No top Genome-crunchers this week, 28.04-05.05.2002

[Rattledagger]

There have been some stats-problems this week, and I'm not sure all results is credited yet.
I'll therefore postpone stats to next week.

 

[Confused]

Thanks Rattledagger

I await the stats next week!

Confused

 

[ShotgunSi]

Yeah I noticed the stats didn't get updated over the weekend for awhile
Darn I hoped that my new AMD XP 1600 would push me over the top too

 

[jchu14]

whew.... I thought my stat would get really ugly this week. I was just putting together my P4 1.6a box this weekend when my main 30gig hard drive on my old computer died on me. . and of course... I didn't have a back up. . So I've been downloading patches, drivers etc this whole weekend while putting together my new computer. Haven't put G@H on yet. But the good thing is, the new P4 1.6a totally rocks! I'm now running at 2.4 with 150fsb and 187mhz memory@1.65v. I could go to 2.5 but then I"ll have to turn down the memory into 156. I would rather have more memory bandwidth than 100mhz of cpu. I'll put G@H back tomorrow. BTW, is G@H dependent on memory speed liek Seti or raw CPU number crunching power?

 

[Assimilator1]

Bummer ,an extra blow when you don't have any backup
Don't know about your memory question btw

Rattledagger
Thanks for telling us ,btw can you answer his^ question?

 

[ShotgunSi]

I remember reading somewhere that G@H is more dependent on RAW cpu speed. So those extra 100Mhz might help.
Seti on the other hand likes really high FSB speeds...

 

[Rattledagger]

Hmm, did anyone say RTFM?

As I wrote in the "excellent" getting-started-faq: "In Genome, it's the raw cpu-speed that counts. Cache size, memory speed, cpu multiplier is important in Seti, but not for Genome."
The only difference I know of, is that AMD-cpu's are faster than p2/p3 for the same MHz. I don't know how p4 is compared to other cpu's.

 

[jchu14]

hehe thanks Rattledagger.

 

[Assimilator1]

Doh! ,I should of remembered that too I RTFM enough times editing it!

 

[Ken g6]

One question I couldn't find answered was how much disk space it takes (on the Linux version). If it's less than 16mb, I might be able to run it on the Wildnet. If it's less than 8, it would be even easier.

Although none of this would happen until August sometime.

 

[ShotgunSi]

My linux version is taking up roughly 12MB, but I think your milage may vary...
Does the files sizes increase with the amount of work you've crunched? I'm not sure...

 

[Rattledagger]

Well, I've not running Linux, but I guess most of the files are the same as in windows.
In windows, the 2 exe-files is using 1.5 MB. These can be smaller in linux.
The rest of the files are text-files, and is probably the same regardless of windows or linux.
An empty install contains 164 KB text-files.
A wu is around 200 KB.
The file made at start of wu, rotamers.lib, depends on aa-size, and can get over 6 MB large.
Every result is around 100 KB, depending on aa-size.
Genome@home is under running echoing most output into a log, scrlog.gah. This file will grow to 5 MB.

Summing it up, 8 MB is too small. 16 MB should be enough, if you're flushing results fairly often.

Lastly, in Windows, Genome@home is using upto 20 MB real memory, but it's also using 50 MB virtual memory. In other words, genome is using over 60 MB hd-space...