Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
Subjunctive
Sep 12, 2006

✨sparkle and shine✨

Desuwa posted:

Based on my experience getting 4x32 to work, I would not trust 4x48 at any reasonable speeds without monitoring for ECC errors. Memtest should pick up errors but it can sometimes miss CPU-side errors especially if you end it early while trying to dial in an overclock.

What did you get 4x32 to run at?

Adbot
ADBOT LOVES YOU

Wibla
Feb 16, 2011

I slapped in 2x32 and 2x16 of DDR4 3200MHz Cl16 and it just works with a 5800X3D. Not sure how happy I'd be trying to do that with DDR5.

Desuwa
Jun 2, 2011

I'm telling my mommy. That pubbie doesn't do video games right!

Subjunctive posted:

What did you get 4x32 to run at?

Without boosting any voltages or playing with timings, 4000mhz has been stable without any corrected errors logged. Even 4200 gave me one after a few days of actively monitoring it. Once all this voltage nonsense settles I might give it another 100mv to see if I can't get it stable at 4800. It boots just fine at 4800 but eventually an uncorrected error will cause a reset. Since I haven't played with voltages I don't really know how well 4 ranks per channel scales with voltage, but 192GB at 6000mhz does not sound believable to me. Even if it appears stable I wouldn't actually trust it myself, and I assume whoever was claiming it wasn't running ECC.

Cygni
Nov 12, 2005

raring to post

I personally haven't seen anyone claiming 192gb at 6000, just 5200. The tweet we've been discussing was using that as an example of memory speed impacts from the new AGESA, it may or may not even be the exact actual impacts.

redeyes
Sep 14, 2002

by Fluffdaddy

Cygni posted:

Uh, losing that much performance months after launch on account of AMDs own gently caress up is pretty bad.

Oh I agree, I just find it funny none of them had any clue about this. They don't do any in-depth testing.. its all for show and selling poo poo. Yes i'm salty.

Twerk from Home
Jan 17, 2009

This avatar brought to you by the 'save our dead gay forums' foundation.

redeyes posted:

Oh I agree, I just find it funny none of them had any clue about this. They don't do any in-depth testing.. its all for show and selling poo poo. Yes i'm salty.

Sure, but it's literally always been like this at new DDR3 generations. I had a launch Phenom II 965, which was found after launch to have a memory controller so hosed up that it couldn't do 4 DIMMs of DDR3-1333. They fixed it with a new stepping of the same model, which also had a 15W lower TDP: https://www.guru3d.com/articles-pages/phenom-ii-x4-965-be-revision-c3-review,1.html

I'd bet that a 7950X manufactured in 2024 is going to support faster memory than one you can buy right now. We don't tend to get reviewers paying much attention to CPU steppings anymore, though.

SwissArmyDruid
Feb 14, 2014

by sebmojo
GN followup to exploding X3D chips: https://www.youtube.com/watch?v=fFNi3YNJXbY

SwissArmyDruid fucked around with this message at 01:22 on May 10, 2023

hobbesmaster
Jan 28, 2008


The only problem is that in the future manufacturers might just wait Steve to do root cause/corrective action for them

edit: holy poo poo the copper near the CCD melted indicating temperature exceeded 1000C after failure. Thats some solid magic smoke!

hobbesmaster fucked around with this message at 01:34 on May 10, 2023

Canned Sunshine
Nov 20, 2005

CAUTION: POST QUALITY UNDER CONSTRUCTION



Paul MaudDib posted:

if it ever was it would require absolutely titanic VSOC voltage

I guess you could say that these enthusiasts were going down with the chip

I’ll see myself out…

Shipon
Nov 7, 2005

hobbesmaster posted:

The only problem is that in the future manufacturers might just wait Steve to do root cause/corrective action for them

edit: holy poo poo the copper near the CCD melted indicating temperature exceeded 1000C after failure. Thats some solid magic smoke!

melted silicon :catstare:

Cygni
Nov 12, 2005

raring to post

Finally got some time to watch the vid and its so good. Even if you aren't interested in the failures themselves, its worth watching just to see the awesome imaging. Learned a lot about how the sand is tricked into thinking.

BlankSystemDaemon
Mar 13, 2009



Yeah, it's a very good video, and also basically confirms my wild speculation above.

Also, the sand isn't thinking, it's just doing maths.

ijyt
Apr 10, 2012

BlankSystemDaemon posted:

Also, the sand isn't thinking, it's just doing maths.

Arguably a greater reason to hate it.

Subjunctive
Sep 12, 2006

✨sparkle and shine✨

BlankSystemDaemon posted:

Yeah, it's a very good video, and also basically confirms my wild speculation above.

Also, the sand isn't thinking, it's just doing maths.

It’s tricked into thinking it’s doing math.

BlankSystemDaemon
Mar 13, 2009



New workstation up, VDDCR_SOC running at 1.248V, memory running at 6kMT/s and drat it runs X4 well.

Loading the EXPO profile also appears to have tuned Infinity Fabric and a bunch of other things (specifically the UCLK-FCLK divider mode) to what I thought would be the optimal values.

ijyt posted:

Arguably a greater reason to hate it.
Eh, hate's a strong emotion.

Subjunctive posted:

It’s tricked into thinking it’s doing math.
Poor thing, doesn't even know what it's gotten itself into.

First it was addition and subtraction, now it's doing hardcore compare and exchanges, vector calculations, and floating points, and even arithmetics directly in the vein.

Kazinsal
Dec 13, 2011


BlankSystemDaemon posted:

First it was addition and subtraction, now it's doing hardcore compare and exchanges, vector calculations, and floating points, and even arithmetics directly in the vein.

*shaking bottle of AVX instructions* Psst. Hey, kid. First one's free.

BlankSystemDaemon
Mar 13, 2009



Kazinsal posted:

*shaking bottle of AVX instructions* Psst. Hey, kid. First one's free.
*cue StarCraft: Broodwar marine stimpack sound*

slidebite
Nov 6, 2005

Good egg
:colbert:

Latest follow up on GN X3D saga.

SCUMBAG ASUS

https://www.youtube.com/watch?v=cbGfc-JBxlY

alex314
Nov 22, 2007

You can see how much fun Steve has roasting companies.

Subjunctive
Sep 12, 2006

✨sparkle and shine✨

Not feeling super good about my AM5 ASUS motherboard right now, but the voltages all look OK with the latest BIOS (even on EXPO) and I’m not going to turn on any of the overclocking stuff so I’m sure(?) it’ll be OK.

It also stopped randomly turning off without me even swapping the PSU (busy/lazy), so I’m going to pretend that is actually fixed too.

slidebite
Nov 6, 2005

Good egg
:colbert:

"Asus has more warranty reversal engineers than electrical engineers"

https://youtu.be/cbGfc-JBxlY?t=1147

lol

But seriously, it does sound scummy as hell

Cygni
Nov 12, 2005

raring to post

slidebite posted:

Latest follow up on GN X3D saga.

SCUMBAG ASUS

https://www.youtube.com/watch?v=cbGfc-JBxlY

This whole series of GN vids is their magnum opus. Ending on the fire and brimstone video is just perfect. The whole segment redoing the testing without expo on is savage.

They briefly referenced issues with Asrock and Gigabyte too, so is MSI really the current leader of the 4 in the competency/doing the right thing department? MSI?? Gods help us all.

wargames
Mar 16, 2008

official yospos cat censor

slidebite posted:

Latest follow up on GN X3D saga.

SCUMBAG ASUS

https://www.youtube.com/watch?v=cbGfc-JBxlY

I can't get expo to be stable with gigabyte b650 so i am below jdec speeds.

https://valid.x86.fr/ba510z

wargames fucked around with this message at 19:34 on May 11, 2023

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast

Cygni posted:

This whole series of GN vids is their magnum opus. Ending on the fire and brimstone video is just perfect. The whole segment redoing the testing without expo on is savage.

They briefly referenced issues with Asrock and Gigabyte too, so is MSI really the current leader of the 4 in the competency/doing the right thing department? MSI?? Gods help us all.

MSI has been taking AMD boards seriously for some time to be fair, and when others were putting lovely VRMs on their budget boards, MSI wasn't. I also remember ASRock just being considered a cheap spin-off of ASUS, although of course now they're known for their quirky (usually in a good way) boards

HalloKitty fucked around with this message at 20:20 on May 11, 2023

runaway dog
Dec 11, 2005

I rarely go into the field, motherfucker.

Subjunctive posted:

Not feeling super good about my AM5 ASUS motherboard right now, but the voltages all look OK with the latest BIOS (even on EXPO) and I’m not going to turn on any of the overclocking stuff so I’m sure(?) it’ll be OK.

It also stopped randomly turning off without me even swapping the PSU (busy/lazy), so I’m going to pretend that is actually fixed too.

me neither, not even running expo, and those non-expo speeds had me eying my amazon return by date of may 20th

CBD Corndog
Jun 21, 2009



Its too bad MSI just had a major data leak after their ransomware attack, making driver downloads from them dubious at best for a long time

https://thehackernews.com/2023/05/msi-data-breach-private-code-signing.html

quote:

Present in the leaked data are firmware image signing keys associated with 57 PCs and private signing keys for Intel Boot Guard used on 116 MSI products. The Boot Guard keys from MSI are believed to impact several device vendors, including Intel, Lenovo and Supermicro.

ijyt
Apr 10, 2012

In hindsight, not having had the funds to build my partner a new PC earlier in the year seems to have saved me a lot of headaches.

Dr. Video Games 0031
Jul 17, 2004

GN aren't the only ones to notice that Zen 4 performance benefits a lot from faster memory.

https://www.youtube.com/watch?v=MOatIQuQo3s

This is more memory scaling than what you see on Intel's 12th and 13th gen chips. You still see some benefit from using high-speed OC profiles with Intel, but it's mostly in the ~10% range as opposed to the 15 - 30% seen here. So you very much want to enable Expo, at the very least. If you're not enabling expo, then you really may as well have bought an Intel chip.

slidebite
Nov 6, 2005

Good egg
:colbert:

CBD Corndog posted:

Its too bad MSI just had a major data leak after their ransomware attack, making driver downloads from them dubious at best for a long time

https://thehackernews.com/2023/05/msi-data-breach-private-code-signing.html

Isn't that more a risk if you're not downloading from MSI themselves? Or is this some automated thing where you have no control where it pulls the file from?

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast

CBD Corndog posted:

Its too bad MSI just had a major data leak after their ransomware attack, making driver downloads from them dubious at best for a long time

https://thehackernews.com/2023/05/msi-data-breach-private-code-signing.html

I don't know what it has to do with downloading drivers from them - surely it would only be a problem if you download from some dodgy site where MSI's keys have been used to sign malware

Klyith
Aug 3, 2007

GBS Pledge Week

slidebite posted:

Isn't that more a risk if you're not downloading from MSI themselves? Or is this some automated thing where you have no control where it pulls the file from?

It's also a risk for potential persistent firmware rootkits -- malware that infects your PC could also install a malicious update to your bios. Now you have malware that's running at an even deeper level than the OS, is impossible to remove, and may be very difficult to even detect. The format & reinstall does nothing.

This stuff is currently difficult enough to execute that it's only state actors and other very sophisticated attackers doing it. So not exactly a huge concern for most people. I wouldn't throw away my MSI mobos over this. But it ain't great.

KodiakRS
Jul 11, 2012

:stonk:

CBD Corndog posted:

Its too bad MSI just had a major data leak after their ransomware attack, making driver downloads from them dubious at best for a long time

https://thehackernews.com/2023/05/msi-data-breach-private-code-signing.html

So as someone who has an MSI x670 motherboard being delivered this afternoon is this something I need to be concerned about? I was feeling pretty smug about choosing MSI over ASUS for my 7800x3d build but apparently I dodged out of the path of one bullet right into another.

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast

Klyith posted:

I wouldn't throw away my MSI mobos over this. But it ain't great.

As far as I could tell when the news hit, it had nothing to do with the mobos you can buy, but just laptops, and seemingly only Intel ones.

https://github.com/binarly-io/SupplyChainAttacks/blob/main/MSI/MsiImpactedDevices.md

I guess most people don't bother to read past the headline to find that list of affected devices. It's definitely poo poo, but we need to be clear about the scope of the issue

KodiakRS posted:

So as someone who has an MSI x670 motherboard being delivered this afternoon is this something I need to be concerned about? I was feeling pretty smug about choosing MSI over ASUS for my 7800x3d build but apparently I dodged out of the path of one bullet right into another.

no

HalloKitty fucked around with this message at 22:00 on May 11, 2023

Klyith
Aug 3, 2007

GBS Pledge Week
^^^ edit: oh lol it's not even some universal MSI keys. Though things are not necessarily safe just because it isn't "detected in the wild".

KodiakRS posted:

So as someone who has an MSI x670 motherboard being delivered this afternoon is this something I need to be concerned about? I was feeling pretty smug about choosing MSI over ASUS for my 7800x3d build but apparently I dodged out of the path of one bullet right into another.

Not really. Right now, if you aren't a target of a national government, this is not a big danger to you.

MSI will be able to remove the danger by putting out bios updates signed by the old key, but containing a *new* safe key. (How long this will take is a good question, right now they're not even admitting their keys got stolen.) So if you have a current gen mobo that's getting bios updates you're in good shape -- it'll be fixed long before average joe hacker can do anything bad.


OTOH, people with mobos that are out of service life, will they get updates? Maybe MSI will do the right thing and issue extra bios updates for old boards. Maybe not. So I have a x370 that's perfectly fine, will MSI update that?

Klyith fucked around with this message at 22:10 on May 11, 2023

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast

Klyith posted:

Not really. Right now, if you aren't a target of a national government, this is not a big danger to you.

MSI will be able to remove the danger by putting out bios updates signed by the old key, but containing a *new* safe key. (How long this will take is a good question, right now they're not even admitting their keys got stolen.) So if you have a current gen mobo that's getting bios updates you're in good shape -- it'll be fixed long before average joe hacker can do anything bad.


OTOH, people with mobos that are out of service life, will they get updates? Maybe MSI will do the right thing and issue extra bios updates for old boards. Maybe not. So I have a x370 that's perfectly fine, will MSI update that?

see my post, as far as we know it's just a number of intel based laptops that are affected, not all their products; unless you've seen something I haven't

Prescription Combs
Apr 20, 2005
   6

wargames posted:

I can't get expo to be stable with gigabyte b650 so i am below jdec speeds.

https://valid.x86.fr/ba510z

Did you happen to check vsoc before/after the update? Pre update mine was solid at 1.245v and post update its still 1.245v. I'm only running 2 sticks of ram instead of 4 on my b650 board though.

https://valid.x86.fr/u0r4zz

wargames
Mar 16, 2008

official yospos cat censor

Prescription Combs posted:

Did you happen to check vsoc before/after the update? Pre update mine was solid at 1.245v and post update its still 1.245v. I'm only running 2 sticks of ram instead of 4 on my b650 board though.

https://valid.x86.fr/u0r4zz

I didn't check before on the vsoc, but as buildzoid has pointed out that doesn't report correctly. I use to be able get 6000mhz/expo speeds with 4 sticks.

Inept
Jul 8, 2003

HalloKitty posted:

As far as I could tell when the news hit, it had nothing to do with the mobos you can buy, but just laptops, and seemingly only Intel ones.

That's just what's known about so far because it leaked on the dark web. It'll be a while before the full scale of the breach is known since they're probably having someone like Mandiant look at it as we speak.

Josh Lyman
May 24, 2009


I had a really frustrating 24 hour period yesterday. I updated my Asus B650E-F to the latest beta BIOS. I initially tried to do it with the BIOS file on my SSD but eventually had to load the file from a flash drive, not using BIOS flashback. I believe this worked fine.

Then I updated a couple mobo drivers using Armoury Crate and also installed the Realtek audio drivers, which also installed Sonic Suite software; the audio drivers that Windows Update installed had been working fine. I also installed Ryzen Master to try to run my Ryzen 7900X in Eco mode. I also moved my left monitor to the onboard video (my right monitor was already there) so as to leave only my middle monitor on my 1070 to lighten its load ahead of D4's final open beta this weekend (though my 4070 arrived today). At some point during this process, the poo poo hit the fan.

I got weird crashes where the bluescreen error was in a small window duplicated in a narrow strip across one of my monitors. It was also distorted, almost like there was a bad cable. I uninstalled the Realtek audio drivers. I undid Eco mode. I then got a Critical Process Died bluescreen on boot. Couldn't even boot into Safe Mode.

I ended up having to reinstall Win10 which isn't the end of the world, though I did lose some updates to my resume and Coachella streams I had encoded. Part of me wondered whether I should partition my SSD for OS vs storage, but it seems like I could run out of space on the OS partition within a year. I've been eyeing the Crucial P3 4TB for $200 as a storage drive; I tried using a 3TB HDD from my previous desktop but it was annoying dealing with it spinning up when waking from sleep. Really though, I should've just sync'd my document to OneDrive. I could've saved the Coachella streams to my external backup SSD but that never would've been in my regular workflow.

The reason I mention the BIOS thing is that when I first updated the BIOS about a month ago, I tried using flashback which meant I had to wipe my old USB 2.0 flash drives as I tried them one by one, one of which had GParted or some Linux-based LiveCD that might have let me access the documents and videos on my Windows install and copy them to another USB drive, though I'm not sure if I could've access them due to Windows password protection--does this only happen if you have BitLocker turned on? I don't think I did.

Adbot
ADBOT LOVES YOU

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast

Inept posted:

That's just what's known about so far because it leaked on the dark web. It'll be a while before the full scale of the breach is known since they're probably having someone like Mandiant look at it as we speak.

Fair, but we have to work with what we know, and not make people unduly worried

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply