Synthetic Benchmarks:
3DMark 11:
The 10.12 drivers actually performed better than the 11.1 and the 11.1a drivers until we overclocked our 6970, which allowed our card to finally beat the 10.12 scores. Also, notice that the 11.1 out performed the 11.1a drivers too. With this being our first test, hopefully this is not a sign of things to come.
3DMark Vantage:
In the Vantage testing, the 11.1 driver won this battle with the 10.12 and 11.1a coming in 2nd and 3rd, respectively. Even overclocked, the 11.1a driver could not keep up with the 11.1 and 10.12 drivers.
3DMark 06:
Once again, the 11.1 drivers come out on top with the 10.12 behind them and the 11.1a following up in 3rd. The 11.1 drivers beat the 11.1a drivers by over a thousand points (6.3%) and also ended up beating the 10.12 drivers by about 5% as well.
With the 11.1a hotfix drivers, they seem to be a let down so far in our tests as the 11.1 drivers out performed them in every test. The 10.12 drivers have also held their own in these tests, which is also a bad sign for the 6970. Hopefully, the game tests prove to be better than what we have seen so far in the synthetic tests.
16 Comments
[...] like they didnt even bench them again so how they got a new total is confusing. Check these out http://www.techreaction.net/2011/01/…0-12-compared/ http://www.techreaction.net/2011/02/…2-99-compared/ They are follow ups for new driver releases. [...]
[...] in every game they are exactly the same as the 6970 release numbers. Drivers didnt change anything http://www.techreaction.net/2011/01/…0-12-compared/ . __________________ NVIDIA GeForce GTX 500 Series Owners [...]
@Juraj
He did not run the benchmark on GTA:IV. If he had you may be right. As it stands I believe in Crysis and Metro a faster processor would only excentuate the changes we see by a few frames.
Not all games are CPU bounded, Some are GPU bounded only and the higher res you go the more of the GPU load it takes. this is from Experience that I have learned over the years from being a Reviewer myself. Now I will have to say I am scratching my head at his results myself, but then again not everyone going to get the same results either and his “review” was based on his experiences and his own results. in this case I say he did a hell of a job. there is no way of going bigger unless he starts overclocking, or goes out and buys a Core i7 970/980 CPU that is the only way of doing it. the 1155 CPUs are based off the same architecture as the current 1156/1366 CPU’s the only difference is that they are 32nm. ie. why other reviews I read state if you are on the current 1366 or 1156 (i7) there is no real need to upgrade to a 1155 based CPU. And if you are currently using a 1366 CPU you be essentially going backwards because You lose the triple channel memory capabilities also the full 16x/16x PCI-E lanes. what part do you not fully understand.
@Dragon
A stock i920 is much more limiting factor than 6970 at 16×10 and lower resolutions.
Higher resolutions would benefit from a faster CPU as well.
It’s very simple. If you want to test GPU you need to use the fastest CPU you can get in order to shift bottlenecks to the tested GPU.
@The Duke
A feedback to you. I hope you take it and learn from it.
Because of the above mentioned reasons I do not consider your tests that much valuable.
SInce I cvan not edit the post, I will add just because the 1366′s are a bit older then the current set up, they are still very much a viable platform and the preferred platform for multiple vid’s and thats been stated over numerous reviews.
Okies cool now it makes a bit more sense to me
maybe put that in your set up portion.
@Jural, the only thing making the newer 1155′s doing so well is that they are all on a 32nm process vs the core i7 920-960 on the 45nm. Also not to mention the 1156/1155′s both came out after the core i7 1366 been out and Intel was able to optimize them a bit more.
The 920 is left at stock throughout the testing.
@all
Have a look here: http://www.anandtech.com/show/4083/the-sandy-bridge-review-intel-core-i7-2600k-i5-2500k-core-i3-2100-tested/20
If you got your i920 to at least 3.33GHz.
Just a dumb question what was your speed on the CPU? because 5k is awfully low. My 5870 Eyefinity Edition barely pulled 4800 on a 4.3 GHz 930 and that was overclocked, the 6970 is pulling 6053 on 3dmark11. While using 1.11a drivers.
@Juraj, I agree with everyone else how is the Core i7 920 out of date? the 1155 CPU’s may be able to take the 920-960 CPU’s on all out frequency but these 1366 CPU’s still support full X16/X16 PCI-E lanes and still maintain’s the highest memory bandwidth. AMD manages to do well, but the 1366 is still vastly superior (Talking Quad core to Quad core, Hex core to Hex core.)
Interesting results, I expected the 11.1a drivers to run better on the 6970 as there were supposed to be tweaks specifically for the 6970/50 that were inundated from the 11.1 WHQL release.
@Juraj I don’t understand. Is it not up to date simply because it is 1366? The performance is higher than anything currently released on the AMD front. 1156 has lower performance chips than the 920 and 1155 is being recalled due to chipset errors. If 1366 is not up to date (and I am going to assume you mean in performance) then what should be used?
@ Juraj:
How is a 920 out of date? What CPU would you consider up to date?
Well, thanks for the testing.
Your test CPU is not up to date. It does not do any justice to this card especially at the low resolution.
Thanks for the write-up Duke, exactly what I needed
As neither are out yet, we have no idea
Hopefully we will be getting both cards to test, then we could answer this for you.
So, with that review, Nvidea or ATI still.
XD
I’m trying to decide between two.
Nivdea 590′s or two 6990′s from ATI.