Hey,

I run BOINC on two computers, my laptop and my desktop.

Desktop: Intel(R) Core(TM) i5 CPU 750 @ 2.67GHz (4 processors)
Laptop: Intel(R) Core(TM) i7-2630QM CPU @ 2.00GHz (8 processors)

I have graphics cards installed on each and notice that the i5 runs 5 tasks at once (4 processors + the gpu?) and the laptop runs 9 (8 processors + the gpu?)

I notice that the desktop i5 is a lot faster in getting through tasks than the laptop i7 . I was wondering if:

1. BOINC processes tasks based on the number of CPU/GPUs?
2. We can change the settings in BOINC so that multiple cores can be used to process individual tasks?
3. Would this even make a difference?
4. Should I expect my laptop to be faster given that the measured floating point speed and measured integer speed are greater compared to the desktop?
5. Am I even making sense or am I just a noob?

Thanks,

Jack.

I was wondering if:

1. BOINC processes tasks based on the number of CPU/GPUs?

This depends on which project you're attached to and whether that project uses multi-threaded applications or not. But by default, BOINC will set 1 task per CPU core and 1 task per GPU.

Then there's the anonymous platform that allows you to fiddle with how many tasks you run on the GPU, although you'll probably need a good ATI or Fermi style Nvidia for that (with lots and lots of memory). But that's all a hands on job, nothing done on the automatic by BOINC or any of the project's science apps.

2. We can change the settings in BOINC so that multiple cores can be used to process individual tasks?

No. To do so the project needs to have a multi-threaded science application and then it will only either use all cores, or none (and thus not work).
Multi-threaded applications do go faster, but then the projects that do use them will use them on their larger tasks, so they still take hours. Milkyway's MT app is an exception, I've seen, it does things in mere minutes.

4. Should I expect my laptop to be faster given that the measured floating point speed and measured integer speed are greater compared to the desktop?

Well, your laptop has a better processor when compared to the desktop, but then it's a laptop, which run hot when just surfing a bit on the net. They're notoriously difficult to cool, and so it doesn't really matter if its benchmarks are better or not, you won't be able to run at full blast all the time due to heat problems (and on battery, it's power problems as well).

Besides that, the benchmarks aren't really being used anymore by any of the main projects. Cosmetics. Easier to keep in than take out. For now.

---

That's the strange thing, I'm only running milkyway@home and it's still processing each task on one processor (are 'core' and 'processor' synonymous?). But yes on my i5 milkyway tasks are finished in minutes (the ones running on my GPU, I have the HD5850).

Thanks for the great info, much appreciated.

GPU applications go faster anyway. A GPU has multiple shader processors, from anywhere between 4 and 960 processors. All these processors will attack the one task at the same time, thereby speeding up the time to do the work in enormously.

While the science application will run on the CPU, that's all it does. All calculations are done inside the GPU.

---