Volunteers needed - Page 3

axn · 2012-05-28, 08:43

Quote:

Originally Posted by firejuggler

Using your latest version, time drop to 3368-3500 for a core busy with a GPU load of 91/92%

Screen responsivness (is there such a word?) is much better

Does that mean that, even now, 1 full CPU core is being used? Dang. I was hoping to avoid that.

firejuggler · 2012-05-28, 09:20

Yes, a full core.
Since most of the modern computer has 4 core minimum, isn't that a non-issue?

axn · 2012-05-28, 09:53

Quote:

Originally Posted by firejuggler

Yes, a full core.
Since most of the modern computer has 4 core minimum, isn't that a non-issue?

The CPU is not actually doing anything useful most of the time. It is just "busy waiting" for GPU kernel to finish. I am trying to see if I can eliminate that -- you know, trying to be a "team player" :). I'd estimate actual useful CPU work to be < 10% of 1 core.

axn · 2012-05-28, 11:39

Quote:

Originally Posted by axn

The CPU is not actually doing anything useful most of the time. It is just "busy waiting" for GPU kernel to finish. I am trying to see if I can eliminate that -- you know, trying to be a "team player" :). I'd estimate actual useful CPU work to be < 10% of 1 core.

Ok. Take 2. It seems I messed up the order of initialization. If this doesn't work, I have couple more options to try.

firejuggler · 2012-05-28, 11:50

hmmm this new exe and the one before that has the exact same number of byte.
Are you sure they are different?

axn · 2012-05-28, 11:58

Quote:

Originally Posted by firejuggler

hmmm this new exe and the one before that has the exact same number of byte.
Are you sure they are different?

Did you do a byte comparison or just a size comparison? They should be different (I have different timestamps). The only difference is that I've swapped the call order of two functions -- so the size should be identical.

firejuggler · 2012-05-28, 12:05

It was a size comparison.
Then I have a bad news and a good one :
the good: there seem to have a difference
the bad : very, very small, less than 1%
(running two instance of pfgw at the same time as your 63 version. Older version was constantly on top, while your newer is sometime third on top cpu usage)

2012-05-28, 12:05	#29
firejuggler "Vincent" Apr 2010 Over the rainbow 2³×5×73 Posts	It was a size comparison. Then I have a bad news and a good one : the good: there seem to have a difference the bad : very, very small, less than 1% (running two instance of pfgw at the same time as your 63 version. Older version was constantly on top, while your newer is sometime third on top cpu usage) Last fiddled with by firejuggler on 2012-05-28 at 12:07

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Call for volunteers: RSA896	jasonp	Operation Kibibit	134	2013-09-03 22:08
call for volunteers: RSA768 polynomial selection	jasonp	Operation Kibibit	200	2011-11-05 21:31
V24.12 QA help needed	Prime95	Software	5	2005-06-17 15:54
Volunteers needed!	Xyzzy	Hardware	23	2003-04-18 23:27
We need two volunteers...	Xyzzy	PrimeNet	8	2003-02-27 02:26

2012-05-28, 09:20	#24
firejuggler "Vincent" Apr 2010 Over the rainbow 2³·5·73 Posts	Yes, a full core. Since most of the modern computer has 4 core minimum, isn't that a non-issue?

2012-05-28, 11:50	#27
firejuggler "Vincent" Apr 2010 Over the rainbow 2³·5·73 Posts	hmmm this new exe and the one before that has the exact same number of byte. Are you sure they are different?