Folding@home cluster?

[7im]

You still fold on the CPU because every PC with a GPU still has a CPU that can make a valuable contribution. Every little bit helps.

[k1wi]

It's not throughput that is the issue, but the latency.

Would It be possible to Use an PCI-Express Slot for Interconection? I don't know PCI-E well enough to tell if PC-PC connections are possible, but PCI-E 3.0 x16 would provide upt to 15Gigabytes/second, about 120x as fast as Gigabit Lan. On the other side why bother with CPU folding anymore, when it is possible to fold anything on a GPU?

[Nathan_P]

Would It be possible to Use an PCI-Express Slot for Interconection? I don't know PCI-E well enough to tell if PC-PC connections are possible, but PCI-E 3.0 x16 would provide upt to 15Gigabytes/second, about 120x as fast as Gigabit Lan. On the other side why bother with CPU folding anymore, when it is possible to fold anything on a GPU?

Because CPU's are still handling the really big jobs, that's why.

As for PCIe memory transfers, doesn't PCIe have a latency problem?

[HaloJones]

You still fold on the CPU because every PC with a GPU still has a CPU that can make a valuable contribution. Every little bit helps.

Got to disagree with you there; I have stopped all my cpu folding except for an i7. Why? because the cores are being used by the new beta GPU client which leaves not enough cores to be worth folding for the points per watt.

If points=science, then the GPUs are producing more science.

[7im]

Yes, but only if you have multiple GPUs, which is not something the typical donor has. Who knows, one day there may be a fahcore that runs on both at the same time. CPU folding is not going away any time soon.

[folding_hoomer]

Got to disagree with you there; I have stopped all my cpu folding except for an i7. Why? because the cores are being used by the new beta GPU client which leaves not enough cores to be worth folding for the points per watt.

If points=science, then the GPUs are producing more science.

I disagree, because:
I´m running a i7 2600K@4400MHz under Linux getting ~32-36K PPD by using about 160W (whole system) - much more efficient than any of your GPU´s . . .

[PantherX]

You still fold on the CPU because every PC with a GPU still has a CPU that can make a valuable contribution. Every little bit helps.

Got to disagree with you there; I have stopped all my cpu folding except for an i7. Why? because the cores are being used by the new beta GPU client which leaves not enough cores to be worth folding for the points per watt.

If points=science, then the GPUs are producing more science.

There are two different points here:
1) CPU folding
2) Points per watt

Those donors who have expensive electric bills or want to be efficient, will optimize their systems based on points per watt. That may or may not include folding on CPUs. However, there are some donors who have really cheap electric bills and will fold on whatever hardware they have as long as it meets the preferred/timeout deadline. I am still folding on Intel Pentium E5500/E5400 which nets me about 2K points each and am very happy since every bit counts

[7im]

HJ also wrongly assumes the GPU will always need a full CPU to feed fahcore_17, not even out of beta yet. That could easily change.