Lunatics Windows Installer v0.39 release notes

Message boards : Number crunching : Lunatics Windows Installer v0.39 release notes

Previous · 1 · 2 · 3 · 4 · 5 . . . 13 · Next

Author	Message
LadyL Volunteer tester Send message Joined: 14 Sep 11 Posts: 1679 Credit: 5,230,097 RAC: 22	Message 1174377 - Posted: 28 Nov 2011, 20:40:19 UTC - in response to Message 1174375.
	...yeah...the WIN41 thing threw me as well...dont know where the he77 that came from. I typoed. should be 32. ok no need to panic, it's the 6.08 appversion entry, so unless somebody on 64 bit was getting 6.08 stock the entry won't be needed. You can manually edit the 41 to 32 and we'll get a fixed copy up.
	ID: 1174377 ·

Richard Haselgrove Volunteer tester Send message Joined: 4 Jul 99 Posts: 7453 Credit: 39,531,665 RAC: 32,412	Message 1174378 - Posted: 28 Nov 2011, 20:42:27 UTC - in response to Message 1174372.
	Windows 41? Edit: Looks as if it'll only affect people who were running Cuda 6.08 with a windows_x86_64 platform tag, Claggy Well done Claggy - yes, that's it. Typo in an AIstub. 64-bit installer only, that version number only. HOLD on 64-bit installs, we'll get that file refreshed and notify y'all when it's ready.
	ID: 1174378 ·

LadyL Volunteer tester Send message Joined: 14 Sep 11 Posts: 1679 Credit: 5,230,097 RAC: 22	Message 1174382 - Posted: 28 Nov 2011, 21:03:46 UTC
	A fresh, fixed copy is now available. Thank you for spotting the error.
	ID: 1174382 ·

QBNServer Volunteer tester Send message Joined: 27 Jul 00 Posts: 24 Credit: 7,225,245 RAC: 74	Message 1174384 - Posted: 28 Nov 2011, 21:09:46 UTC
	..just a comment...in the next go-around, why not have a user-entry field asking for how many iterations per GPU to run. That way when the app_info file is generated, a user doesnt have to go in and modify it? Thanks for all you have done so far! ____________
	ID: 1174384 ·

Richard Haselgrove Volunteer tester Send message Joined: 4 Jul 99 Posts: 7453 Credit: 39,531,665 RAC: 32,412	Message 1174387 - Posted: 28 Nov 2011, 21:27:59 UTC - in response to Message 1174384.
	..just a comment...in the next go-around, why not have a user-entry field asking for how many iterations per GPU to run. That way when the app_info file is generated, a user doesnt have to go in and modify it? Thanks for all you have done so far! That's a good suggestion, but it would be quite clumsy to do using the present installer code - we just copy complete files at the moment, without having any tools to modify their contents as we go along. The current thinking is perhaps to ship a separate tool which could modify not only the <count> number, but also <flops>, and the wider range of command-line settings used for the ATI OpenCL applications: we could then launch that tool at the end of the main installation. But it all depends on having enough volunteer programmers to do that, on top of the optimisation work. Thanks - keep the ideas coming.
	ID: 1174387 ·

janneseti Send message Joined: 14 Oct 09 Posts: 73 Credit: 379,838 RAC: 444	Message 1174395 - Posted: 28 Nov 2011, 22:02:17 UTC - in response to Message 1174361.
	Tests on V7 tasks everytime gave false positive autocorrelation signals on each V7 task tested. On both Cat 11.10 and 11.11. So I consider these driver badly broken (there is at least one another application for wich invalid computations already reported under these drivers). And recommend to stay away from them if possible. AMD trying to fix, but they do any software improvements VERY slow (and break working things very quickly...). But looks likeon V6 tasks app give correct results. So, if you absolutely need to install new crappy Catalyst you may be as "after release" tester and check if all v6 processed OK or I was just lucky... I'm confused over this. V7 tasks vs V6 task? Is there several types of tasks (WU's)? Maybe I shouldn't worry cause I haven't got any problem with any of the Cat drivers, but perhaps that might change.
	ID: 1174395 ·

Mike Volunteer tester Send message Joined: 17 Feb 01 Posts: 19452 Credit: 21,045,795 RAC: 26,652	Message 1174396 - Posted: 28 Nov 2011, 22:10:04 UTC - in response to Message 1174395.
	Tests on V7 tasks everytime gave false positive autocorrelation signals on each V7 task tested. On both Cat 11.10 and 11.11. So I consider these driver badly broken (there is at least one another application for wich invalid computations already reported under these drivers). And recommend to stay away from them if possible. AMD trying to fix, but they do any software improvements VERY slow (and break working things very quickly...). But looks likeon V6 tasks app give correct results. So, if you absolutely need to install new crappy Catalyst you may be as "after release" tester and check if all v6 processed OK or I was just lucky... I'm confused over this. V7 tasks vs V6 task? Is there several types of tasks (WU's)? Maybe I shouldn't worry cause I haven't got any problem with any of the Cat drivers, but perhaps that might change. V7 will come but is still in beta phase. ____________
	ID: 1174396 ·

skildude Send message Joined: 4 Oct 00 Posts: 8940 Credit: 36,337,368 RAC: 52,716	Message 1174399 - Posted: 28 Nov 2011, 22:15:50 UTC - in response to Message 1174396.
	and by beta he means Seti@home beta not on Lunatics. ____________ Proud member of TSWB. End terrorism by building a school
	ID: 1174399 ·

Fred E. Volunteer tester Send message Joined: 22 Jul 99 Posts: 432 Credit: 12,798,394 RAC: 32,815	Message 1174403 - Posted: 28 Nov 2011, 22:32:43 UTC Last modified: 28 Nov 2011, 22:33:45 UTC
	I ran into a problem when I edited the new app_info.xml to add the portion for the AP for Nvidia Beta. On startup, BOINC reported it could not find the application files for the Beta app: 11/28/2011 3:40:30 PM \| SETI@home \| File referenced in app_info.xml does not exist: ap_5.06_win_x86_SSE3_OpenCL_NV_r521.exe 11/28/2011 3:40:30 PM \| SETI@home \| File referenced in app_info.xml does not exist: AstroPulse_Kernels_r521.cl I suspended the project before I exited BOINC for the install/edit, but still got 5 of these for the tasks on hand: 11/28/2011 3:40:30 PM \| SETI@home \| [error] No application found for task: windows_intelx86 505 cuda; discarding Found both missing files in the backup directory and moved them to the SETI Project directory, restarted and it looks okay although I can't tell about the AP/Nvidia app yet. I just got resends for the 5 trashed tasks, so should be okay. Anything else I need to fix? (The .pbd file for the beta app was still present.) I also reinserted my flops entries - I assume those are still recommended per the v0.37 release notes. Minor thing - the link on the Lunatics website to the v0.39 readme gets the v0.38 readme. However, the v0.39 readme came with the installation package. ____________ Another Fred Support SETI@home when you search the Web or shop online with GoodSearch and GoodShop
	ID: 1174403 ·

Richard Haselgrove Volunteer tester Send message Joined: 4 Jul 99 Posts: 7453 Credit: 39,531,665 RAC: 32,412	Message 1174407 - Posted: 28 Nov 2011, 22:39:42 UTC - in response to Message 1174395.
	I'm confused over this. V7 tasks vs V6 task? Is there several types of tasks (WU's)? To explain in a little more detail. First, for the time being, forget about Astropulse. Nothing I'm going to write here concerns Astropulse at all. Neglecting Astropulse, there's currently only one type of task: that's the good old search for ET, using an application called setiathome_enhanced, which is currently at version 6 (and various sub-versions - 6.03 runs on CPUs, 6.08/9/10 runs on GPUs, etc.) There are long-term plans afoot to increase the sensitivity of the search for ET, by using a new version 7 of the application to search for an extra type of possible signals in the same data. The new v7 application(s) are currently being written, and those that are ready are being tested at the SETI Beta project. Lunatics are already working on optimised versions of the new applications. The changes between v6 and v7 aren't all that great, so the Lunatics applications, when they're finished, will be able to work on either type of task. But they have to be tested separately for each type of task. Gradually, new apps introduced here will be of that dual v6-v7 capable variety. But here, on the Main project, they will operate exclusively in version 6 mode, to suit the tasks being issued by the project. In due course, the Main project will be upgraded to run the new, more sensitive search. When that happens, there'll be lots of announcements and changes needed. But until then, things carry on as normal. Hope that helps.
	ID: 1174407 ·

Richard Haselgrove Volunteer tester Send message Joined: 4 Jul 99 Posts: 7453 Credit: 39,531,665 RAC: 32,412	Message 1174409 - Posted: 28 Nov 2011, 22:42:24 UTC - in response to Message 1174403.
	Minor thing - the link on the Lunatics website to the v0.39 readme gets the v0.38 readme. However, the v0.39 readme came with the installation package. Thanks for the report - depending on who's awake in which time-zone, we'll get that fixed asap - should be within 12 hours from now.
	ID: 1174409 ·

Richard Haselgrove Volunteer tester Send message Joined: 4 Jul 99 Posts: 7453 Credit: 39,531,665 RAC: 32,412	Message 1174412 - Posted: 28 Nov 2011, 22:45:54 UTC - in response to Message 1174403.
	I ran into a problem when I edited the new app_info.xml to add the portion for the AP for Nvidia Beta... Sorry about that. We overlooked that case in the release notes: again, we'll have another look at it in the morning. Yes, if you're used to using <flops> entries, you can put them back in (from the backup files or fresh from APR) as before.
	ID: 1174412 ·

JohnDK Send message Joined: 28 May 00 Posts: 756 Credit: 20,333,796 RAC: 37,074	Message 1174419 - Posted: 28 Nov 2011, 23:32:04 UTC
	If you have flops and/or command line in the app_info and if you know how to use 7-zip, then extracting the files needed and edit the app_info seems the easiest.
	ID: 1174419 ·

perryjay Send message Joined: 20 Aug 02 Posts: 3359 Credit: 9,605,823 RAC: 9,877	Message 1174437 - Posted: 29 Nov 2011, 0:42:02 UTC
	Okay, things seem to be running right. My thanks to Fred E. for pointing out the problem about the missing files for NVIDIA APs. It will be awhile before I get to the one I have but at least it wasn't discarded. Of course I can't do these things without doing at least one thing wrong. I forgot to change my count to run two MBs at a time before starting up again. Quick easy fix and I am now on my way. Thanks Lunatics, great job as usual. ____________ PROUD MEMBER OF Team Starfire World BOINC
	ID: 1174437 ·

dcappello Send message Joined: 3 Apr 99 Posts: 249 Credit: 164,808,223 RAC: 16,466	Message 1174449 - Posted: 29 Nov 2011, 1:31:40 UTC
	Loading it up on one of my hosts now. Well done guys!!! ____________
	ID: 1174449 ·

skildude Send message Joined: 4 Oct 00 Posts: 8940 Credit: 36,337,368 RAC: 52,716	Message 1174463 - Posted: 29 Nov 2011, 4:02:18 UTC - in response to Message 1174449.
	can I also assume that the Linux apps weren't affected by the "bug" that was found? ____________ Proud member of TSWB. End terrorism by building a school
	ID: 1174463 ·

jason_gee Volunteer developer Volunteer tester Send message Joined: 24 Nov 06 Posts: 4037 Credit: 60,102,061 RAC: 59,008	Message 1174467 - Posted: 29 Nov 2011, 4:23:28 UTC - in response to Message 1174463.
	can I also assume that the Linux apps weren't affected by the "bug" that was found? Yes, Urs Echternacht verified for me that the bug has not propagated to Linux AKv8 builds. ____________ "It is not the strongest of the species that survives, nor the most intelligent that survives. It is the one that is the most adaptable to change." Charles Darwin
	ID: 1174467 ·

S@NL - eFMer - www.efmer.eu/boinc Volunteer tester Send message Joined: 7 Jun 99 Posts: 497 Credit: 110,928,131 RAC: 171,168	Message 1174496 - Posted: 29 Nov 2011, 7:54:32 UTC - in response to Message 1174387.
	The current thinking is perhaps to ship a separate tool which could modify not only the <count> number, but also <flops>, and the wider range of command-line settings used for the ATI OpenCL applications: we could then launch that tool at the end of the main installation. But it all depends on having enough volunteer programmers to do that, on top of the optimisation work. Thanks - keep the ideas coming. I could do that. ____________ TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.
	ID: 1174496 ·

Richard Haselgrove Volunteer tester Send message Joined: 4 Jul 99 Posts: 7453 Credit: 39,531,665 RAC: 32,412	Message 1174497 - Posted: 29 Nov 2011, 8:02:56 UTC - in response to Message 1174496.
	The current thinking is perhaps to ship a separate tool which could modify not only the <count> number, but also <flops>, and the wider range of command-line settings used for the ATI OpenCL applications: we could then launch that tool at the end of the main installation. But it all depends on having enough volunteer programmers to do that, on top of the optimisation work. Thanks - keep the ideas coming. I could do that. That's a very welcome offer, Fred. Let's discuss it at Lunatics.
	ID: 1174497 ·

Terror Australis Volunteer tester Send message Joined: 14 Feb 04 Posts: 1415 Credit: 181,170,907 RAC: 159,303	Message 1174543 - Posted: 29 Nov 2011, 15:13:15 UTC Last modified: 29 Nov 2011, 15:17:47 UTC
	Are the files available without the installer ? EDIT: Please ignore - I got them out with 7-Zip afterall T.A.
	ID: 1174543 ·

Previous · 1 · 2 · 3 · 4 · 5 . . . 13 · Next

Message boards : Number crunching : Lunatics Windows Installer v0.39 release notes