Message boards :
Number crunching :
BOINC 5.4.9
Message board moderation
Previous · 1 · 2 · 3 · 4 · Next
Author | Message |
---|---|
David@home Send message Joined: 16 Jan 03 Posts: 755 Credit: 5,040,916 RAC: 28 |
Some of the Changes in 5.4.9: Had a look at the changes detail but I can not see any mention of allowing column sort back in BOINC Manager. I have not upgraded yet could somebody confirm is this is now available or not? With the new variable deadlines I would like to sort by deadline so I can see what WU is next, at the moment the scheduler is scheduling by deadline date but it is very difficult to see this in the long list of WUs. |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
I have not upgraded yet could somebody confirm is this is now available or not? It's on the to do list. Or for a volunteer developer or cracker to try to fix. ;) |
Gecko Send message Joined: 17 Nov 99 Posts: 454 Credit: 6,946,910 RAC: 47 |
Gents, I upgraded my core client to 5.4.9 and am running Crunch3r's 4.11 Ap (dated 12.24.05)and his S. Enhanced 5.12 Ap in "switching mode" w/ edited app_info.xml. I'm a little confused at the results below. I thought this work unit and a few remaining like it were seti-standard WUs, yet it appears that it's being crunched by 5.12? I double checked my install files. In my Boic folder, The boinc exe. files show 5.4.9, in Boinc Mgr., the message tabs shows "starting 5.4.9...." In my projects folder, the application files (4.11 exe & pdb, & 5.11 exe) look correct. App_info.xml looks correct. Will core 5.4.9 not process Crunch3r's 4.11 ap? A standard WU typically crunches @ 3200 secs on my rig. These Seti-standard(?)units are running 9K. SE WUs appear fine at 14-16K. Sorry, I must be getting X-eyed w/ all the changes the last few days. This just doesn't look right to me. Ideas? Result ID 321267704 Name 06ja99ab.28689.27232.790908.3.251_1 Workunit 77138294 Created 7 May 2006 14:07:39 UTC Sent 9 May 2006 5:00:46 UTC Received 12 May 2006 1:30:23 UTC Server state Over Outcome Success Client state Done Exit status 0 (0x0) Computer ID 1997635 Report deadline 17 May 2006 8:15:01 UTC CPU time 9649.505307 stderr out <core_client_version>5.3.12.tx36</core_client_version> <stderr_txt> Windows optimized S@H Enhanced application by Crunch3r Version info: Windows SSE2 V5.12 by Crunch3r Work Unit Info: ............... WU true angle range is : 1.0421 Flopcounter: 6980519465498.209000 Spike count: 0 Pulse count: 0 Triplet count: 10 Gaussian count: 0 </stderr_txt> Validate state Valid Claimed credit 27.1464645880486 Granted credit 26.9241356586832 application version 5.12 |
rhakala Send message Joined: 3 Apr 99 Posts: 2 Credit: 353,655 RAC: 0 |
I upgraded to 5.4.9 Universal Binary on my Intel iMac (Still not used to saying that!) - anyway, I just looked at my list of my computers and it says i386 for the processor family: i-m-intel.local 97.33 2,200.14 i386 iMac4,1 Darwin 8.6.1 I expected to see i686 - not important just wondered. thanks......Richard Richard "Age and treachery overcome youth and skill" |
Zap de Ridder Send message Joined: 9 Jan 00 Posts: 227 Credit: 1,468,844 RAC: 1 |
Gents, According to the "ab" bit in that result it's an enhanced result. To see what you are actualy crunching use CTRL-ALT-DEL and see in the taskmanager under processes wich app is crunching. |
Gecko Send message Joined: 17 Nov 99 Posts: 454 Credit: 6,946,910 RAC: 47 |
Gents, Thanks, Guess i just needed someone else to confim It appears, every WU I have downloaded since first installing Enhanced has been Enhanced WUs, even though some appear closer in time to standard WUs. |
Bob Guy Send message Joined: 7 Sep 00 Posts: 126 Credit: 213,429 RAC: 0 |
I had some trouble updating from 5.2.13 to 5.4.9, I was using the truXoft client. When I ran the 5.4.9 installer it wanted the Boinc.msi from the 5.2.13 which no longer existed in the folder it named - windows/downloaded installations/... I had to find the old 5.2.13 download and fake an installation of it to get the old .msi file on disk - then the 5.4.9 used the 5.2.13 .msi to uninstall the old Boinc. Without the old file the old Boinc could not be uninstalled and the new 5.4.9 could not be installed - this is unacceptable. The old .msi file is also in the windows/installer folder (I think) and the 5.4.9 installer should look there, not in the downloaded installations folder which is really a temporary folder - I regularly clean this folder out! You should be able to uninstall Boinc without having the install files present. It would seem reasonable to keep a list of the files (and other changes) in the registry so that Boinc could be uninstalled when a user does not have access to the original installer or when there has been a loss of files on disk. At the very least the .msi file should be automatically looked for in the windows/installer folder since this folder is more or less permanent and not ordinarily tampered with. I just happen to keep copies of the Boinc downloads so it was just some extra work for me to do this but most users would be stumped by this install failure. And a note: The truXoft client (tx36) does not work at all with the 5.4.9 Boinc. I use it now only for Einstein so I'll miss it - I hope there will be a new 5.4.9 compatible truXoft client available soon. |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
I had to find the old 5.2.13 download and fake an installation of it to get the old .msi file on disk - then the 5.4.9 used the 5.2.13 .msi to uninstall the old Boinc. Without the old file the old Boinc could not be uninstalled and the new 5.4.9 could not be installed - this is unacceptable. The old .msi file is also in the windows/installer folder (I think) and the 5.4.9 installer should look there, not in the downloaded installations folder which is really a temporary folder - I regularly clean this folder out! The installer for 5.4.9 looks in the C:\\WINNT\\Installer or C:\\WINDOWS\\Installer directory for an old .MSI file it can uninstall. That you installed a different version of Boinc afterwards, is not a problem that UCB can anticipate. The Truxoft client is another BOINC version. If you have any complaints about where the new Boinc is looking for its uninstall files, after you installed an optimized Boinc version, then give them to Trux. You should be able to uninstall Boinc without having the install files present. You should be able to remember what you did. You replaced boinc.exe by another version. Since the uninstaller keeps MD5 hash tabs on the client you installed, it couldn't find the old boinc.exe anymore and thus you ran into trouble. Instead of overwriting the file, renaming the original boinc.exe to boinc.old at the time, and before trying to uninstall (or have 5.4.9's installer uninstall it) the old files, deleting the Truxoft boinc.exe and renaming the boinc.old file again was too much trouble, I guess. And a note: The truXoft client (tx36) does not work at all with the 5.4.9 Boinc. I use it now only for Einstein so I'll miss it - I hope there will be a new 5.4.9 compatible truXoft client available soon. That one will then only be compatible for Einstein, as Seti Enhanced doesn't use the calibration for your credits and benchmarks. Anyway, you will need to ask Trux about it. I am sure he's got an email address on his site? |
Bob Guy Send message Joined: 7 Sep 00 Posts: 126 Credit: 213,429 RAC: 0 |
That's not accurate - the truXoft client was not installed with an installer. The files were simply unzipped and replaced. I did, in fact, rename and keep the originals.
That's also not accurate - I had the original files in place and I did rename them prior to attempting an install. Instead of overwriting the file, renaming the original boinc.exe to boinc.old at the time, and before trying to uninstall (or have 5.4.9's installer uninstall it) the old files, deleting the Truxoft boinc.exe and renaming the boinc.old file again was too much trouble, I guess. There is no truXoft uninstaller. The 5.4.9 installer looked for the 5.2.13 installer to uninstall the files and, since it wasn't present, the installer stalled. I did not overwrite any file! I renamed and kept the old 5.2.13 files and renamed them back before attempting the 5.4.9 install. The 5.4.9 installer never looked in the windows/installer folder or never found the proper files. I've briefly looked in the registry for the proper installer entries but didn't find anything at first look. I'll look again and look at my install log for 5.2.13. I run a file/registry log program before every install and a compare afterward. This lists ALL the registry changes made by an install as well as listing all file changes. At first look I think the 'file source' installer entry is the culprit but I'm not an installer expert. That one will then only be compatible for Einstein, as Seti Enhanced doesn't use the calibration for your credits and benchmarks. Anyway, you will need to ask Trux about it. I am sure he's got an email address on his site? I already know that, that's exactly why I mentioned it. I have done a number of seti-beta WUs so I'm completely familiar with the new credit scheme. Did you hear me ask about calibration for seti? Please, I don't mean this response to be mean-spirited. I just thought I would clear up what I consider a misunderstanding of the problem. |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
I've briefly looked in the registry for the proper installer entries but didn't find anything at first look. I'll look again and look at my install log for 5.2.13. I run a file/registry log program before every install and a compare afterward. This lists ALL the registry changes made by an install as well as listing all file changes. At first look I think the 'file source' installer entry is the culprit but I'm not an installer expert. The only files written to the registry are for the Boinc Manager. You can find them at HKEY_CURRENT_USER\\Software\\Space Sciences Laboratory, U.C. Berkeley If you are looking for installer files, BOINC uses the Windows Installer. So the installer/uninstaller files are written underneath a string value of 32 characters. And wherever that is in the registry, it coincides with the position BOINC takes in on Add/Remove Programs. That's not accurate - the truXoft client was not installed with an installer. The files were simply unzipped and replaced. I did, in fact, rename and keep the originals. Yes, but did you reinstall 5.2.13 after you renamed the files? Renaming a file takes the MD5 hash off of it (as far as I know). So you rename the file, then reinstall 5.2.13, and then immediately try either an install over it with 5.4.9 or uninstall 5.2.13 first. Please, I don't mean this response to be mean-spirited. I just thought I would clear up what I consider a misunderstanding of the problem. I hope I didn't come across too cross either (yes, I do ;)), it's just a thing that a lot of people will wonder about without a clear answer. With your help I hope we have given it to that part that's reading this. :) |
Bob Guy Send message Joined: 7 Sep 00 Posts: 126 Credit: 213,429 RAC: 0 |
I know where the registry entries are - I have a complete listing of all the changes. Yes, but did you reinstall 5.2.13 after you renamed the files? Renaming a file takes the MD5 hash off of it (as far as I know). So you rename the file, then reinstall 5.2.13, and then immediately try either an install over it with 5.4.9 or uninstall 5.2.13 first. I am sure that the name of the file has nothing to do with the md5 - the md5 is computed from the contents of the file. I never re-installed 5.2.13 - I did a 'fake' install. You run the installer just to the first screen so that the .msi gets extracted then cancel. At that point all that you have done is unpack the .msi file - nothing has been installed, replaced or changed - even the registry remains untouched. |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
No, but what you did so far was tell for the Windows Installer/Uninstaller to see where the file it was looking at was. And at that, the complete file, as it had installed it. Meaning that boinc.exe at that time has the correct hash and the correct size. |
Bob Guy Send message Joined: 7 Sep 00 Posts: 126 Credit: 213,429 RAC: 0 |
No, but what you did so far was tell for the Windows Installer/Uninstaller to see where the file it was looking at was. And at that, the complete file, as it had installed it. All that was needed was to have the 5.2.13 .msi installer file available to the 5.4.9 installer to uninstall the 5.2.13 application. Unpacking the 5.2.13 .msi file restored the proper directory and file. I really don't think it matters what the md5 is/was of the original Boinc files - I think the installer runs properly even if the files are not present. You just have to have the 5.2.13 .msi file available. |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
So, what happened on your system that you didn't have the correct .msi files present then? Isn't that something you need to investigate first, before blaming Boinc 5.4.9's installer not being capable of finding the files? :) |
Bob Guy Send message Joined: 7 Sep 00 Posts: 126 Credit: 213,429 RAC: 0 |
So, what happened on your system that you didn't have the correct .msi files present then? Isn't that something you need to investigate first, before blaming Boinc 5.4.9's installer not being capable of finding the files? :) The problem is that the folder used is generally considered to be a temporary folder. This is not a problem for installing - it is only a problem when uninstalling. And in any case, the file (.msi) was also installed into the windows/installer folder (as is proper) but the installer's uninstaller would not look in the windows/installer folder. The installer's uninstaller insisted on looking in that temporary folder. And, yes there are too many same sounding words in the previous statement. It's enough to make your brain hurt. It begins to sound like Monty Python. |
jedimstr Send message Joined: 23 Oct 00 Posts: 33 Credit: 16,828,887 RAC: 0 |
hmmm.. what I don't get is what you did different that messed with 5.4.9's install/uninstall routine. I had the Truxoft 5.2.13tx37 files copied over an install of 5.2.13 on 6 different systems so that they would calibrate Crunch3r's 4.11. I was able to install 5.4.9 cleanly over them today without any problems whatsover now that most of my WU's are running on Crunch3r's 5.12. No hiccups, just went straight into the client upgrade messages and re-benchmark. Did you try copying the 5.2.13tx3? files over 5.4.9 at some point??? |
jedimstr Send message Joined: 23 Oct 00 Posts: 33 Credit: 16,828,887 RAC: 0 |
hmmm.. what I don't get is what you did different that messed with 5.4.9's install/uninstall routine. I had the Truxoft 5.2.13tx37 files copied over an install of 5.2.13 on 6 different systems so that they would calibrate Crunch3r's 4.11. I was able to install 5.4.9 cleanly over them today without any problems whatsover now that most of my WU's are running on Crunch3r's 5.12. No hiccups, just went straight into the client upgrade messages and re-benchmark. Did you try copying the 5.2.13tx3? files over 5.4.9 at some point??? |
Bob Guy Send message Joined: 7 Sep 00 Posts: 126 Credit: 213,429 RAC: 0 |
Did you try copying the 5.2.13tx3? files over 5.4.9 at some point??? No, and why would I want to do that? I've got 5.4.9 running just fine now - it only took a couple of minutes to find and fix the problem. I just wanted to report my trouble so that others who might have the same trouble might learn something useful from my experience. And now for something entirely different... The Boinc 5.4.9 reports a very low benchmark - looks like a penalty when used on a P4 Prescott. It seems to be reporting a benchmark divided by 2. That's fine if you have a dual core, but I don't - I have a Prescott HT. I can never get better than a ~%10..%15 increase in throughput even when running 2 apps at the same time. And, I currently have Boinc set to use only 1 CPU (for general useability, temp moderation and power consumption) so my machine is actually running a given app faster since there's no competition. I remember that this is an old problem with HT and one of the reasons why I used a calibrating client - which you can't do with the current tx36 client and Boinc 5.4.9. Now don't write back and tell me that Seti doesn't use the benchmark - I already know that. I do run other Boinc projects that still do use the benchmark and the low numbers give my machine a big disadvantage. |
Jack Gulley Send message Joined: 4 Mar 03 Posts: 423 Credit: 526,566 RAC: 0 |
The problem is that the folder used is generally considered to be a temporary folder. Some people are reading too fast, missing details, seeing what they dislike and jumping bottom end first.. Bob, Are you sure it was looking in the Windows\\Downloaded Installations for the BOINC.MSI file? That would be a temporary download folder (on Windows 98/ME systems at least) and its contents often deleted. Your only listed system is a Windows XP system, and the uninstaller should have been looking in the C:\\WINNT\\Installer folder as pointed out, so I have to agree with you! It was looking in the wrong place for some reason. Replacing one of the files with another version would not have had any effect on where the uninstaller should be looking, unless that installer was very confused by something on your system. |
jedimstr Send message Joined: 23 Oct 00 Posts: 33 Credit: 16,828,887 RAC: 0 |
Did you try copying the 5.2.13tx3? files over 5.4.9 at some point??? No reason, just diagnosing what could have gone wrong. I was just trying to figure out why your install would have problems when my similar install parameters did not. As for the benchmarks...noticed that too, and not only on P4 based systems. From lowly Coppermine PIII's to my P4s, PentD's, Athlon64's, G4's and a G5 Quad, all the benchmarks were much lower than I remember for the official clients. I'm hoping Truxoft comes out with a new calibrating client based off the new source soon. Even on his beta and alpha release list, his clients are all 5.3.x based. |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.