I quickly got hit with the hard truth that my Cooler Master V1000 PSU is not enough for my dual Vega 64's like I thought initially . I had no problem stability testing them by mining on them for 24H together with the CPU, when I tested about 3 weeks ago. Which basically is the same as putting them throught furmark powerwise from what I saw, a ~800-900W powerdraw)
Even if I run them on the powersave settings now (or downclock them manually to 1200 Core/945 mem @ 1.000V), I am having stability issues in games. Eventhough my powermeter reads only around 400-600W from the wall for the whole system, games are unstable/crashing (I am on custom watercooling). And if I try some more demanding games the systems shuts off completely. Really disappointing as my PSU is atleast goldrated and got great reviews when I first bought it a couple of years back.
Looking at an EVGA Superonove P2 1600W right now, as I want to be safe and will be upgrading to Threadripper sometime in the near future, and overclocking it.
I think the content may have been moderated when you tried to acess it. Could you try the link again please?
If you still have problems in accessing it I will will repost the results plot here if that is o.k.?
Getting closer to putting my 64's go up for sale.........i'm sorry, but it's simply pathetic!
Just got back from being away for work for 5 days, to see a new driver update. Did the AMD clean up utility.
Installed new driver from scratch. Now I now can't even play most games and I lost 2000 points in DX11 benchmark (ran 5 times).
ROTTR crashes and tells me the Video Driver crashed as my Threadripper 1950x with 32GB of ram doesn't have enough resources.
I am going to try a complete system reset and start again, if I have no luck I guess I will have to stick with single GPU or seriously look into nVidia GPUS.
P.S. At least there is a good second hand market for the 64's in the mining community.
Update 2 :
So after nearly half a day of stuffing around with Wattman settings I managed to get something decent from the Vega 64's.
I am still unable to play ROTTR and Battlefront due to frame drops and artefacting......but at least they are not crashing like earlier.
Here's FireStrike score :
Sorry to hear that you are having issues with crossfire, after I bought a new PSU that could handle my cards I am a very happy camper. Have had zero proble, I was hopeing that 17.9.3 would sort these problems out for you but, I just noticed that you had already tried them
Thanks for your reply.......it has been rather puzzling as I have seen YouTubers with fairly solid framerates with X-Fire Vega setups.
I certainly thought that perhaps a fresh install would fix the issues....but to no avail. 😞
What size PSU do you have? I have a 1000watt Corsair HX1000i.......i thought this maybe the issue, but I don't get frame drops in FireStrike or TimeSpy.
So I don't think it's the cause.....also I don't wanna lash out for a new PSU unless I can be certain that's the cause.
I had a coolermaster Vanguard 1000, a 1000W Gold certified PSU that has served me well for a couple of years.
The issues I had was that games crashed to the desktop without warning, sometimes they wouldn't even start. The games main process would just quit instantly after I had double clicked the games exec (which shouldn't have anything to do with the PSU, as there was no load at the time).
The games actually that would start could make my computer shutdown completely or reboot it. So I either had to turn it back via the powerbutton, or wait for it to boot back into windows. This would happen during certain benchmarks or an especially GPU taxing sequence in game (The witcher 3 would run for 5-10 minutes before my system restarted).
The strange with all this was that I was nowhere close to the limit of what my powersupply should have been able to handle, when playing the witcher 3 for example the powermeter I had connected to my walloutlet would show that I never drew more than ~650-700W from the wall for the entire system, yet the system would reboot. I would run 3DMARK firestrike ultra, and that only drew around ~650W at peakload (it went between 550-650W), but my computer would turn off completely after a while. Although it would let me finish runs from time to time (but games would always reboot or shut the computer down).
I tried running all powersave, balanced and turbo states but it all ended in the same way (even tried removing the overclock of my CPU just to let the GPU's have another ~100W of power). One thing that did work which shouldn't was when I tried mining on both GPU's with the turbo powerstate. Which gave me the biggest powerdraw I've seen, around ~870W from the wall. That ran for 24 hours without any issues at all, and my cards never downclocked or anything like that. And from what I've heard mining on a GPU is almost as taxing as running furmark.
I tried cooling my PSU in case the issue was heat, by directing 2 140mm fans directly on it, and opening up my case's sidepanel + the window to the room (quit chilly in Sweden right now). At this point I thought this had to be driver related, as I wasn't getting consistent shutdowns or reboots that I expected. But I had to try something as I was seeing other people having great success with the same benchmarks, and the one thing they had in common was that almost everybody I saw had pretty overpowered PSU's (1200W+).
And I was correct, since I installed my new EVGA SuperNova P2 1600W PSU, I've had none of the issues I was previously having. And it works like a dream, and the scaling is surprisingly good for being the first driver out enabling crossfire.
Edit: Oh I am running custom watercooling on CPU and both GPU's so heat was never the issue.
Your scores ,like I mentioned before, are way off to what they should be....
Something very wrong here.
My vega 56-s can do this : https://www.3dmark.com/fs/13697215
I am sure you have seen my scores before, but something is probably limiting your GPUs.
Most likely the PSU like many suspect. Had similar issues when I accidentally put 3 of my 8pin connectors under 1 rail on my PSU. Had massive frame drops as the PSU didn`t want to surrender and somehow managed, but didn`t provide nearly enough juice. Scores similar to yours. Sometimes it would shut down and I was blaming my OC for that, but then I started to look at my power draw from the wall and also used to log my FS runs with GPU-Z. Saw some very weird power draws and behaviours and then I remembered that I was tired when I connected my cables and mixed up the rails on late night.
1. Check the PSU rails first. (I in no means think you are stupid and that is too obvious, it can happen:)).
2. If no.1 is not the issue, your PSU might be dying or simply can`t cope. What do you use to measure your power draw by the way?
3. are your scores separately okay with one GPU installed, or does it behave similarly?
4. Have you repasted the original thermal paste? These cards thermal throttle like hell.....
5. could be a faulty card(s). Very unlikely....
Going 1080Ti is definitely less troublesome:)
Hang in there buddy!
Thanks for you reply......woah, that's awesome score. I appreciate it as I haven't found anything to compare mine with.
I'm certainly getting toward the end of my tether with these cards 😞
Thanks for all the tips.......I'm going to put in a day today and try to get this sorted.