Vega64 PSU recommendation

Associate
Joined
5 Sep 2010
Posts
756
I have just got a brand new Seasonic GX650 as a replacement for my Focus+ under RMA (Focus+ had a manufacturing issue that caused it to trip when it shouldn't) and I am still tripping the PSU when the graphics card causes power spikes.

AMD recommend a 750w for this, but I was told a 650w that is decent quality would be enough. Well, it's not lol. Soon as a fancy effect(like an explosion) occurs on my screen, the screen goes black, fans hit 100% and it requires a hard reset.

Before anyone suggests it could be my GPU, it's not. If I limit my GPU power draw so that it only pulls 150-180, this never happens (been running like this for 2 weeks waiting for RMA replacement). Even if I force the GPU to run at throttling temperature it's not happening.

So I am after a PSU that is more than enough for this GPU and any future GPUs I may get, preferably overkill.

Can someone spec me a 850w-1000w PSU for the following:

  • Vega64
  • Ryzen 2700x
  • x2 Crucial MX500
  • x1 3TB Western Digital Green
  • Some corsair AIO I forget the model of
Looking for it to be between £150-200. I'll sell on this GX650 to recuperate the cost.
 
Associate
OP
Joined
5 Sep 2010
Posts
756
Before you go buying anything how are you powering the Vega64? Is it from a pair of seperate pci-e leads or from a single lead with the pair of daisy chained connectors? If it's the latter then switching to a pair of seperate leads may cure this problem as it did for many other people when vega came out. If it doeasn't fix it then the Corsair RM850 is a very good psu and a steal at it's current price.

2 separate cables.

That's some kind hard crashed state.

It's not hard crashed though. I can still hear audio, skip music etc etc. Everything is still running. The behaviour I am describing is the same behaviour that others have experienced with this card during power spikes, my particular PSU (Focus+) was known to have a problem with the fail safes tripping when it didn't have to (I believe it was set to tight?).

If I reduce the cards power load (put power limit to -15%) this never happens.

In that situation PSU would shut down completely and your PC wouldn't make even a beep untill you power cycle PSU to clear that "error state"

Well that sucks. In that case I am not sure what to do, I am not inclined to blame the card, as it runs flawlessly if I reduce the power draw. A new PSU is not expensive and will last the life of another 1-2 builds, so I am prepared to rule it out.
 
Associate
OP
Joined
5 Sep 2010
Posts
756
Vega 64 isn't anymore any new card, so who knows if something in it has degraded.

Afraid not. As far as I remember, this card has always done this. I just never came across it frequently as my day to day games were things such as Terraria, Minecraft and Heroes of the Storm. So nothing that would heavily tax this GPU. I have recently started playing new titles, hence why I've sorta raised it more of a persistent problem than random event like I assumed it was in the past.

So it's some kind graphics card/its driver crash.
And check Event viewer and Reliability history for any errors coinciding with disappearance of image.

Could this be due to lack of power delivery? The GPU can run perfect without allowing it to draw massive amounts of power. If I reduce the graphics down in games, it never occurs. It occurs every time without fail whenever something on the screen suddenly happens, like an explosion. Again, if I limit the power draw, this doesn't happen (but I lose FPS, a lot).

Could try changing monitor connection to another output just to change variables.

Already tried this, was the first thing I tried :D.
 
Associate
OP
Joined
5 Sep 2010
Posts
756
My Vega was rock solid on RM850X, RM850 is pretty much the same spec, have one of those also in a different system. I think it skips braided cables and other unnecassary 'features'.
Either of those should be more than fine.

The card by default is limited to 250W unless you have the liquid cooled version so shouldn't be causing an issue with a 2700X.
If you maxed the power slider then it can take up to +50% power, that may be enough to challenge a 650W PSU with transient loads as the card can pull close to 400W maxed. Unlikely the cooling will cope but enough to cause a crash.

Vega performance is actually way better with an undervolt, then it should be nowhere near 250W power limit.
From the factory the voltage is way high so it hits the power limit and throttles, the 64 is really bad for this hence why the 56 can often perform nearly as well.
However if you undervolt ~ 1.1V, perhaps a little lower, from memory you should be able to run at ~1600+Mhz core continuously without hitting any power throttling.

So I currently have it set to 1075 on P6/P7 and +50% power limit. It pulls 220-250 during gaming, but it's obviously much higher during spikes. I remember the first day I got the card and seen the stupid voltage/heat numbers on the factory setup.
My version is the RedDevil, so it's actually got a better aircooler than other cards except the Nitro.

It's also worth noting that even with the preset "Turbo" mode in the AMD software, this issue still occurs.
 
Associate
OP
Joined
5 Sep 2010
Posts
756
Thanks for the input everyone!

I know there is still a chance my card could be faulty, unfortunately this machine is also my work machine (independent contractor) and PowerColors process is 6-8 weeks and they won't advance me a temporary card. So this is the cheapest, fastest way for me to possibly solve it. I would rather remove all the variables before I deem my card ******. It's also out of warranty, and they have specified if it does need repairs and it can be repaired, I would be footing the bill. Not sure what could possibly be wrong with it that they can repair to be honest, normally when these things are busted they get tossed in the bin anyway.
 
Back
Top Bottom