FAH-Addict: Folding@home's premier news site.

[toTOW]

Sorry for the lack of news ... we've been all too busy to work on the site recently, but here is an attempt to catch up

FahMon reaches version 2.3.99.3 and supports BigAdv/SMP2 bonuses

New nVidia drivers released: 260.89 WHQL

-smp 128 inside

GTS 450 and GTX 460; great performance, and more to come?

AMD Radeon 6850 and 6870 announced

ATI Catalyst 10.10 drivers out with OpenCL support

[toTOW]

GTX 580 rumours accumulate

GPU3 client reaches version 6.40r1; featuring Compute Capability support

[PantherX]

It looks like the GTX 580 won't be very good for folding

...According to sources it can be as many as 300 million transistors that NVIDIA has been able to cut in this way. The effect is that GF110 will be a GPU targetting only retail and will not be as efficient for GPGPU applications as the older siblings of the Fermi Tesla family...GPU clockspeed is said to be 775MHz, with 1544MHz shaders and 1536MB GDDR5 clocked at 4008MHz...

Source

EDIT -> I found an interesting response:
"They took out all the redundant, switched off HPC caches, but are ADDING CUDA cores. HPC functionality was only active in multi-GPU Tesla + rack units anyway AFAIK."
So I am guessing that making assumptions won't be useful until the real product is released.

[toTOW]

That will require a review

In the meantime, on the red side of the force : Folding with a Radeon HD6800? Possible, but underwhelming!

[arcon]

It looks like the GTX 580 won't be very good for folding

...According to sources it can be as many as 300 million transistors that NVIDIA has been able to cut in this way. The effect is that GF110 will be a GPU targetting only retail and will not be as efficient for GPGPU applications as the older siblings of the Fermi Tesla family...GPU clockspeed is said to be 775MHz, with 1544MHz shaders and 1536MB GDDR5 clocked at 4008MHz...

Source

EDIT -> I found an interesting response:
"They took out all the redundant, switched off HPC caches, but are ADDING CUDA cores. HPC functionality was only active in multi-GPU Tesla + rack units anyway AFAIK."
So I am guessing that making assumptions won't be useful until the real product is released.

"the older siblings of the Fermi Tesla family" = GF100
The GF110 is just a extended GF106. So no probs for F@H

[toTOW]

New uniprocessor projects: 10200-10205

[toTOW]

Further information about the GTX580

It looks like a great card to fold

[HaloJones]

It looks like a card that costs 4x what a 450GTS costs and is not 4x more productive. I know where I'm putting my folding cash.

[toTOW]

New GPU3 Fermi project: p6800

[toTOW]

It looks like a card that costs 4x what a 450GTS costs and is not 4x more productive. I know where I'm putting my folding cash.

I think the hierarchy will be back to the correct order with this :

New GPU3 project for Fermi: the serious work begins

[toTOW]

Pande Group at work ...

New Gromacs A4 project: p10425

New BigAdv Gromacs A3 project: p6900

[toTOW]

A little summary :

Three new uniprocessor projects: p11289, p6880, p6886

[toTOW]

Is the Sandy Bridge GPU compatible with OpenCL?

[tinokun]

http://software.intel.com/en-us/article ... pencl-sdk/

There's no mention of it running on the GPU, only the CPU (for now?).

[codysluder]

So my with SSE4 on my i7 ( or "i9" ), I will be able to run some future OpenCL client rather than bigadv but the hardware will only process maybe 1/64th as many simultaneous instructions as my GPU from ATI on NV. Oh, how will I ever choose.