Only see your AMD K7-XP, Athlon XP 2200+ MP, 1799Mhz, (Palomino-662-0) machine using the client so far.
I checked out the source from berkely as the public_419 release source files...then applied the cpu naming and project benchmarking code...so I haven't touched any parts that work with "windows interface", example "hover text over icon in taskbar".
I can't think of why having multiprocessors as opposed to hyperthread would make any significant difference in the way it interfaces with windows (mouse clicking, display, etc.)
How does this machine behave with unmodified 4.19? I could download 4.19 from berkeley into a separate folder and compile it unmodified...see if behavior is there also.
> Only see your AMD K7-XP, Athlon XP 2200+ MP, 1799Mhz, (Palomino-662-0) machine
> using the client so far.
Right, I started both the first & now this second version of your BOINC only on my MP computer so far. When I started the first version, I think it continued processing the WUs it had started under the previons BOINC until they finished & were uploaded. Then it was given a new computer number (but the same name) & posted later results to the web site under that new number. So it could take as long as 14 hours to complete a WU of its own, & could take almost 21 hours before the web site actually allows it to upload a new result, & still a few hours longer for the web site to update itself & actually post that new result. So if it is actually running, we are talking about a 24 hour delay before we might see anything on the web.
I obtained the version I am using by clicking the link in your earlier note here.
> How does this machine behave with unmodified 4.19? I could download 4.19 from
> berkeley into a separate folder and compile it unmodified...see if behavior is
> there also.
As far as I know, the original BOINC 4.19 & BOINC 4.25 & then 4.19 again & then your first version of 4.19 all ran normally on this computer. I have never seen the type of "touchiness" I described earlier with any software on this or any computer.
If you think it would be best, I could now stop BOINC & uninstall it & delete its folder, & then begin again from a clean installation of the original BOINC 4.19.
> I have it running on 14 machines I use without these hover problems...windows
> 2K and windows XP mixed, and one Win 98.
> http://setiweb.ssl.berkeley.edu/hosts_user.php?userid=548863
>
> I can't think of why having multiprocessors as opposed to hyperthread would
> make any significant difference in the way it interfaces with windows (mouse
> clicking, display, etc.)
It just occurred to me that I install BOINC in the directory E:/BOINC/, whereas the default installation directory is Program Files/BOINC/ If you happened to make some reference to the default installation directory, that might cause a problem on my computer but not yours. Just grasping at straws here.
> It just occurred to me that I install BOINC in the directory E:/BOINC/,
> whereas the default installation directory is Program Files/BOINC/ If you
> happened to make some reference to the default installation directory, that
> might cause a problem on my computer but not yours. Just grasping at straws
> here.
>
Nope, as I say, no file changes, no code changes beyond project benchmark, CPU ID code. I have several installed on D:/boinc or F:/data/boinc.
Perhaps some error when uploading or downloading the exe
hmm...no thats not probable as it is compressed with an exe compression program...checks its checksum after decompressing...
I will download it from the website and put it into my folder see if it works as expected.
...nope there..downloaded..renamed...works fine.
Perhaps try launching it by double clicking on it. In that way the gui display should show (File menu, Work Tab, etc.) If it doesnt hmm.
As I say, I don't have an actual dual processor machine to fool around with.
Try the same file copied over (on your network) to one of your single CPU machines (already with 4.19)...does it behave the same way?
> I will download it from the website and put it into my folder see if it works
> as expected.
>
> ...nope there..downloaded..renamed...works fine.
A file of exactly 254,464 bytes, huh?
> Perhaps try launching it by double clicking on it. In that way the gui
> display should show (File menu, Work Tab, etc.) If it doesnt hmm.
Nope, tried, doesn't help.
> As I say, I don't have an actual dual processor machine to fool around with.
>
> Try the same file copied over (on your network) to one of your single CPU
> machines (already with 4.19)...does it behave the same way?
OK, I tried it on my single-CPU computer named "ecollege". Same problem. Has nothing to do with dual processors.
I have now tried downloading the exefile three times to floppies. All have the same problem.
I tried loading 4.19 on a clean laptop that had never seen any boinc before, & then replaced the boinc_gui.exe with your latest version, Same problem. Makes no difference which directory I install it on.
On all these computers, I can later go back to the original 4.19 boinc_gui.exe & then everything works fine again. The problem is reversible.
I don't have a clue now.
In pure desperation, I suppose we could swap the exefiles we are using back & forth by email.
You helped me break the code on the LIGO aiming thing, but now this, too. I know this effort isn't about the "credit", so if it will help make a better program....
I'm also running a dual 3.0 GHz P4 with HT:
Windows benchmarks
.... 3953 whet MIPS per CPU
.... 7874 dhry MIPS per CPU
BOINC benchmarks
.... 1364 whet MIPS per CPU
.... 1150 dhry MIPS per CPU
For the 15 WUs I've processed on this computer:
avg CPU time = 38960.301 secs
avg claimed = 55.96
avg granted = 73.29
However, my single CPU 2.2 GHZ P4 laptop....
BOINC benchmarks
.... 1117 whet MIPS per CPU
.... 2281 dhry MIPS per CPU (Twice that of my dual 3GHz computer!!)
5 WU's completed on this one....
avg CPU time = 38725.878 secs
avg claimed = 74.55
avg granted = 74.96
Granted this is not a large sample of data, but.... it's, ummm... all I got. Both computers are within 250 secs of each other on avg CPU time (my laptop is actually FASTER). And, although I complete twice as many WUs on the dual CPU computer, I claim less credit for each. Using BOINC 4.25.
Cheers,
Glenn
"No, I'm not a scientist... but I did stay at a Holiday Inn Express."
ADDMP, Just read boards
)
ADDMP,
Just read boards this evening...California here.
Only see your AMD K7-XP, Athlon XP 2200+ MP, 1799Mhz, (Palomino-662-0) machine using the client so far.
I checked out the source from berkely as the public_419 release source files...then applied the cpu naming and project benchmarking code...so I haven't touched any parts that work with "windows interface", example "hover text over icon in taskbar".
I have it running on 14 machines I use without these hover problems...windows 2K and windows XP mixed, and one Win 98.
http://setiweb.ssl.berkeley.edu/hosts_user.php?userid=548863
I can't think of why having multiprocessors as opposed to hyperthread would make any significant difference in the way it interfaces with windows (mouse clicking, display, etc.)
How does this machine behave with unmodified 4.19? I could download 4.19 from berkeley into a separate folder and compile it unmodified...see if behavior is there also.
ben wrote, > Only see your
)
ben wrote,
> Only see your AMD K7-XP, Athlon XP 2200+ MP, 1799Mhz, (Palomino-662-0) machine
> using the client so far.
Right, I started both the first & now this second version of your BOINC only on my MP computer so far. When I started the first version, I think it continued processing the WUs it had started under the previons BOINC until they finished & were uploaded. Then it was given a new computer number (but the same name) & posted later results to the web site under that new number. So it could take as long as 14 hours to complete a WU of its own, & could take almost 21 hours before the web site actually allows it to upload a new result, & still a few hours longer for the web site to update itself & actually post that new result. So if it is actually running, we are talking about a 24 hour delay before we might see anything on the web.
> I have it running on 14 machines I use without these hover problems...windows
> 2K and windows XP mixed, and one Win 98.
> http://setiweb.ssl.berkeley.edu/hosts_user.php?userid=548863
I obtained the version I am using by clicking the link in your earlier note here.
> How does this machine behave with unmodified 4.19? I could download 4.19 from
> berkeley into a separate folder and compile it unmodified...see if behavior is
> there also.
As far as I know, the original BOINC 4.19 & BOINC 4.25 & then 4.19 again & then your first version of 4.19 all ran normally on this computer. I have never seen the type of "touchiness" I described earlier with any software on this or any computer.
If you think it would be best, I could now stop BOINC & uninstall it & delete its folder, & then begin again from a clean installation of the original BOINC 4.19.
ADDMP
ben wrote, > I have it
)
ben wrote,
> I have it running on 14 machines I use without these hover problems...windows
> 2K and windows XP mixed, and one Win 98.
> http://setiweb.ssl.berkeley.edu/hosts_user.php?userid=548863
>
> I can't think of why having multiprocessors as opposed to hyperthread would
> make any significant difference in the way it interfaces with windows (mouse
> clicking, display, etc.)
It just occurred to me that I install BOINC in the directory E:/BOINC/, whereas the default installation directory is Program Files/BOINC/ If you happened to make some reference to the default installation directory, that might cause a problem on my computer but not yours. Just grasping at straws here.
ADDMP
> It just occurred to me that
)
> It just occurred to me that I install BOINC in the directory E:/BOINC/,
> whereas the default installation directory is Program Files/BOINC/ If you
> happened to make some reference to the default installation directory, that
> might cause a problem on my computer but not yours. Just grasping at straws
> here.
>
Nope, as I say, no file changes, no code changes beyond project benchmark, CPU ID code. I have several installed on D:/boinc or F:/data/boinc.
Perhaps some error when uploading or downloading the exe
hmm...no thats not probable as it is compressed with an exe compression program...checks its checksum after decompressing...
I will download it from the website and put it into my folder see if it works as expected.
...nope there..downloaded..renamed...works fine.
Perhaps try launching it by double clicking on it. In that way the gui display should show (File menu, Work Tab, etc.) If it doesnt hmm.
As I say, I don't have an actual dual processor machine to fool around with.
Try the same file copied over (on your network) to one of your single CPU machines (already with 4.19)...does it behave the same way?
ben wrote, > I will
)
ben wrote,
> I will download it from the website and put it into my folder see if it works
> as expected.
>
> ...nope there..downloaded..renamed...works fine.
A file of exactly 254,464 bytes, huh?
> Perhaps try launching it by double clicking on it. In that way the gui
> display should show (File menu, Work Tab, etc.) If it doesnt hmm.
Nope, tried, doesn't help.
> As I say, I don't have an actual dual processor machine to fool around with.
>
> Try the same file copied over (on your network) to one of your single CPU
> machines (already with 4.19)...does it behave the same way?
OK, I tried it on my single-CPU computer named "ecollege". Same problem. Has nothing to do with dual processors.
I have now tried downloading the exefile three times to floppies. All have the same problem.
I tried loading 4.19 on a clean laptop that had never seen any boinc before, & then replaced the boinc_gui.exe with your latest version, Same problem. Makes no difference which directory I install it on.
On all these computers, I can later go back to the original 4.19 boinc_gui.exe & then everything works fine again. The problem is reversible.
I don't have a clue now.
In pure desperation, I suppose we could swap the exefiles we are using back & forth by email.
ADDMP
Ok, I will also perform an
)
Ok,
I will also perform an md5sum on the file..and include that.
Send me an email...with your return address to "gladry1 at yahoo.com".
Of course replace "at" with the symbol.
Forgive me, Ben.... You
)
Forgive me, Ben....
You helped me break the code on the LIGO aiming thing, but now this, too. I know this effort isn't about the "credit", so if it will help make a better program....
I'm also running a dual 3.0 GHz P4 with HT:
Windows benchmarks
.... 3953 whet MIPS per CPU
.... 7874 dhry MIPS per CPU
BOINC benchmarks
.... 1364 whet MIPS per CPU
.... 1150 dhry MIPS per CPU
For the 15 WUs I've processed on this computer:
avg CPU time = 38960.301 secs
avg claimed = 55.96
avg granted = 73.29
However, my single CPU 2.2 GHZ P4 laptop....
BOINC benchmarks
.... 1117 whet MIPS per CPU
.... 2281 dhry MIPS per CPU (Twice that of my dual 3GHz computer!!)
5 WU's completed on this one....
avg CPU time = 38725.878 secs
avg claimed = 74.55
avg granted = 74.96
Granted this is not a large sample of data, but.... it's, ummm... all I got. Both computers are within 250 secs of each other on avg CPU time (my laptop is actually FASTER). And, although I complete twice as many WUs on the dual CPU computer, I claim less credit for each. Using BOINC 4.25.
Cheers,
Glenn
"No, I'm not a scientist... but I did stay at a Holiday Inn Express."
And I just saw your 4.30
)
And I just saw your 4.30 post....
"No, I'm not a scientist... but I did stay at a Holiday Inn Express."
I can tell you all that this
)
I can tell you all that this problem does _not_ happen under MacOS.
Power Mac Dual G5/2.0 Ghz: Benchmarks 1119 FP 3725 Integer
Time per unit ~ 28,250 claimed credit is 79.xx
This is with boinc 4.19
BTW, barkster, I love you avatar photo. As you may know Highway 666 has now been renamed to something boring like 178. Shame.
> As you may know Highway 666
)
> As you may know Highway 666 has now been renamed to something boring like 178. Shame.
Rat farts... hoped I'd get to go back to that stretch between Cortez, CO and Moab, UT someday on my way back west. Love that part of the country.
Shame, indeed.
"No, I'm not a scientist... but I did stay at a Holiday Inn Express."