Potential AMD RX Vega benchmarks appear online
Potential AMD RX Vega benchmarks appear online
Published: 1st May 2017 | Source: Guru of 3D |
Potential AMD RX Vega benchmarks appear online
3DMARK results for an unknown AMD GPU have been posted online, showcasing the performance of what could be AMD's RX Vega GPU.
These results contain the hardware driver ID 687F:C1, which is the same code that has been seen in previous RX Vega leaks, as well as in AMD's early Vega demo which showed DOOM gameplay at 4K 60FPS.
The results are still on Futuremark's 3DMARK Database, with Guru3D being the first to notice these benchmark results. At this time it is unknown what this new GPU is, be it an old engineering sample, a lower-end Vega model or using old driver versions.
Below is a table from Guru3D which showcases these leaked benchmark scores against currently available GPUs. Here we can see that this GPU is only able to beat Nvidia's GTX 1070 and beat AMD's ageing R9 Fury X by a mere 9.6%, which isn't enough of a performance increase to justify 2+ years of development since Fiji.
These results are simply not that impressive for what is expected to be AMD's upcoming flagship GPU, though it is likely that this GPU will be a cut-down version of Vega that is designed to compete with the GTX 1070 and sit above the RX 480/580.
3DMARK's database also shows GPU clock speeds of 1200MHz, which is lower than what is expected from Vega, especially given the clock speeds available on AMD's older Polaris GPUs. These clock speeds could improve with future samples, again increasing performance. This benchmark result was from April 12th, so it is likely that performance will improve in the weeks coming up to Vega's launch.
Like most early benchmark results, these performance numbers are indicative of nothing, as many factors could change the performance of this GPU before launch. Let's hope that this is a lower-end Vega variant, as we really need some competition in the high-end GPU market.
You can join the discussion on these alleged AMD RX Vega benchmarks on the OC3D Forums.
Most Recent Comments
Quote:
|
Originally Posted by AlienALX
Hmm. See, this brings us back to the initial rumour which was that it was to compete with the 1070.
I really hope it isn't true, because if it is I can get to within 200 points of that with a Fury X. |
If the clock speeds of 1200MHz are true it should be at least 20% faster than the R9 Fury X, not including other areas for potential gains. These results make very little sense if this is full-fat Vega.
The Fury X is almost three years old, these results are simply too low for that much development time and a process node shrink.Quote
Lets just hope this is the lower end modelQuote
Quote:
|
Originally Posted by WYP
I could overclock my Fury to get these kinds of numbers, this must be a cut-down model.
If the clock speeds of 1200MHz are true it should be at least 20% faster than the R9 Fury X, not including other areas for potential gains. These results make very little sense if this is full-fat Vega. The Fury X is almost three years old, these results are simply too low for that much development time and a process node shrink. |
And that is pretty much where I got the "it's as fast as a 1070" from. Lately people's minds have been going into overdrive and a lot of "what ifs" have been said. Like "What if AMD deliberately locked their games to 60 FPS to slow the card down so that Nvidia wouldn't know the exact performance" and so on. Which of course is not anywhere near as accurate as scientific data.
Also. The Fury X was a metric mile behind the 1070 back then too. It's only been with a large dose of DX12 and some pretty amazing drivers that it's started to play with the 1070.
I really hope I am wrong but if Vega is what it was penned out at then yeah, could be in for a huge disappointment.Quote
Quote:
|
Originally Posted by AlienALX
Thing is when Vega was first penned out and somewhat announced a load of the "pros" on OCUK basically tore it to bits and predicted it would be as fast as a 1070. When I say pros I mean "Heavy science dudes" who know exactly how it all works. IE - the cores, clock speeds etc etc. These weren't just rumours but somewhat scientifically predicted.
And that is pretty much where I got the "it's as fast as a 1070" from. Lately people's minds have been going into overdrive and a lot of "what ifs" have been said. Like "What if AMD deliberately locked their games to 60 FPS to slow the card down so that Nvidia wouldn't know the exact performance" and so on. Which of course is not anywhere near as accurate as scientific data. Also. The Fury X was a metric mile behind the 1070 back then too. It's only been with a large dose of DX12 and some pretty amazing drivers that it's started to play with the 1070. I really hope I am wrong but if Vega is what it was penned out at then yeah, could be in for a huge disappointment. |
That being said, there are a lot of games where AMD GPUs fall far behind, which is why AMD's changes to Vega's front-end are so important, as they allow AMD to better load Vega and eliminate internal bottlenecks that cause these performance issues.
TBH with Vega I am not expecting a GTX 1080 Ti killer, but at a minimum, I want to see higher clock speeds. Nvidia's gains in recent years from Maxwell and Pascal have been mostly due to clock speed increases rather than deeper architectural changes.
AMD needs higher clock speeds and a strong GPU front-end to do well, AMD cannot afford to release a high-end GPU that offers lower clock speeds than most current Polaris GPUs.
What I really want to see with Vega is 1500MHz clock speeds. Since Keplar Nvidia's clock speeds have gone through the roof and AMD really needs to get their arse in gear in this regard.Quote
CSS Valid
I really hope it isn't true, because if it is I can get to within 200 points of that with a Fury X.Quote