Is WINE useful, if you can run it native? Bruce and the rest
of the Einstein team worked hard, I think, to make the Linux
application as fast as the Windows one. Why not honour it?
>Is WINE useful, if you can run it native? Bruce and the rest
>of the Einstein team worked hard, I think, to make the Linux
>application as fast as the Windows one. Why not honour it?
>
>Michael
i didn't know that the Linux executable had been brought up to Win speed.
I'll give it a try.
For What Its Worth, I have now compared the speed of the latest native Linux E@H with the Windows E@H running under WINE under Linux for one Intel processor & one AMD processor.
For an Intel P4 hyperthreading socket 778,
Native Linux version takes 45000 sec per work unit
Windows version under WINE under Linux takes 43000 sec per work unit
That makes the latest native Linux version 5% slower.
For an AMD Athlon MP
Native Linux version takes 31000 sec per work unit
Windows version under WINE under Linux takes 26200 sec per work unit.
That makes the latest native Linux version 18% slower.
That is much better performance for native Linux than earlier, but I am still inclined to go with WINE for Athlon processors.
If anyone has much different comparisons to report, I would be interesting in hearing them.
Does anyone happen to know whether E@H uses SS3 multimedia extensions with Intel processors?
For What Its Worth, I have now compared the speed of the latest native Linux E@H with the Windows E@H running under WINE under Linux for one Intel processor & one AMD processor.
For an Intel P4 hyperthreading socket 778,
Native Linux version takes 45000 sec per work unit
Windows version under WINE under Linux takes 43000 sec per work unit
That makes the latest native Linux version 5% slower.
For an AMD Athlon MP
Native Linux version takes 31000 sec per work unit
Windows version under WINE under Linux takes 26200 sec per work unit.
That makes the latest native Linux version 18% slower.
That is much better performance for native Linux than earlier, but I am still inclined to go with WINE for Athlon processors.
If anyone has much different comparisons to report, I would be interesting in hearing them.
Does anyone happen to know whether E@H uses SS3 multimedia extensions with Intel processors?
ADDMP.
My old Pentium II at 400 MHz takes 180000 s to complete a WU running SuSE Linux 9.3. Since I have a dual boot system and can run also WIN98SE once I started BOINC on Windows and it was definitely faster. From what I see of computer results it seems to me that AMD single CPUs are faster that Intel double CPUs. Should I go for a new box, I would certainly choose AMD for running programs such as Seti@home and Einstein@home.
Tullio
Does anyone happen to know whether E@H uses SS3 multimedia extensions with Intel processors?
ADDMP.
ADDMP,
Einstein makes no use of SSE extensions of any vintage. From what I've read, they at one point considered coding for SSE (Plain, not -2 or -3) and concluded that there would be little if any gain, so dismissed the idea. There has been some sentiment on the boards for coding optimizations for SSE2 and 3, especially after the AltiVec-optimized app for Mac G4 and G5 was released, halving processing times, but so far nothing has come of it. We can only keep wishing, but don't expect anything soon. Right now, concentration is focused on a new app to look deeper into the latest search data.
Michael
microcraft
"The arc of history is long, but it bends toward justice" - MLK
This from BoincSynergy site - "New version 5.2.14 available (but not as recommended version yet. MAY BE UNSTABLE - USE ONLY FOR TESTING)!
# Changes since 5.2.13
* The manager should now correctly detect if the core client is currently running on the machine. "
Thought you'd like to know that it appears more relief is on the way. :-)
(edited for the usual typos)
microcraft
"The arc of history is long, but it bends toward justice" - MLK
Yes, I have seen it being discussed over on Seti/NC. I even saw a post from Rom so he must have had some time over the weekend to do a bit more playing with BOINC instead of getting out into the Australian sunshine and letting go for once :).
It still says 5.2.13 in the "official" download place and all standard platforms were synced up at 5.2.13 so I suspect they wont change that immediately in case there is a need for 5.2.15 or more :).
I've been noticing an upsurge of posts here and on the Boinc boards where there is some communication problem between the Boinc modules at start. I was about to suggest to some of them that they remove Boinc from the Startup folder (temporarily), to see if that cured it. It's my theory that all the stuff loading at Win startup might confuse Boinc, or get parts of it loaded out-of-sync.
(edited for klaritee)
microcraft
"The arc of history is long, but it bends toward justice" - MLK
For What Its Worth, I have now compared the speed of the latest native Linux E@H with the Windows E@H running under WINE under Linux for one Intel processor & one AMD processor.
For an Intel P4 hyperthreading socket 778,
Native Linux version takes 45000 sec per work unit
Windows version under WINE under Linux takes 43000 sec per work unit
That makes the latest native Linux version 5% slower.
For an AMD Athlon MP
Native Linux version takes 31000 sec per work unit
Windows version under WINE under Linux takes 26200 sec per work unit.
That makes the latest native Linux version 18% slower.
That is much better performance for native Linux than earlier, but I am still inclined to go with WINE for Athlon processors.
If anyone has much different comparisons to report, I would be interesting in hearing them.
Does anyone happen to know whether E@H uses SS3 multimedia extensions with Intel processors?
ADDMP.
My old Pentium II at 400 MHz takes 180000 s to complete a WU running SuSE Linux 9.3. Since I have a dual boot system and can run also WIN98SE once I started BOINC on Windows and it was definitely faster. From what I see of computer results it seems to me that AMD single CPUs are faster that Intel double CPUs. Should I go for a new box, I would certainly choose AMD for running programs such as Seti@home and Einstein@home.
Tullio
Hello
I don't have exact seconds to mention, but in a recent install of Kubuntu I found that Einstein does a WU around 55 minutes faster than in WinXP - using BOINC 5.2.13 Kubuntu 5.10 with updates.
5.2.13 (Windows and Linux) just posted
)
Is this version useful with WINE?
ADDMP
RE: Is this version useful
)
Is WINE useful, if you can run it native? Bruce and the rest
of the Einstein team worked hard, I think, to make the Linux
application as fast as the Windows one. Why not honour it?
Michael
Team Linux Users Everywhere
>Is WINE useful, if you can
)
>Is WINE useful, if you can run it native? Bruce and the rest
>of the Einstein team worked hard, I think, to make the Linux
>application as fast as the Windows one. Why not honour it?
>
>Michael
i didn't know that the Linux executable had been brought up to Win speed.
I'll give it a try.
Thanks,
ADDMP
For What Its Worth, I have
)
For What Its Worth, I have now compared the speed of the latest native Linux E@H with the Windows E@H running under WINE under Linux for one Intel processor & one AMD processor.
For an Intel P4 hyperthreading socket 778,
Native Linux version takes 45000 sec per work unit
Windows version under WINE under Linux takes 43000 sec per work unit
That makes the latest native Linux version 5% slower.
For an AMD Athlon MP
Native Linux version takes 31000 sec per work unit
Windows version under WINE under Linux takes 26200 sec per work unit.
That makes the latest native Linux version 18% slower.
That is much better performance for native Linux than earlier, but I am still inclined to go with WINE for Athlon processors.
If anyone has much different comparisons to report, I would be interesting in hearing them.
Does anyone happen to know whether E@H uses SS3 multimedia extensions with Intel processors?
ADDMP.
RE: For What Its Worth, I
)
My old Pentium II at 400 MHz takes 180000 s to complete a WU running SuSE Linux 9.3. Since I have a dual boot system and can run also WIN98SE once I started BOINC on Windows and it was definitely faster. From what I see of computer results it seems to me that AMD single CPUs are faster that Intel double CPUs. Should I go for a new box, I would certainly choose AMD for running programs such as Seti@home and Einstein@home.
Tullio
RE: Does anyone happen to
)
ADDMP,
Einstein makes no use of SSE extensions of any vintage. From what I've read, they at one point considered coding for SSE (Plain, not -2 or -3) and concluded that there would be little if any gain, so dismissed the idea. There has been some sentiment on the boards for coding optimizations for SSE2 and 3, especially after the AltiVec-optimized app for Mac G4 and G5 was released, halving processing times, but so far nothing has come of it. We can only keep wishing, but don't expect anything soon. Right now, concentration is focused on a new app to look deeper into the latest search data.
Michael
microcraft
"The arc of history is long, but it bends toward justice" - MLK
Gary, This from
)
Gary,
This from BoincSynergy site - "New version 5.2.14 available (but not as recommended version yet. MAY BE UNSTABLE - USE ONLY FOR TESTING)!
# Changes since 5.2.13
* The manager should now correctly detect if the core client is currently running on the machine. "
Thought you'd like to know that it appears more relief is on the way. :-)
(edited for the usual typos)
microcraft
"The arc of history is long, but it bends toward justice" - MLK
Yes, I have seen it being
)
Yes, I have seen it being discussed over on Seti/NC. I even saw a post from Rom so he must have had some time over the weekend to do a bit more playing with BOINC instead of getting out into the Australian sunshine and letting go for once :).
It still says 5.2.13 in the "official" download place and all standard platforms were synced up at 5.2.13 so I suspect they wont change that immediately in case there is a need for 5.2.15 or more :).
Anyway, thanks for the "heads up".
Cheers,
Gary.
I've been noticing an upsurge
)
I've been noticing an upsurge of posts here and on the Boinc boards where there is some communication problem between the Boinc modules at start. I was about to suggest to some of them that they remove Boinc from the Startup folder (temporarily), to see if that cured it. It's my theory that all the stuff loading at Win startup might confuse Boinc, or get parts of it loaded out-of-sync.
(edited for klaritee)
microcraft
"The arc of history is long, but it bends toward justice" - MLK
RE: RE: For What Its
)
Hello
I don't have exact seconds to mention, but in a recent install of Kubuntu I found that Einstein does a WU around 55 minutes faster than in WinXP - using BOINC 5.2.13 Kubuntu 5.10 with updates.
Gray