|
shrike82 posted:Was thinking of using it for a multi GPU compute home setup. Stuffing 3-4x blowers into a case is tough to manage from a heat and power standpoint. Is that what we're calling Bitcoin mining rigs these days? =P edit: pagesnipe, spoke too soon.
|
#
?
Dec 2, 2019 18:29
|
|
|
|
| # ? Jun 5, 2024 04:55 |
|
|
MaxxBot posted:Yeah games can sometimes be the best stress test, I had a weird issue where my PC would shut off randomly during certain games. I suspected an issue with the PSU but it could run Prime95 Small FFT + Furmark simultaneously for days. It turned out it was the PSU and The Witcher 3 was a bigger stress test than Prime95 and Furmark were. orcane posted:Yeah I had plenty of attempts at (Intel) CPU OC and recent RAM OC with a Zen+ CPU that looked fine in Memtest and Prime95 but would reliably become crash happy in recent demanding games. As someone who runs Prime95 regularly, (more as a hobby/interest in the project itself, rather than OC stability testing), I just wanted to point out something, as far as I understand: Small FFT is great for stress testing CPU overclock because it can fit the whole FFT in cache, and the CPU can do more "work" per clock, saturating the AVX/FPU/ALU etc. with very little latency. On the other hand if you are actually wanting to test RAM OC stability, you're gonna want larger FFTs which don't fit in cache, and saturate your memory interface instead. So yeah, I don't know what a good FFT size would be for that considering AMDs ever increasing cache sizes, but maybe testing larger sizes would help finding unstable configs a bit faster/more reliably.
|
|
#
?
Dec 2, 2019 19:03
|
|
|
shrike82 posted:That's my current setup: 1950X TR + 4x 1080Ti hooked up by 16x riser cables. It's fine but the heat dispersion is terrible (every GPU at load is >10 degrees higher than when using one GPU) and I'm using a paperclip jury-rigged setup to kickstart 2x 1kw PSUs to power the system. I wouldn't mind paying to move to a case and shifting at least two GPUs to eGPU enclosures. What're we mining these days? 
|
|
#
?
Dec 2, 2019 19:30
|
|
|
Crunchy Black posted:Yeah Win 10 is amazingly resilient in that fashion compared to previous OSes. I've taken a w10 boot drive from Cascade Lake SMP to Skylake-S to Ryzen with no issues. It might take a few minutes to boot while it figures itself out but it won't poo poo a brick like previous versions of Windows might. drat, I figured we were still in sysprep generalize hell with a huge list of drivers hoping that the right one gets detected so it can boot
|
|
#
?
Dec 2, 2019 19:58
|
|
|
If it makes you feel better, my boot loader was on a different drive than the system drive for some reason, so I got to spend an hour and a half giving out how to rebuild that on the correct drive and lost my license activation in the process. Also, it was an upgrade license and I don't have the original any more, which meant I had to buy a new windows license for the first time in 12+ years.
|
|
#
?
Dec 2, 2019 20:38
|
|
|
My initial reason for installing Linux on this machine was so I could have an OS that could go from Intel to AMD without rebooting. Windows might be getting better, but it isn't monolithic so [ascii shrug] Besides which, even if the kernel can adapt and fetch drivers, undoubtedly the registry has been made a horrible place.
|
|
#
?
Dec 2, 2019 21:09
|
|
|
I mean, it's never a bad idea to do a fresh install.
|
|
#
?
Dec 2, 2019 21:15
|
|
|
While we're on this subject. Anyone switch to an AMD system from Intel with a Windows 10/Linux duel boot? Just plug the drives in and it'll be good?
|
|
#
?
Dec 2, 2019 23:14
|
|
|
OhFunny posted:While we're on this subject. Anyone switch to an AMD system from Intel with a Windows 10/Linux duel boot? Just plug the drives in and it'll be good? Probably not, they're going to shoot each other
|
|
#
?
Dec 2, 2019 23:15
|
|
|
Shouldn't make any difference, both will boot just fine.
|
|
#
?
Dec 3, 2019 00:00
|
|
|
BangersInMyKnickers posted:Probably not, they're going to shoot each other i guess burr/hamilton is basically amd/intel tbh
|
|
#
?
Dec 3, 2019 00:14
|
|
|
BangersInMyKnickers posted:Probably not, they're going to shoot each other  
|
|
#
?
Dec 3, 2019 01:01
|
|
|
ItBreathes posted:I mean, it's never a bad idea to do a fresh install. the next feature update you do is basically a fresh install for what matters. not really worth it in win10. feature updates fix all sorts of hosed up poo poo that goes wrong and doing them at a regular interval (even if you stay 45 days behind or whatever) basically keeps windows 10 running great forever. the boot drive in my now retired gaming PC started its life as a win7 drive, got upgraded to 8, upgraded to 8.1, shut down for a year, upgraded to 10, works great to this day. one of those 120gb SSDs from when they were a fortune but worth it. never reinstalled/wiped. bus hustler fucked around with this message at 01:07 on Dec 3, 2019 |
|
#
?
Dec 3, 2019 01:05
|
|
|
Tfw u nod ur head approvingly at a hard drive for being a good boy
|
|
#
?
Dec 3, 2019 01:10
|
|
|
ItBreathes posted:I mean, it's never a bad idea to do a fresh install. of course not. Doesn't mean for dev environments its not a possibility. No one is suggesting you *should* do it, just that you probably can.
|
|
#
?
Dec 3, 2019 01:14
|
|
|
Statutory Ape posted:Tfw u nod ur head approvingly at a hard drive for being a good boy Write a thank you note, and save it.
|
|
#
?
Dec 3, 2019 01:15
|
|
|
shrike82 posted:I see setups with 4x 1080 Tis slotted right next to each other in a case and I wonder how the owners manage the heat.
|
|
#
?
Dec 3, 2019 07:09
|
|
|
Mr Shiny Pants posted:I don't know the specific TI's you have, but mine sucks air from the front and blows it out of the back instead of sucking air from the bottom, so theoretically it should be ok. yeah, this is one of the few use cases for blower designs
|
|
#
?
Dec 3, 2019 07:10
|
|
|
One thing to mention about reinstalling Windows is that some OEMs create a partition on the drive with a full image of "their" install of Windows 10. I freed up 15GB on the last dell laptop I bought by deleting that since it wasn't needed. You don't HAVE to reinstall Windows to get rid of this, you can always delete the partition and expand the C drive, but Windows 10 installs so fast on an SSD I just go ahead and re-do the OS when I first get a laptop anyway.
|
|
#
?
Dec 3, 2019 14:43
|
|
|
CFox posted:One thing to mention about reinstalling Windows is that some OEMs create a partition on the drive with a full image of "their" install of Windows 10. I freed up 15GB on the last dell laptop I bought by deleting that since it wasn't needed. You don't HAVE to reinstall Windows to get rid of this, you can always delete the partition and expand the C drive, but Windows 10 installs so fast on an SSD I just go ahead and re-do the OS when I first get a laptop anyway. this is good advice if you are one of the people that  in horror when you click disk management
|
|
#
?
Dec 3, 2019 15:37
|
|
|
I always secure erase the drive and install the latest version of Windows fresh with any laptop I buy. Don't want to run any OEM software.
|
|
#
?
Dec 3, 2019 16:07
|
|
|
You can't cut back on McAfee Security Manager Trial, you will regret this.
|
|
#
?
Dec 3, 2019 16:18
|
|
|
Lambert posted:I always secure erase the drive and install the latest version of Windows fresh with any laptop I buy. Don't want to run any OEM software. Doesn't necessarily fix it. UEFI has a feature where it can store an executable which will automatically be run and installed by windows. Either the crapware itself or a loader that pulls the crapware down. It's intended for things like driver packages but there's nothing stopping manufacturers from stuffing crapware in there, and they have in the past (eg Superfish). https://www.howtogeek.com/226308/the-windows-platform-binary-table-why-crapware-can-come-back-after-a-clean-install/
|
|
#
?
Dec 3, 2019 18:04
|
|
|
I have an MSI x570-A PRO, running the September BIOS that updated it to 1.0.0.3abba. There's a newer one with "1.0.0.4 Patch B (SMU v46.54)", do I need and/or want this? I don't want to gently caress with any beta BIOS or drivers, system is stable and not giving me any problems, but I'd upgrade if there's a significant performance increase.
|
|
#
?
Dec 3, 2019 20:47
|
|
|
Azuren posted:I have an MSI x570-A PRO, running the September BIOS that updated it to 1.0.0.3abba. There's a newer one with "1.0.0.4 Patch B (SMU v46.54)", do I need and/or want this? I don't want to gently caress with any beta BIOS or drivers, system is stable and not giving me any problems, but I'd upgrade if there's a significant performance increase. You do want it. But there's an even newer version available for your mainboard, H6, that includes the newer microcode as well.
|
|
#
?
Dec 3, 2019 20:55
|
|
|
KillHour posted:If it makes you feel better, my boot loader was on a different drive than the system drive for some reason, so I got to spend an hour and a half giving out how to rebuild that on the correct drive and lost my license activation in the process. Also, it was an upgrade license and I don't have the original any more, which meant I had to buy a new windows license for the first time in 12+ years. That's weird - I had to reinstall Windows 10 on someone's laptop the other day that had been pretty obviously updated from 7 and it was perfectly happy to go without a new license.
|
|
#
?
Dec 3, 2019 22:54
|
|
|
It will, but it won't let you access some settings and it will bark at you periodically.
|
|
#
?
Dec 3, 2019 23:27
|
|
|
Munkeymon posted:That's weird - I had to reinstall Windows 10 on someone's laptop the other day that had been pretty obviously updated from 7 and it was perfectly happy to go without a new license. A lot of laptop bioses have a certificate in them that Windows will read to active and Windows 10 will usually activate with a Windows 7 cert.
|
|
#
?
Dec 3, 2019 23:38
|
|
|
Lowen SoDium posted:A lot of laptop bioses have a certificate in them that Windows will read to active and Windows 10 will usually activate with a Windows 7 cert. I also helped with one from some white label OEM that I had helped swap out the motherboard on in February. I thought they used some kind of device fingerprinting now.
|
|
#
?
Dec 4, 2019 00:56
|
|
|
Munkeymon posted:That's weird - I had to reinstall Windows 10 on someone's laptop the other day that had been pretty obviously updated from 7 and it was perfectly happy to go without a new license. It either used the key stored in the UEFI or the hash value stored on Microsoft's servers As soon as you activate Windows 10, a hash value of your system will be stored on Microsoft's servers wich will be used to activate automatically even without user intervention (even without a key) after reinstallation.
|
|
#
?
Dec 4, 2019 01:15
|
|
|
I have an MSI x570 Gaming Plus board and tried updated to the A60 firmware, and almost immediately Firefox and Chrome tabs crashed constantly while Windows decided I have only Security Intelligence version 1.0.0.0 installed somehow (current version is 1.305....) Reverting back to A50 made all the issues go away. What the hell could have caused that? All I did was update using the downloaded file from the MSI website over USB/M-Flash and then turn XMP back on. Changes in XMP causing ram issues maybe? I don't know a whole lot about this kind of stuff.
|
|
#
?
Dec 5, 2019 03:11
|
|
|
Did you happen to have the 1909 update fire on the same reboot that you updated bios? That's the only thing I can think of that would cause the software rollback.
|
|
#
?
Dec 5, 2019 14:52
|
|
|
No, I got that update last week. I don't think it actually rolled back, because when I reverted to A50 the Security Intelligence didn't have to update itself back or anything, it just recognized again that it was the proper version. The tab crashing I can see being ram related but how the hell Windows became convinced Defender was out of date doesn't make any sense to me.
|
|
#
?
Dec 5, 2019 15:51
|
|
|
If you were having ram stability issues and bitflips then all bets are off. Something could have been corrupted as the definitions were loaded to memory from disk and make it go wonky. Big question is if that behavior is consistent through reboots.
|
|
#
?
Dec 5, 2019 16:22
|
|
|
Well the first boot blue screened with a critical process failure, then the second had all the tab crashing issues. That's when I decided to revert instead of trying a 3rd time.
|
|
#
?
Dec 5, 2019 16:43
|
|
|
Anyone else with the same issue on MSI support forums? Is it passing memory tests? GRINDCORE MEGGIDO fucked around with this message at 01:07 on Dec 6, 2019 |
|
#
?
Dec 6, 2019 01:05
|
|
|
Not that I saw. It passes on the A50 so I'm just going to skip A60 for now, it doesn't seem especially important.
|
|
#
?
Dec 6, 2019 01:32
|
|
|
New 3900x build, using ryzen master it is currently displaying an average core voltage of ~1.48 despite the fact that I have peak core voltage set at 1.125 on the auto overclocking screen. Am i reading this wrong? Is it borked? Anybody have any handy tips?
|
|
#
?
Dec 12, 2019 07:47
|
|
|
DonkeyHotay posted:New 3900x build, using ryzen master it is currently displaying an average core voltage of ~1.48 despite the fact that I have peak core voltage set at 1.125 on the auto overclocking screen. Am i reading this wrong? Is it borked? Anybody have any handy tips? It is normal. The 3900x can bounce up to 1.5v while boosting for single core stuff. It often will switch between boost voltage and sleep as frequently as every 1ms, which can be outside the ability of monitoring software to detect, leading to it looking like the chip is stuck at 1.4v or whatever even though most of the cores are asleep. Try CPUZ and see if it gives you a different result. Your light load single core voltages will usually always be high. Under heavy load with multiple cores voltages should drop. But this is a good read, particularly WRT to the CPU ignoring a lot of your configuration options. https://community.amd.com/thread/244684 NickBlasta fucked around with this message at 13:25 on Dec 12, 2019 |
|
#
?
Dec 12, 2019 13:05
|
|
|
|
| # ? Jun 5, 2024 04:55 |
|
|
Make sure you're on the latest BIOS for your board. Early AGESA versions for Ryzen 3000 were really aggressive about boosting the voltage even when the CPU wasn't highly loaded and newer ones are more chill.
|
|
#
?
Dec 12, 2019 15:54
|
|
