Generation 0 speed in new version?

[Ned]

I notice also that it takes a very LONG TIME to shut down this client. dfGui deleted the "lock" file 5 minutes ago, but trajfoldlite.exe is still running according to W2K's task manager. So, be careful when working with this client. I guess it has to complete a structure before it checks for termination.

My two machines which completed the Gen 0 part in 3-5 hours last time, have estimates of 26 - 58 hours with this client.

One more reason to convert to Linux..

Ned

[willebenn]

The bench mode works ok on my windows system, have not tried it on Linux yet.
On a P4 2.4 the bench shows the new client is 5 times slower !!!!!!!!!!!
Old client
10.9 .4
42.2 15.0
New
10 .55
230.1 15.9

[erk]

Originally posted by willebenn
The bench mode works ok on my windows system, have not tried it on Linux yet.
On a P4 2.4 the bench shows the new client is 5 times slower !!!!!!!!!!!
Old client
10.9 .4
42.2 15.0
New
10 .55
230.1 15.9

P4 2.8C clocked to 2,996Mhz, Redhat linux 9.0

Usr time Sys time
-------- --------
Maketrj 2.640 0.810
Foldtraj 24.920 11.430

[Jammy]

Heh . . .after 14 hours:

XP 2600 - 6029 structures til upload (service install)
XP 1900 - 7457 structures til upload
1.33 T-Bird - 7065 structures til upload

This client is a real pleaser ain't it? I am using W2K . ..I wish I knew what programs I could get rid of that run in the background.

Jammy

Originally posted by pfb
After 12 hours of Folding I have:

1 Barton 2800+, running Mandrake on gen 63 (15%)
1 P4 2.7, running Windows XP on gen 0 (56%)

[Rebels Haven]

Is there any word if this slooooooooooo client is going to be fixed?

Our small team has about 50 systems committed to the project, but unless something happens soon, it's gonna' all be over for us...:shocked:

[pfb]

It doesn't seem to be just gen 0 the Windows client is slow on...although a direct comparison can't really be made, I copied gen 84, structure 33 from my Linux box (Barton 2800+) to the Windows (P4 2.7) box and kicked both off - it's taken 10 mins to do 1 structure on the P4 and in the same time the Linux client has finished that gen...

Benchmarking is also showing a severe drop (around a 6x increase in time taken on the P4)...

[mighty]

I just re-downloaded the text-client from the download-section and it seems the problem have been fixed. The foldtrajlite.exe is dated today at 10.48 and has a size of 3MB. It is also back to normal speed on my w2k box.

How can one tell for sure if this is indeed the new algorithm?

[Jammy]

I hear ya Rebels Haven! I only have three boxes here at home that I am running, but this new client does not appeal to me at all because it is so darned slow and also because it does nothing to appeal to those with a "competitive" nature (like me, even though I have only three boxes, I run them 24/7 and try to catch the "big" guys, lol.)!

After this first slow-ass generation 0 finishes on the three I am moving back to Seti and saying bye-bye to this df team (unless something drastic happens and the client miraculously speeds up!). At least Seti runs faster than df now, lol.

Jammy

Originally posted by Rebels Haven

Our small team has about 50 systems committed to the project, but unless something happens soon, it's gonna' all be over for us...:shocked:

[Jammy]

Did the download page say anything about it being an updated new version?

Originally posted by mighty
I just re-downloaded the text-client from the download-section and it seems the problem have been fixed. The foldtrajlite.exe is dated today at 10.48 and has a size of 3MB. It is also back to normal speed on my w2k box.

How can one tell for sure if this is indeed the new algorithm?

[gOhAsE]

I've seen no message.

But this new version, I've just downloaded, runs faster by far.
Not as slow as the "old" new one.


Stats, I'm coming!