• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

The RX Vega 64 Owners Thread

Razor Time said:
As a guess it might be something to do with the coolant temp and damaging the pump, most watercooling pumps have a max coolant temp tolerance of about 55-60c. A gpu core temp of 71c is fairly toasty for a loop which would mean your coolant temps would in turn be fairly high.
Click to expand...
Didn't know that. Useful, thanks.
 
Been running Time Spy on the new driver to see if there is any performance improvement. On turbo profile I got a better score on the new driver, increasing from 7231(total) and 7610(graphics) to 7353 and 7778 although this could be within the margin of error I suppose. Not so great however is than Timespy is now crashing when the memory is over 1050 where on the previous drivers it ran 1075 no problem...
 
heartburnron said:
Been running Time Spy on the new driver to see if there is any performance improvement. On turbo profile I got a better score on the new driver, increasing from 7231(total) and 7610(graphics) to 7353 and 7778 although this could be within the margin of error I suppose. Not so great however is than Timespy is now crashing when the memory is over 1050 where on the previous drivers it ran 1075 no problem...
Click to expand...

I don't understand. You're moaning you would rather get better clocks then better scores?
I would go the other way and favour better scores with lower clocks as that should mean power which means less costly to run.
 
heartburnron said:
Was just playing Shadow Warrior 2 with my overclock of 1907/1075 and got a warning from the Radeon Settings that the GPU temp had exceeded it's limit. It didn't affect performance or crash or anything but after checking Wattman the peak temperature it hit was 71 degrees. Still not hot compared to the air cards but certainly the highest I've seen it. I'm guessing the warning limit for the AOI cards is set to 70 degrees then? AMDMatt - can you confirm this? Is there any reason why the AOI cards would be at some sort of risk for running higher than that given the air cards go up to 85?
Click to expand...

Yep the temp limit in bios is 70, the reason AMD gave for the temp limit was "just because we can" ...lol. So probably not the real reason, and yes as pointed out above, could be to do with coolant temp, and/or the coolant needing to be below a certain temp for that pressure reservoir to work or something
 
Twyst said:
I don't understand. You're moaning you would rather get better clocks then better scores?
I would go the other way and favour better scores with lower clocks as that should mean power which means less costly to run.
Click to expand...
Not moaning at all, just sharing the early experiences and information with my fellow Vega owners. It's also of particular significance as due to the apparent buggy behaviour of Wattman on occasion it's difficult to ascertain whether the new drivers have affected the limitation on the memory or whether the higher clocks achieved on the previous driver where in fact true or not. Obviously the better scores for lower clocks would always be preferable however I can no longer achieve scores as high as I could on the previous driver now that I cannot increase the clocks as much. Surely it's not hard to understand why it would be perceived as less then favourable when it appears as if you can no longer extract the same level of performance from the card?
 
run unigine heaven or some other 3d load, adjust voltage until something breaks; when it breaks go back a step.

heres me searching for the voltage for state 6: aso42s0.png
 
Last edited:
I just had a quick play with the new drivers on my vega 64 blower. Colour me impressed with the under volting!

so far ive set state 6 to 1000, state 7 to 1050, power limit to 30%, mem frequency to 1050. Nothing else changed.

It's gone from throttling at 85c and dropping down to 1536mhz to................... rock stable lines through heaven, time spy extreme... Heaven only got it upto 76c max. Timespy 70c, zero throttling as its well under the 85c temp limit.

Timespy extreme graphics score 7470 and heaven 1440p everything maxed is 1416 score.

Ive only run a few passes of each benchmark so it might be unstable or could have some more headroom. I'll have more of a play later and report back.

Matt.
 
Been doing more testing on the new drivers this morning and whilst I can no longer reach the frequencies I could on the previous driver I am getting better scores at lower clocks which suggests that the frequencies may not have been applied correctly on the original drivers? The highest stable overclock I managed this time was 1802/1045 although this produced a total Time Spy score of 7607 and graphics score of 8061 which is in line with the scores I got with 1907/1075 on the previous drivers.

I tried (only once mind) to run Time Spy at 1802/1045 slightly undervolted (1100 and 1150 on state 6 & 7) but it crashed.
 
Smudge said:
so far ive set state 6 to 1000, state 7 to 1050, power limit to 30%, mem frequency to 1050. Nothing else changed..
Click to expand...

would be interested if that is stable my card isnt happy undervolting state 7 so far...!

edit yep i'm getting nowhere on stock clocks at 50% power limit which is strange....

certainly seems like it needs every last mv of the 1.2v to get the 1600+ clocks!
 
Last edited:
Chewie said:
That's great news Matt, thanks for confirming.

Any chance you could share a quick overclocking and undervolting guide for Vega newbies like myself (I also own a Liquid Edition)? :D
Click to expand...

Try -0.050v on State 7, with +50% power limit, HBM at stock. Game for 30+ minutes at 99% GPU usage.

If that passes, try -0.075v and so forth.

My 1st sample does -0.075v @1752/1050Mhz.

My second sample does -0.050mv 1752/945Mhz.

Increasing memory on my second sample caused a GPU crash, but i will need to do more testing on GPU 2 to confirm those clocks.
 
@AMDMatt any known issues with voltages? if I set core to 1000mv the HBM clock seems to be bugged at 800mhz (and the temp sensor seems to say 1deg c)!

*edit* .. no idea how I broke the HBM but recalled you saying reset if that happens!

my card seems fine at 1537Mhz/1000mv at 1100HBM (stock voltage not sure if thats just mem controller like polaris or what)

pretty close to full power 1630Mhz on firestrike: http://www.3dmark.com/fs/13423936

the core voltage required for higher clocks seems to be ridiculous...
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom