Yeah, it does (to me) as it represents what Boinc actually reports, rather than what NVidia say it should be. Two reasons; not all cards are reference design and Boinc knows about CC. Removed above, as it's probably not applicable to here, because it's just a guide for comparison of like cards and because the 'CC2.1-like' issues will not appear in GF600 anyway (supposedly), when it turns up.
When the WCG apps arrive you won't necessarily be able to go by NVidia GFlops peak and AMD GFlops peak and get reliable comparisons. So we are going to have to construct a mixed table of AMD and NVidia cards based on performance, rather than reported GFlops peak. Then it's going to have to be adapted for different drivers and apps (continuously) to keep the information live. GPU's are not P&P!

IB is another reason not to start public testing before mid-April. By mid/late April people will stat building around IB and GF600. However, the GF600 gains most of it's performance from Physics, so any apps would need to be Physics-coded to be well optimized. If not then performance could be poor. It's the norm to facilitate the most powerful cards even at the expense of lesser cards, due to overall contribution - unless several generations of apps will be built and maintained. For AMD, the HD 7770 and 77750's are due out this month and the HD 7870 and 7850's are due out in March. So a lot to accommodate, and they won't all be GCN.

JEklund, the reasons of public testing not having commenced is server side work AFAIAI. No one need to run to the shop and shell out hi dollar just for crunching WCG GPU units, either on a red, green or blue card.

--//--

skgiven,
I noticed on your list my 560 Ti card specs are a little different than what Boinc is reporting for the GPU.
1311 -vs- 1428

NVIDIA GPU 0: GeForce GTX 560 Ti 448 (driver version 290.10 , CUDA version 4010, compute capability 2.0, 1279MB, 1428 GFLOPS peak)
About double GFLOPS of my old 285 GPU and uses less Watts
NVIDIA GPU 1: GeForce GTX 285 (driver version 290.10, CUDA version 4010, compute capability 1.3, 1023MB, 708 GFLOPS peak)

I am running x64 Linux 2.6.28-11 and NV driver 290.10
I wonder if the difference is due to EVGA stock overclocking or OS or what ?

(I added the ones in red cuz Boinc didn't state it.)

By the way here is a SAD case to add to your list of NV GPUs
This is what Boinc says about the GPU in my laptop. 179 GFLOPS
NVIDIA GPU 0: Quadro 1000M (driver version 290.10, CUDA version 4010, compute capability 2.1, 2047MB, 179 GFLOPS peak)

Not that I would use it to crunch with , just thought it was kind of funny and kind of sad at the same time...

Those hanging out for the GPU version of HCC might want to check out this review of two new Radeon cards.

http://www.anandtech.com/show/5541/amd-radeon...7770-ghz-edition-review/1

While not the greatest games cards, they look promising for price/performance GPU computing. The review specifically looks at GPU computing in several of the later pages.