....after due consideration I think I will stick to just one GPU.....which one? I will decide later when I can monitor what happens on WCG. I am not really concerned about other projects, Milkyway is just one that I can test a couple of different cards upon.

When we get GPU here I will test the same cards and see the results, listen to others who arer also doing it here, then go for a new graphics card.....

Thanks for all the help & advise guys.....now awaiting progress here.

Learning something NEW every day, that is what gpu crunching all about! I just learned something, THANK YOU!!! I knew about MW and DP but not the rest!!!

I was on PrimeGrid this morning and saw this:
"For the most part, all of the software that we use is highly optimized and hand-tuned for the particular hardware it's designed to run on. The best example of this is George Woltman's gwnum libraries, which is inside LLR and a lot of other software. It's hand-written assembly language code designed specifically for the various CPU improvements that have been created over the years.

It squeezes every last drop of performance out of whatever CPU you have.

Opening up the CUDA architecture to more platforms makes it easy to write portable code. It won't necessarily make it easy to write fast code for all platforms. You can have portable code, or you can have fast code, but fast portable code is almost an oxymoron.

Take a look at our GPU sieving software, for example.

On Nvidia GPUs, the software is written in CUDA, which is sepcifically designed for the Nvidia CPUs. It's blazingly fast.

On ATI/AMD, it's written in OpenCL, which is portable and is not specifically designed for ATI's GPUs. It's pitifully slow by comparison to the Nvidia version of the software. The hardware itself is not inferior; it's the fact that a portable, cross platform software architecture is being used that makes it so slow."

I am sure WCG is making the same hard choices!

mikey159b, please provide us with a link to the relevant webpage at PrimeGrid.

Thanks
;

The WCG decided to opt for OpenCL. While the merits of this decision are debatable (CUDA probably being faster for NVidia, Stream for AMD/ATI and C++ possibly for CPU's), the researchers wanted an app that was universal, or as close to it as possible. Unfortunately they still had to design 3 somewhat separate apps; for CPU, AMD and NVidia. There are however some positives; WCG can at least compare performances from a relatively level playing field when the apps are released and award credit accordingly, hopefully setting the standard Boinc-wide. It's not necessarily a bad thing to have a low performing app either; lower GPU utilization reduces running costs and heat problems, making the app more affordable and compatible.

"On Nvidia GPUs, the software is written in CUDA, which is sepcifically designed for the Nvidia CPUs. It's blazingly fast."
To nit-pick; CUDA is fast, low level code designed specifically for NVIDIA GPU's. Perhaps the snippet was taken from a thread discussing the topical use of CUDA on CPU's, supported by recent NVidia tools? This just facilitates future uses of CUDA and app development; some CUDA code can be used now on CPU's (development/testing/early releases) and when at a later date a GPU is added that CUDA code would move to the GPU, and work much faster.

"CAL ATI Radeon HD 3800 (RV670) (512MB) and NVIDIA GeForce GT 430 (961MB)"
I suspect your 3800 may not be of any use here. I don't think it supports OpenCL1.0 (not 100% sure though). It's about 4years old, but does have 320 stream processors. Still works with MW's streaming ati14 app. ATI Radeon HD 4800's (RV770) on the other hand do support OpenCL1.0, so from that up, probably for here. Possibly 4700's too (RV740). I'm sure someone here will construct a list, prior to or during Beta testing (unlike elsewhere).

Your GT430 is somewhat shader (cuda core) limited (96), so while it will work here, don't expect fantastic results from it. Until it can be used here you could use it at Einstein, and might be able to use it soon at PEOM (a single precision Bio-science project).

The ATI stream code used at MW is designed for ATI GPU's. It's more akin to CUDA on NVidia than OpenCL on NVidia, ATI, or a CPU.
There's also a CUDA_OpenCL app for NVidia GPU's, but desktop/gaming NVidia GPU's are not designed with double precision in mind, and dp performance tends to be around 1/8th to /12th for recent NVidia GPU's (compared to their sp performance). Hence NVidia GPU's are poor at MW.

OpenCL is NVIDIA 'compatible'; as much as it is ATI compatible.