new protein

[FoBoT]

129AA? slower?

looking for confirmation

only a little slower? or much slower?

[bwkaz]

I don't know, as I haven't been able to get the update yet.

There are 4 sets buffered on one machine, and 2 on the other... luckily, these will still count.

[pfb]

haven't got the update but the update email mentions 'A new version of the Distributed Folding Project Client (Lysozyme - 135L)'...

[FoBoT]

from the last 105 protein

------------------------------------------------------------
Distributed Folding Windows dfGUI v1.8 Benchmark

Sample Size : 19670 structures over 10396 seconds.
Protein Size: 105AA

Structures Per Second: 1.89
Structures Per Minute: 113.5
Structures Per Hour : 6811
Structures Per Day : 163475

OS : Windows XP MHz: 1063
CPU: Intel(R) Pentium(R) III Mobile CPU 1066MHz
Client Switches: -df -qt -i f -rt -s 10000
------------------------------------------------------------





compared to the NEW 129 AA protein

same laptop, the new benchmark is of course a much shorter period of time/smaller sample size, so some inaccuracy is expected


------------------------------------------------------------
Distributed Folding Windows dfGUI v1.8 Benchmark

Sample Size : 600 structures over 629 seconds.
Protein Size: 129AA

Structures Per Second: 0.95
Structures Per Minute: 57.2
Structures Per Hour : 3434
Structures Per Day : 82417

OS : Windows XP MHz: 1063
CPU: Intel(R) Pentium(R) III Mobile CPU 1066MHz
Client Switches: -df -qt -rt -s 3000
------------------------------------------------------------

[runestar]

I got the 129 aka 135L on my machine... didn't see any errors about the server being down or unreachable in the log. Looks like I just started a new set a few minutes ago too.

I thought Brian said they had some more maintenance they had to finish up. Things must have gone really well or they didn't finish the maintenance.

RS½

[FoBoT]

oh, yeah, in Windows XP
it is using about 106 MB of RAM with -rt

so those using the -rt switch on 128MB boxes are probably going to be SOL, unless they are running a super lite *nix OS, or maybe Windows 95

[pointwood]

Looks like it's a lot slower.

[bwkaz]

It "feels" slower, that's for sure (this box got the update while I was at class). I haven't run the df-linux-bench.pl script, though, because the progress.txt file didn't reset itself. So if I were to run it, I'd get 99% of the data from the old 105AA protein, and about 1% from the new one. Definitely not ideal... :p

Anyway, once the 3 that have piled up again since then get uploaded, I'll restart the client and get some numbers.

[pfb]

was getting around 3.4 / sec with the 105, and getting 1.8 with this one...will post some 'proper' benchies once it's run for a while

[Pascal]

On my Athlon 1.2 GC/s I'll do

something around 1.15 structures per second or 100k per day with the option -rt.

[DGROMS.com]

Much slower. previous protein was doing 320,000 on my Athlon @1904Mhz now it is only doing 170,000.:(

[Paratima]

DGROMS, that matches my looks at it... just about 1/2 the speed. But WOW, those uploads!

FoBoTski, you oughta love that!

[FoBoT]

some more benchmarks

------------------------------------------------------------
Distributed Folding Windows dfGUI v1.8 Benchmark

Sample Size : 19640 structures over 17261 seconds.
Protein Size: 129AA

Structures Per Second: 1.14
Structures Per Minute: 68.3
Structures Per Hour : 4096
Structures Per Day : 98308

OS : Windows XP MHz: 1795
CPU: Intel(R) Pentium(R) 4 CPU 1.80GHz
Client Switches: -df -qt -rt -s 1500
------------------------------------------------------------

------------------------------------------------------------
Distributed Folding Windows dfGUI v1.8 Benchmark

Sample Size : 20405 structures over 17457 seconds.
Protein Size: 129AA

Structures Per Second: 1.17
Structures Per Minute: 70.1
Structures Per Hour : 4208
Structures Per Day : 100991

OS : Windows XP MHz: 1794
CPU: Intel(R) Pentium(R) 4 CPU 1.80GHz
Client Switches: -df -qt -rt -s 1500
------------------------------------------------------------

[DGROMS.com]

Originally posted by Paratima
DGROMS, that matches my looks at it... just about 1/2 the speed. But WOW, those uploads!

FoBoTski, you oughta love that!

Yeah this machine is an axe. Last protein it ran for the entire length non stop and produced 11.4 million structures. The uploads are much better, I have been putting off bring lots more of my machines over to DF from F@H and now with the problem F@H has and the reduced DF uploads I have taken heaps more machines across. :D :cheers: