I am currently facing an issue with my Vega 64 Air Cooled! I bought it used on Craigslist and I am beginning to think the seller sold me a bad GPU. Here is the TLDR: My GPU will “cut-out” at random times causing a black screen only to come back. However, when it comes back I notice that the “tachometer” on the Vega card sometimes shows it is “maxed out.” However, Task Manager in Windows shows low usage on the GPU software side. Keep in mind, it happens at seemingly random times! The only consistent case in when trying to play League of Legends with a group of friends.
Things I have done:
-Reinstalled GPU Drivers (using DDU to remove them before re-installing)
-Reinstalled Windows 10
-Upgraded to a different power supply (750w Thermaltake Power Supply 80 Plus Bronze)
Things to consider:
-I am using two 8 pin pci-e cable extensions to help the aesthetic of my build [I have no idea if these could be causing the issue; I don’t really see how]
-System Specs: https://pcpartpicker.com/b/MWJ8TW
w/e, that ClockBlocker was nice tool for the time I needed it, and HBM crapping itself resembles odd Mantle crashing, even next boot may show red screen and have to restart flush it out? Then the damn thing wont crash although its not fully stable OC, which is why I found Warframe in the first place as it crapped my hours stable OC within few seconds, and that game turned out to be great RAM test too
That ClockBlocker used to work so that you just set it similarly to how driver has it, but for idle/boost and there was even option for boosting when foreground
So you want to change that idle mode to just keep on boosting, for example
Aside from fixing things, useful feature from that would be to idle when browsing
EDIT: Changed downclock term to idle, more clear this way
The ‘easiest’ way to test if the tool works at all.
Is to after installation edit the “DEFAULT” rule to “Block”, instead of “Downclock”.
If everything is working right, at this point your Graphics Card would clock itself up to the maximum frequency (0% load, at idle on the Windows desktop).
That did not work, but interestingly Radeon settings keeps crashing a lot. I imagine this is a driver issue, and it probably has something to do with a conflict between my motherboard and GPU?
is there anything in the windows event log? I saw this exact behavior testing the level1 kvm and the display port cables were dodgy or worn out. try a different dp connector on the card, try a different cable. or a bunch. the constant handshake/fail/handshake would totally crash the software, too.
I wound up testing an HDMI cord instead of the DP that was connected, and it rendered the same result. However, after entering my BIOS and resetting it to optimized defaults I was able to finally run Witcher 3 without crashing. I am beginning to wonder if my overclock was causing too much power draw on my power supply? It is a 750-watt power supply, so I cannot imagine how that would be the case. Could it somehow have been an unstable CPU/RAM overclock that was causing these issues? I don’t know?! -_- I quit
Right, it is a Thermaltake Smart Series 750w rated at 80+ Gold efficiency. However, I have a third-party LED controller running two LED strips off of Molex. I suspect that thing might be sucking more power than it needs to be.
Try ditching the pcie extensions and see what happens. What was the OC on the 1700? 1.35v all the time ~~ 175w power draw +/-
graphics card could be 350-400 peak so I I’d expect the psu could keep up but if its older it could be closer to 650w in reality just from loss of electrolyte/etc.
since this is a second hand card it might not be a bad idea to flash the rom on the card to a stock rom. you never know if they put a modded rom onto it for mining purposes.
