BOINC benchmarks

[Bateluer]

Originally posted by: Assimilator1


That P4 isn't running on PC133 RAM is it?
For some stupid reason (budget systems I guess) Intel thought it would be a good idea to run the P4/Cel on PC133 RAM on some variants of the 845 chipset for a while:roll: ,this absolutely cripples the P4s performance.
If not then the dual channel RAM on the Cel must be the answer (P4s & their Cel variants love bandwidth)

Seems you are right, the P4 is running on single channel PC133 memory. No wonder it feels so sluggish in comparison to the Cely. I wonder if I should switch the CPUs around? These were both cheap systems though, sub 200 dollar for each.

Edit - I'd also like to add that I just passed a personal milestone in S@H. 50,000 credits completed.

 

[The Borg]

Hi all,

With Seti Classic, I had PC's as as slow as P1-133's running. But I decided about a year ago that from a cost point of view it was just not worth keeping those slow machines running. I have made a break point for myself. Nothing below 1GHz. Although I do have a dual P3-600 running (1.2 GHz) and a 950MHz OC to 1263MHz. It really is not worth the hassels, especially if they are built up machines with old parts that may be doddgy.

 

[Assimilator1]

Fair enough.

Bateluer
Yea swaping the P4 over with the Cely would give ya a big boost
Congrats on hitting 50k btw

 

[Bateluer]

Originally posted by: Assimilator1
Fair enough.

Bateluer
Yea swaping the P4 over with the Cely would give ya a big boost
Congrats on hitting 50k btw

So I opened the HP system to discover that it actually does have DDR memory, PC2100 memory to be exact. Unfortunately, I could not lift the clips on the CPU cooler. HP's using some fandangled design with dual clips on it. I don't have the thumb strength to lift the stupid things.

I did install my PCI FX5500 video card though, which lets the P4 have all the memory bandwidth from the RAM. Seems a bit snappier, and the BOINC benches go up slightly.

 

[bluestrobe]

I've installed the "optimized" clients for all of my systems and haven't noticed that much, if any improvement.

edit: I read up on a few websites and overlooked I didn't have the proper "optimized" software for the version of BOINC I'm using. Currently using 5.4.11. Anyone have links to the updated optimized clients?

 

[AstroGuardian]

OK

My 3,06Ghz CeleronD gave this result:
1442 Whetstone
2382 Drystone.

At home i have an Athlon64 3500AM2, a Prescott at 3,0ghz, and a 1,6 ghz Intel Centrino on my laptop.

I will benchmark it later and wite tomorrow

 

[Dulanic]

Originally posted by: bluestrobe
I've installed the "optimized" clients for all of my systems and haven't noticed that much, if any improvement.

edit: I read up on a few websites and overlooked I didn't have the proper "optimized" software for the version of BOINC I'm using. Currently using 5.4.11. Anyone have links to the updated optimized clients?

It only helps the benchmark, and not crunching the units themselves.

 

[Assimilator1]

And theirs a link for it further up in this thread.

AstroGuardian
Which BOINC version is that on?

 

[AstroGuardian]

Originally posted by: Assimilator1
And theirs a link for it further up in this thread.

AstroGuardian
Which BOINC version is that on?

The version is 5.4.11.

 

[dennilfloss]

10/9/2006 2:36:01 PM||Running CPU benchmarks
10/9/2006 2:37:00 PM||Benchmark results:
10/9/2006 2:37:00 PM|| Number of CPUs: 2
10/9/2006 2:37:00 PM|| 2552 double precision MIPS (Whetstone) per CPU
10/9/2006 2:37:00 PM|| 4763 integer MIPS (Dhrystone) per CPU
10/9/2006 2:37:00 PM||Finished CPU benchmarks

 

[Insidious]

10/9/2006 3:58:33 PM||Starting BOINC client version 5.5.0 for windows_intelx86
10/9/2006 3:58:33 PM||libcurl/7.15.3 OpenSSL/0.9.8a zlib/1.2.3
10/9/2006 3:58:33 PM||Data directory: C:\Program Files\BOINC
10/9/2006 3:58:33 PM||Processor: 2 AuthenticAMD AMD Athlon(tm) 64 X2 Dual Core Processor 3800+
10/9/2006 3:58:33 PM||Memory: 1023.48 MB physical, 2.40 GB virtual
10/9/2006 3:58:33 PM||Disk: 74.53 GB total, 30.91 GB free

...
...

10/9/2006 4:27:39 PM||Suspending network activity - running CPU benchmarks
10/9/2006 4:27:41 PM||Running CPU benchmarks
10/9/2006 4:28:40 PM||Benchmark results:
10/9/2006 4:28:40 PM|| Number of CPUs: 2
10/9/2006 4:28:40 PM|| 4192 floating point MIPS (Whetstone) per CPU
10/9/2006 4:28:40 PM|| 13589 integer MIPS (Dhrystone) per CPU
10/9/2006 4:28:40 PM||Finished CPU benchmarks
10/9/2006 4:28:41 PM||Resuming computation

 

[apriest]

BOINC client version 5.6.5 for windows_intelx86
Intel(R) Pentium(R) 4 CPU 3.06GHz
Memory: 2.00 GB physical, 3.85 GB virtual
Disk: 74.52 GB total, 26.10 GB free
Microsoft Windows XP Professional Edition, Service Pack 2, (05.01.2600.00)
Number of CPUs: 2 (Hyperthreading)
1277 floating point MIPS (Whetstone) per CPU
1283 integer MIPS (Dhrystone) per CPU


BOINC client version 5.6.5 for windows_intelx86
AMD Athlon(tm) Processor (1800+ 266MHz FSB I think)
Memory: 1023.48 MB physical, 1.67 GB virtual
Disk: 11.72 GB total, 1.99 GB free
Microsoft Windows 2003 Standard Server Edition, Service Pack 1, (05.02.3790.00)
Number of CPUs: 1
1047 floating point MIPS (Whetstone) per CPU
1812 integer MIPS (Dhrystone) per CPU


The newer Pentium D's and Core 2's are at remote sites and I can't check the benchmarks results until I'm back there. I'm building a couple 2x Dual Core Xeon 5130 servers soon, I can't wait to see what they can crank out. 8 new cores to add to the mix baby!

 

[Bateluer]

Originally posted by: apriest
I'm building a couple 2x Dual Core Xeon 5130 servers soon, I can't wait to see what they can crank out. 8 new cores to add to the mix baby!

You have too much money. Send some my way.

 

[Assimilator1]

Eight new cores?! sweet!

v5.6? is that a new one?
I see it still seems to penalise the P4

Originally posted by: dennilfloss
10/9/2006 2:36:01 PM||Running CPU benchmarks
10/9/2006 2:37:00 PM||Benchmark results:
10/9/2006 2:37:00 PM|| Number of CPUs: 2
10/9/2006 2:37:00 PM|| 2552 double precision MIPS (Whetstone) per CPU
10/9/2006 2:37:00 PM|| 4763 integer MIPS (Dhrystone) per CPU
10/9/2006 2:37:00 PM||Finished CPU benchmarks

Hi Den
So what CPU is that then?

 

[apriest]

Here are statistics on the whole fleet! I thought it would be easier to make a spreadsheet and graph of the floating and integer ops/sec of all 34 cores or threads (several are Hyperthreaded) currently crunching away, sorted by floating and then integer benchmark results. Some of these benchmark numbers make absolutely no sense to me though, like the two Pentium 4 2.80GHz Prescotts at the way bottom, each generating only 10. Huh?!

Anyway, check it out, see how they compare to your numbers. These are all currently running BOINC 5.6.5.

http://www.coastaltech.org/rosetta_summary.htm

(Note: It might be easier to read in IE since FireFox doesn't shrink the columns.)

 

[dennilfloss]

Originally posted by: Assimilator1

Hi Den
So what CPU is that then?

The one in my sig.

 

[Assimilator1]

Wow awesome rig Den!:Q

apriest
Like the graph
And yea something definatly went wrong with the P4 2.8GHz scores ,wierd.

Of course the other unknown factor is how much that new BOINC version changes the scores.Could you try 5.4.9 on 1 rig for comparision?

 

[dennilfloss]

Thanks. I did my three-year upgrade in April. This is my Oblivion-playing machine but I haven't started the game yet.

 

[Tweakin]

For what it's worth:

BOINC 5.4.11
Measured floating point speed 2297.32 million ops/sec
Measured integer speed 4281.4 million ops/sec

See sig for system...

 

[dennilfloss]

I'm using 5.2.13. Your numbers seem correct to me (I'm at 2.7GHz on a 4200+ Manchester).

 

[dennilfloss]

10/12/2006 3:58:05 PM||Running CPU benchmarks
10/12/2006 3:59:04 PM||Benchmark results:
10/12/2006 3:59:04 PM|| Number of CPUs: 2
10/12/2006 3:59:04 PM|| 4588 floating point MIPS (Whetstone) per CPU
10/12/2006 3:59:04 PM|| 15048 integer MIPS (Dhrystone) per CPU
10/12/2006 3:59:04 PM||Finished CPU benchmarks


Updating my BOINC client to 5.5.0 made a hell of a difference. :Q

I am using the optimised 5.15 enhanced SETI application for crunching btw.

 

[Assimilator1]

Yea 5.5.x seems to massively boost the benchmark score ,how realistic that boost is I've no idea.....