I cannot see in Task Manager whether these are using SSE as the complete file descriptor is not given. It stops at einstein_S5R4_6 - so I cannot use the data Archae86 gave.
If your Task Manager is like my Win XP Pro Task Manager you can drag the boundary between "Image Name" and "User Name" farther to the right to expose more of the actual task name. If not, I suggest you download Process Explorer (formerly a small independent company, now owned by Microsoft, but still _much_ nicer than Task Manager).
Quote:
Once I am well established crunching the S5R4s then my RAC will drop to below 700, I guess?
I'm down a little less than a factor of two, and am not sure I am stable yet, so something like that seems plausible.
My dual P3 (Win2K Pro) is just coming up to complete it's first 2 S5R4 WUs at 51-52 hours. Fortunately I set the local machine queue to1 day. But despite this I have over 8 days in the cache (based on the 50+ hour WU x 2).
I cannot see in Task Manager whether these are using SSE as the complete file descriptor is not given. It stops at einstein_S5R4_6 - so I cannot use the data Archae86 gave.
The other rig (Win XP Pro) does show I am using SSE, and the crunch times (4 at a time) are about 40 hours. In this case my 1 day cache, for the local machine, resulted in me still having about 7 days worth to crunch.
At least all of them are within the given deadline for results return.
Once I am well established crunching the S5R4s then my RAC will drop to below 700, I guess?
In an other thread somewhere in the boinc-universe i found some words like that: if your host is older than 2 years your costs-usage-ratio turns to negativ and from that point you only wasting energy.
I had shutdown my DL380-G2 with 2x P3s-1400MHz past the last S5R-3-units with crunching times around 30hrs. A Core2-CPU crunched such a unit in less than 10hrs. For a S5R-4-unit you need around 50hrs on your P3? I believe the time for P3 or P3s is really over now.
I think
-for the science or
-to checkout if it runs or
-sentimental motives
every host is needed!
By the way Process Explorer is my favorite tool to check a system. It gives me much more needed infos about all than the original taskmanager.
I cannot see in Task Manager whether these are using SSE as the complete file descriptor is not given. It stops at einstein_S5R4_6 - so I cannot use the data Archae86 gave.
If your Task Manager is like my Win XP Pro Task Manager you can drag the boundary between "Image Name" and "User Name" farther to the right to expose more of the actual task name. If not, I suggest you download Process Explorer (formerly a small independent company, now owned by Microsoft, but still _much_ nicer than Task Manager).
Quote:
Once I am well established crunching the S5R4s then my RAC will drop to below 700, I guess?
I'm down a little less than a factor of two, and am not sure I am stable yet, so something like that seems plausible.
Well! The dual P3 reported while I was asleep and both WUs gave 185,000 seconds, now updated in BOINC. This is almost exactly 100,000 seconds longer than the S5R3 WUs under SSE.
My Win2k Task Manager has the boundary, but no other details were revealed. So,. I will look at acquiring a copy of Process Exploder.
I run the dual P3 for 2 main, business related, reasons. It is my data store, using a RAID-1 mirror, and has one of my 2 Email clients active. The latter uses an Email address I have has for 14 years now.
As the machine runs 24/7 for other reasons I might as well run Einstein (despite it's slowness and inefficiencies).
Oh! PS: There goes the RAC downwards, downwards, downwards!
Shih-Tzu are clever, cuddly, playful and rule!! Jack Russell are feisty!
Ah well, I just made a 'powerapp' change... hardware style. ;)
I had a server with a xeon 5110 (2x1.6Ghz) that had been misbehaving so I've made a little switch. Oddly enough the CPU would generate a therm_trip signal randomly despite the cores running at like 36C... therm_trip should go up at around 100C since it's the processor's signal that possible damage to the silicon is coming and it halts.
In any case, my primary production server had a 5120 (2x1.83Ghz) so I ordered a spiffy new e5345 (4x2.33Ghz), put it in the production server and replaced the 5110 with the 5120 from it so between the two, they should be a wee bit more productive. ;-D
I'm also at 4+ hours without the troublemaker throwing a therm_trip, so I'm happy all 'round!
I'll look at the binary for "AMD" being in it in just a minute... ...
There is indeed an "AuthenticAMD" string in the "1" executable... :(
What?
Is that the "Naughty Intel" cripple the AMD CPUs trick resurrected?
Cheers,
Martin
No idea Martin... I know that at one point it was an issue. I have not moved my AMD over to working on S5R4 due to the potential of it though. I'll move over when I can't get anything from any other of my projects (except SETI) or Bernd makes a power users / beta app available to test getting the improvements that exist in the Linux app over into the Windows app...
Until then, someone else will need to check to see if there is an AMD penalty again...
I'll look at the binary for "AMD" being in it in just a minute... ...
There is indeed an "AuthenticAMD" string in the "1" executable... :(
What?
Is that the "Naughty Intel" cripple the AMD CPUs trick resurrected?
Cheers,
Martin
No idea Martin... I know that at one point it was an issue. I have not moved my AMD over to working on S5R4 due to the potential of it though. I'll move over when I can't get anything from any other of my projects (except SETI) or Bernd makes a power users / beta app available to test getting the improvements that exist in the Linux app over into the Windows app...
Until then, someone else will need to check to see if there is an AMD penalty again...
Your turn. (I did SSE/Intel/Windows).
Why not run the project's app as supplied, build some baseline stats (taking note of the variation over sequence numbers), then download a free hex editor and re-run the test with 'AuthenticABC' or other well known vendor strings.
Or are you afraid that your search for gravity waves will be contaminated by WIMPs?
No idea Martin... I know that at one point it was an issue. I have not moved my AMD over to working on S5R4 due to the potential of it though. I'll move over when I can't get anything from any other of my projects (except SETI) or Bernd makes a power users / beta app available to test getting the improvements that exist in the Linux app over into the Windows app...
Until then, someone else will need to check to see if there is an AMD penalty again...
Your turn. (I did SSE/Intel/Windows).
Why not run the project's app as supplied, build some baseline stats (taking note of the variation over sequence numbers), then download a free hex editor and re-run the test with 'AuthenticABC' or other well known vendor strings.
Or are you afraid that your search for gravity waves will be contaminated by WIMPs?
The mere presence of the AuthenticAMD string is not an indicator of any problem, it comes as part of the CPU feature detection code in the runtime library that the MS VC compiler links to the app. The parts of the code that were affected by the AMD specific "feature detection" in this library were later replaced by other code , but if anyone wants to check by hex-editing, please be aware that BOINC has a checksum mechanism that would reject a modified version. However, this mechanism does not apply to executables downloaded manually and run with an app_info.xml file. So for example if you are using the "dual run" app_info.xml file posted in the sticky thread, yes, you could do that hex-editing experiment.
We will see... I do have a 3-day weekend coming up...
Quote:
Or are you afraid that your search for gravity waves will be contaminated by WIMPs?
Slower single-core systems make it to where observations like that will take significantly longer to chart. Had this been 2-3 weeks ago, maybe, but right now, I have about 200 pages of reading to do by Monday night and an assignment to turn in...
RE: I cannot see in Task
)
If your Task Manager is like my Win XP Pro Task Manager you can drag the boundary between "Image Name" and "User Name" farther to the right to expose more of the actual task name. If not, I suggest you download Process Explorer (formerly a small independent company, now owned by Microsoft, but still _much_ nicer than Task Manager).
I'm down a little less than a factor of two, and am not sure I am stable yet, so something like that seems plausible.
RE: Thanks for the
)
In an other thread somewhere in the boinc-universe i found some words like that: if your host is older than 2 years your costs-usage-ratio turns to negativ and from that point you only wasting energy.
I had shutdown my DL380-G2 with 2x P3s-1400MHz past the last S5R-3-units with crunching times around 30hrs. A Core2-CPU crunched such a unit in less than 10hrs. For a S5R-4-unit you need around 50hrs on your P3? I believe the time for P3 or P3s is really over now.
I think
-for the science or
-to checkout if it runs or
-sentimental motives
every host is needed!
By the way Process Explorer is my favorite tool to check a system. It gives me much more needed infos about all than the original taskmanager.
RE: RE: I cannot see in
)
Well! The dual P3 reported while I was asleep and both WUs gave 185,000 seconds, now updated in BOINC. This is almost exactly 100,000 seconds longer than the S5R3 WUs under SSE.
My Win2k Task Manager has the boundary, but no other details were revealed. So,. I will look at acquiring a copy of Process Exploder.
I run the dual P3 for 2 main, business related, reasons. It is my data store, using a RAID-1 mirror, and has one of my 2 Email clients active. The latter uses an Email address I have has for 14 years now.
As the machine runs 24/7 for other reasons I might as well run Einstein (despite it's slowness and inefficiencies).
Oh! PS: There goes the RAC downwards, downwards, downwards!
Shih-Tzu are clever, cuddly, playful and rule!! Jack Russell are feisty!
Ah well, I just made a
)
Ah well, I just made a 'powerapp' change... hardware style. ;)
I had a server with a xeon 5110 (2x1.6Ghz) that had been misbehaving so I've made a little switch. Oddly enough the CPU would generate a therm_trip signal randomly despite the cores running at like 36C... therm_trip should go up at around 100C since it's the processor's signal that possible damage to the silicon is coming and it halts.
In any case, my primary production server had a 5120 (2x1.83Ghz) so I ordered a spiffy new e5345 (4x2.33Ghz), put it in the production server and replaced the 5110 with the 5120 from it so between the two, they should be a wee bit more productive. ;-D
I'm also at 4+ hours without the troublemaker throwing a therm_trip, so I'm happy all 'round!
RE: I'll look at the binary
)
What?
Is that the "Naughty Intel" cripple the AMD CPUs trick resurrected?
Cheers,
Martin
See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)
RE: RE: I'll look at the
)
No idea Martin... I know that at one point it was an issue. I have not moved my AMD over to working on S5R4 due to the potential of it though. I'll move over when I can't get anything from any other of my projects (except SETI) or Bernd makes a power users / beta app available to test getting the improvements that exist in the Linux app over into the Windows app...
Until then, someone else will need to check to see if there is an AMD penalty again...
RE: RE: RE: I'll look
)
Your turn. (I did SSE/Intel/Windows).
Why not run the project's app as supplied, build some baseline stats (taking note of the variation over sequence numbers), then download a free hex editor and re-run the test with 'AuthenticABC' or other well known vendor strings.
Or are you afraid that your search for gravity waves will be contaminated by WIMPs?
RE: RE: No idea
)
The mere presence of the AuthenticAMD string is not an indicator of any problem, it comes as part of the CPU feature detection code in the runtime library that the MS VC compiler links to the app. The parts of the code that were affected by the AMD specific "feature detection" in this library were later replaced by other code , but if anyone wants to check by hex-editing, please be aware that BOINC has a checksum mechanism that would reject a modified version. However, this mechanism does not apply to executables downloaded manually and run with an app_info.xml file. So for example if you are using the "dual run" app_info.xml file posted in the sticky thread, yes, you could do that hex-editing experiment.
CU
Bikeman
RE: So for example if you
)
Thanks...
RE: Your turn. (I did
)
We will see... I do have a 3-day weekend coming up...
Slower single-core systems make it to where observations like that will take significantly longer to chart. Had this been 2-3 weeks ago, maybe, but right now, I have about 200 pages of reading to do by Monday night and an assignment to turn in...