• 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.

Anyone running Vega 64 in a custom loop?

ive got two 240 thick rads with the blocks (cpu and gpu) between; as I said the HBM voltage setting can cause the driver to go haywire so I just left it alone as wattman is annoying enough without adding to the bugs

I'm using the stock backplate, its actually Alu so not seeing any value in the EK ones; certainly not the price they are asking anyway!
 
As far as I've noticed in all my tests so far (liquid 64), adjusting hbm voltage does nothing at all. I don't think it is active yet.

GPU caps on 1000mV, going any lower changes nothing. However going a bit below 1100mV drops average clocks a bit too. It's apparently automatic system when clocks are adjusted all the time according to not only limits (temp. and power) but also power available (which depends of ampers and voltage).
 
Yes, PowerPlay comes into effect controlling the target voltages and clock speeds. Some reliable reporting software would be quite invaluable but I don't think there's anything out there yet.

I don't see third party app support coming until AMD have nailed the driver control itself.

The early adopter process as usual.
 
The latest GPU-Z (beta) has core voltage monitoring, hwinfo seems to work quite well too although the HBM voltage is just stuck at 1.356 currently; one of Raja's post vacation tweets was about the core clocking down to avoid vdroop if something is using more power.
 
Tinek said:
As far as I've noticed in all my tests so far (liquid 64), adjusting hbm voltage does nothing at all. I don't think it is active yet.

GPU caps on 1000mV, going any lower changes nothing. However going a bit below 1100mV drops average clocks a bit too. It's apparently automatic system when clocks are adjusted all the time according to not only limits (temp. and power) but also power available (which depends of ampers and voltage).
Click to expand...

Try increasing the GPU state 7 up to something much higher. I had this. My card I can set everything to 950 if I want but when doing so it wouldnt boost past state 6. I slapped my state 7 up to 1702 (vega 64 air) and now its boosting to 1656! temps are great too for the blower cooler.
 
I set the fan to 100% with stock core clocks/voltage, then set 1100 HBM and ran Doom. FPS increase versus stock was on par with 1020 HBM and 1552 core but card got to 74C and started artifacting (core and HBM OC was stable for 30 minutes). I need to get the thermals under control before the card can truly shine, it appears. Still waiting on my fittings so nothing I can do about that for now.

1599 MHz is the highest boost temp I've gotten. As soon as it hits 75C the core clock speed rapidly drops along with the performance. I'll revisit once the block is on.
 
I've settled on 1050 mV and 1677 core, 1020 HBM. This gave the maximum performance with no throttling, core boosts to 1610 MHz.

Is anyone else experiencing a performance loss when going over 1020 on HBM? It's bloody weird. Would be interesting if those running 1100 or thereabouts could lower theirs to 1020 and see what the metrics demonstrate.
 
think its some silicon lottery stuff, seems to be an increase in perf for me; it does need another 20w ish according to GPU-Z though so you'll need to up the power limit to rule that out?
 
So I played some PUBG tonight and was getting a shutdown just before pulling the parachute. This happened consistently without fail (3 times in a row). I did a quick Google and a suggestion was not to use a daisy chained PCI-E connector (I have an RM1000i) and use two dedicated cables from the PSU and a single connector. This resolved the shut down issues.
 
Waterblock is on... and the results are good! I'll post some pics tomorrow. It's bloody quieter; that was necessary in every way. Can't get the core past 1699 without a crash. It needs more voltage but no joy putting anything over 1200 mV (only hitting 51C peak).
 
aaronyuri said:
Waterblock is on... and the results are good! I'll post some pics tomorrow. It's bloody quieter; that was necessary in every way. Can't get the core past 1699 without a crash. It needs more voltage but no joy putting anything over 1200 mV (only hitting 51C peak).
Click to expand...

Excellent. I flashed an LC bios to mine and now hitting 1752mhz @ 1210. Well I was, just hit solid 1785mhz gaming tonight with 1230mv only 41c
 
Haven't had a chance to do much with this (I've been grinding away at Destiny instead) but I'll be picking this up soon. LC BIOS is flashed and working fine. Slight performance increase at stock but doesn't compare to the OC performance.

Results to come once I've spent some time. :)
 
You must log in or register to reply here.
Back
Top Bottom