AP wu's?

Message boards : Number crunching : AP wu's?

1 · 2 · 3 · Next

Author	Message
Miklos M. Send message Joined: 5 May 99 Posts: 597 Credit: 14,701,058 RAC: 84	Message 1164488 - Posted: 22 Oct 2011, 14:17:58 UTC
	I have not been getting any AP wu's lately and my cue has only one left, crunching six at this time. Has anyone else had problems receiving AP units lately? Will my six cores crunch seti enhanced units if the AP runs dry? Thanks. ____________
	ID: 1164488 ·

BilBg Send message Joined: 27 May 07 Posts: 2023 Credit: 3,054,590 RAC: 6,590	Message 1164490 - Posted: 22 Oct 2011, 14:30:33 UTC - in response to Message 1164488.
	Will my six cores crunch seti enhanced units if the AP runs dry? Depends on: 1) Your SETI@home preferences: http://setiathome.berkeley.edu/prefs.php?subset=project Are all the options at the bottom of page set to yes?: SETI@home Enhanced: yes Astropulse v5: yes (irrelevant/obsolete) Astropulse v5.05: yes If no work for selected applications is available, accept work from other applications? yes 2) Since you are running anonymous platform it depends also on what apps you have in your app_info.xml ____________ - ALF - "Find out what you don't do well ..... then don't do it!" :)
	ID: 1164490 ·

Miklos M. Send message Joined: 5 May 99 Posts: 597 Credit: 14,701,058 RAC: 84	Message 1164494 - Posted: 22 Oct 2011, 14:59:19 UTC - in response to Message 1164490.
	They are all yes and I will find out about app_info.xml pretty soon. Hope it works. Thank you ____________
	ID: 1164494 ·

Donald L. Johnson Send message Joined: 5 Aug 02 Posts: 3938 Credit: 362,030 RAC: 824	Message 1164511 - Posted: 22 Oct 2011, 15:46:44 UTC Last modified: 22 Oct 2011, 16:02:03 UTC
	See this message in the HE Connect Problems thread. They let the AP splitters run dry so they could test the new router memory. Should be more AP tasks available later today (Saturday 10/22). ____________ Donald Infernal Optimist / Submariner, retired
	ID: 1164511 ·

msattler Volunteer tester Send message Joined: 9 Jul 00 Posts: 34289 Credit: 346,110,133 RAC: 554,111	Message 1164514 - Posted: 22 Oct 2011, 15:54:48 UTC - in response to Message 1164511.
	See this message in the HE Connect Problems thread. They let the AP splitters run dry so they could test the new router memory. Should be more Ap tasks available later today (Saturday 10/22). AP splitting started back up Friday already. But it is going out about as fast as it is being split, so not all work requests for AP are going to get WUs allocated. ____________ ****** "Ask not, what your kitty can do for you. Ask what you can do for your kitty." As it is kitten, so shall it be done.
	ID: 1164514 ·

Sten-Arne Volunteer tester Send message Joined: 1 Nov 08 Posts: 2690 Credit: 7,766,719 RAC: 25,568	Message 1164516 - Posted: 22 Oct 2011, 15:56:41 UTC - in response to Message 1164511.
	See this message in the HE Connect Problems thread. They let the AP splitters run dry so they could test the new router memory. Should be more Ap tasks available later today (Saturday 10/22). The AP splitters started getting new files to split, just a couple of hours after that message, and I've been getting new AP units for almost 24 hours now. ____________ /The grumpy old Swede. "I'm so old, that 98% of all trees in the forest, are younger than I am"
	ID: 1164516 ·

Donald L. Johnson Send message Joined: 5 Aug 02 Posts: 3938 Credit: 362,030 RAC: 824	Message 1164517 - Posted: 22 Oct 2011, 16:01:13 UTC - in response to Message 1164514.
	See this message in the HE Connect Problems thread. They let the AP splitters run dry so they could test the new router memory. Should be more Ap tasks available later today (Saturday 10/22). AP splitting started back up Friday already. But it is going out about as fast as it is being split, so not all work requests for AP are going to get WUs allocated. That's what happens when I don't check in for a day or two, and then don't look at the message date stamp. (8{) Yes, as the MB splitters finish a file (they take longer than the AP splitters), we'll see more new data files for the AP splitters to work through. Patience, as usual..... ____________ Donald Infernal Optimist / Submariner, retired
	ID: 1164517 ·

Mike Volunteer tester Send message Joined: 17 Feb 01 Posts: 19511 Credit: 21,237,653 RAC: 26,726	Message 1164523 - Posted: 22 Oct 2011, 16:18:05 UTC
	I´m getting APs constantly even i´m hitting the limits. ____________
	ID: 1164523 ·

Miklos M. Send message Joined: 5 May 99 Posts: 597 Credit: 14,701,058 RAC: 84	Message 1164524 - Posted: 22 Oct 2011, 16:19:47 UTC - in response to Message 1164511.
	Looking forward to that. One of my machines is getting them already, but the cuda is still waiting. ____________
	ID: 1164524 ·

Mike Volunteer tester Send message Joined: 17 Feb 01 Posts: 19511 Credit: 21,237,653 RAC: 26,726	Message 1164529 - Posted: 22 Oct 2011, 16:34:08 UTC - in response to Message 1164524. Last modified: 22 Oct 2011, 16:34:32 UTC
	Looking forward to that. One of my machines is getting them already, but the cuda is still waiting. I wouldn´t run APs on your Phenom Miklos. I would rather run them on your 460 would be more efficient. My host is normally above 22000 RAC without those limits and hiccups. And your 460 is a little faster than my 5850. ____________
	ID: 1164529 ·

Miklos M. Send message Joined: 5 May 99 Posts: 597 Credit: 14,701,058 RAC: 84	Message 1164543 - Posted: 22 Oct 2011, 17:40:37 UTC - in response to Message 1164529.
	How would I do that Mike? ____________
	ID: 1164543 ·

Josef W. Segur Volunteer developer Volunteer tester Send message Joined: 30 Oct 99 Posts: 3932 Credit: 934,327 RAC: 15	Message 1164612 - Posted: 22 Oct 2011, 21:19:55 UTC - in response to Message 1164543.
	Mike wrote: I wouldn´t run APs on your Phenom Miklos. I would rather run them on your 460 would be more efficient. How would I do that Mike? I'm not Mike, but see the Public beta for nVidia AstroPulse, rev 521 thread for information. Joe
	ID: 1164612 ·

Mike Volunteer tester Send message Joined: 17 Feb 01 Posts: 19511 Credit: 21,237,653 RAC: 26,726	Message 1164620 - Posted: 22 Oct 2011, 21:58:26 UTC - in response to Message 1164612. Last modified: 22 Oct 2011, 21:59:27 UTC
	Mike wrote: I wouldn´t run APs on your Phenom Miklos. I would rather run them on your 460 would be more efficient. How would I do that Mike? I'm not Mike, but see the Public beta for nVidia AstroPulse, rev 521 thread for information. Joe Thanks Joe. If you decide to try r521 i suggest following settings for your 460. <cmdline>-ffa_block 8192 -ffa_block_fetch 2048 -unroll 10 -instances_per_device 2</cmdline> <count>0.5</count> That will run 2 units a time and should be most efficient on your card. ____________
	ID: 1164620 ·

Fred J. Verster Volunteer tester Send message Joined: 21 Apr 04 Posts: 3232 Credit: 31,585,541 RAC: 1	Message 1164634 - Posted: 22 Oct 2011, 23:05:52 UTC - in response to Message 1164620. Last modified: 22 Oct 2011, 23:06:53 UTC
	I now use 16384 ffa_bock and 8192 block_fetc, unroll 16, 1 instance_per_device, also since I've too little AstroPulse work, this on 2 ATI 5870 GPUs. Rev.521 for AMD ATI HD5. ____________ Knight Who Says Ni N!, OUT numbered.................
	ID: 1164634 ·

arkayn Volunteer tester Send message Joined: 14 May 99 Posts: 3096 Credit: 40,540,108 RAC: 66,751	Message 1164637 - Posted: 22 Oct 2011, 23:13:38 UTC - in response to Message 1164620.
	Mike wrote: I wouldn´t run APs on your Phenom Miklos. I would rather run them on your 460 would be more efficient. How would I do that Mike? I'm not Mike, but see the Public beta for nVidia AstroPulse, rev 521 thread for information. Joe Thanks Joe. If you decide to try r521 i suggest following settings for your 460. <cmdline>-ffa_block 8192 -ffa_block_fetch 2048 -unroll 10 -instances_per_device 2</cmdline> <count>0.5</count> That will run 2 units a time and should be most efficient on your card. I go slightly different on my GTX460/768 I run 0.49 count for MB and 0.51 for AP, that way I can run either 2 MB or 1 of each on the card. ____________
	ID: 1164637 ·

LadyL Volunteer tester Send message Joined: 14 Sep 11 Posts: 1679 Credit: 5,230,097 RAC: 14	Message 1164986 - Posted: 24 Oct 2011, 9:40:15 UTC
	Miklos, I suggest you drop the NVidia driver to below 270.x unless you really want to test how excessively the CPU core grabbed by the GPU under newer drivers is actually used.
	ID: 1164986 ·

Mike Volunteer tester Send message Joined: 17 Feb 01 Posts: 19511 Credit: 21,237,653 RAC: 26,726	Message 1164987 - Posted: 24 Oct 2011, 9:46:12 UTC - in response to Message 1164986.
	Miklos, I suggest you drop the NVidia driver to below 270.x unless you really want to test how excessively the CPU core grabbed by the GPU under newer drivers is actually used. He was running 267x drivers when i suggested it. ____________
	ID: 1164987 ·

Miklos M. Send message Joined: 5 May 99 Posts: 597 Credit: 14,701,058 RAC: 84	Message 1165111 - Posted: 24 Oct 2011, 20:06:36 UTC
	I am sorry guys, but you lost me in the technical language. ____________
	ID: 1165111 ·

Blurf Volunteer tester Send message Joined: 2 Sep 06 Posts: 6357 Credit: 5,680,177 RAC: 4,655	Message 1165113 - Posted: 24 Oct 2011, 20:16:37 UTC
	I pulled 12 today...since I'm now doing Astropulse only I'm thrilled. ____________
	ID: 1165113 ·

LadyL Volunteer tester Send message Joined: 14 Sep 11 Posts: 1679 Credit: 5,230,097 RAC: 14	Message 1165250 - Posted: 25 Oct 2011, 8:26:00 UTC - in response to Message 1165111.
	I am sorry guys, but you lost me in the technical language. You have upgraded the graphics driver to 270.61 With that and newer drivers the OpenCL applications show excessive CPU usage. That may or may not be 'real' usage but nobody has up to now tested whether throughput on CPU suffers and to what extent. I suggest you downgrade your graphic driver to 266.xx if you want to run OpenCL applications such as the r521 AP for NVidia beta. Installing a beta app requires manual editing of app_info.xml. If you do not wish to do that I personally would recommend upgrading to 280.xx driver. Yes, I know, the 285.xx is out, but nobody has thoroughly checked performance yet.
	ID: 1165250 ·

1 · 2 · 3 · Next

Message boards : Number crunching : AP wu's?