Can you manually ask for different WU's?

TheLostSwede

Since they are on different ip adresses, i thought that may be a possibility. This is for benchmarking purposes only btw, not to reject wu's for better points.

I'm doing P2124 and P2125 at about 23 minutes per frame at 3.7 ghz with a 4mb cache Conroe. Is that very slow or is it on par? I use the advethod flag on both. Should i add more flags?

Datsun-1600

TheLostSwede wrote:

Since they are on different ip adresses, i thought that may be a possibility. This is for benchmarking purposes only btw, not to reject wu's for better points.

Version 6 of the client is supposed to give you that option.

TheLostSwede wrote:

I'm doing P2124 and P2125 at about 23 minutes per frame at 3.7 ghz with a 4mb cache Conroe. Is that very slow or is it on par? I use the advethod flag on both. Should i add more flags?

My Conroe @ 3 Ghz is doing a p2125 in just under 28 minutes, so you are fine.

You should have the -local flag, if you are running more than one instance and if you want more imformation you can run -verbosity 9 flag (This increases the amount of information that F@H writes to the log file. Not required but is a useful diagnostic switch, if you are having a problem with F@H you may be able to pin point it faster).

Datsun 1600

TheLostSwede

Version 6 out?

csimon

TheLostSwede wrote:

Since they are on different ip adresses, i thought that may be a possibility. This is for benchmarking purposes only btw, not to reject wu's for better points.

I'm doing P2124 and P2125 at about 23 minutes per frame at 3.7 ghz with a 4mb cache Conroe. Is that very slow or is it on par? I use the advethod flag on both. Should i add more flags?

if you remove -advmethods flag you should get different wu's from a different server. Don't forget to restart.

If you're merely benching I would suggest ORTHOS. You can stress test w/ gromacs or prime on both cores at the same time and it works great. That's what I use.

Datsun-1600

TheLostSwede wrote:

Version 6 out?

It is on the list to do, by Stanford and it has supposedly been coming for a while, but still no sign of it.

Datsun 1600

FoldingAddict

TheLostSwede wrote:

I'm doing P2124 and P2125 at about 23 minutes per frame at 3.7 ghz with a 4mb cache Conroe. Is that very slow or is it on par? I use the advethod flag on both. Should i add more flags?

And my 2.52GHz (2MB cache) E6300 is running the P2124 at 33 minutes per frame.

~FA

TheLostSwede

Wow! Just got the 2414 Protein now and considering that it is worth a lot of credit, things are going fast. Each fram takes 13 and a half minute and if my headcount is still ok, that's pretty much 23-24 hours for the wu. Theoretically, 1200 points per day for one cpu is a lot more than when i was active a few years back.

csimon

TheLostSwede wrote:

Wow! Just got the 2414 Protein now and considering that it is worth a lot of credit, things are going fast. Each fram takes 13 and a half minute and if my headcount is still ok, that's pretty much 23-24 hours for the wu. Theoretically, 1200 points per day for one cpu is a lot more than when i was active a few years back.

Yeah those are real nice and fold very quickly. Wait til you get a double gromacs ...you'll go thru the ceiling!

Datsun-1600

TheLostSwede wrote:

Wow! Just got the 2414 Protein now and considering that it is worth a lot of credit, things are going fast. Each fram takes 13 and a half minute and if my headcount is still ok, that's pretty much 23-24 hours for the wu. Theoretically, 1200 points per day for one cpu is a lot more than when i was active a few years back.

Add the -advmethods flag and you will pick up p1495_tet_new, 1450ppd here, remembering to restart the instance to activate the flag.

Datsun 1600