the difference
5600X/6800XT system, this ancient game will drop down to 30fps in that area because of how awful they coded it
I am less happy with my 7600X 😂
your pc would love a 5800x3d
yea but my VRMs wouldnt
x3d really just works wonders with mmorpgs
Well, your system has some really weird issues...
care to elaborate?
Blame AMD.
88W max?
basically. B450 gigabyte
As in the 5800X3D is very low power.
I had to downclock FLCK to get it stable. Auto FLCK is 2000MHz. I have to run it at 1733MHz.
Else I get hard crashed, black screen, no BSOD, no logs.
if it drops to $250, maybe
Still not 100% sure that this is the real issue. But something funky is going on with my CPU.
Unstable in some exact things.
And might have VSOC damage for running with old BIOS when new, because ASRock seems to have very stupid defaults.
On their AM5 boards.
but i'm betting something in 14th gen intel or ryzen 8000 series will beat it before it gets cheap enough, & i'd probably go for that upgrade
Reboots without warning, which points to either uncorrectable ECC fault, or in theory MB VRM protections kicking in.
aah.. if you've seen my auto gigabyte pbo settings you would cry
And only ECC protected thing that doesn't send warnings is FCLK/IF.
Point was the MB was setting 6000MT/s RAM settings on voltages and FCLK by default, always.
Can G-Sync cause crashes like this?
Not just when the RAM was set to specific speed.
the first thing i did was manual pbo settings and curve optimizer negative on all cores
No, shouldn't.
so in theory this already limits the voltage
That's not what I asked 😄 Can it. In theory.
my pc still runs perfectly with very old bios
Unless the GPU is doing something very weird, and in that case it should be the GPU crashing.
and it never goes higher than 1.2V if i rememebr correct
I ran prime95 yesterday for 15 minuts. All stable.
Which shouldn't autoreboot the computer.
ASRock in latest BIOS was setting VSOC to 1.25V with RAM on auto.
So the original BIOS might have set it to 1.35V on first boot, before BIOS update to the VSOC limited ones.
As in when CMOS was cleared or "Set defaults" used inside BIOS.
since I tuned it back to 1733MHz it ran stable for a day. But way too soon to tell if it's "solved" or not.
It did crash when I switched from that weird 1750MHz setting to 1733MHz. But only once. Since then it didn't happen again.
and the prime95 readme file says 10-15 minuts should crash flacky system.
Not really, depends how flaky.
Can take days in cases.
For the IF crashes I had, took few hours each.
Yeah I've read that, but since it happens sometimes frequently I though to gave it a shot. 🤷♂️
i'm not even sure if i changed anything in fclk manually
really freaking weird.
i only changed pbo settings, curve optimized and since i have expo kit, i just enabled it
My current plan is this: keep everything as is for a month. If stable after a month. Re-enable G-sync on my monitor. If it's still stable. Never touch bios again.
everything works wonders
1733MHz is the AMD default.
Most MBs raise it up to 2000MHz as you raise memory clock, hitting that 2000MHz with 6000MT/s RAM.
ASRock is setting it to 2000MHz by default, always.
If my CPU is dying: hopefully by then I can buy X3D CPU.
If it wasn't the CPU, I will replace my motherboard. But that's for when it really starts to die on me.
i will remmeber it to check what is my fclk ocne i go back, but my kit also is 60000MT/s one
In the meantime I make sure to offload screenshots to my NAS so my data is safe.
but probably it's what you said, since it's an expo profile, it should also automaticall change that value
2000MHz FCLK is the number AMD says that (almost) all AM5 CPUs should be able to do.
speaking of which, going to do meta-event in GW2 that made the setup crash a lot. 🤞
But the VSOC damage causes it to start dropping with time.
At least based on my own personal experience.
in theory, if you have high negative curve optimizer, dosn't this also limit the voltage upper limit?
That is for the CCDs, VSOC is for the IOD.
afaik, vsoc is core voltage
Got my first 7800X3D replaced by AMD, it couldn't do 1933MHz FCLK stable anymore.
but baldur should know better
that's weird, then i guess i were lucky that my Voltage never exceed 1.2V?
i'm not even sure, it also has to do with me limiting the pbo settings to eco?
One mistake on that one, the VDDCR_GFX is from VDDCR, not from VDDCR_SOC:
So, if it's VSOC damage, then it's CPU damage. Just want to know what to replace first 😄
And I mean, at this point I have no case to make to AMD. Since it's so hard to reproduce. I cannot go for a replacement on a hunch can I? 😒
The VSOC debacle was that AMD originally said that VSOC can be 1.4V max.
But in reality that killed CPUs.
And was limited to 1.3V in later AGESA versions.
And all MB makers were using 1.35V SOC when the memory was set to 6000MT/s.
not an issue with Intels 😄
Intel
Well, it seems in the end I will paying the same price as a Intel CPU tbh. Since I might have to buy two now 😂
fuck it, when I think like that, I might regret the switch. But honestly. I might gotten really unlucky.
you probably got really unlucky
Yeah, I think so too.
just to make sure i will update my bios once i go back
i think you just got unlucky if its hardware failure
Part of the problem was/is that the VDDCR_GFX (IGP voltage) and the VDDCR_SOC (VSOC/IOD main voltage) got too large difference when the CPU cores & IGP were idle, and that causes/caused damage inside the IGP, that killed/kills CPUs.
On all Raphael CPUs, just killed the X3Ds quicker.
Asus had another bug and continued to feed 300+A into CPU that had shorted IOD, melting the thing.
But all MB manufacturers killed CPUs, and first ones were from over 2 months before X3Ds were released, when people went back.
No over current protection on VRM when CPU wasn't booted up yet, except what the VRM could feed.
So the computer shut down when the IOD died, then if you tried to reboot with Asus board, you got melted IOD, CPU substrate and socket.
Other manufacturers just didn't boot anymore.
AMD replaced the CPUs, Asus replaced the MBs (after some fighting for North American market).
If in reality AMD really fucked up with their max Voltage values then it's on them
even then, I don't have a spare CPU. So sending my CPU to AMD isn't really an option.
I did suggest trying to ask the CPU seller, which should be couple of days turnover.
It was AMD giving out bad specs, and then all MB makers using the max "allowed" for "reasons".
I know, and I said: I have no case. "PC crashes. Mostly (99%) with GW2. I lowered my FLCK and that seemed to make it stable. Can I have new CPU?" 😄
How to reproduce it? "I don't know, play GW2 for hours and hours? Download something in the background"
Try with the 1800MHz FCLK with TestMem5, Prime95 and Y-Cruncher for nights.
its strange what triggers something ... like it only seems to happen in GW2 for you
for my 980ti, the ONLY thing that triggered its failure point, was standing near shores in ESO
every other game was 'fine'
Only once did it crash on the desktop. And that when steam was downloading BG3. Also CPU related.
but that was when FLCK was on 2000MHz
the problem is you need multiple stuff avaialbe if you want to detect what is faulty
so what Baldur suggested is the easiest fix but requires time and patience
When you are heading to bed next time, set the FCLK to 1800MHz, then boot into Windows and set Prime95 running for the night.
And see what is the status in the morning.
send it back to seller so they test it
its been years now ?
feels like years 😆
mhm, can try.
Also it's weird, but alt + tabbing caused a lot of crashes. And now that I disabled G-sync on monitor. Alt+Tab feels so smooth, no black screens for 1ms or anything.
So I am still keeping that open as a possible culprit. With very low % that it's actually the issue.
when you say crash, do you mean an application crash or a system crash?
Non-warned HW level reset.
Like you had pressed reset button.
g-sync isn't software only 😒 afaik.
read reports of hard crashes because of bad HDMI cables.
I mean, it has a hardware side to it. (monitor needs to support it correctly)
So needs to be something that causes MB to trigger it.
VRM overload or resonance in them.
Or uncorrected ECC error somewhere, as AM5 AGESA doesn't report corrected errors for some types yet and hard-reboots on uncorrected ones, instead of reporting to OS for BSOD.
And IF (and possibly PCIe) is the only one I know of.
The CPU cache ECC errors do get reported.
And RAM ECC errors would need ECC RAM.
i'm not sure how exactly gsync works in hardware level but if your issue is 100% related to that, i would look how it works on that part
but even then that's really complicated
Should only at max crash the GPU, GPU driver, or OS.
None of which should cause HW reset.
that's also what i thought
atm the easiest fix is to have 2nd pair of everything in hardware
a mobo, new memory kit, new cpu
heck, it could even be a psu problem
the whole point of this build was no to blow money. and don't have 2nd pairs since previous is Intel/Nvidia.
and that PC went to my niece.
And IIRC there isn't the clicks that would be there for PSU protections kicking in.
Or for PSU off then back on power cycle by MB either.
well tbh if he takes his pc to a good repair guy, he will 99% figure out the problem
i can't imagine my current desktop breaking though
it's heavy af, has tons of fans, a 420mm aio
Yup, would need the system tested with known good CPU, and the CPU tested in otherwise known good system.
i can barely lift it
To be sure.
I don't know. That costs me also money.
i mean sadly, that's the best way for you to find the culprit
But they might figure it out after weeks.
Which is why we can try to pin down a specific case where it crashes near enough to AMD standard, that it should be able to be replaced with RMA.
a similar topic to yours
Like that 1800MHz FCLK with some test program, in realistic time frame.
1 important note: his were all software crashes
and guess what was the solution?
Not sure if MB & CPU replacement, or just MB replacement.
But again ASRock board.
or both, in a really unlucky day
Originally, and the new one isn't.
But that doesn't really help, it is most likely the CPU, but need to prove it to suitable level to get quick RMA with the seller.
Because direct AMD one takes ages.
sadly he didint test his old cpu with new mobo first
so we can't know what exactly was at fault
but mostly a faulty mobo is a higher risk than a faulty cpu if im right
can a faulty cpu fry a mobo?
Usually not, in theory yes.
But the instability can be unstable VRM output, or just insanity by the MB otherwise.
But need to try to confirm one way or another.
yeah, mostly it's the otherwise that stupidly high voltage/Ampere frying the cpu
so that's why mostly the first thing that is tested is the mobo
sadly that's it
there is a high chance your pc randomly stop working totally too
if it's a vrm/cpu related thing
if you are lucky, you will only have to replace your mobo
i got quite lucky in that aprt, i chose gigabyte mobo for my am5 setup and it's working quite nice
afaik gigabyte ones are quite decent in am5 lineup
other than the really cheap ones
the crash from 1750MHz to 1733MHz. Could it been a fluke, Baldur?
No idea.
And still no idea how ASRock managed that 1750MHz FCLK.
I set it to 1750MHz. It crashed. I set it back to 1733MHz. It crashed. (the last crash?)
Who was it that posted the same screenshot? 🤔 Of the weird bumps.
ASRock doing something very weird.
fireworker or fight or flight
In both the MBs, yours and his.
yeah, seems so.
the craters of dead?
there are many topic
ASRock MB FCLK settings going in weird steps.
aah
Should be in 33.3... MHz steps, (1/3 divider from 100MHz base clock).
Both ASRock boards (one AM5, one AM4) had them all over the place.
With some 1x40MHz and 1x50MHz settings for example.
that's weird
For Letterdief, it jumps from 1800MHz direct to 2000MHz, nothing in between.
And there are 1700, 1733, 1750, 1800 MHz settings, instead of 1700, 1733, 1767, 1800 etc.
but can this cause a crash?
If the settings work right, shouldn't.
But is weird and no idea what the ASRock BIOS team is doing...
The defaults were all over the place.
Well, I went from 1733MHz running stable for two weeks to 1750MHz. It crashed 3 times that day.
Switched back to 1733MHz. Played 10 hours of GW2 yesterday and no crash. Today I have been playing for about 4-5 hours without a crash,.
then just stay there i guess?
Which is why I expect this might be VSOC damage from the initial BIOS, without the VSOC limiter in place, running on 1.35V by default when the new ones run at 1.25V VSOC by default.
When the default should be 1.05V.
I am going to all the zones/events (in GW2) that made the PC crash before. The events with most people, mobs and spell efects.
Now slowly progressing with the lower voltage too.
Like my CPU did.
Because the crash type fits uncorrected IF ECC errors.
But trying to get more data.
Other MB makers set the VSOC to 1.05V when the RAM is on auto.
That MB with the latest BIOS sets it to 1.25V fixed.
And so there is reasonable chance that the original BIOS the MB came with set it to 1.35V until the BIOS was updated with CPU installed.
do you remmeber at what version this issue has fixed?
if i remember im on F6 bios in my b650 aorus elite and have this one
and mine never vent above 1.25
ah that sohuld be the case
VCore or VSOC?
What is Vcore on Motherboard?
Asus added the limit to first BIOS with AGESA 1.0.0.6 for example on the BIOS side.
The voltage that feeds the CPU cores.
1.4V max is normal for non-X3D CPUs.
tying to find vsoc 😛
aaah you are right
with X3D:s having 1.2VCore max.
In the CPU temperature & power section.
Between the temperatures and powers.
CPU VDDCR_VDD Voltage (SVI3 TFN) for VCore and same with _SOC for the VSOC.
Right below the CPU IOD hotspot temp.
Which is what we manually forced it to.
i love discord, i found an image i posted in another server
When the default by ASRock was 1.25V.
I think I put it on auto. I cannot remember. Only value hardcoded I saw in Bios was the 1733MHz.
With the 2.02 BIOS.
It was forced to 1.25V, and auto set it to same IIRC.
Same with the FCLK, was set to 2000MHz, and Auto set it to same, so had to be forced down.
so yeah it runs at 1.025
Right. The FLCK cannot be put on auto.
not even 1.25
You probably have your RAM on Auto frequency, not XMP enabled.
The only thing I whished asrock provided is: set everything to default option. Because hunting in the BIOS for value set is annoying.
auto expo
not touched anything
EXPO isn't on by default.
Needs to be manually set, and usually at that point the MB cranks the VSOC up too.
ASRock was cranking it even without RAM frequency being set higher.
what i mean is that i enabled it in bios
and havnet touched anything else regarding those values
Ok.
a really tight fight 420mm aio
😂
i wonder why amd is sleeping on x3d cpus on laptop side
i need a new laptop and the only x3d laptop avaiable is with a 4090
Because most of the laptop CPUs are with different die(s) that don't have the taps etc. for extra cache die to be added.
imagine paying 4k euros for a 150w tdp 4090
there is already one that asus released
but i dont understand the 4090 part
AMD released that single desktop replacement laptop CPU with v-cache, almost no-one just uses it yet.
that is also correct tbh, but they gotta start partnering with manufacturers if they want bigger market share
that was always the case with AMD, they don't really want to compete in consumer side with Intel nor with Nvidia
my current laptop is with Intel cpu, im not buying another intel laptop
Want, don't have the resources to put into it yet.
in terms of efficiency it's shit
And are also limited on how many units they have die allocations for.
From TSMC:s limited capacity.
yeah, now that i rethink that is correct
intel has the capacity to fabricate their own chips even if they are not top level
I guess that problem will be semi fixed with new fabs comming online in the future
both in US, EU and Japan
they dont have to make the newest ones, as long as they can make the current ones that's quite the deal
Yes.
And once they have enough wafers to fill their current demand on server space, they can use some of the money to form the design help department(s) needed.
For increased standing in laptop space.
When they can actually fill the laptop chip demand there already is.
they still have quite the time
unless intel does a total 180 and fixed their issue with their new architecture
it seems they also are having problems with it
what was it, Arrow lake or something?
The fact that Intel currenly fabs some of their chips on TSMC...
All the ARC GPUs, some of the chiplets for the new 14th gen laptop chips.
yeah.. It's kind of inevitable to happen, TSMC is pouring shitton of money on R&D every year
it's kind of hard to catch up to them
not forgetting that 10+ years where Intel did nothing
Intel exec team just failed, and intentionally let go the more experienced engineers 10 years or so ago...
As cost cutting measure...
And the CPU design team had too much power over the fab tech team.
that's what happens when a tech company has a economics CEO
or whatever was that guys prof.
it never works it will never work
1 side always prioritizes short term profit
while engineers know that long term is always the best goal
"Why do we pay these old engineers this much when the new people fill the same spot with half the pay?"
exactly, i couldn't even name it better
same thing is happening with Amazon on Twitch right now
they literally fired most of the experienced stuff
in name of cost cutting
it baffles me how these big companies never learn
the only exception here is Nvidia for some reason
they are greedy but they are the leader in pushing new technologies
atleast in their department
they never stop
So at same time the 10nm Intel fab tech was step too far at once, because the CPU design team didn't have realistic wishes and the fab team was pushed to try to fill them.
And at same time the experienced people were let go on both teams via big voluntary severance packages etc.
Twitch has been digging it's own grave for a while.
Corporate life is a weird thing tbh
not even people in Corporate world understand it
MBA was supposed to be just additional thing you went through after some real world experience.
Not be the only thing you had.
a lot of people switched to Youtube for streaming. The whole real life streaming is the nail in the coffin for twitch (for me).
Now it teaches that the MBA people know best about everything and is taught as only thing for people and then they go to leadership positions.