guru3d Radeon RX 480 Performance Benchmarks with 16.7.1 driver

[csbin]

http://www.guru3d.com/news-story/radeon-rx-480-performance-benchmarks-with-16-7-1-driver.html

With the new AMD Radeon 16.7.1 driver freshly released I understand that there is concern among you guys that the power distribution fix might have done something that makes the Radeon RX 480 throttle or downclock e.g. have a negative effect on performance. Hence I ran some benchmarks.

So as it stands right now, the effect on the power distribution fix is not measurable. By saying that I mean the performance is has remained the same. There are 1 FPS differences here and there, but overall it's all similar and here and there the performance even increased a little, 3DMark FireStrike is such an example.

I ran 3DMark FireStrike, Hitman 2016, Rise of the Tom Raider, Total War: Warhammer and Far Cry Primal. You can check out the benchmarks inbetween the AMD Radeon Software Crimson Driver 16.6.2 without power fix and the new AMD Radeon Software Crimson Driver 16.7.1 below. I've made a full spread in resolution, meaning Full HD (1920x1080), WQHD (2560x1440) and Ultra HD (3840x2160). I've also added a core clock recording of the card in a 3DMark FireStrike loop. Though there is less varying throttling, the card runs below 1200 MHz with that benchmark (which is normal behaviour). I verified and compared this with against Unigine Heaven (also included), which runs steady at 1240~1250 MHz in an Extreme mode. So the card does still reach its maximum boost frequency.

Now in the driver you'll notice a new 'Compatibility mode' as well as a new 'Power Efficiency mode'. These modes are irrelevant in the sense that you do not have to change them. Compatibility mode” toggle will further reduce overall board power. All modes have the new power distribution applied, we confirmed this with AMD moments ago, so you do not have to switch or alter anything in the drivers.

Click the thumbnails to watch the benchmarks, and obviously the benchmarks themselves don't confirm if the power distribution actually was fixed and within normal parameters.

 

[railven]

Other charts are more useful.

Without power draw it sort of seems pointless. Since the driver is basically a fix for the power draw issue.

Beside that performance seems flat, which I guess is good. If the power draw issue is resolved and there is no performance degradation, kudos AMD!

EDIT:

Even Guru admits their testing is rather pointless since it doesn't test the main purpose of the driver update!

Click the thumbnails to watch the benchmarks, and obviously the benchmarks themselves don't confirm if the power distribution actually was fixed and within normal parameters.

Oh well, still prompt response from AMD. So they keep the kudos!

 

[Silverforce11]

Toms & PCPER tested the power thoroughly. Power through PEG slot is <75W, ~70W on default and ~66W or less on compatibility mode.

 

[railven]

Toms & PCPER tested the power thoroughly. Power through PEG slot is <75W, ~70W on default and ~66W or less on compatibility mode.

Has the Tom's English article come out yet? Was having a hard time with the translated one haha.

 

[Silverforce11]

http://www.tomshardware.com/reviews/amd-radeon-rx-480-polaris-power-fix,4668.html

 

[IEC]

http://www.tomshardware.com/reviews/amd-radeon-rx-480-polaris-power-fix,4668.html

Heh, they left out the best line in the original German article:

Drama, Baby, das funktioniert nicht.

 

[railven]

http://www.tomshardware.com/reviews/amd-radeon-rx-480-polaris-power-fix,4668.html

Welps, can't win em all. Still kudos for the quick fix. RTG is definitely more responsive than AMD ever was.

 

[guskline]

I agree on how quick and responsive RTG was. Ryan Shrout should ship SVP Raja a bottle of Kentuckey bourbon!&#128541;

 

[DarkKnightDude]

Wait, so they fixed the power issue and increased performance a tad?

 

[Yakk]

Wait, so they fixed the power issue and increased performance a tad?

Yup, both at the same time.

 

[Silverforce11]

Wait, so they fixed the power issue and increased performance a tad?

They did it by undervolting slightly. My cards now have lower default voltages for Stages 4 to 6 of boost clocks. This results in holding their higher boost clocks instead of throttling due to power limit at stock settings.

How effective it is depends on how your card behaves, what it's stock voltages are etc.

 

[guachi]

Whoever decided to roll the performance increases, however slight, into the power fix deserves kudos.

 

[casiofx]

Whoever decided to roll the performance increases, however slight, into the power fix deserves kudos.

It is very simple, just slightly undervolt the card compared to previous driver, it'll reduce the amount of throttling and bring the performance back to where it should be

 

[Thala]

It is very simple, just slightly undervolt the card compared to previous driver, it'll reduce the amount of throttling and bring the performance back to where it should be

Almost. If you in addition increase the power target slightly it will stay at 1266MHz all the time. As 3dGuru pointed out, it still goes below 1266MHz due to power target.