Underclock your pc and benchmark it to increase credit claimed, then crank it up.

This won't work either because the boinc client benchmarks the computer periodically throughout the day. So you'll have to sit watching your computer 24/7 to catch it doing the benchmarks.

Well... it actually benchmarks every five days, so you could just check a couple times every fifth day... But it's still pointless and way more difficult than it needs to be, if all you're after is increased benchmarks. Many simpler ways to do that; install an optimized BOINC client, compile your own with whatever benchmark you like, or the simplest "cheating" method, which I'm not going to spell out, but it requires no rebooting, under/overclocking, or any of the rest of this complicated stuff. I still come back to "what's the point"?

Yes, Linux benchmarks are known to be wrong, considerably lower than the benchmarks for the same computer running Windows. That's why there are dozens of Linux optimized BOINC clients. Install one and be done with it!

---

I posted some results over a month ago and the benchmarks were about a 50% difference between Windows and Linux nonoptimized.


I will go and get new data since version 5.2.6 is out.

---

I stand corrected :)

---

I stand corrected :)

Well I might as well start my newbie questions here.I'm trying to find the best way to set my systems up. I can go with windows or xandros.I'm running both amd and intel processors. I got about 50 systems to setup and I want to find what works best on both processors.
Most of what I've read on the optimized linux is way over my head and explains nothing. Is there any links that put it in plain English that a truck driver can understand??

---

Well I might as well start my newbie questions here.I'm trying to find the best way to set my systems up. I can go with windows or xandros.I'm running both amd and intel processors. I got about 50 systems to setup and I want to find what works best on both processors.
Most of what I've read on the optimized linux is way over my head and explains nothing. Is there any links that put it in plain English that a truck driver can understand??

The Boinc-Wiki has a lot of info in it. It also has a search feature if you're looking for something specific like errors codes and stuff.

Also, as you've read below, the boinc linux client reports less credit, so if you're a "credit whore" then you'll want to install windows. (for now) :)

As for optimized boinc clients, they are versions for linux and windows so once you decide on the platform you can get binary distributions of both.

If you want optimized project clients, then you're pretty limited. For instance Rosetta doesn't provide the source of its client yet, so there isn't an optimized client available.

EDIT: If you decide on Windows, this might help install boinc on your 50 systems: http://boinc.berkeley.edu/win_install.php

---

OK thanks Andrew.I want the program to be optimized to do the best work per cpu time.These systems are mine and at my home so I want the machines to get the best bang for the buck cause I pay the bills on them.

---

Did I answer your questions ok?

---

I find this really interesting...

Out of curiosity if someone has a dual boot win/linux on the machine, maybe they could posts some benchmarks numbers to see if the boinc client is reporting similar Whetstone and Dhrystone numbers. If it's a boinc client problem then the linux benchmark will be significantly lower than the window's ones.

just for you :)

Computer : Dual boot Win XP home / Mandrake 10  
         : P4 3.0 GHz HT 
         : 1.00 GB RAM

Averages 
Win
   1309.67 Whetstone
   1298 Dhrystone

Linux 
   583 Whetstone
   1028.67 Dhrystone   

Actual
Windows XP Home Service pack 2
Boinc version 5.2.2

6/11/2005 8:08:17 AM||Benchmark results:
6/11/2005 8:08:17 AM||   Number of CPUs: 2
6/11/2005 8:08:17 AM||   1293 double precision MIPS (Whetstone) per CPU
6/11/2005 8:08:17 AM||   1295 integer MIPS (Dhrystone) per CPU
6/11/2005 8:08:17 AM||Finished CPU benchmarks
6/11/2005 8:08:57 AM||Running CPU benchmarks
6/11/2005 8:09:56 AM||Benchmark results:
6/11/2005 8:09:56 AM||   Number of CPUs: 2
6/11/2005 8:09:56 AM||   1331 double precision MIPS (Whetstone) per CPU
6/11/2005 8:09:56 AM||   1293 integer MIPS (Dhrystone) per CPU
6/11/2005 8:09:56 AM||Finished CPU benchmarks
6/11/2005 8:10:59 AM||Running CPU benchmarks
6/11/2005 8:11:58 AM||Benchmark results:
6/11/2005 8:11:58 AM||   Number of CPUs: 2
6/11/2005 8:11:58 AM||   1305 double precision MIPS (Whetstone) per CPU
6/11/2005 8:11:58 AM||   1306 integer MIPS (Dhrystone) per CPU
6/11/2005 8:11:58 AM||Finished CPU benchmarks

Mandrake 10
Boinc version 5.2.4

Sun 06 Nov 2005 08:40:49 EST||Running CPU benchmarks
Sun 06 Nov 2005 08:41:48 EST||Benchmark results:
Sun 06 Nov 2005 08:41:48 EST||   Number of CPUs: 2
Sun 06 Nov 2005 08:41:48 EST||   578 double precision MIPS (Whetstone) per CPU
Sun 06 Nov 2005 08:41:48 EST||   1044 integer MIPS (Dhrystone) per CPU
Sun 06 Nov 2005 08:41:48 EST||Finished CPU benchmarks
Sun 06 Nov 2005 08:42:13 EST||Running CPU benchmarks
Sun 06 Nov 2005 08:43:12 EST||Benchmark results:
Sun 06 Nov 2005 08:43:12 EST||   Number of CPUs: 2
Sun 06 Nov 2005 08:43:12 EST||   583 double precision MIPS (Whetstone) per CPU
Sun 06 Nov 2005 08:43:12 EST||   1055 integer MIPS (Dhrystone) per CPU
Sun 06 Nov 2005 08:43:12 EST||Finished CPU benchmarks
Sun 06 Nov 2005 08:43:20 EST||Running CPU benchmarks
Sun 06 Nov 2005 08:44:19 EST||Benchmark results:
Sun 06 Nov 2005 08:44:19 EST||   Number of CPUs: 2
Sun 06 Nov 2005 08:44:19 EST||   588 double precision MIPS (Whetstone) per CPU
Sun 06 Nov 2005 08:44:19 EST||   1047 integer MIPS (Dhrystone) per CPU
Sun 06 Nov 2005 08:44:19 EST||Finished CPU benchmarks

---

An interesting thread on the seti boards: http://setiathome.berkeley.edu/forum_thread.php?id=22079


So if Rosetta started to count flops there would be less of a variance :)


EDIT: Also found this post on seti, from David Baker

---

An interesting thread on the seti boards: http://setiathome.berkeley.edu/forum_thread.php?id=22079


So if Rosetta started to count flops there would be less of a variance :)


EDIT: Also found this post on seti, from David Baker

That'a good idea. I hope they will introduce it soon. :-)

---

LUE