A new benchmark is out in the wild.  If you want to try it out, go to the following link.  ENJOY


http://www.ngohq.com/news/16735-dire...rk-0-25-a.html



Oh my goodness.  My GPUs score 10x better than my i7 running @ 4.1ghz with 8 threads.  Yah!

DX11 benchmarks are just coming out from everywhere now

kscott190

DX11 benchmarks are just coming out from everywhere now

I know.  I'm extremely excited.  I'm looking forward to the GPGPU aspect of DX11 even moreso than the gaming!!!  DX11 is going to enhance to speed and quality of computing over the next few years.  It is unbelievable.  For the first time in history, we are going to live through a complete transformation in computing.  Sure, we've had ATI Stream and CUDA over the last couple years, but DX11 is going to exponentiate the GPU's role in everyday computing and bring everything together, FINALLY!

That is a heck of a score mate!  I wonder if if Crossfire is working for it?  I'll test and get back with you.

Also, I wonder if it makes a difference since you are testing DirectCompute 10?

garnetandblack

luv2increase

That is a heck of a score mate!  I wonder if if Crossfire is working for it?  I'll test and get back with you.

Also, I wonder if it makes a difference since you are testing DirectCompute 10?

Yeah - my first thought was that it must not be using crossfire on your rig.  The HD 5XXX series have far too many shader mudules to not blow my setup out of the water.
 
It could also be that my result was bugged, but I ran it twice...  Both were within 100 points of each other.

Yup... CF does not work on that benchmark.  That is another thing that ATI "must" fix with their drivers.  Darn ATI

Also, I believe the DX10 cards will use a different kernel than the DX11 cards in the test.  This is based off reading the Tom's Hardware review of DirectCompute 11 and the wiki which states:


http://en.wikipedia.org/wiki/DirectCompute

DirectCompute defines several versions of Compute Shader capabilities to ensure that applications can take advantage of GPU Computing on hundreds of millions of consumer systems already in the market today, and also have the flexibility to adopt new features exposed in DirectX 11 capable GPUs when they are available. DirectCompute shader versions 4.0, 4.1, and 5.0 correspond to DirectX 10.0, 10.1, and 11 respectively. Developers can start using Compute Shader 4.x kernels in their applications today, and add Compute Shader 5.x capabilities as the installed base of DirectX 11 GPUs becomes a significant portion of target market (typically millions of units).

Wow - I'm using 3 GTX 280s...  I wonder if more gpus actually hurt direct compute.  Sweet score Darius!
 
EDIT:  OC'ing my cores from 650 to 700 (and shaders from 1300 to 1400) yielded a score increase of over 10,000 points for me.  New GPU score is M75633...  Clocks have a huge impact on this test, it appears.
 

Darias,  remember, we are talking about scaling in a premiere release of the first DirectCompute benchmark.


Let us also not forget that we may be comparing the results which are irrelevant due to them being based off of two different (APIs).  Take note that yours says D3D_FEATURE_LEVEL_10_0 while mine says D3D_FEATURE_LEVEL_11_0.  D3D10 uses Compute Shader 4.0 kernel while D3D11 uses Compute Shader 5.0 kernel.


About the upgrading to 285s from 280s, I hope that was a step-up.  And, you need to put an OC on your rig to take more advantage of your cards in SLI.

it was a step up from a 280 XFX card EVGA ftw

well cant you force your 5xxx to run in dx 10

get somescores that way

also you gotta admit if the 285s get this kinda scoring you gotta wonder what a 380 will get with new architecture

still a SSC is at 702 vs your 700 dont you think 3 would beat 2 285s?

or are 285s that great of a revision o.O

OmegaBlade

Time for a new CPU, am I right? =P

No, I'd say you got another couple years before an upgrade

It's one 260. And I'm as surprised at the scores as you are. I think I won the "Highest Difference in Scores" award.

Now as to why my GPU score is so high, your guess is as good as mine. =\

Haha, I'm cracking up at the 260 having nearly twice the score of three 5870s. Yeah, I know crossfire doesn't work and something else must be wrong there, but it's hilarious nonetheless

On a side note, does anyone know why the test says my 275 doesn't support direct compute? I do have Windows 7 with the latest DirectX installed... Not that big of a deal, but it refuses to run on my GPU...

i condem this benchmark for being unusable and say that the scores are not countable

i was just about to ask that same question!

cause otherwise this thread has lost its relevance imo

unless the gtxs are some sort of computational .... devils..

qiaowenqian

seems the scoring algorithm is biased

well no its not. nvidia has a better gpgpu architecture and thats a fact. hell the guy who wrote the benchmark has an ATi card.