Hi Folks,

I've just uploaded a new SETI client for AMD64 processors, version 4.07.3a (see link in my signiture). This should be the fastest client available for Athlon 64, Athlon FX and Opteron processors.

The client is a 32-bit client so should run on either a 32-bit (x86) or 64-bit (x86-64) Linux installation. It features suport for SSE and SSE2 instructions and includes the FFTW3 Fast Fourier Transform enhancements as used previously. A true 64-bit version is still much slower - we think the 64-bit compiler is still buggy producing very slow binaries.

I've also updated the i686 client (now at V4.07.3a) with a minor bug fix and an updated Athlon XP client should also be available sometime during the weekend.

Some users (spacemeat) had previously reported that the i686 client didn't work on K6-2/3 processors, so I would appreciate if anyone could confirm whether or not the new i686 client runs on these processors?? (try it in standalone mode with the reference WU to check).

The i686 client is also very fast on PIII CPU's according to Metod's benchmarks:

http://www.mkx.si/boinc/#timing

If Maverick's clients are no faster (see this thread: http://setiweb.ssl.berkeley.edu/forum_thread.php?id=15443) I may also consider doing a PIII (and maybe a PII) specific version to see if it's any faster than the generic i686 version (if people want it).

As usual, back up first or test in standalone mode as I've not thoroughly tested these clients - they are purely made available for your enjoyment :)

Regards,

Ned

---

*** My Guide to Compiling Optimised BOINC and SETI Clients ***
*** Download Optimised BOINC and SETI Clients for Linux Here ***

I've also updated the i686 client (now at V4.07.3a) with a minor bug fix and an updated Athlon XP client should also be available sometime during the weekend.

Good to see you still on the compilings.

Let us know the update on this one please!

Regards,
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

Hi Martin,

Unfortunately I'm unable to test amd64 clients before release (I have no amd64 hardware yet!), and Chris reported it was quite slow - a lot slower than the current athlon-xp client for some reason.

A fellow forum member has sent me an amd64 client that is faster, so I'll try a recompile using the exact procedure he followed and try again with that.

If anyone would like to test an amd64 client before release, please e-mail me to let me know :)

Also, FalconFly has suggested we could do an SSE3 version for the latest venice cores - I'll look into that too.

Ned

---

*** My Guide to Compiling Optimised BOINC and SETI Clients ***
*** Download Optimised BOINC and SETI Clients for Linux Here ***

no errors are thrown when running -standalone on a test work unit.
this is the same P2 with the 4.07.3a i686

The AMD64 client is now fixed and is back up. FalconFly reports +71.4% on a So754 3000+ and Chris Bosshard reports +80% on a So939 3500+. This seems to be the fastest client currently available for AMD64 processors. I'm planning a SSE3 version for the new venice core that supports SSE3 SIMD instructions.

no errors are thrown when running -standalone on a test work unit.
this is the same P2 with the 4.07.3a i686

Thanks spacemeat. Do you have an AMD K6-2/3 system you could test on? I'm not sure if it will run on a K6, although it should in theory.

Ned

---

*** My Guide to Compiling Optimised BOINC and SETI Clients ***
*** Download Optimised BOINC and SETI Clients for Linux Here ***

again

Thanks Ned for your really Good and hard Work here

Greetings from Germany NRW
Ulli S@h Berkeley's Staff Friends Club m7 ©