updated distributed.net win32 Beta Client

[Moose]

The Windows 95/98/NT [x86/Zipped] beta client has been updated to v2.8011.464c

The Pre-Release Page can be found at:
http://www.distributed.net/download/prerelease.html

Please report bugs to Cyrus Patel cyp@fb14.uni-mainz.de

Moose

 

[RC]

Thanks Moose

 

[SSP]

Cool! Thanks Moose.

I hope you guys optimized the Athlon cores again.. hehe

So anyone tried it yet?

 

[vss1980]

This will have to wait until tomorrow, sorry.

 

[stappel]

OGR core still slow on a Classic Athlon 700@750.
Getting 5000 nodes/sec on the new client (C version) was getting 6100 on
the 463 client.

 

[RC]

stappel - thanks for the update. Please be certain to email Cyrus with your results so that he can correct the problem.

 

[SSP]

Old Client

dnetc v2.8010-463-CTR-00071214 for Win32 (WindowsNT 5.0).

[Oct 26 04:18:10 UTC] Benchmark for RC5 core #6 (RG/HB ath)
0.00:00:16.20 [2,264,998.01 keys/sec]


New Client
dnetc v2.8011-464-GTR-00102500 for Win32 (WindowsNT 5.0).

[Oct 26 04:21:31 UTC] Benchmark for RC5 core #6 (RG/HB re-pair II)
0.00:00:17.65 [2,249,667.59 keys/sec]

I took the lowest score from the old client and the highest score on the new client (ran both benchmarks [long] 3 times).

This is on an Athlon 600@642 running Win 2K.

Hope this helps.

BTW, I went back to the older client (v2.8010-463-CTR-00071214).

 

[vss1980]

Got it, tried it and it works fine connecting through my Win98se ICS (NAT) box also.

Benchmarked it: P3-600 (512k cache), 192MB 100MHz CAS-2, Intel BX (Gigabyte BX-2000), Win32 client

463 client:
Average (best of 3) - Percentage performance lost from 463 client
RC5 Average - 1,675,295.5 keys/sec
OGR Average - 3,994,161.25 nodes/sec

464b client:
RC5 Average - 1,674,176.55 keys/sec - -0.067%
OGR Average - 3,829,154.90 nodes/sec - -4.309%

464c client:
RC5 Average - 1,674,422.71 keys/sec - -0.052% (not as bad)
OGR Average - 3,842,158.16 nodes/sec - -3.956% (not as bad)

This of course assumes that the benchmarks are telling the full story - which is unlikely. But the benchmarks still indicate a slower client - looks like if you dont have a proxy problem, stick with 463

 

[BGod]

I changed all 5 machines over lastnight and let it run. The results after almost 24 hours are: All machines are doing the same output as they were on v2.8010-463.

The upside to this client is that it has NO problems finding the pproxy on my ICS machine. I actually killed the Task Scheduler on 3 of my machines that I had to run to force them to flush blocks to the pproxy (dnetc.exe -update).

Now I get this: [Oct 26 22:33:24 UTC] Detected IP-link on 'eth0/lan0'...

Moose, Thank You!

 

[Viztech]

Heyhey, that sounds like great news BGod! I assume that was with the 464c version.

As for the benchmark vs actual speed, I let the 464 a or b version crunch a full OGR stub on my K6 233. The speed overall was .993 Mnodes/sec, which is down from 1.5 Mnodes/sec with the good old 463b client.
The 464 benchmarked at .973 Mnodes/sec, so the actual is quite close.

I'll drop Cyp an email and switch back to 463b for OGR cracking.

viz