• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

AMD RX Vega 64 Clock Speeds and Benchmarks

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
A

ArranDaniel

Registered
‍
Joined
Sep 18, 2017
Just wanted to share and see what other people are getting.

This was the best i could get. my chip crashes if it goes over 1670mhz, and tbh the 1060 mem clock wasnt perfectly stable, crashed every so often, but 1055 was fine. The card is on EK Waterblock in a loop with the cpu 1x360mm + 1x120mm radiators. running fans at 100% as id been benching for a while but normally get 55C at 40% fan speed during gaming.

i7 6700k @4.4ghz boost
32GB Vengeance Ram
750w EVGA G3 PSU

50% Power Limit which you cant see in the screen. For some reason no matter what i set the Core clock to, whether i use Radeon or MSI or Asus software it runs a lot below. the average speed as i was watching it was arounf 1645-1655. I tried undervolting and i could get 1630 @ 1100mV and 25% Power but the frame rate actually suffered for it.. interested to see how others are getting on.

Untitled.jpg
 
I think it has something to do with the HBM as the Fury X gets low scores to, 60fps, lower even than the 290X. Real world it performs much better. Doesn't drop a frame on 1440p Witcher 3 @Ultra, stays at my 60fps target constantly.
 
ArranDaniel said:
I think it has something to do with the HBM as the Fury X gets low scores to, 60fps, lower even than the 290X. Real world it performs much better. Doesn't drop a frame on 1440p Witcher 3 @Ultra, stays at my 60fps target constantly.
Click to expand...

Wattman is completely broken, the only options that work are: power target, fan control, temperature set and HBM speed. The best benchmark achievable right now is +50% Power Target, Fan @ Max, Max stable HBM and leave everything else alone. I also assume you're aware of the 65C* issue with the HBM but that's what I bench at. My normal settings are either Power Save or Custom @ 960mhz, Power Target -20%, Temp Target 65C*, and Fan Speed set to either extreme, ie min-min/max-max.

Also pull up GPUz and watch your temps to make sure it's not something unusual going on. I repasted mine with liquid metal and kept the stock cooler. Currently mine are +0 Core, +5 Hotspot, +12 HBM meaning if I am at 65C* in wattman I am showing: 65/70/77. HBM normally doesn't (in my experience) have a soft area where it becomes unstable, it either is or isn't. My Vega56 hard locks at 965 but I can mine all day at 960.
 
I don't know about the issues over 65C no but running a custom loop i rarely see over 60 anyway. Only when running repeat stress and benchmarks. If I do 3 full benches on Heaven in a row I'll hit 70 but after 1 no more than 55. 45-50 in games like Witcher. 40max in wow/gw2 and other easier to run games.
I noticed when it got to 70C it started to drop the coreclock speed so I assume the issue is the temp target set is not the temp that we see on software and it's in fact running hotter and throttling earlier than we think?
But at those settings there it ran happily at 1640-1650 and 1060hbm

At 1060 on the memory 1/10 times I'll get a crash but not hardlock.
 
ArranDaniel said:
I don't know about the issues over 65C no but running a custom loop i rarely see over 60 anyway. Only when running repeat stress and benchmarks. If I do 3 full benches on Heaven in a row I'll hit 70 but after 1 no more than 55. 45-50 in games like Witcher. 40max in wow/gw2 and other easier to run games.
I noticed when it got to 70C it started to drop the coreclock speed so I assume the issue is the temp target set is not the temp that we see on software and it's in fact running hotter and throttling earlier than we think?
But at those settings there it ran happily at 1640-1650 and 1060hbm

At 1060 on the memory 1/10 times I'll get a crash but not hardlock.
Click to expand...

The HBM timings go into a throttle like position above 65C* and so for best performance it should be kept below that.
 
I can control everything on my Wattman. AMD have already pushed the chip to its limit so clock speeds won't rise without more power like on other cards.. i can get higher clocks with undervolting still. I feel like this is a good thing.. other GPUs are Bethesda, releasing a half finished product knowing that we'll finish it for them. Any effort in overclocked this card isn't worth it. The real world gains are tiny. Driver updates and time will be the only performance upgrades.
 
Whats the pricing of Vega 64 of in the US now? im still hearing about the prices being inflated but over here i got mine for £450 about 2-3 weeks ago, which is RRP. so with the current performance and price, i have no complaints whatsoever.
 
ArranDaniel said:
Whats the pricing of Vega 64 of in the US now? im still hearing about the prices being inflated but over here i got mine for £450 about 2-3 weeks ago, which is RRP. so with the current performance and price, i have no complaints whatsoever.
Click to expand...

Most are going for about $650-700 USD, you can occasionally find a 56 for $550

ArranDaniel said:
I can control everything on my Wattman. AMD have already pushed the chip to its limit so clock speeds won't rise without more power like on other cards.. i can get higher clocks with undervolting still. I feel like this is a good thing.. other GPUs are Bethesda, releasing a half finished product knowing that we'll finish it for them. Any effort in overclocked this card isn't worth it. The real world gains are tiny. Driver updates and time will be the only performance upgrades.
Click to expand...

My understanding was that those functions did not work
 
Its overall perfomance on dx12 is great, theres very few games still using it and even less with dedicated benchmarks. Just good to see its performing right in the middle of 1080ti and 1080 be interesting to see how it fairs in the coming months. FuryX gained significant performance ever big driver update it recieved in the early months.
 
AMD almost always does. Support isn't usually where it should be at launch. Not bashing it just is what it is.
 
Johan45 is right, I've used AMD cards from the 5xxx line to the 290x (skip Fury and, for now Vega), and through all their lines, there was always major improvements over the first year after launch.

One big example is the 7970, it was 10% slower than the GTX 680 at launch, and 10% faster a year after, thanks to drivers and games optimizations.

Edit: below, 3dm11 and 3dmfs with the same computer. Run one 7970, run 2, 780 ti. Difference in graphic score for both is like 20% for the 780 ti. While that was the difference 2 years before between the 7970 and the 780 "non ti".

3dm11_780ti.jpg 3dm11_7970.jpg 3dmFS_780ti.jpg 3dmFS7970.jpg
 
Last edited:
My Monitor is freesync and I now couldn't go without free/G-sync so I have to buy AMD cards it's that simple for me. So having it perform well is nice aha. Also the money saved, The extra cost of a G-sync monitor for the extra performance just isn't worth it. 1080ti + 3440x1440 gsync is about £1800. My monitor (about a year ago) and card were £1100. I've effectively built the rest of my system on the money saved.
 
It's a good choice, I have a Fury and an RX580, I like AMD but I know what to expect just takes them a while to optimize. My HTPC/Gamer is a Ryzen rig hooked to my TV. So this G-sync free-sync is lost on me. My eyes really can't tell much diifference at 1080 VS 1440 just everything looks smaller ha ha
 
You must log in or register to reply here.

Similar threads

Evilsizer
AMD and a lost chance?
Replies
16
Views
2K
EarthDog
EarthDog
custom90gt
Over Powered 15-LP2 (TONGFANG GK5CN6Z) benchmarks + temps
Replies
0
Views
3K
custom90gt
custom90gt
G
Can someone explain clock speeds and "boost" clocks with Pascal?
Replies
5
Views
391
gsrcrxsi
G
Bustos
Just Upgrade or Build From Scratch?
2 3
Replies
59
Views
5K
ronss
ronss
storm-chaser
HexaCore Restoration Project underway (AMD Phenom II 970 Zosma) *Pending 960T*
4 5 6 7 8
Replies
159
Views
12K
storm-chaser
storm-chaser
Back