It's also about stability; too many people were trying to run high end cards while using every last percentage of their CPU's. The result was system and Boinc instability, task failures and more problems for the projects to deal with. Failures aside, this CPU over-commitment also resulted in GPU performance reductions and in some cases a decline in CPU project performances.

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

As I said above: "GPU-Grid decided that the 600 series GPUs were becoming so fast, that not reserving an entire core would slow the GPU down too much."

I missed that. MrS, thanks for the explanation. Question, why is it saying 0.481C and really reserving 1 CPU core? How does that work?

And "I don't think it's a good choice for GK107 based cards"

I would agree. I would much rather manage and reserve the core myself if it helps the speed.

A few posts above:

Mumak wrote:

I still don't understand where the "0.594 CPUs" number comes from.. Any idea?

It's automatically generated, I think by the GPU-Grid app by some algorithm, and passed to BOINC to display it.

MrS

The number shown is probably based on some estimation used for the older cards with SWAN_SYNC and not yet updated for Keplers.

Reserving a CPU core your self is fine, but not everyone running a fast GPU (in need of this) will know he/she should do this. I guess that's where the idea came from to do it automatically.

MrS

---

Scanning for our furry friends since Jan 2002

Thanks guys, now it makes sense to me :-)
That number should be updated to match real scenario, however I'm not sure who or what decides that estimation.. Maybe it's not easy to implement such change in the system (to have different estimations for pre-Kepler and later GPUs).