Einstein@Home for 64-bit Linux on AMD Athlon 64 X2

James Clark
James Clark
Joined: 17 Jun 05
Posts: 1
Credit: 683,354
RAC: 0

yes, whether it's ultimately

yes, whether it's ultimately worthwhile or not, i spent some time trying to get E@H taking full advantage of my shiny new 64bit processor. I managed to get a 64bit boinc client compiled but, as has been pointed out here, didn't have any success with einstein due to the unrecognisable hostname.

That aside, I'm a little concerned about the benchmarks I'm getting on my home machine (AuthenticAMD AMD Athlon(tm) 64 Processor 3500+):
Measured floating point speed 10 million ops/sec
Measured integer speed 10 million ops/sec

My work machine (GenuineIntel Intel(R) Pentium(R) 4 CPU 3.00GHz) gets:
Measured floating point speed 657.54 million ops/sec
Measured integer speed 1962.79 million ops/sec

I wouldn't be surprised at getting higher benchmarks on the P4, it is a faster processor afterall, but should it really be *this* much higher?

Apologies if this is slightly off-topic but it seemed like a reasonably sensible place to ask.

ps. for clarity, i should point out that it's a 32bit boinc client on the AMD64 running a 32bit E@H...

ebahapo
ebahapo
Joined: 22 Jan 05
Posts: 47
Credit: 755,276
RAC: 0

RE: That aside, I'm a

Message 43286 in response to message 43285

Quote:

That aside, I'm a little concerned about the benchmarks I'm getting on my home machine (AuthenticAMD AMD Athlon(tm) 64 Processor 3500+):
Measured floating point speed 10 million ops/sec
Measured integer speed 10 million ops/sec

My work machine (GenuineIntel Intel(R) Pentium(R) 4 CPU 3.00GHz) gets:
Measured floating point speed 657.54 million ops/sec
Measured integer speed 1962.79 million ops/sec

I wouldn't be surprised at getting higher benchmarks on the P4, it is a faster processor afterall, but should it really be *this* much higher?


I have a 2.2GHz Athlon 64 and a 3.2GHz P4 running the 64-bit client and the benchmark reports 1.9/5.2 and 1.1/3.0, respectively. So, first off, the AMD processor is clearly the fastest of both, which can be confirmed by the average WU times. And, on to your issue, I've seen the benchmark report bogus numbers. Have you tried running the benchmark again?

HTH

PS: the same AMD system gets 1.1/2.1 in the benchmark using the 32-bit client, as you can see here.

ebahapo
ebahapo
Joined: 22 Jan 05
Posts: 47
Credit: 755,276
RAC: 0

RE: RE: RE: RE: So

Message 43287 in response to message 43284

Quote:
Quote:
Quote:
Quote:
So far, two projects have added support for x86-64: SETI sends the x86 application and SIMAP a true x86-64 application. Would you consider supporting x86_64-pc-linux-gnu or x86_64-unknown-linux-gnu similarly?

Chess960 now supports AMD64 with a native Linux application and HashClash with a 32-bit Linux application...

HashClash now supports AMD64 on Windows with a 32-bit application...

Guess what? Leiden Classical now supports AMD64 on Linux with a 32-bit application. ;-)


Yay! Malaria now supports AMD64 on Linux with a 32-bit application too.

ebahapo
ebahapo
Joined: 22 Jan 05
Posts: 47
Credit: 755,276
RAC: 0

RE: RE: RE: RE: Quote

Message 43288 in response to message 43287

Quote:
Quote:
Quote:
Quote:
Quote:
So far, two projects have added support for x86-64: SETI sends the x86 application and SIMAP a true x86-64 application. Would you consider supporting x86_64-pc-linux-gnu or x86_64-unknown-linux-gnu similarly?

Chess960 now supports AMD64 with a native Linux application and HashClash with a 32-bit Linux application...

HashClash now supports AMD64 on Windows with a 32-bit application...

Guess what? Leiden Classical now supports AMD64 on Linux with a 32-bit application. ;-)

Yay! Malaria now supports AMD64 on Linux with a 32-bit application too.


ABC ß now provides a 64-bit application for AMD64 on Linux yielding double the performance.

Trog Dog
Trog Dog
Joined: 25 Nov 05
Posts: 191
Credit: 541,562
RAC: 0

RE: RE: RE: RE: Quote

Message 43289 in response to message 43288

Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
So far, two projects have added support for x86-64: SETI sends the x86 application and SIMAP a true x86-64 application. Would you consider supporting x86_64-pc-linux-gnu or x86_64-unknown-linux-gnu similarly?

Chess960 now supports AMD64 with a native Linux application and HashClash with a 32-bit Linux application...

HashClash now supports AMD64 on Windows with a 32-bit application...

Guess what? Leiden Classical now supports AMD64 on Linux with a 32-bit application. ;-)

Yay! Malaria now supports AMD64 on Linux with a 32-bit application too.

ABC ß now provides a 64-bit application for AMD64 on Linux yielding double the performance.

I'll add my name to the petition for 64bit linux support.

ebahapo
ebahapo
Joined: 22 Jan 05
Posts: 47
Credit: 755,276
RAC: 0

RE: RE: RE: RE: Quote

Message 43290 in response to message 43288

Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
So far, two projects have added support for x86-64: SETI sends the x86 application and SIMAP a true x86-64 application. Would you consider supporting x86_64-pc-linux-gnu or x86_64-unknown-linux-gnu similarly?

Chess960 now supports AMD64 with a native Linux application and HashClash with a 32-bit Linux application...

HashClash now supports AMD64 on Windows with a 32-bit application...

Guess what? Leiden Classical now supports AMD64 on Linux with a 32-bit application. ;-)

Yay! Malaria now supports AMD64 on Linux with a 32-bit application too.

ABC ß now provides a 64-bit application for AMD64 on Linux yielding double the performance.


Docking just added support for AMD64 Linux clients through a 32-bit application for the time being.

Trog Dog
Trog Dog
Joined: 25 Nov 05
Posts: 191
Credit: 541,562
RAC: 0

Einstein doesn't support

Einstein doesn't support x86_64 but you should be able to crunch with the 32bit app. You need to create an app_info.xml containing the following



einstein_S5RI


einstein_S5RI_4.17_i686-pc-linux-gnu



einstein_S5RI_4.17_i686-pc-linux-gnu.so


einstein_S5RI
417

einstein_S5RI_4.17_i686-pc-linux-gnu



einstein_S5RI_4.17_i686-pc-linux-gnu.so

and download the following files -

http://einstein.phys.uwm.edu/download/einstein_S5RI_4.17_i686-pc-linux-gnu
http://einstein.phys.uwm.edu/download/einstein_S5RI_4.17_i686-pc-linux-gnu.so

If you haven't already attached your 64bit box to E@H you need to.

Create the app_info.xml file, and copy the two downloaded files into you E@H project directory, you'll need to set the executable bit on the first file (the one that doesn't end in .so). Stop BOINC, then restart it.

If all went well there will be a message early on in the log stating app_info found for Einstein, and soon you'll be downloading wu's and crunching away.

It would still be nice if the project supported the x86_64 platform though, without us users having to resort to the anonymous platform.

ebahapo
ebahapo
Joined: 22 Jan 05
Posts: 47
Credit: 755,276
RAC: 0

RE: Einstein doesn't

Message 43292 in response to message 43291

Quote:

Einstein doesn't support x86_64 but you should be able to crunch with the 32bit app. You need to create an app_info.xml containing the following...

It would still be nice if the project supported the x86_64 platform though, without us users having to resort to the anonymous platform.


I tried this once for Einstein and Rosetta, but it's quite annoying, especially when these projects have a fairly high application update frequency and one has several systems.

If the developers are not willing to go through the trouble of porting the application, they could easily configure the server to send the x86 application to x86-64 clients. Others projects have done it without much effort.

Trog Dog
Trog Dog
Joined: 25 Nov 05
Posts: 191
Credit: 541,562
RAC: 0

RE: I tried this once for

Message 43293 in response to message 43292

Quote:


I tried this once for Einstein and Rosetta, but it's quite annoying, especially when these projects have a fairly high application update frequency and one has several systems.

If the developers are not willing to go through the trouble of porting the application, they could easily configure the server to send the x86 application to x86-64 clients. Others projects have done it without much effort.

Couldn't agree more Augustine - it's bloody annoying having to create app_info's, but it's better than having the box sit idle.

The best solution though is for the project to support the growing number of 64bit crunchers.

ebahapo
ebahapo
Joined: 22 Jan 05
Posts: 47
Credit: 755,276
RAC: 0

RE: Couldn't agree more

Message 43294 in response to message 43293

Quote:
Couldn't agree more Augustine - it's bloody annoying having to create app_info's, but it's better than having the box sit idle.


But that's not the alternative when there are projects such as SIMAP, Chess960, ABC ß, SETI & SETI ß, HashClash, Leiden, Malaria and Docking that do support AMD64 clients. ;-)

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.