I checked the graphics of my task running with 4.32 for Windows and BOINC 6.1.8b
I noticed the following: I only have the sphere with the L's, dots and lines. It's not showing my data, where it's looking on the sphere and the orange marker is also nowhere to be seen.
Something in the graphics themselves? I can remember that it worked in 4.26
stdoutdae.txt is empty about this. No error messages about it either.
Hi,
I'll try the new app on my AMD Phenom in a few moments. But two questions meanwhile:
1. Is the official version E@H 4.32 the BOINC client has tried to download for 2 hours now the same as the 4.32 on the power-apps site?
2. The K10 architecture of the Phenom has a doubled SSE path so that 128 bit SIMD data can be loaded within one clockcycle. Does the handcoded SSE-part of the application make use of it?
Hi,
I'll try the new app on my AMD Phenom in a few moments. But two questions meanwhile:
1. Is the official version E@H 4.32 the BOINC client has tried to download for 2 hows now the same as the 4.32 on the power-apps site?
Strange, 4.32 is definitely not the official version. Please make sure to unpack ALL files in the ZIP archive to the */BOINC/projects/einstein.phys.uwm.edu subdirectory.
Quote:
2. The K10 architecture of the Phenom has a doubled SSE path so that 128 bit SIMD data can be loaded within one clockcycle. Does the handcoded SSE-part of the application make use of it?
The doubled path width is an internal performance optimization, this is nothing that a programmer has to care about "actively" by writing code in a certain way. Any SIMD SSE code will benefit from this.
Strange, 4.32 is definitely not the official version. Please make sure to unpack ALL files in the ZIP archive to the */BOINC/projects/einstein.phys.uwm.edu subdirectory.
Well, I haven't installed the power app, yet. I've resetted Einstein@Home project since I haven't took part for a while and after that the BOINC client immediately tried to download a new E@H client with the version 4.32. So I thought the 4.32-SSE-power-app might have become the official client meanwhile.
Strange, 4.32 is definitely not the official version. Please make sure to unpack ALL files in the ZIP archive to the */BOINC/projects/einstein.phys.uwm.edu subdirectory.
Well, I haven't installed the power app, yet. I've resetted Einstein@Home project since I haven't took part for a while and after that the BOINC client immediately tried to download a new E@H client with the version 4.32. So I thought the 4.32-SSE-power-app might have become the official client meanwhile.
Nope. The client might try to find App version 4.32 because it finds tasks it had already assigned that version. Are you sure you never had installed a 4.32 App previously?
Strange, 4.32 is definitely not the official version. Please make sure to unpack ALL files in the ZIP archive to the */BOINC/projects/einstein.phys.uwm.edu subdirectory.
Well, I haven't installed the power app, yet. I've resetted Einstein@Home project since I haven't took part for a while and after that the BOINC client immediately tried to download a new E@H client with the version 4.32. So I thought the 4.32-SSE-power-app might have become the official client meanwhile.
Maybe you been gone a very long time and used the 4.32 app for S5R2? That was in July -07. Then you need to delete the app_info.xml, or maybe detach and reattach to the project. (Reset Project doesnt delete app_info i think??)
Nope. The client might try to find App version 4.32 because it finds tasks it had already assigned that version. Are you sure you never had installed a 4.32 App previously?
Yes, I am sure. But it doesn't matter anymore. Since the official client is no SSE-client I installed the 4.32-SSE and it is working now.
We have to wait and see now how the Phenom performs with this client...
On both my 2.4 GHz P-IV machine, and my 3500+ machine, there's been a dramatic speed-up, even though they're now both crunching on the "b" workunits. (In fact, the 2.2 GHz 3500+ with this Windows app isn't that much slower than the 3.0 GHz 6000+ with the Linux power-users app.)
Yep, great speedup. The Windows app on a E6600 need between 7-8hours and Linux 4.21 power-app need 10hours but on E6320.
Why 4.21? Because 4.27 [edit] and 4.28 [/edit] are much slower on Suse64 but i don't know why?
I checked the graphics of my task running with 4.32 for Windows and BOINC 6.1.8b
I noticed the following: I only have the sphere with the L's, dots and lines. It's not showing my data, where it's looking on the sphere and the orange marker is also nowhere to be seen.
Something in the graphics themselves? I can remember that it worked in 4.26
stdoutdae.txt is empty about this. No error messages about it either.
Also, is this with an IGP or an actual card? Does it work on 5.x.x builds? I'm sure you probably already guessed that it works fine on my system with the 6800GT card and 5.8.16...
After testing the new pp for about one week, I really must say. It is a beautiful speedup. Average runtime goes down from 7. hours with 4.26 to 5. hours with 4.32.
Only one problem is there. But this could be because of the new app or the longer Deadlines. I have on my single host over 6800 Credits in pending. This is much more than normal (2500 Credits)...
I checked the graphics of my
)
I checked the graphics of my task running with 4.32 for Windows and BOINC 6.1.8b
I noticed the following: I only have the sphere with the L's, dots and lines. It's not showing my data, where it's looking on the sphere and the orange marker is also nowhere to be seen.
Something in the graphics themselves? I can remember that it worked in 4.26
stdoutdae.txt is empty about this. No error messages about it either.
Link to example picture.
Hi, I'll try the new app on
)
Hi,
I'll try the new app on my AMD Phenom in a few moments. But two questions meanwhile:
1. Is the official version E@H 4.32 the BOINC client has tried to download for 2 hours now the same as the 4.32 on the power-apps site?
2. The K10 architecture of the Phenom has a doubled SSE path so that 128 bit SIMD data can be loaded within one clockcycle. Does the handcoded SSE-part of the application make use of it?
Thanks and bye
RE: Hi, I'll try the new
)
Strange, 4.32 is definitely not the official version. Please make sure to unpack ALL files in the ZIP archive to the */BOINC/projects/einstein.phys.uwm.edu subdirectory.
The doubled path width is an internal performance optimization, this is nothing that a programmer has to care about "actively" by writing code in a certain way. Any SIMD SSE code will benefit from this.
Happy crunching
Bikeman
RE: Strange, 4.32 is
)
Well, I haven't installed the power app, yet. I've resetted Einstein@Home project since I haven't took part for a while and after that the BOINC client immediately tried to download a new E@H client with the version 4.32. So I thought the 4.32-SSE-power-app might have become the official client meanwhile.
RE: RE: Strange, 4.32 is
)
Nope. The client might try to find App version 4.32 because it finds tasks it had already assigned that version. Are you sure you never had installed a 4.32 App previously?
BM
BM
RE: RE: Strange, 4.32 is
)
Maybe you been gone a very long time and used the 4.32 app for S5R2? That was in July -07. Then you need to delete the app_info.xml, or maybe detach and reattach to the project. (Reset Project doesnt delete app_info i think??)
Team Philippines
RE: Nope. The client might
)
Yes, I am sure. But it doesn't matter anymore. Since the official client is no SSE-client I installed the 4.32-SSE and it is working now.
We have to wait and see now how the Phenom performs with this client...
RE: This must be doing me
)
Yep, great speedup. The Windows app on a E6600 need between 7-8hours and Linux 4.21 power-app need 10hours but on E6320.
Why 4.21? Because 4.27 [edit] and 4.28 [/edit] are much slower on Suse64 but i don't know why?
RE: I checked the graphics
)
Reboot help?
Also, is this with an IGP or an actual card? Does it work on 5.x.x builds? I'm sure you probably already guessed that it works fine on my system with the 6800GT card and 5.8.16...
After testing the new pp for
)
After testing the new pp for about one week, I really must say. It is a beautiful speedup. Average runtime goes down from 7. hours with 4.26 to 5. hours with 4.32.
Only one problem is there. But this could be because of the new app or the longer Deadlines. I have on my single host over 6800 Credits in pending. This is much more than normal (2500 Credits)...