Discussion in 'Videocards - AMD Radeon' started by OnnA, Jan 20, 2019.
TimeSpy - 9423 GPU pts.
Yes that is the correct drive.
I am also using this one.
I got the card yesterday.
At default setting (1802MHz @1082mV) TimeSpy Graphic Score: 8837
Working on under volt. I still have not found any good configuration.
Mine seems to not like voltages below 1020mV @1802MHz.
In Firestrike stress test, I see artifacts if I reduce the voltage below 1020mV and left the card run for a time under stress test.
I have some problems with 3DMark suite at the moment, every time I want to run any test I get error at the beginning.
After getting this error at the beginning the benchmarks work. The stress tests does not end. They go into infinite loop of testing.
I have steam version. I have used it in the history to do benchmarks on Vega64.
I don't know if this error has to do with the 3DMark suite itself or is it not recognizing the Radeon VII properly?
I tried reinstalling it but it did not help.
The Firestrike test does not end at the moment. It goes into infinite testing loop.
Has anybody else experience such problems with 3DMark suite (Steam version)?
What I have also experienced is that the temperature on the core goes shortly after stress tests to 110°C and cards starts to throttle.
We need a good water cooler and I think then it is possible to extract more performance out of this and stabilize the clock in range of 1780MHz at default settings.
Ordered one of these but don't think it will be here for a couple of weeks
Reading on OCUK forum: https://forums.overclockers.co.uk/threads/the-radeon-vii-owners-thread.18845346/page-10
Sounds like bad drivers and app are making overclocking / undervolting difficult at the moment, but people are seeing memory up to 1200+ which is a big boost, not seeing massive differences on core yet. Still early days
I have been playing around the voltage in WattMan. Leaving frequency at 1802MHz, I have dropped the voltage to 1046mV instead of 1082mV. The stress test runs with out issues with these settings.
Going below 1030mV is not stable. I see artifacts in Firestrike stress test.
With 1046mV the clock seems to be more stable and core temperature does not hit 110°C as fast as at default settings.
TimeSpy score @1046mV=8794
TimeSpy score @ 1082mV=8830
I have played Battlefield V for 40minutes using 1046mV @ 1802MHz and logged the system power consumption with my Corsair HX1200i power supply through iCUE.
Note: This is power draw at output side of HX1200i which is delivered to the system and not on the wall side.
The input power draw is higher and depends on the efficiency at which HX1200i is running during different load phases.
Idle at desktop average system power draw=127W
Battlefield average power draw=418W
This is Delta of about=291W
I don't exactly know how much in this 291W comes from Ryzen 2700x during gaming.
Looking at the online reviews for Ryzen 2700x gives around 50W consumption for 2700x at gaming load and about 12W at idle.
If above assumption is true then, I can assume that the Radeon VII is consuming around 265W in Battlefield @ 1046mV.
I also tried to do the same test in COD Black OPS 4. In that game the frame rate was not stable for some reason.
It was varying very fast between 70 to 120 all the time with every fighting move and going up to 144 when standing idle.
The Radeon VII clock frequency was fluctuating in the same way and were not having a stable curve.
I did not check before with my old GPU (Vega 64). I am not sure if this also happens there.
The power consumption was fluctuating also very fast. The average system power consumption was at 370W.
Quieted it down... and heated it up to 95c . Best wait for water block and LC it.
Heated, because it runs at VeryHigh clocks ~1800MHz for Vega is a little too much IMO (even at 7nm - looking at HotSpot and it's ~99-105Deg !).
Best is to set it at 1750MHz (max) with good UV.
Vega uArch is efficient enough, but ATI can't clock it lower cuz' the benches & reviews
We need more DX12/VLK games.
As for Vega LiQuiD, best is to set it at ~1717-1737MHz with 1.094->1.137mV
But it can run with 1680Mhz/1125 1.087v/937v not touching ~200tW spike (~165tW in BFV at 70FPS FS)
Check the bit with the washers meybe that will help.
Brutal Vega Se7eN
That's still unbearably loud! And those temps are a joke.
AMD's Radeon VII Packs A Lot More FP64 Performance Than Expected
When AMD first announced their Radeon VII graphics card, they stated that the graphics card would feature throttled FP64 performance,
a downgrade that would prevent their gaming-oriented graphics card from cannibalising the sales of their enterprise-grade Radeon Instinct M50 and M60 offerings.
Early on AMD said that the Radeon VII would provide 0.88 TFLOPS of FP64 performance, a sixteenth of the GPU's FP32 performance and around an eighth of what is offered on the Radeon Instinct MI50.
Since CES AMD has had a change of heart, deciding that their Radeon VII users deserved a little more FP64 performance from their new gaming flagship,
making the graphics card more appealing to professional users while maintaining the performance advantages of their Radeon Instinct Lineup.
Not the Radeon V offers 4x more FP64 performance (3.5 TFLOP) than what AMD originally planned, making the graphics card an FP64 monster given its £649.99 price point.
While FP64 is not useful for gaming applications, professional users will know that FP64 compute is a useful tool for a range of calculations, making the Radeon VII extremely appealing to some workstation users.
AMD Radeon VII (gfx906) Delivers ~90 MHS Ethash Out of the Box
Thats it... good luck with ever seeing the R VII in stock again... or at a reasonable price.
Looking at the temperatures and the clock speed made me scratch my head.
After I saw the original post saying that the test was done outside in the cold weather at 100% fan speed, it came clear to me.
This is but a good indication that a decent custom water cooler can bring nice gains.
Don't worry, at current prices and network difficulty that 90 Mh/s is not at all profitable.