Re: 64bit Linux client slower than 32bit client?

I've added an entry in the faq to address this common question, I hope  
it helps, if anything remains unclear, please let me know.

http://faq.distributed.net/cache/291.html

On 25-Oct-08, at 2:45 PM, David Bolt wrote:

> Having just noticed the release of the latest clients, I decided it
> might be a good idea to benchmark the 64bit clients before starting to
> use them on my 64bit systems, rather than continuing to use the 32bit
> clients as I am now. However, on the two systems I've performed the
> benchmarks, both have shown that the 32bit client when benchmarking  
> the
> RC5-72 core is at least 500kkeys faster than the 64bit client on the
> same system.
>
> The first system, an AMD Sempron based system running the 64bit  
> version
> of openSUSE 10.3 at 1.6GHz, is consistently 500kkeys/s faster with the
> 32bit client.
>
> The second system tested, an AMD 64 X2 4200+ running the same 64bit
> version of openSUSE 10.3 but at 2.2GHz per core, is also consistently
> 700kkeys/s faster with the 32bit client.
>
> So, my question is does anyone have an idea why there should be this
> drop in speed?
>
> Also, as a matter of interest, while doing the benchmarking of the
> latest clients, the latest 64bit client on the X2 is shown to be  
> almost
> 1Mnode/s faster than the latest 32bit client, and a little more than
> 1Mnode/s faster than the 32bit 504 client I was previously using.
>
> dav2:/usr/lib/dnetc # ls -l dnetc-2.90*
> -rwxr-xr-x 1 root root 463952 Sep 26  2005 dnetc-2.9008.490
> -rwxr-xr-x 1 root root 983576 Jun  8  2005 dnetc-2.9011.496
> -rwxr-xr-x 1 root root 983576 Oct 13  2006 dnetc-2.9011.496.64bit
> -rwxr-xr-x 1 root root 512816 Dec 28  2005 dnetc-2.9012.497
> -rwxr-xr-x 1 root root 521300 Nov 27  2006 dnetc-2.9013.498
> -rwxr-xr-x 1 root root 523348 Feb  8  2008 dnetc-2.9015.504
> -rwxr-xr-x 1 root root 531624 Oct 25 19:58 dnetc-2.9016.506.32bit
> -rwxr-xr-x 1 root root 972256 Oct 25 19:58 dnetc-2.9016.506.64bit
> dav2:/usr/lib/dnetc # ./dnetc-2.9016.506.64bit -benchmark ; ./ 
> dnetc-2.9016.506.32bit -benchmark ; ./dnetc-2.9015.504 -benchmark
>
> distributed.net client for Linux Copyright 1997-2008, distributed.net
> Please visit http://www.distributed.net/ for up-to-date contest  
> information.
> Start the client with '-help' for a list of valid command line  
> options.
>
>
> dnetc v2.9016-506-CFR-08081716 for Linux (Linux 2.6.22.17-0.1- 
> default).
> Please provide the *entire* version descriptor when submitting bug  
> reports.
> The distributed.net bug report pages are at http://bugs.distributed.net/
>
> [Oct 25 19:22:48 UTC] Automatic processor type detection found
>                      an AMD Athlon 64 X2 Dual Core processor.
> [Oct 25 19:22:48 UTC] RC5-72: Running micro-bench to select fastest  
> core...
> [Oct 25 19:23:19 UTC] RC5-72: using core #1 (KBE-64 3-pipe).
> [Oct 25 19:23:40 UTC] RC5-72: Benchmark for core #1 (KBE-64 3-pipe)
>                      0.00:00:17.36 [8,813,006 keys/sec]
> [Oct 25 19:24:00 UTC] OGR-P2: Benchmark for core #0 (GARSP 6.0-64)
>                      0.00:00:18.12 [29,896,330 nodes/sec]
>
> distributed.net client for Linux Copyright 1997-2008, distributed.net
> Please visit http://www.distributed.net/ for up-to-date contest  
> information.
> Start the client with '-help' for a list of valid command line  
> options.
>
>
> dnetc v2.9016-506-CFR-08081716 for Linux (Linux 2.6.22.17-0.1- 
> default).
> Please provide the *entire* version descriptor when submitting bug  
> reports.
> The distributed.net bug report pages are at http://bugs.distributed.net/
>
> [Oct 25 19:24:00 UTC] Automatic processor type detection found
>                      an AMD Athlon 64 X2 Dual Core processor.
> [Oct 25 19:24:00 UTC] RC5-72: using core #6 (GO 2-pipe).
> [Oct 25 19:24:22 UTC] RC5-72: Benchmark for core #6 (GO 2-pipe)
>                      0.00:00:17.57 [9,442,506 keys/sec]
> [Oct 25 19:24:22 UTC] OGR-P2: using core #4 (GARSP 6.0-asm-rt1-mmx- 
> amd).
> [Oct 25 19:24:40 UTC] OGR-P2: Benchmark for core #4 (GARSP 6.0-asm- 
> rt1-mmx-amd)
>                      0.00:00:16.27 [28,904,560 nodes/sec]
>
> distributed.net client for Linux Copyright 1997-2007, distributed.net
> Please visit http://www.distributed.net/ for up-to-date contest  
> information.
>
>
> dnetc v2.9015-504-CFR-07110717 for Linux (Linux 2.6.22.17-0.1- 
> default).
> Please provide the *entire* version descriptor when submitting bug  
> reports.
> The distributed.net bug report pages are at http://www.distributed.net/bugs/
>
> [Oct 25 19:24:40 UTC] Automatic processor type detection found
>                      an AMD K8-11 processor.
> [Oct 25 19:24:40 UTC] RC5-72: using core #6 (GO 2-pipe).
> [Oct 25 19:25:01 UTC] RC5-72: Benchmark for core #6 (GO 2-pipe)
>                      0.00:00:17.56 [9,409,983 keys/sec]
> [Oct 25 19:25:01 UTC] OGR-P2: using core #4 (GARSP 6.0-asm-rt1-mmx- 
> amd).
> [Oct 25 19:25:20 UTC] OGR-P2: Benchmark for core #4 (GARSP 6.0-asm- 
> rt1-mmx-amd)
>                      0.00:00:16.24 [28,813,607 nodes/sec]
> dav2:/usr/lib/dnetc #
>
>
> Regards,
>        David Bolt
>
> -- 
> Team Acorn: http://www.distributed.net/ RC5-72 @ ~70Mkeys/s
> SUSE 10.1 32 |                   | openSUSE 10.3 32b | openSUSE 11.0  
> 32b
>             | openSUSE 10.2 64b | openSUSE 10.3 64b | openSUSE 11.0  
> 64b
> RISC OS 3.6  | TOS 4.02          | openSUSE 10.3 PPC | RISC OS 3.11
> _______________________________________________
> rc5 mailing list
> rc5-Ra3b/QYEcJ3d140v2zMXi0fjHoOT/h/[email protected]
> http://lists.distributed.net/mailman/listinfo/rc5

---
Steven Nikkel
steven_nikkel-WeBCG/[email protected]




_______________________________________________
rc5 mailing list
rc5-Ra3b/QYEcJ3d140v2zMXi0fjHoOT/h/[email protected]
http://lists.distributed.net/mailman/listinfo/rc5