Small update - the initial estimation for Einstein and Rosetta seems accurate once several WUs have been done, but the time until completion needs to be sorted as this is often substantially incorrect, with figures like 1,5 hours to go, shown when the WU is 97% completed, and when the work unit has taken 8 hours, 45 mins so far, with a normal total of 8 hours 50 minutes. Possibly this will improve as BOINC works out the way my PC is used etc.
Gray
The "CPU time remaining" estimate has been fixed, but it was after BOINC 5.2.6 came out. It should be in the next version.
Walt
Gray,
You (and others) may like to know that Boinc 5.2.7 is now out. Hopefully, it incorporates the Time Remaining fix that Walt has mentioned.
Regards,
Michael
microcraft
"The arc of history is long, but it bends toward justice" - MLK
Small update - the initial estimation for Einstein and Rosetta seems accurate once several WUs have been done, but the time until completion needs to be sorted as this is often substantially incorrect, with figures like 1,5 hours to go, shown when the WU is 97% completed, and when the work unit has taken 8 hours, 45 mins so far, with a normal total of 8 hours 50 minutes. Possibly this will improve as BOINC works out the way my PC is used etc.
Gray
The "CPU time remaining" estimate has been fixed, but it was after BOINC 5.2.6 came out. It should be in the next version.
Walt
Gray,
You (and others) may like to know that Boinc 5.2.7 is now out. Hopefully, it incorporates the Time Remaining fix that Walt has mentioned.
Regards,
Michael
Thank You Michael !! Funny - I've been checking the dev list for new updates pretty much every day, and the day I dont do that...
Will be getting it soon, and this time it's gonna be a straight uninstall of the one and an install of the newer one: no formatting !! (well not for a day or two anyhow)
Just to let you knwo that Boinc seems to be much better at judging time remaining to finish a WU. For example:
cpu time 2 hours 9 minutes
to completion 6 hours 54 minutes
guestimated time per WU 9 hours 3 minutes
so it looks like the devs are spot on this time !!
Gray
Gray,
I updated to 5.2.7 about 10 hours ago, and waited until it processed a "virgin" WU (freshly downloaded and processed by 5.2.7) and when I returned home, verified that the fix is indeed effective, and posted in Gary's thread in Problems and Bug Reports/Discussion and Comments...
Now that's done, I'm going back "retro" to 4.19, for the better Integer benchmarks and creds. :-)
I don't know what Boinc uses for benchmarking, but when I use SiSoft Sandra 2005, my rig hits ~10300 MIPS, while 4.19 reports ~5700, and newer Boincs give me only ~4000. FPops are ~4000M in Sandra and ~2400M in any version of Boinc I've tried. (shrug)
Michael
microcraft
"The arc of history is long, but it bends toward justice" - MLK
You mention just one PC in your specs at the top - are you shoving RAM in and out and doing multiple re-installs, or are you the lucky possessor of lots of PCs (if this seems a daft statement - sorry I just woke up and the coffee hasn't kicked in yet..... Where I come from PCs are amazingly expensive).
Did you do any tweaking to the system/s at all, or are we talking about a stock "as recomended by Mr Bill Gates" install in all cases. While I don't suggest ripping files out of the Windows folder without any forethought, certainly the smaller that folder, the better (hence mine-own re-installs, much to the amusement of my friends) - there are certain files I remove as a matter of course tho'.
Also dropping the amount of running services in XP makes a difference, obviously, tho' again caution (and research) is required. There are one or two registry tweaks too, and things like no wall-paper, bare minimum desktop stuff (shove large folders elsewhere and make a shortcut rather) - hey I'm sure you know all this stuff, if you run all those boxes.
Gray
Hi Gary,
If you are talking to me...it is just one computer with a bunch of harddrives that I swap in and out. All of my install's are as bare bones as I can make it without resorting to third party software.
If you lookup my computers you will find two listed...they are one and the same puters...just the OS is different, and I can't merge them.
Now that's done, I'm going back "retro" to 4.19, for the better Integer benchmarks and creds. :-)
Michael,
Don't do that - we need you up-to-date with the latest versions so that you can continue the excellent work of helping newcomers. Nobody will be starting up with 4.19 these days.
If all you need is to be able to claim a more appropriate amount of credit, then go here to get an optimised boinc.exe that will do about the same as 4.19. The full instructions are in a recent post of mine on the Problems board in a thread started by Fred Plett. It's this message and if your cpu has SSE2 instructions you need exactly the same one that Fred has installed.
Now that's done, I'm going back "retro" to 4.19, for the better Integer benchmarks and creds. :-)
Michael,
Don't do that - we need you up-to-date with the latest versions so that you can continue the excellent work of helping newcomers. Nobody will be starting up with 4.19 these days.
If all you need is to be able to claim a more appropriate amount of credit, then go here to get an optimised boinc.exe that will do about the same as 4.19. The full instructions are in a recent post of mine on the Problems board in a thread started by Fred Plett. It's this message and if your cpu has SSE2 instructions you need exactly the same one that Fred has installed.
Gary,
I have an Athlon XP-Mobile 2600 - SSE2 set was not implemented by AMD until the 64-bit processor line. :-{ I'll take a look at the link you provided, though, see if I can shop around for something I can use. I'm pretty weary now, it's past 2:20 a.m. here, and I had a rough day working at the garage. I'm a tired little cowboy, lol. Speaking of overworked, I'm surprised this cpu is holding up so well - CPU-Z reports it at 2545 MHz, 27+% overclocked, but stably crunching.
Cya later today,
Michael
(edit) I went there, chose the SSE (not SSE2) client, d/l ed, uninstalled 4.19, tried to install the opt cc, and it refused, said "Missing component", reckoned maybe what was missing was a lib from 5.x, so installed 5.2.7, tried again to install opt, and it refused again, same message, so I'm back with 4.19. Oh, well.
microcraft
"The arc of history is long, but it bends toward justice" - MLK
(edit) I went there, chose the SSE (not SSE2) client, d/l ed, uninstalled 4.19, tried to install the opt cc, and it refused, said "Missing component", reckoned maybe what was missing was a lib from 5.x, so installed 5.2.7, tried again to install opt, and it refused again, same message, so I'm back with 4.19. Oh, well.
You are trying to make it too hard when it should be easy :). You are not installing anything. You are replacing 5.2.7's boinc.exe with the one you snagged.
* Reinstall 5.2.7
* Start it and observe that the science app starts crunching
* Stop BOINC completely
* Delete the single file boinc.exe from your BOINC folder
* Copy the SSE version that you snagged earlier - just one file, boinc.exe, into place in the BOINC folder
* Start BOINC Manager and watch new benchmarks get done in the messages window
* EAH will resume crunching from where it left off.
* Enjoy your new higher benchmarks
(edit) I went there, chose the SSE (not SSE2) client, d/l ed, uninstalled 4.19, tried to install the opt cc, and it refused, said "Missing component", reckoned maybe what was missing was a lib from 5.x, so installed 5.2.7, tried again to install opt, and it refused again, same message, so I'm back with 4.19. Oh, well.
You are trying to make it too hard when it should be easy :). You are not installing anything. You are replacing 5.2.7's boinc.exe with the one you snagged.
* Reinstall 5.2.7
* Start it and observe that the science app starts crunching
* Stop BOINC completely
* Delete the single file boinc.exe from your BOINC folder
* Copy the SSE version that you snagged earlier - just one file, boinc.exe, into place in the BOINC folder
* Start BOINC Manager and watch new benchmarks get done in the messages window
* EAH will resume crunching from where it left off.
* Enjoy your new higher benchmarks
Gary,
OK, I'll do that straightaway.
Michael
(update) Yes, well that was simple, wasn't it? OK, done. Benchmarks are now 1930/7190 vs 2384/4020 for 5.2.7, 19% lower on Whetstone and 79% higher on Dhrystone, composite 42% higher. Should translate to ~95 credits/WU, and I can still evaluate 5.2.7, sort of. :-)
Many thanks, and did I mention that it will be my pleasure to pick up the tab for the first keg, when you get around to that party?
(2nd update) Hey! When I laid that in with 5.2.7, it retained the To Completion time fix, as well.
Oops, To Completion for queued WUs is 5:45, when reality is 5:02 or 5:03. Still, I'm happy. First WU completion in about 6 minutes, so we'll see.
microcraft
"The arc of history is long, but it bends toward justice" - MLK
(update) Yes, well that was simple, wasn't it? OK, done. Benchmarks are now 1930/7190 vs 2384/4020 for 5.2.7, 19% lower on Whetstone and 79% higher on Dhrystone, composite 42% higher. Should translate to ~95 credits/WU, and I can still evaluate 5.2.7, sort of. :-)
Yes, it will be close enough so that you can pretend it's 5.2.X :).
Quote:
Many thanks, and did I mention that it will be my pleasure to pick up the tab for the first keg, when you get around to that party?
At ~95/WU you will soon have enough credits to buy all the kegs :).
Quote:
(2nd update) Hey! When I laid that in with 5.2.7, it retained the To Completion time fix, as well.
Sorry, no it wont retain that fix but it's a small price to pay for all the "benefits" you are getting :).
Quote:
Oops, To Completion for queued WUs is 5:45, when reality is 5:02 or 5:03. Still, I'm happy. First WU completion in about 6 minutes, so we'll see.
It's going to look a bit worse when it starts a new result. Back to the previous wildly inaccurate values unfortunately. I'll let you know if I see a later optimised BOINC get announced.
RE: RE: Hello Small
)
Gray,
You (and others) may like to know that Boinc 5.2.7 is now out. Hopefully, it incorporates the Time Remaining fix that Walt has mentioned.
Regards,
Michael
microcraft
"The arc of history is long, but it bends toward justice" - MLK
RE: RE: RE: Hello Smal
)
Thank You Michael !! Funny - I've been checking the dev list for new updates pretty much every day, and the day I dont do that...
Will be getting it soon, and this time it's gonna be a straight uninstall of the one and an install of the newer one: no formatting !! (well not for a day or two anyhow)
Gray
Hello Michael Just to let
)
Hello Michael
Just to let you knwo that Boinc seems to be much better at judging time remaining to finish a WU. For example:
cpu time 2 hours 9 minutes
to completion 6 hours 54 minutes
guestimated time per WU 9 hours 3 minutes
so it loks like the devs are spot on this time !!
Gray
RE: Hello Michael Just to
)
Gray,
I updated to 5.2.7 about 10 hours ago, and waited until it processed a "virgin" WU (freshly downloaded and processed by 5.2.7) and when I returned home, verified that the fix is indeed effective, and posted in Gary's thread in Problems and Bug Reports/Discussion and Comments...
Now that's done, I'm going back "retro" to 4.19, for the better Integer benchmarks and creds. :-)
I don't know what Boinc uses for benchmarking, but when I use SiSoft Sandra 2005, my rig hits ~10300 MIPS, while 4.19 reports ~5700, and newer Boincs give me only ~4000. FPops are ~4000M in Sandra and ~2400M in any version of Boinc I've tried. (shrug)
Michael
microcraft
"The arc of history is long, but it bends toward justice" - MLK
RE: Hello You mention just
)
Hi Gary,
If you are talking to me...it is just one computer with a bunch of harddrives that I swap in and out. All of my install's are as bare bones as I can make it without resorting to third party software.
If you lookup my computers you will find two listed...they are one and the same puters...just the OS is different, and I can't merge them.
Greg
RE: Now that's done, I'm
)
Michael,
Don't do that - we need you up-to-date with the latest versions so that you can continue the excellent work of helping newcomers. Nobody will be starting up with 4.19 these days.
If all you need is to be able to claim a more appropriate amount of credit, then go here to get an optimised boinc.exe that will do about the same as 4.19. The full instructions are in a recent post of mine on the Problems board in a thread started by Fred Plett. It's this message and if your cpu has SSE2 instructions you need exactly the same one that Fred has installed.
Cheers,
Gary.
RE: RE: Now that's done,
)
Gary,
I have an Athlon XP-Mobile 2600 - SSE2 set was not implemented by AMD until the 64-bit processor line. :-{ I'll take a look at the link you provided, though, see if I can shop around for something I can use. I'm pretty weary now, it's past 2:20 a.m. here, and I had a rough day working at the garage. I'm a tired little cowboy, lol. Speaking of overworked, I'm surprised this cpu is holding up so well - CPU-Z reports it at 2545 MHz, 27+% overclocked, but stably crunching.
Cya later today,
Michael
(edit) I went there, chose the SSE (not SSE2) client, d/l ed, uninstalled 4.19, tried to install the opt cc, and it refused, said "Missing component", reckoned maybe what was missing was a lib from 5.x, so installed 5.2.7, tried again to install opt, and it refused again, same message, so I'm back with 4.19. Oh, well.
microcraft
"The arc of history is long, but it bends toward justice" - MLK
RE: (edit) I went there,
)
You are trying to make it too hard when it should be easy :). You are not installing anything. You are replacing 5.2.7's boinc.exe with the one you snagged.
* Start it and observe that the science app starts crunching
* Stop BOINC completely
* Delete the single file boinc.exe from your BOINC folder
* Copy the SSE version that you snagged earlier - just one file, boinc.exe, into place in the BOINC folder
* Start BOINC Manager and watch new benchmarks get done in the messages window
* EAH will resume crunching from where it left off.
* Enjoy your new higher benchmarks
Cheers,
Gary.
RE: RE: (edit) I went
)
Gary,
OK, I'll do that straightaway.
Michael
(update) Yes, well that was simple, wasn't it? OK, done. Benchmarks are now 1930/7190 vs 2384/4020 for 5.2.7, 19% lower on Whetstone and 79% higher on Dhrystone, composite 42% higher. Should translate to ~95 credits/WU, and I can still evaluate 5.2.7, sort of. :-)
Many thanks, and did I mention that it will be my pleasure to pick up the tab for the first keg, when you get around to that party?
(2nd update) Hey! When I laid that in with 5.2.7, it retained the To Completion time fix, as well.
Oops, To Completion for queued WUs is 5:45, when reality is 5:02 or 5:03. Still, I'm happy. First WU completion in about 6 minutes, so we'll see.
microcraft
"The arc of history is long, but it bends toward justice" - MLK
RE: (update) Yes, well
)
Yes, it will be close enough so that you can pretend it's 5.2.X :).
At ~95/WU you will soon have enough credits to buy all the kegs :).
Sorry, no it wont retain that fix but it's a small price to pay for all the "benefits" you are getting :).
It's going to look a bit worse when it starts a new result. Back to the previous wildly inaccurate values unfortunately. I'll let you know if I see a later optimised BOINC get announced.
Cheers,
Gary.