"running, high prioity (0.06 CPUs, 1 CUDA)"

What does the 0.06 CPUs mean? I am guessing the 1 CUDA means that it is running a WU through the GPU.

My system is set to run at 75%, however, even when I changed it to 100% it did not change.

Vista HP Pentium D @ 3.6
EVGA Nvidia 9600 GT KO

---

"running, high prioity (0.06 CPUs, 1 CUDA)"

What does the 0.06 CPUs mean? I am guessing the 1 CUDA means that it is running a WU through the GPU.

My system is set to run at 75%, however, even when I changed it to 100% it did not change.

Vista HP Pentium D @ 3.6
EVGA Nvidia 9600 GT KO

Yes, you are running a CUDA app. The requirement for running GPUs is 1 CPU per GPU app. This is so that the CPU can properly load and feed the GPU its data. So in a quad core system with a single CUDA complaint card, 3 CPUs will crunch and the 4th CPU will feed the GPU, hence 75% system load (3 CPUs) and the 4th will feed the GPU.

---

Correct me if I am wrong, but I don't think you have answered the question.

I have 0.04 CPUs running, and I recall a time when I had 0.90 CPUs running, and CUDA was crunching faster. GPUGrid seems to take twice as long since the change. So if someone know what it mean, maybe someone will know how I can change CPU's running.

---

Q6600 @ 2.40GHz
Vista Ultimate x64 SP1
Mem: 4GB
8800GT 181.22
BOINC 6.6.3 Win64

ACTIVE PROJECTS

Correct me if I am wrong, but I don't think you have answered the question.

I have 0.04 CPUs running, and I recall a time when I had 0.90 CPUs running, and CUDA was crunching faster. GPUGrid seems to take twice as long since the change. So if someone know what it mean, maybe someone will know how I can change CPU's running.

It has been my understanding that the 0.04 represents how much of the cpu is used by CUDA in your system.

---

But does anyone know how that could be increased?

---

Q6600 @ 2.40GHz
Vista Ultimate x64 SP1
Mem: 4GB
8800GT 181.22
BOINC 6.6.3 Win64

ACTIVE PROJECTS

But does anyone know how that could be increased?

my guess would be in your client_state file under <avg_ncpus>0.04000</avg_ncpus>
<max_ncpus>0.04000</max_ncpus>
though I'm not sure it's something you should increase.

---

Thanks...have now set too 0.25 CPUs, 1 CUDA. Will test to see how this works.

---

Q6600 @ 2.40GHz
Vista Ultimate x64 SP1
Mem: 4GB
8800GT 181.22
BOINC 6.6.3 Win64

ACTIVE PROJECTS

Found this CUDA and multi CPU?. It explains how to get boinc to use a specific number of cores (including the core from gpu).
Its a temp-fix for boinc v6.4.5 and below.

---

Thanks...have now set too 0.25 CPUs, 1 CUDA. Will test to see how this works.

The value 0.06 seems to come from the following calculation

// the following is for an app that uses a CUDA GPU
00066         // and some CPU also, and gets 50 GFLOPS total
00067         //
00068         for (unsigned int i=0; i<sreq.coprocs.coprocs.size(); i++) {
00069             COPROC* cp = sreq.coprocs.coprocs[i];
00070             if (!strcmp(cp->type, "CUDA")) {
00071                 COPROC* cu = new COPROC (cp->type);
00072                 cu->count = 1;
00073                 hu.coprocs.coprocs.push_back(cu);
00074                 double x = 1e9/sreq.host.p_fpops;
00075                 if (x > 1) x = 1;
00076                 hu.avg_ncpus = x;
00077                 hu.max_ncpus = x;
00078                 hu.flops = 5e11;
00079                 return true;
00080             }
00081         }

So a huge number of fpops means you are only going to use a fraction of the number of cpus (or threads) available. That is assuming you got at least 1e9 flops.

I got this by grep'ing "CPUs" in the boinc source and then backtracking avg_ncpus to http://boinc.sourcearchive.com/documentation/6.2.18/sched__plan_8C-source.html

I am guessing the value is calculated when the cuda driver is loaded which could account for a value near 0.95 that you mentioned. That could happen if the cpu is really busy doing something when the CUDA code calculates avg_ncpus. That then gets stored in the resource file (the xml file) and the boinc manager picks it up. It assume it gets updated later on (??) to a better value. Basically the value you see is 1e9 divided by what it think the flops are. This is just a guess as I am not privi to the goings on and dont see a lot of comments in the code. I once complained about source code comments and was asked to contribute time and resources so I dont complain anymore. This is perfectly understandable in a grad school environment.

Thanks...have now set too 0.25 CPUs, 1 CUDA. Will test to see how this works.

I'm guessing it's going to reset back to 0.04 in a short time.
It's worth mentioning if your intention is just to speed up the app, using the Modified SETI MB CUDA + opt AP package for full GPU utilization might be a better alternative if it's something you haven't already looked into. There is also an update to the exe of the CUDA portion of the app in this message

---

It would be cool if you could run 2 boinc cores.. one using the CPUs cores doing either/and MBs and APs and the other just handling cuda.
I have been digging in the XML files but I can't seem to find a config that sets the file paths. As far as I know there is NO boinc.ini file. ini files have gone away pretty much in the newer versions of windows. Now its all stored in the registry. I searched it and found nothing usable.
I tried creating a short cut and changing the path and startup folder.. NO GOOD!
Oh well... I hope they get this worked out on the sharing issues.. A simple patch to allocate on slot JUST FOR CUDA would solve a bunch of issues!

---

Not Ready Reading BRAIN. Abort/Retry/Fail?