Making use of multiple cores on i7 processor with FAH client(s)

Hey guys, nuB folder here

I searched and found info on running multiple instances with HT processors, and how to put them in separate directories and what-not, but I'm curious about something...

I'm running an i7 920; 4 physical cores, each with hyper threading; so Windows sees it as 8 cores, but that's neither here nor there...

I use Everest Ultimate and monitor my CPU usage, memory usage, and get a general idea of my temps, and keep them up on my desktop on the sidebar.

Now, I noticed that if I start up a few FAH clients and change the Machine ID's to 1, 3, 5, and 7 (all the odd number's cores; just so happens to be the actualy physical cores), it really starts to make use of ALL the cores, which is good, that's what I was going for.

On top of that, I'm running a GPU client on machine ID 8, but that really doesn't make too much use of the CPU as it does the GPU.

But what I'm wondering is, will this really change my PPD much?

It's looking like just off the one CPU client and the one GPU client, this computer's going to turn out just under 800 PPD; about 400 from the CPU, and 400 from the GPU, roughly. But can I stack these up and really use this i7 and turn out some serious numbers like this?


Proof of concept here; obviously not efficient, having all the displays up and showing but you see what I'm doing here; I showed the hidden icons in the lower right to show the running FAH clients as well. Note that I had just made the changes to the configuration, so most of them had restarted and were getting work from the server; it doesn't stay at 29%, lol, it's running at about 54% right now with all of them going.






I haven't noticed and ill side effects yet; the computer's still running just fine; no stumbles or hesitations yet, but the CPU's only running at about half tilt




But I was just curious if anyone's experimented with this yet on the i7's. I mean, I'm sure there's probably a bunch of you that have, but I'm just wandering if I can expect a good size growth in PPD running this setup. After all, that's what it's all about right? the competition? haha, jk

alrighty, well I restarted the whole process :roll:


I kinda skipped the obvious step of creating separate directories, even though I mentioned it with my own words in my first post :roll:


Now having it running out of 4 different directories for the CPU client, and one more for the GPU client


So far today, it looks like it boosted my PPD just a little, about 300 woopty do, right... but I had to restart the computer, and with it being all out of one directory, none of the progress on the current work units was saved to be continued out, so who knows how many points that threw away


But we'll see what happens now

Two comments.

1. Don't use the viewer, it slows down the processing and is known to be buggy.

2. Instead of running four single-core clients, you will get much higher PPD with a single SMP client. Someone else here may help you do that on Windows, I only know how on Linux.

You should be able to get at least 6k PPD with a GPU client and a Windows SMP client together on your hardware (possibly more, I don't know ATI either).

Ben N1NP

Thanks for the advice!

I'll look into the SMP client right now!


REP!!!

'tis up and running

let's see how this changes things

seems to be working

3k points today, and it's only 6:45pm yet





I might add that I'm only running the processor at 60%, only because I'm on the stock cooler...

I have rejected yours. Another post of this nature and you will banned.