PC Graphics

[MintChipper]

Hey I got an issue, my Gigabyte RX 580 8gb in my new pc has been acting up crashing in the middle of games/some editing software. Usually just the application crashes but sometimes the whole computer blue screens and restarts because of it.
I've already tried:
-turning off hardware accel in certain background apps
-checking memory
-scanning for bad files
-updating the radeon drivers
-uninstalling then reinstalling radeon drivers with auto-detect(have only tried windows' uninstalling method not DDU or AMD cleanup utility)

What do you think is going on? I've seen an older post on here that solved a similar problem saying their VRAM voltage was too low but they didn't explain how exactly they changed it.

also worth noting that all of the blue screen error codes I've been getting so far are somewhat display driver related, when I uninstalled and reinstalled before there were no crashes for about 24 hours and thought I was good but then I got another game crash so I'm back to square one. My power supply is bronze 600w and using high performance plan btw.

Any help is appreciated!

[mackbolan777]

Full system specs please. Aside from that, try the card in another PC. Try a different PSU, they go bad all the time and that would be a symptom. Yes, use DDU and reinstall the driver by downloading it first. Try using the 20.8.3 or 20.9.2 vs. the newer drivers. Those older versions are more compatible for your card. 

DDU: Display Driver Uninstaller Download version 18.0.3.6 (guru3d.com)

Driver: Radeon™ RX 580 Previous Drivers | AMD

[MintChipper]

Thanks for the reply! I'll try those older drivers you recommended sometime today. 
Here's my specs: 

CPU: AMD Ryzen 7 2700X 3.7 GHz 8-Core Processor
Motherboard: ASRock B450M Steel Legend Micro ATX AM4 Motherboard 
Memory: G.Skill Ripjaws V Series 16 GB (2 x 8 GB) DDR4-3200 CL16 Memory
Storage: Samsung 970 Evo Plus 1 TB M.2-2280 NVME Solid State Drive 
Video Card: Gigabyte Radeon RX 580 8 GB GAMING 8G Video Card
Case: Fractal Design Focus G ATX Mid Tower Case 
Power Supply: EVGA BQ 600 W 80+ Bronze Certified Semi-modular ATX Power Supply

[mackbolan777]

Power supply could be shady since it's a bronze and I'd go 650W or higher in gold for future reference. Ignore PSU calculators, they're only a guide. Quite a few complaints for that model starting in 2019 on Reddit. If yours is a 2017 I can pretty much tell you it's done.

[MintChipper]

I don't know how I would identify that mine is from 2017 but here's the page i bought it from. I'm gonna try the driver thing before trying anything physical.

[mackbolan777]

[PSU] EVGA BQ 80 Plus Bronze, Semi Modular, 600W Power Supply - $39.99 (after $15.00 MIR) + FS with ...

I didn't want to go there but... there's some of the complaints. It's more a matter of when you bought it. I was thinking it could be 2017ish because of the rest of the build. Any bronze rated PSU is PU, if you catch my drift. Never go cheap on the PSU, it's heart of the PC.

[MintChipper]

sorry it was a miscommunication I put "new pc" in the OP to try to communicate that I made this build around 4 weeks ago lol

[MintChipper]

I tried playing Devil May Cry 5 with the 20.9.2 drivers, I got 1 crash at first because I had the discord overlay on and forgot the game doesn't like that, then I got another crash about 10 minutes later, I tried running the game in DirectX 11 and the game ran fine and I played until I got bored and closed it.

So I have no clue if it's actually working now but I'll update if it keeps working or not. How did you know if these driver versions were friendlier to my card?

[mackbolan777]

Complaints from others and a problem I had with my RX 5600 XT I was running, giving a similar issue. About 4 months of research and stumbled on a video or two that placed the issue squarely on those 2 drivers being stable. Then finding out that AMD marked the 5000 series GPU's EOL (End Of Life) and the fact no new fixes are in any driver made past 20.9.2 for the 5000 series cards. Your card is based off the Polaris GPU line, however the newer Navi GPU driver is made the same and this idea has worked for others that have an RX 400-500 series card, as well as the 5000 series. 

Something of interest: RE Engine | Resident Evil Wiki | Fandom

That's the game engine and since it uses PAK files, it's much like doing a lot of file extraction, much like "WinZip" at high speed. This makes the game hugely CPU intensive but in your card's case, your GPU can handle those calculations relatively well, since you card was also one of the better "mining" cards. 

I would recommend just running in DX11 mode if that works well as DX12 is going to use way more VRAM and tax the CPU harder. I'd also stick with that card, unless you move up to the 500/570 chipset and a 5000 series CPU or your FPS will really suffer. You won't get much benefit from trying to OC into DX12 either, it may even cause more issues.

Discord has caused crashes on my PC but that was before doing a clean install of Windows 20H2 x64 PRO vs. Home. I also went to the RX 6800, so I'm not sure exactly what stopped the issue. Another thing many are missing is that today's games run much better on SSD's. Use the spin drives for backup's.

[MintChipper]

Thank you for the extra info, I was going back to this page because the game was crashing again today and I just did a clean to reinstall the other driver you recommended, but out of suspicion I went and checked the config again and it reverted itself back to DX12 lol, so maybe the issue's fixed and I have to go check the config every time I feel like playing and go try some other demanding games to see if it crashes there. 

[blackgear25]

what your gpu temperature ?

[mackbolan777]

You can alter the "config.ini" file to make the game always use DX11. Here's a video with someone with you nearly identical specs running the game both ways: Devil May Cry 5 | DX 11 vs DX 12 | How to switch to DX11 - YouTube

As you can see, very little visual difference but there is a 10-15 FPS to the better running in DX11, so DX12 isn't worth the effort. DX12 uses much more VRAM as well. Also you can see that the CPU is hovering near 70% at times, like I said previously a CPU bottleneck. 

GPU temps as mentioned by another OP, could be an issue if you're at 90c, optimally you want under 80c. Right on 80c is "ok" but a bit hot. Like the video shows the GPU/VRAM sticking to the 65c area which is fine. Check your cooling situation out, like use a side fan if temps are higher and your case has that option. Space out any PCIe cards so the card can get air. Manage wires better to not restrict front airflow if they are in the way at all. 

See if other games are affected. If the card is staying in good thermal range and the CPU as well, you might want to RMA the card. Looking at that video your system should be able to handle the game on DX12 but your FPS would be lower by ~10-15 FPS.

If the card is a "blower" card meaning one squirrel cage fan, it could be getting hot. You might need to get creative getting more air to it and also change the fan curve to be more aggressive. Some cases allow for a bottom mounted fan and that would be ideal if it has a fairly unobstructed path to blow air to the card. You could experiment by running the PC with the side cover off if you don't have any fans lying around. Again, this is only if your card is at 80c or above. 90c is the max but issues can happen as one gets closer to that threshold.

[MintChipper]

I was playing Jedi Fallen Order last night which tends to get pretty hot and it crashed after about 40 minutes but I wasn't looking at my GPU temps during that, playing more Devil May Cry today I was checking temps and the most it got to was 80c sticking at or below 70c so I don't think GPU temp is the issue, but I got another blue screen unfortunately, this time DRIVER_IRQL_NOT_LESS_OR_EQUAL with dxgmms2.sys which I figured out might've been me running it in DX12 again, in most blue screen reports I get the files shown are either amdkmag.sys, dxgmms2.sys and ntoskrnl.exe, if that helps at all. 

[mackbolan777]

I found a post that might help describing a registry edit. You need to open the registry editor by pressing Win+R and typing "regedit" or "start", "run" then type "regedit" in the box. Follow the instructions below.

HKEY_LOCAL_MACHINE\SYSTEM\CURRENTCONTROLSET\CONTROL\GRAPHICSDRIVERS

Then, right-tap on the right-side panel and tap on New > DWORD (32-bit) Value.
Now set the name of this latest created DWORD as TdrDelay.
Double-tap on the newly created DWORD and then set its value as 10. However, it’ll set the response time for your GPU to be 5 to 10 Seconds, modifying it from 2 seconds which is the default.
Then leave the Registry Editor and then reboot your PC for the modifications to take effect.

Other options would include reinstalling the OS fresh from USB made from the Windows Media Creation Tool, try to update Direct X (never had luck with that), try driver 20.8.3 but use DDU in safe mode. PSU could be defective. Retest the RAM using: MemTest Manual (hcidesign.com) to at least 800% coverage or overnight. Final thought is a defective GPU or CPU in that order.

[MintChipper]

I'm starting to think reinstalling windows might be what I need to fix it, because I installed windows 10 on this using a USB with a setup resource from 2018 because I didn't have access to the creation tool and updated from there which might have messed with it somehow. 

[mackbolan777]

It could because it would add a ton of useless updates to the useful ones that can corrupt things. Also, 20H2 is an entirely different OS build, meaning not much was carried over to the new build.

[MintChipper]

Hey I'm back, sorry for not updating for a while. I reinstalled windows and was good for a few days then I got crashes again, so I did the standard routine I'm used to at this point trying different drivers, even the newer versions, so far no luck. I was beginning to wonder if there was something I was missing when setting up this PC cause I'm new to this in general and realized I didn't have AMD chipset drivers so I went and downloaded those, the PC seems to use the SSD faster now but the crashes are still around,

I admittedly still don't want to accept that I could have to RMA or replace something just yet so I'm still looking for software solutions, here's my motherboard's driver page, I haven't downloaded anything there besides the AMD chipset so if you think one of them might be something I need please tell me.

I also want to bring attention again to this post I had put in the OP, if someone can help me find out how to check my VRAM voltage to see if that is the issue I'd really like that because I don't have a great understanding of the software.

If at this point you do think it's just my hardware is there anything I can do to check whether it's my PSU or GPU or CPU specifically? I understand the PSU isn't ideal already but I'd rather make sure which exact part is the issue here before I wait on a new or replaced part only to find out that wasn't the problem.

Sorry for the wall of text, I just want to cover all bases I could think of, thank you for still checking up on the thread.

[delevero]

Try this.

1). go into bios  ( i recommend to load default factory settings. )
2). press F2 and open advanced mode
3). Somewhere on that page ( depend on manufacturer if its on this page or other )
4) Find CPU-Vcore and it is configured to be Auto
5). change auto to  1,375V.
6). Exit and save your bios.

Maybe this will make your computer stable.  ( the above have made my own computer stable i have a 5900x cpu and 6900xt cpu and it crashed all the times in games after 5-15-30 minutes the computer would shut off with the error Kernel-power ID 41 ( 63) but after doing the above i had 0 crashes.

[mackbolan777]

Did you try the registry edit mentioned previously? What error is it in Event Viewer? Have you tried running the games in DX11 only with no crashes?

VRAM voltage should be fine if not too high out of the box. You can't adjust VRAM voltage, just the main voltage to the card, VRAM voltage is regulated via the cards vBIOS and usually floats around the 980mv mark. You can check all voltages using HWiNFO: Free Download HWiNFO Sofware | Installer & Portable for Windows, DOS

While running that, run OCCT: OCCT : Free, all-in-one stability test & stress test for your PC (ocbase.com)

Try Prime95, using small FFT's: Prime95 download version 30.4 build 9 (guru3d.com)

Check your PSU by running OCCT and looking at HWiNFO for the 12v area. If it drops below 11.4v under load, you have a bad PSU. Check the the 5v if that drops below 4.75v the PSU is bad. The 3.3v below 3.135v, bad PSU. Those are the main 3 "rails" to look at that are tied to the PSU exclusively. The 12v negative reading is one that can indicate excessive vDROOP, then you'll want to set LLC higher in the BIOS, to 3 or the middle, then higher if needed to get stable. Voltage should remain within 5% of spec or something is wrong.

The OCCT test is extensive, takes about an hour to run in full and finds most problems. Prime95 will test the CPU and if it tosses a lot of windows saying the worker has stopped, etc. Then the CPU has an issue like something in the BIOS set wrong or it could wind up as bad. I'll repeat before, there are MANY complaints about that PSU model specifically.

[MintChipper]

just tried your TdrDelay fix and still got a crash, I haven't shared the crash data from Event Viewer before because I didn't know what to make of the codes and looking them up didn't help but if you can see something from it be my guest, if there's something specific in Event Viewer I didn't include please say what:

" Faulting application name: DevilMayCry5.exe, version: 0.0.0.0, time stamp: 0x5fd3490d
Faulting module name: DevilMayCry5.exe, version: 0.0.0.0, time stamp: 0x5fd3490d
Exception code: 0xc0000005
Fault offset: 0x00000000045ab374
Faulting process id: 0x3560
Faulting application start time: 0x01d7035a9aa98ff2
Faulting application path: C:\Program Files (x86)\Steam\steamapps\common\Devil May Cry 5\DevilMayCry5.exe
Faulting module path: C:\Program Files (x86)\Steam\steamapps\common\Devil May Cry 5\DevilMayCry5.exe
Report Id: c949e524-7d67-401a-abb6-92a8bbcaa72a
Faulting package full name: "

I'll try to run the OCCT and prime95 tests you mentioned by tomorrow, thank you for providing links and instructions, also I'm not sure if this is important but I use 2 monitors with this pc, one is hdmi at 60hz and one is displayport at 144hz using FreeSync, if that's known to cause crashes or instability please let me know.

[mackbolan777]

I would definitely just try the DP monitor first by itself, leave the other disconnected until this is sorted or use the HDMI one if the problem still exists.

The code you show is telling me that you should run the "repair" option in STEAM for your game or reinstall the game completely. It's not the first time I've heard or experienced a game fault due to STEAM getting corrupted or a game download.

If you run your games on a different drive than the OS, like games are on a 4TB regular HDD and OS on a 256GB SSD, this can be a problem. It can also be a problem running a regular HDD vs. an SSD period with newer games, especially MMO's or "open world" type games. Most run strictly an SSD these days to avoid things like micro stutter, or FPS drops. It's come down to more than quicker load times for gaming to consider using an SDD for all.

[MintChipper]

Ok I started the OCCT CPU test (I realized afterward you probably wanted me to do the Power test and not this) and something happened, about 5 minutes after the test started errors came piling in like it quickly shot up to thousands of errors in a very short amount of time, and then OCCT closed on its own. HWinfo still showed my CPU usage at 100% and my CPU was at 94°, I started writing this post initially but tried to open OCCT again and got a blue screen (DRIVER_IRQL_NOT_LESS_OR_EQUAL with ntoskrnl.exe & USBXHCI.SYS according to BlueScreenView). OCCT didnt say anything about the problems it found when the PC finished restarting. HWinfo also didn't have any PSU voltage drops but this was the CPU test so that was probably why.

Should I make sure the PC is not running anything else before trying another test? should I unplug my USB devices too like the blue screen implies? Am I correct that I made a mistake and you wanted me to do the Power test and not the CPU? I wanna make sure things are safe and know what to do before trying again.

[mackbolan777]

If your CPU hit 94c it probably overheated. Check your CPU cooler. OCCT uses HWiNFO to monitor power, so no need to run both at once. Yes, make sure most of the back round stuff is off while testing, because it will go faster. But the CPU shouldn't be heating up that fast. If the cooler is ok, fans, AIO, whatever (AIO's need to be mounted on top or hoses down in a front/side mount), then re-run the test in regular form. No extra FFT's or any setting changed, just the regular, full OCCT test run.

[MintChipper]

Alright thank you, I'll try to run another test today doing what you said. I only have about 3 fans being the stock cooler that came with the 2700x and the 2 that came with the case which has an awful lot of holes for ventilation so I keep it eyes level at my desk and it's usually pretty cool though it isn't much yeah. To clarify is there a specific name for the "full test"? I only see the kinds that test specific components besides the Power test that pushes both CPU and 3D at once.

[MintChipper]

Hey providing an update, still haven't run more OCCT tests but I did try just running games on an individual monitor instead of running both at once, I thought it worked the first day I tried it then the next I got another crash, I also tried reinstalling games like you recommended and nothing changed. So we can rule out if the monitors or that were an issue I guess. 

[MintChipper]

By the way, my Windows settings has in the optional updates a driver update called "Advanced Micro Devices, Inc. -Display- 2/20/2019 12:000:00 AM - 25.20.15025.1002". Should I care about this? Is it something I've been missing?

[MintChipper]

Hi again, another update, good news and bad news, bad news is OCCT still closes on it's own about 10 minutes into the test, it did stop giving blue screens but it still does that weird thing I mentioned before where the app closes but the cores are still under load, I did manage to start the app again but it didnt say it was still in a test so I just restarted my computer to get rid of the load. I found out that it closes because it gets too hot but i don't have any more cooling solutions on me, so I don't know how I can still try to use this test.

The good news is the Prime95 Small FFTs test seemed to work just fine, I let it run for about an hour and it didnt have or detect any problems in any threads.

So where should I go from here? I know I said I didn't want to wing it with hardware before but now that I know my CPU at least should be good I'm wondering if I should just try replacing my PSU first since that's been blamed the most in this thread, though the issue there is in what I could get done of the OCCT test it still didn't drop to the voltages you said could be bad though it did drop a teeny bit compared to the last test I reported.

Thank you for checking up on the thread!

[mackbolan777]

Uninstall the GPU driver and try running OCCT. In "event viewer" it should say "app crashed" or something to indicate if OCCT crashed or the GPU driver or what. If OCCT runs fine without the GPU driver installed. Try the newest driver out 21.2.3 just to see if it works. If you can, swap the card into another PC to see if it works there. If it does, try a PSU. You're running out of road though. RMA the card might be an option or take the PC to a shop and let a pro look at it in person. It gets hard to try to work through issues as complex as this without being "hands on".

[MintChipper]

Yeah sorry I know it's difficult to see what's really happening from just words. In event viewer I've got these 2 things when OCCT closes, from sources Application and .NET Runtime: 
"Faulting application name: OCCT7.3.2.exe, version: 7.3.2.99, time stamp: 0x601e570c
Faulting module name: atiumd64.dll, version: 27.20.14535.1000, time stamp: 0x601cd20b
Exception code: 0xc0000005
Fault offset: 0x0000000000097dd9
Faulting process id: 0x32e8
Faulting application start time: 0x01d708f6b4956426
Faulting application path: C:\Users\username\Downloads\OCCT7.3.2.exe
Faulting module path: C:\WINDOWS\System32\DriverStore\FileRepository\u0364033.inf_amd64_f36ad498a78de343\B363860\atiumd64.dll
Report Id: 0208b786-fa50-4342-a017-e05be3e1b8b5
Faulting package full name:
Faulting package-relative application ID:"
and
"Application: OCCT7.3.2.exe
Framework Version: v4.0.30319
Description: The process was terminated due to an unhandled exception.
Exception Info: exception code c0000005, exception address 00007FFCF6F37DD9
Stack:"

I'll try to test with the driver uninstalled and do what follows as needed, thank you for suggesting that. I thankfully do know someone who works on PCs and can try out my GPU on a different machine, I know that's strange for me to say now and not before but they just helped me build this PC so I wanted to try to figure out the exact problems before risking covid to ask them again.

[MintChipper]

Alright I'm back, I uninstalled the driver and started OCCT, after 5 minutes it started getting dozens of errors again and I got scared about another blue screen so I cancelled the test, I realized that doesn't achieve much so I tried to start the test again and after 3 minutes I got a blue screen so there was no crash report for OCCT in event viewer. At this point I think I should just ask that guy I was talking about in my last post above for help.

Sidenotes:
- I installed the latest driver anyways afterward just to try it out, as expected didn't change much
- When cold booting the PC I hear some kind of funny thud noise for a moment, I figured it was normal but looking it up I found out it could be one of the fans in the GPU or PSU being strange, I don't think it's the case fans in the front or the stock ryzen cooler since those are already going when I press the button, I can't see the GPU or PSU fan very well to find out if or which one could be making the noise.

[mackbolan777]

You need the beta version of OCCT, that version has problems with AMD GPU's for some reason. I found out that it uses a known problem version of HWiNFO. As far as the noise, pull the side off and look? Fans typically whine, grind, but thud no unless the blades hit something. Did the CPU cooler fall off? A mouse? Capacitors sizzle or pop, but it wouldn't boot at all. DVD drawer could do a test routine where the door makes noise, similar to I guess a "thud". 

Crash report had to have something in it besides code 41 Kernel Power System, from shutting down wrong. Application crash or hang might be one, driver time out, etc.

Also, there's been a few BIOS updates for that board. Even if you haven't done any, it's time to look at the BIOS settings. Things like SOC voltage should be at 1.10v not auto, if you're using XMP or DOCP it could be a RAM timing problem that either a BIOS update or using DRAM Calc and manually entering the "safe" timing for your RAM might be a plan. "Creeping" problems can often be linked back to bad settings or something else needing a tweak in BIOS. Visit AsRock's Discord, it's all typing but there's a vast wealth of knowledge specifically for AsRock products and others, that's pretty much live in time frame. Past midnight EST there's people on there that an help. "Stoney" is really good with funky problems like this. We even get in depth on PBO and all the other AMD related settings there.

[MintChipper]

Sorry, thud was probably the wrong word to use, looking up more about it I guess it's more of a grind though it only lasts a second. I don't have any HDDs or DVD drives that could make the noise either.

Well the bluescreen specifically listed UNEXPECTED_KERNEL_TRAP with ntoskrnl.exe attached only, looking in event viewer there was this DirectX thing but i assumed it was unrelated because it happened around 15 minutes before the blue screen but I never saw that specifically say it had crashed before so I guess I could have mentioned it:
"Faulting application name: GPU3DDX11.exe, version: 0.0.0.0, time stamp: 0x5f66fa01
Faulting module name: GPU3DDX11.exe, version: 0.0.0.0, time stamp: 0x5f66fa01
Exception code: 0xc0000005
Fault offset: 0x00010a0e
Faulting process id: 0x15c0
Faulting application start time: 0x01d70c1963b62b8a
Faulting application path: C:\Users\username\AppData\Local\Temp\OCCT\DX11\GPU3DDX11.exe
Faulting module path: C:\Users\username\AppData\Local\Temp\OCCT\DX11\GPU3DDX11.exe
Report Id: e681701a-784e-49cc-bb18-d98cb08606a9
Faulting package full name:
Faulting package-relative application ID:"
copy-pasting that just now I didn't know it was OCCT related, but I guess I do now.

My SOC voltage wasn't set to that before so thank you for telling me, almost thought it solved it but I got a crash the next day, my BIOS version is P3.30 which looking at the website is somewhat recent. As for DRAM, I think there could be something to worry about, I do have XMP settings using two 8GB sticks inserted in the right spots, I installed the DRAM calc and Thaiphoon to get the data but I noticed something in Thaiphoon with my RAM that the tutorials didn't seem to have, my sticks are ddr4-3200, but both had their frequency listed at the bottom as 1600mhz, is this normal? or did I do something wrong? Task Manager lists my ram speed at 3200mhz but is that cause that's 1600x2? Is it supposed to read 6400?

Thank you for the bios help/suggestions, if you'd still like me to try that OCCT beta build or run to that Discord already please let me know.

[mackbolan777]

Your RAM is reading correctly in Task Manager. Thaiphoon is reading the half speed as per SPD/JDEC spec. 1600 X 2=3200, which would be right if that's what the XMP or box says. The noise could be coil whine from the GPU or a fan. Have you tested this GPU in another PC? If you didn't or can't, might be time for an RMA on that card. Seems like everything has been covered aside from a tech looking at this hands on.

What's cooling the CPU? I looked back since this is a 4 week thread, and you mentioned a CPU temp of 94c during OCCT. That should never happen unless you had run the highest level stress test. The "regular" full test is just leave the options stock and hit "start". If that's how you ran it and the CPU went to 94c, something is up with the cooler. Those temps cause the CPU to throttle or even shutdown the PC and then you'll get GPU errors and all kinds of others due to the improper shutdown.

Since you have an AsRock board, you might find the AsRock Discord beneficial as there's guys on nearly 24/7 live to help with all issues. It's a mostly chat thing, in fact I've never used voice there but there's good info there for all builds. I found the link online, either via Redditt or the AsRock site. 

[MintChipper]

Hearing examples of coil whine I don't think it's that, I still think it sounds like a fan but I could be wrong.

(sorry just saw your edit so I'm editing my own post to add this paragraph now)I have a wraith prism cooler that came with the ryzen cooling it, when I made that post about 94c I didn't know I was looking at the Tctl temp and not the Tdie temp which would be lower. Is the "regular" full test a certain data set or something or is it just the large data set?

Yeah at this point I'm gonna have to organize getting this card or the whole pc sent to that computer guy I was talking about to see if they can figure out what's up. I'll try to remember to update again if something happens but for now thank you so much for helping me along the way!

[mackbolan777]

Send the whole deal in if you can. As a former business owner of a PC sales and service, I can't stress enough how important it is that the tech can see the whole picture. Nothing gets missed that way. The 94c in any of the 3 CPU temps is abnormally high. In any case, glad to hear someone's going to look at it hands on. 

[kalckamlo]

Power supply could be shady since it's a bronze and I'd go 650W or higher in gold for future reference.

[kalckamlo]

Power supply could be shady since it's a bronze and I'd go 650W or higher in gold for future reference.

Speed Test