I now have a version of app_info.xml which should deal with both the previous versions of albert/einstein if you are worried about keeping these for any remaining S4 work you might be offered. If anyone is interested I'll post it, but I've yet to see any PC get an S4 unit after it has started getting S5. Anyone had that experience? Mike
Post it.
thanks!
Things are moving fast - I point you at Udo's superior looking version just posted. Note that if you do not have all the applications mentioned in these files you will probably get some error messages under the "Messages" tab at startup. These should not affect the performance of the application though. If you find you need the older S4 applications the most relevant are still available for download, see the "Einstein@Home Applications" link on the Einstein home page. Mike
@Udo:
Why we need the einstein_S5R1_4.10_windows_intelx86.pdb?
I think the client didn't use this file.
My client works without the pdb-file.
... I just did the package the same as albert_450 and albert_462...
and they contain the corresponding .pdb file...
It is requested within client_state.xml for 4.02, and if you change this to 4.10 you also have to have a .pdb file for 4.10.
Guys and gals, if Bernd said he wanted us to pretend we have a 4.10 application, then the verion info in app_info.xml should state so. In order no to wait for running down the cache, bot application version should be in the app_info.xml.
Bernd probably want's to have different versions in order to differentiate the results of Akos' app from thoe of official application.
This app_info.xml should do the trick (if we forget the S4 work - let the rest do it, OK?):
[pre]
einstein_S5R1
einstein_S5R1_4.02_windows_intelx86.exe
einstein_S5R1
402
einstein_S5R1_4.02_windows_intelx86.exe
einstein_S5R1
einstein_S5R1_4.10_windows_intelx86.exe
einstein_S5R1
410
einstein_S5R1_4.10_windows_intelx86.exe
[/pre]
It looks quite simmilar to those already posted.
This one will process hte remaining WUs already downloaded using whatever version you have under name of einstein_S5R1_4.02_windows_intelx86.exe. My gues is that Bernd would have wanted it to be official app. The app with name einstein_S5R1_4.10_windows_intelx86.exe is your favourite pathed Akos' application.
[edit]
Regarding the .pdb file: it's a file with debug information - in case anything wrong happens, OS looks into this file to fetch the names of functions ... as Akos changed the important stuff, the .pdb file no longer applies.
Here is a first result for S5T0712 patch on a box running short workunits. App is renamed to version 4.10 and patching worked without a hitch.
Stock App speed = 3696 seconds
S5S0007 Speed = 3085 seconds (speedup is 19.8%)
S5T0712 Speed = 2709 seconds (speedup is 36.4%)
Note: %Speedup was calculated = [Stock secs - Patched secs] / Patched secs X 100
CPU is AMD Barton 2500+ running at 3200+ speed. It has SSE only. All results have validated. There is only one so far for S5T0712 but there are multiple ones (more than 10) for S5S0007.
The data file is h1_0216.0.
Obviously with only one result, you should take the numbers with a grain of salt but it's all very impressive nevertheless.
RE: RE: I now have a
)
Things are moving fast - I point you at Udo's superior looking version just posted. Note that if you do not have all the applications mentioned in these files you will probably get some error messages under the "Messages" tab at startup. These should not affect the performance of the application though. If you find you need the older S4 applications the most relevant are still available for download, see the "Einstein@Home Applications" link on the Einstein home page. Mike
Why we need the
)
Why we need the einstein_S5R1_4.10_windows_intelx86.pdb?
Okay, I´ve taken the app_info
)
Okay, I´ve taken the app_info from MRAO and shortened it. I deleted the 4.37 part of it. Thanks Zero.
Now it looks so:
einstein_S5R1
einstein_S5R1_4.10_windows_intelx86.exe
einstein_S5R1
402
einstein_S5R1_4.10_windows_intelx86.exe
There are no problems. I changed it with WU´s in the cache.
And don´t forget to rename the einstein.exe. ;)
I don´t also know any usage for the einstein_S5R1_4.10_windows_intelx86.pdb???
standard app / 3140,11 sec
)
standard app / 3140,11 sec / 52,3min / result
S5T0712 SSE3 / 2122,34 sec / 35,4min / result
S5T0712 SSE3 / 2132,47 sec / 35,5min / result
All Results are valid.
gain:
S5T0712 = 32,4%
I have tested all available
)
I have tested all available apps from akosf with the same short WU.
I backed up the whole BOINC directory and ran the tests:
Pentium 930D (3000MHz) @ 3600MHz:
stock : 4171 sec
0302 (SSE3): 3263 sec - 21,76% speedup
0303 (SSE3): 3212 sec - 22,99% speedup
0304 (SSE) : 3674 sec - 11,91% speedup
0305 (SSE) : 3323 sec - 20,33% speedup
0306 (SSE) : 3265 sec - 21,72% speedup
0307 (SSE) : 3186 sec - 23,61% speedup
0709 (SSE) : 3199 sec - 23,30% speedup
0711 (SSE) : 3025 sec - 27,47% speedup
0712 (SSE) : 2712 sec - 34,97% speedup
AthlonXP 2700+ (2166MHz) @ 2166MHz:
stock : 3704 sec
0304 (SSE) : 3439 sec - 7,15% speedup
0305 (SSE) : 3324 sec - 10,23% speedup
0306 (SSE) : 3234 sec - 12,68% speedup
0307 (SSE) : 3166 sec - 14,52% speedup
0709 (SSE) : 3164 sec - 14,57% speedup
0711 (SSE) : 3058 sec - 17,44% speedup
So, thats impressive work, akosf!!!!
And now a new ZIP file
)
And now a new ZIP file containing S5T0712...
more info some posts below...
Udo
@Udo: Why we need the
)
@Udo:
Why we need the einstein_S5R1_4.10_windows_intelx86.pdb?
I think the client didn't use this file.
My client works without the pdb-file.
RE: @Udo: Why we need the
)
... I just did the package the same as albert_450 and albert_462...
and they contain the corresponding .pdb file...
It is requested within client_state.xml for 4.02, and if you change this to 4.10 you also have to have a .pdb file for 4.10.
Udo
[Edit] typo errors corrected [/edit]
Udo
Guys and gals, if Bernd said
)
Guys and gals, if Bernd said he wanted us to pretend we have a 4.10 application, then the verion info in app_info.xml should state so. In order no to wait for running down the cache, bot application version should be in the app_info.xml.
Bernd probably want's to have different versions in order to differentiate the results of Akos' app from thoe of official application.
This app_info.xml should do the trick (if we forget the S4 work - let the rest do it, OK?):
[pre]
einstein_S5R1
einstein_S5R1_4.02_windows_intelx86.exe
einstein_S5R1
402
einstein_S5R1_4.02_windows_intelx86.exe
einstein_S5R1
einstein_S5R1_4.10_windows_intelx86.exe
einstein_S5R1
410
einstein_S5R1_4.10_windows_intelx86.exe
[/pre]
It looks quite simmilar to those already posted.
This one will process hte remaining WUs already downloaded using whatever version you have under name of einstein_S5R1_4.02_windows_intelx86.exe. My gues is that Bernd would have wanted it to be official app. The app with name einstein_S5R1_4.10_windows_intelx86.exe is your favourite pathed Akos' application.
[edit]
Regarding the .pdb file: it's a file with debug information - in case anything wrong happens, OS looks into this file to fetch the names of functions ... as Akos changed the important stuff, the .pdb file no longer applies.
Metod ...
Here is a first result for
)
Here is a first result for S5T0712 patch on a box running short workunits. App is renamed to version 4.10 and patching worked without a hitch.
Stock App speed = 3696 seconds
S5S0007 Speed = 3085 seconds (speedup is 19.8%)
S5T0712 Speed = 2709 seconds (speedup is 36.4%)
Note: %Speedup was calculated = [Stock secs - Patched secs] / Patched secs X 100
CPU is AMD Barton 2500+ running at 3200+ speed. It has SSE only. All results have validated. There is only one so far for S5T0712 but there are multiple ones (more than 10) for S5S0007.
The data file is h1_0216.0.
Obviously with only one result, you should take the numbers with a grain of salt but it's all very impressive nevertheless.
Cheers,
Gary.