I'm surprised to see the asus card compared to those leaked 480 nitro graphs. Although I'm assuming those aren't fake, it showed the 480 performing a bit better than the 1060.
Can anyone explain why ComputerBase.de and TPU have such wildly varying numbers?
What am I missing here?
1420 is 18% over stock.. stock card is power throttled at just below 1200Mhz.My personal guess is that those nitro graphs are in fact almost certainly fake.
They showed an average performance improvement of 15.5% over the reference 480, even though the nitro is only overclocked by 6% (1342 vs. 1266). Even if we assume the reference 480 is throttling down to 1200, it's still only a 12% overclock.
And then there's the overclocked results where they overclocked the nitro to 1420 MHz (6% over stock) and got an average performance improvement of 8%
There could be any number of reasons for why there might be differences, which is why people should stop relying on individual sites and start using 3DCenter.org instead (they collate a large number of sites and calculate averages from there):
http://www.3dcenter.org/artikel/lau...aunch-analyse-nvidia-geforce-gtx-1060-seite-3
Can anyone explain why ComputerBase.de and TPU have such wildly varying numbers?
1420 is 18% over stock.. stock card is power throttled at just below 1200Mhz.
Actual 3D performance gained from overclocking is 3.0%.
These are early days, but the Asus overclock is the highest we've managed on any GTX 1060 and end users who are willing should have no trouble teasing an extra 10 per cent out of their cards.
Ahh I didn't realize you were comparing to the factory OC. Did they also OC the VRAM?1420 MHz is 6% over the 1342 MHz reference clock of the Nitro+.
Gotcha so ignore TPU from now on. Thank you!The biggest difference is Computerbase runs benches inside a regular gaming case, and they do 10 minute pre-warm up before starting benches. This is the true result for any GPU.
"These are early days, but the Asus overclock is the highest we've managed on any GTX 1060 and end users who are willing should have no trouble teasing an extra 10 per cent out of their cards."
http://hexus.net/tech/reviews/graphics/94714-asus-rog-strix-geforce-gtx-1060-oc/?page=12
Ahh I didn't realize you were comparing to the factory OC. Did they also OC the VRAM?
I think those leaks might be fake too but rx480 does scale well from memory OC as well. But I will wait for official benchmarks from reputable sources.Sadly "highest we've ever managed" is still only a very mediocre 8% overclock.
I don't know if they overclocked the memory for the 1420 MHz run, but the stock Nitro+ doesn't have overclocked memory as far as I can tell.
The biggest difference is Computerbase runs benches inside a regular gaming case, and they do 10 minute pre-warm up before starting benches. This is the true result for any GPU.
I think those leaks might be fake too but rx480 does scale well from memory OC as well. But I will wait for official benchmarks from reputable sources.
Custom GTX 1060 in stock right now for $249 @ Newegg (Zotac) and Best Buy (PNY):
http://www.newegg.com/Product/Produ...-na-_-na&cm_sp=&AID=10446076&PID=6202798&SID=
http://www.bestbuy.com/site/pny-nvi...D=1&siteID=AKGBlS8SPlM-4ziqveJqla5Wa3sgiYylzQ
Edit: Zotac gone.
Didn't change much it seems. GTX 1060 FE still 8% faster than the long awaited custom RX 480, MSI Gaming X ~11.3% faster based on their other review.
FYI that image was wrong. He didn't calculate them correctly.
Thanks for that. Just seems so unscientific and wildly prone to cherry picking.Each site uses his own settings and location for benchmark (if they don't use the in-game build in one which usually suck), also settings like hairworks can impact in the W3 as the performance hit should be lower on nvidia's hardware.
Guru3D noted the settings used
still they might do the capture on different location that can tank the card more then others, like on RoTR, the geothermal valley is one of most demanding areas.
FYI it's 8% faster with the updated chart.
sirmo said:Can anyone explain why ComputerBase.de and TPU have such wildly varying numbers?
Power consumption is certainly not as bad as many claim for rx480.. considering it's also powering more VRAM (10 watts). Still 1060 is better in that department but a pretty big improvement over previous generation for AMD.MSI GeForce GTX 1060 Gaming vs GeForce GTX 1060 FE vs ASUS Strix RX 480
Computerbase benchmarked the above in FHD on five DX11 games.
Anno 2205
COD : Black Ops III
Fallout 4
Far Cry Primal
The Witcher
https://www.computerbase.de/2016-07/asus-radeon-rx-480-strix-test/2/#abschnitt_benchmarks_in_full_hd
They found that:
the Strix 480 in OC Mode is 2% faster than the 1060 FE
the MSI 1060 was 4% faster than the strix 480 and 6% faster than the 1060 FE.
Regarding power consumption, they showed that during Anno the OC-Mode Strix used 50 Watts more than the 1060 FE and just 22 Watts more than the MSI 1060.
Also, maybe of interest, the 1060 FE was faster by 1% over the Strix RX 480 in "gaming" mode while the RX consumed around 30 Watts more.
Note that these tests were only conducted on those five DX 11 games.
MSI GeForce GTX 1060 Gaming vs GeForce GTX 1060 FE vs ASUS Strix RX 480
Thanks for that. Just seems so unscientific and wildly prone to cherry picking.
Does not matter if its updated, calculation is still wrong.
Not that i would care much for TPUs outdated benchmark collection...
Gotcha so ignore TPU from now on. Thank you!
Thanks, that's my observation as well.I wouldn't say ignore, it's more a best case scenario for GPUs that rely on boosting and can benefit from an open bench setup without a warm up.
Examples include: Fury Nano and every NV GPU since Kepler.
Also, TPU didn't use Doom Vulkan. They used Hitman DX11 and Rise of the Tomb Raider DX12. It's backwards actually when RotTR DX12 even in their previous testing, runs faster in DX11 for everybody. While Hitman DX12, AMD's lead grows.
At Computerbase, they included Doom w/ Vulkan and the result is well, close.
https://www.computerbase.de/2016-07/asus-radeon-rx-480-strix-test/2/#abschnitt_benchmarks_in_full_hd
One game can swing the results when the rest are pretty close. Examples: Anno swings wildy to NV, where a 970 is faster than Fury X. Likewise Project Cars. On AMD swing games, it's definitely Hitman DX12 and Doom w/ Vulkan.
So looking at it objectively, TPU's current testing, favors NV strongly while Computerbase's testing is neutral, because they include games that swing to NV and others that swing to AMD along with neutral titles.
Ultimately, these cards are very close, so you can skew it either way. If you did testing only DX12/Vulkan, you will swing it big time to the RX 480's favor. But that isn't an accurate picture of gaming, because DX11 games are still popular.