|
Klyith posted:MSI somehow neither bad nor stupid, what????? kliras posted:as said before, be ... extra careful with downloading bios updates, especially for msi motherboards there we go, that's the MSI I know. lmao.
|
#
?
May 6, 2023 15:04
|
|
|
|
| # ? May 30, 2024 06:10 |
|
|
Enos Cabell posted:New Gigabyte F8d bios today, but still on AGESA 1.0.0.6. It does explicitly mention the SOC voltage changes though. I've got everything fully safe and stable now with my current gigabyte bios, so I really don't think I'm updating until all the beta versions and poo poo are over and people have had time to thoroughly test these bioses. Kazinsal posted:as if this platform wasn't loving cursed enough That "be careful about how you get your BIOSes" warning goes for both platforms. Intel's in double trouble actually because of this bit of fun: https://videocardz.com/newz/intel-bootguard-keys-leaked-through-msi-data-breach
|
|
#
?
May 6, 2023 15:08
|
|
|
yeah i stopped using asus' online updater in bios after the whole thing was compromised a few years ago. usb flashback or bust
|
|
#
?
May 6, 2023 15:18
|
|
|
I'm way late with this question, but can I get some guidance on acceptable Zen 3 temperatures? Is 85C fine? I'm not seeing any throttling, it maintains the 4.45 GHz boost speeds at those temps. I did a dirt cheap bargain bin build late last year, with a $130 5600G in a 2L Asrock Deskmini X300 case. Here's the kicker: The stock lovely 5600G cooler is too big for the X300 case, so I'm using the cooler that came with the X300, that looks like something out of 1998. I'm running Ubuntu 22.04 on it, and I recall that there may be some positive offset in the numbers reported to lm-sensors and the BIOS, to help get bad fan curves up to speed faster. I'm pretty sure that it must be actual numbers though, because I'm seeing idle temperatures get drat near room temperature, mid 20s and low 30s C. My load temps are 75-85C, where the fan spins up to 100% finally and catches up. I'm looking specifically at "Tctl" temperature. I have PBO disabled, because when I first used it with PBO default on it was kicking itself up to 85C all the time. I've tried to reduce the PPT to 45W to further reduce temps and noise, but I've seen some users reporting that setting PPT Is broken on this Asrock X300 board, and I don't have a great way to check it from Ubuntu without installing kernel extensions. Anyway, I'm only asking because I've found myself using this machine differently than I expected, and it's spending a lot of time at 85C. Fine and dandy for a 5600G, or is it time to think about a more aggressive fan curve or finding some way to lower PPT from within the OS?
|
|
#
?
May 6, 2023 16:03
|
|
|
I feel like they talked about that msi hack on nexus at least a month ago, how msi wasn't going to pay the ransomware and that ppl should be wary of downloading msi files from 3rd party websites.Klyith posted:If you are annoyed by windows installing drivers you didn't ask for, you can fix that with settings -> search for "device installation" -> change to No. I couldn't remember for the life of me how to turn this off or what key words to search for, so you're my hero for...at least the next 90 minutes. runaway dog fucked around with this message at 16:49 on May 6, 2023 |
|
#
?
May 6, 2023 16:06
|
|
|
edit: quote is not edit
runaway dog fucked around with this message at 16:49 on May 6, 2023 |
|
#
?
May 6, 2023 16:14
|
|
|
Twerk from Home posted:I'm way late with this question, but can I get some guidance on acceptable Zen 3 temperatures? Is 85C fine? I'm not seeing any throttling, it maintains the 4.45 GHz boost speeds at those temps. It's fine for CPU lifespan. You probably are losing a tiny sliver of performance, but not worth caring about. Hearing damage from the that tiny fan screaming away is maybe the only concern.
|
|
#
?
May 6, 2023 17:01
|
|
|
Twerk from Home posted:I'm way late with this question, but can I get some guidance on acceptable Zen 3 temperatures? Is 85C fine? I'm not seeing any throttling, it maintains the 4.45 GHz boost speeds at those temps. The 5600G isn't hitting it's Tjmax, which is 95c, so you do have some headroom and the cooler the system came with is doing it's job. That said, there are a few things you could try. The Noctua NH-L9i is a low profile cooler that gets great reviews across the board. Combine that with a good thermal paste (like Thermal Grizzly Kryonaut) could drop your temps down 5-10c . I found a video of the Deskmini X300. It's just this side of a NUC isn't it? Worse case you could try a small desk fan blowing at the side of the computer to add some air-flow, since it looks like the only fan in the thing is on the CPU Cooler. You might want to try and adjust the fan hysteresis a bit, make the fan ramp up a bit faster to adjust to the temperature swings (should be a setting in the BIOS or even ASRock's software). With the stock fan you will probably get some more noise, with the Noctua you might not even notice.
|
|
#
?
May 6, 2023 17:08
|
|
|
So, I ran a test with the new case fans in - peak CPU temps went down from 97° to 92°. Not being quite satisfied with that, I pulled off the 110R's front panel yesterday to see what that would do. CPU temps peaked at 80.3° from there on. Well there's your problem... The panel's staying off for now, when I find the time I'll be taking a jigsaw to it, maybe aim for an insert from thin black fabric. Might be best to just fab a full-size frame around the front face from veneer (to cover and wrap around in thin black fabric). Klyith posted:Arctic Freezers are fine. In the US they're more expensive than other budget solutions, but where you are they might be cheaper -- arctic are based in germany. Thanks! If the above doesn't ultimately result in something I'm happy with, It'll be Freezer time. Duuk fucked around with this message at 19:14 on May 6, 2023 |
|
#
?
May 6, 2023 19:05
|
|
|
new gigglebite bios allows me to reach good memory speeds again
|
|
#
?
May 6, 2023 21:13
|
|
|
Koskun posted:The 5600G isn't hitting it's Tjmax, which is 95c, so you do have some headroom and the cooler the system came with is doing it's job. https://www.amazon.com/Noctua-NH-L9a-AM4-Premium-Low-Profile-Cooler/dp/B09NRL9SCJ (there's an AM5 version but that one requires more dexterity)
|
|
#
?
May 7, 2023 03:13
|
|
|
what's the best cheap B560 in terms of DDR5 OCing potential
|
|
#
?
May 7, 2023 06:30
|
|
|
Palladium posted:what's the best cheap B560 in terms of DDR5 OCing potential Assuming you mean B650, the norm is 6000 and the ceiling for golden Ryzen 7000 samples is 6400. Almost any board can support this, so it's basically down to CPU lottery. The ASRock B650M-HDV will support at least 6000 according to HUB's review (they didn't test 6400). It's $125 in the US. I'd avoid all the A620 boards, they seem to cut way too many corners to end up at sub-$100 price points. https://www.newegg.com/asrock-b650m-hdv-m-2/p/N82E16813162114
|
|
#
?
May 7, 2023 11:51
|
|
|
surely can't be worse than the status quo https://twitter.com/anandtech/status/1654598178505728007
|
|
#
?
May 7, 2023 15:37
|
|
|
kliras posted:surely can't be worse than the status quo Ah yes. Just like that time AMD was going to push coreboot.
|
|
#
?
May 7, 2023 15:59
|
|
|
Klyith posted:It's fine for CPU lifespan. Koskun posted:The 5600G isn't hitting it's Tjmax, which is 95c, so you do have some headroom and the cooler the system came with is doing it's job. Thank you to both of you! None of this was a surprise to me, I knew what I was getting into with a setup this tiny. I'm going to make the fan curve a little more aggressive on the stock heatsink, and think about that Noctua if it bothers me. I'm really happy with this system overall. $600 all in for a 5600G, 64GB DDR4-3200, 2TB Intel NVMe disk. It's only hitting 85C under sustained AVX2 all-core loads, but it turns out that Scikit-Learn scales terribly with threads and this little $130 AMD CPU trains models faster than a 1 generation old 18 core Xeon.
|
|
#
?
May 7, 2023 16:02
|
|
|
kliras posted:surely can't be worse than the status quo so they're on track to release before the asus 1410 bios comes out of beta.
|
|
#
?
May 7, 2023 17:02
|
|
|
kliras posted:surely can't be worse than the status quo i think the amd open source programers do a super good job with the gpu driver stack so them doing stuff on the bios side, hell yeah.
|
|
#
?
May 7, 2023 17:11
|
|
|
wargames posted:i think the amd open source programers do a super good job with the gpu driver stack so them doing stuff on the bios side, hell yeah. Are they the same people? I would expect that this will be “BIOS programmers working on something open source” and not some pool of existing open source developers that they will retask on this.
|
|
#
?
May 7, 2023 18:29
|
|
|
Subjunctive posted:Are they the same people? I would expect that this will be “BIOS programmers working on something open source” and not some pool of existing open source developers that they will retask on this. Yeah, I suspect there is little domain knowledge overlap in those two areas. Unless you just happen to have a tiny number of people on staff who are interested in both. This is already out there for Genoa EPYC though, so it's not being announced as vaporware*. Whoever is working on it, is already working on it. *Do not misconstrue the factual statement that opensil is not vaporware as a forward-looking statement of belief that AMD's timeline and future goals will come to fruition.
|
|
#
?
May 7, 2023 20:21
|
|
|
I’m going to end up having recurring nightmares that “open bios” somehow becomes giving up completely and x86 becoming the typical no UEFI ARM board
|
|
#
?
May 7, 2023 23:44
|
|
|
LinuxBIOS went through this in the early 2000s, surely we’ve figured it out by now!
|
|
#
?
May 8, 2023 00:11
|
|
|
hobbesmaster posted:I’m going to end up having recurring nightmares that “open bios” somehow becomes giving up completely and x86 becoming the typical no UEFI ARM board
|
|
#
?
May 8, 2023 08:31
|
|
|
Scuttlebutt is that AGESA 1.0.0.7 fucks with memory speeds, dropping things to DDR5-4400. Not sure if it’s on profiles where things are kinda automagic, or if it overrides entirely manual settings, too. AMD’s ostensibly working on v1.0.0.9.
|
|
#
?
May 8, 2023 10:54
|
|
|
Oh no, you can't use 192GB of memory at 6000MT/s: https://twitter.com/g01d3nm4ng0/status/1654080860201598979 Nevermind that there aren't a whole lot of 48GB DIMMs on the market, or that you'd need to use four of them which is known to have a bunch of other issues. BlankSystemDaemon fucked around with this message at 11:09 on May 8, 2023 |
|
#
?
May 8, 2023 11:05
|
|
|
PROCHOT not in 1.0.0.7? That’d be kinda pointless.
|
|
#
?
May 8, 2023 11:59
|
|
|
Combat Pretzel posted:PROCHOT not in 1.0.0.7? That’d be kinda pointless. This damage to the SMU is what then lets the CPU pull 37A from the EPS12V, meaning that some 440W of thermal energy needs to be dissipated by a cooler where the CPU is incapable of asserting PROCHOT - because the firmware responsible for it has already given up the ghost (otherwise it wouldn't be pulling 37A from a 12V rail). I'm not really sure whether it matters if it's implemented or not, and I doubt that it's the only form of overheating protection that the CPU has.
|
|
#
?
May 8, 2023 13:42
|
|
|
It's going to be hilarious when the usual tubers crap their panties over the loss of memory speeds and prolly overclocking.
|
|
#
?
May 8, 2023 14:36
|
|
|
How on earth was a 4x48GB config hitting 6000 MT/s before? That doesn't seem possible given what we already know about 2r2dpc performance on Zen 4.
|
|
#
?
May 8, 2023 14:54
|
|
|
Dr. Video Games 0031 posted:How on earth was a 4x48GB config hitting 6000 MT/s before? That doesn't seem possible given what we already know about 2r2dpc performance on Zen 4. More voltage solves all problems except too much voltage, right? I remember seeing insanely high speeds QVLed on AM4 boards running like 1.65V through the RAM (and memory controller).
|
|
#
?
May 8, 2023 14:56
|
|
|
redeyes posted:It's going to be hilarious when the usual tubers crap their panties over the loss of memory speeds and prolly overclocking. Uh, losing that much performance months after launch on account of AMDs own gently caress up is pretty bad.
|
|
#
?
May 8, 2023 14:56
|
|
|
Cygni posted:Uh, losing that much performance months after launch on account of AMDs own gently caress up is pretty bad. Maybe someone found four separate 48GB DIMMs capable of running at 4800MT/s and somehow managed to overvolt them enough to run at 6000MT/s by using an open bench setup with a couple of high static pressure server fans pointing directly at the DIMMs, but I have my doubts. BlankSystemDaemon fucked around with this message at 15:29 on May 8, 2023 |
|
#
?
May 8, 2023 15:25
|
|
|
BlankSystemDaemon posted:I don't think we can know anything about whether the systems with 4x48GB memory at 6000MT/s were running stable at any point, considering that these sets of DIMMs haven't even hit the market. The 48GB DIMMs have been on the market since February. https://www.corsair.com/us/en/Categories/Products/Memory/VENGEANCE-DDR5-Memory---Black/p/CMK48GX5M2B5200C38 I posted ITT about the 48gig sticks getting 6000 speeds last month. There are lots of people on Reddit that claim it works fine out of the box with EXPO voltages. Even if it only ends up impacting that small group, it still sucks.
|
|
#
?
May 8, 2023 15:41
|
|
|
Cygni posted:The 48GB DIMMs have been on the market since February. You accidentally linked the 2x24 kit. Assuming that is single rank; I would guess that 4 single rank will hit 6000 with fiddling but 4 dual rank will not come anywhere close. Or double those numbers because each ddr5 dimm is kinda like two ranks right, but w/e.
|
|
#
?
May 8, 2023 16:41
|
|
|
hobbesmaster posted:You accidentally linked the 2x24 kit. Assuming that is single rank; I would guess that 4 single rank will hit 6000 with fiddling but 4 dual rank will not come anywhere close. Redditors were reporting 5200 with 4x48 with just Expo. I personally haven’t seen anyone getting 6000 with that config though, just the 2x48.
|
|
#
?
May 8, 2023 16:52
|
|
|
I previously set a -25 offset through the crazy Ryzen Master and just now I opened it up again to check on it. I ran “validate offset” and it did its little toy stress test, and then when it rebooted after I pressed “apply”, well, it didn’t reboot. Nothing on the console after 10 mins, and I tried a hard reboot as well. I guess I need to reset the CMOS and try again? I had just updated to the ASUS 1410 beta earlier this morning and then turned on Expo but otherwise left at defaults.
|
|
#
?
May 8, 2023 17:30
|
|
|
Cygni posted:The 48GB DIMMs have been on the market since February. Also, I'm not sure I'd trust redditors with anything.
|
|
#
?
May 8, 2023 17:34
|
|
|
Subjunctive posted:I previously set a -25 offset through the crazy Ryzen Master and just now I opened it up again to check on it. I ran “validate offset” and it did its little toy stress test, and then when it rebooted after I pressed “apply”, well, it didn’t reboot. Nothing on the console after 10 mins, and I tried a hard reboot as well.
|
|
#
?
May 8, 2023 21:18
|
|
|
Combat Pretzel posted:Scuttlebutt is that AGESA 1.0.0.7 fucks with memory speeds, dropping things to DDR5-4400. Not sure if it’s on profiles where things are kinda automagic, or if it overrides entirely manual settings, too. AMD’s ostensibly working on v1.0.0.9. I rather doubt 192GB was stable at DDR5-6000, and if it ever was it would require absolutely titanic VSOC voltage which is exactly the thing that was killing chips
|
|
#
?
May 8, 2023 22:25
|
|
|
|
| # ? May 30, 2024 06:10 |
|
|
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.
|
|
#
?
May 9, 2023 01:01
|
|
