Welcome to the MacNN Forums.

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

You are here: MacNN Forums > Community > Team MacNN > Benchmarks for intel macs vastly different?

Benchmarks for intel macs vastly different?
Thread Tools
zombie67
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Jul 7, 2006, 11:46 PM
 
Hello again.

Check out the differences here:

My 17 MBP (dual 2.17ghz w/ 2gb): 1727 FP, 4444 Int
http://boinc.bakerlab.org/rosetta/sh...?hostid=269065

JustinHoMi's iMac (dual ?gbz w/ 512mb): 4809 FP, 13675 Int
http://boinc.bakerlab.org/rosetta/sh...?hostid=269093

That's about 3 times faster than mine. And because Rosetta uses the benchmarks to award credits, I am getting about 1 credit per ~277 seconds, while he is getting 1 credit per ~93 seconds.

As far as I know, there are no optomized clients for rosetta.

So, what am I doing wrong here? I'm getting completely destroyed by a computer that should be, if anything, slower.
     
JustinHoMi
Forum Regular
Join Date: Oct 2000
Location: Raleigh, NC, USA
Status: Offline
Reply With Quote
Jul 8, 2006, 12:23 AM
 
Hehe, I was just asking myself that question last night! I'm dissapointed that you noticed... I was enjoying the rapid increase in rank

To answer your question... yes, there are optimized clients for rosetta. I found them here:

http://members.dslextreme.com/~reade...boincbeta.html

I got the non-superbench 5.4.9 universal binary.
     
mduell
Posting Junkie
Join Date: Oct 2005
Location: Houston, TX
Status: Offline
Reply With Quote
Jul 8, 2006, 03:16 AM
 
I don't think they're "optimized for Rosetta" but rather "compiled for x86 in the first place."
     
zombie67  (op)
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Jul 8, 2006, 09:37 AM
 
Originally Posted by JustinHoMi
Hehe, I was just asking myself that question last night! I'm dissapointed that you noticed... I was enjoying the rapid increase in rank

To answer your question... yes, there are optimized clients for rosetta. I found them here:

http://members.dslextreme.com/~reade...boincbeta.html

I got the non-superbench 5.4.9 universal binary.
Very cool! How does one use them? The non-superbench version is a single binary called "boinc_5.4.9ExtraCreditUniversal". Do I just replace the "BOINCManager" file in /Applications/BOINCManager/Contents/MacOS ?

Edit: Ah, I see this is CLI. Okay. I can do that (if I must). Do I need to run the benchmark every time I start it?

What strings do people use for CLI? This is what I used. Does it look right? Should I change anything?

/Applications/boinc_5.4.9ExtraCreditUniversal -run_cpu_benchmarks -dir /Library/Application\ Support/BOINC\ Data/

Finally, I just want to make sure this is kosher. Is this client really crunching faster, and therefore earning the higer credit? Or is it just messing with the benchmark numbers, which artificially inflates the scores? I don't want to spend weeks/months using this, only to be told that it is cheating somehow. And then I get all those credits yanked, or yelled at, or whatever.

Thanks,
Erik
( Last edited by zombie67; Jul 8, 2006 at 10:43 AM. )
     
JustinHoMi
Forum Regular
Join Date: Oct 2000
Location: Raleigh, NC, USA
Status: Offline
Reply With Quote
Jul 8, 2006, 12:06 PM
 
Very cool! How does one use them? The non-superbench version is a single binary called "boinc_5.4.9ExtraCreditUniversal". Do I just replace the "BOINCManager" file in /Applications/BOINCManager/Contents/MacOS ?
I think you can use the GUI. You probably just have to open up the package like you were and replace the appropriate files. I could be wrong.

Edit: Ah, I see this is CLI. Okay. I can do that (if I must). Do I need to run the benchmark every time I start it?
No, you don't need to run the benchmark each time. Just once.

What strings do people use for CLI? This is what I used. Does it look right? Should I change anything?
I don't use any params.

Finally, I just want to make sure this is kosher. Is this client really crunching faster, and therefore earning the higer credit? Or is it just messing with the benchmark numbers, which artificially inflates the scores? I don't want to spend weeks/months using this, only to be told that it is cheating somehow. And then I get all those credits yanked, or yelled at, or whatever.
I personally do think it's cheating. I don't know the whole story behind it, and my information may be skewed, but the main guy behind optimized binaries in the windows and linux world (crunch3r) just got accused of cheating by the Rosetta(?) staff, which is why he left the project. Maybe he doesn't think these binaries are cheating, and maybe he has a good reason for fudging the benchmarks like he does, but apparantly some people disagree. In the future, I have heard that some things may change in the way the credit is calculated.
     
zombie67  (op)
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Jul 8, 2006, 12:21 PM
 
Originally Posted by JustinHoMi
I think you can use the GUI. You probably just have to open up the package like you were and replace the appropriate files. I could be wrong.
Ah. Right. Rename it to just "boinc" and replace. Cool.



Originally Posted by JustinHoMi
I personally do think it's cheating. I don't know the whole story behind it, and my information may be skewed, but the main guy behind optimized binaries in the windows and linux world (crunch3r) just got accused of cheating by the Rosetta(?) staff, which is why he left the project. Maybe he doesn't think these binaries are cheating, and maybe he has a good reason for fudging the benchmarks like he does, but apparantly some people disagree. In the future, I have heard that some things may change in the way the credit is calculated.
I think the crunch3r thing was a bit different. SETI doesn't use benchmarks anymore. Credits are now calculated on the number of FP (?) operations, straight up, I think. Aparrently, the crunch3r app had an error in an earlier version, that was fixed in the final version. But accusations were made, and it all blew up.


For the app we are talking about, it *is* changing a piece of the credit calculation (benchmarks). But that *may* be okay, if it really is doing more work. I just can't tell.
     
JustinHoMi
Forum Regular
Join Date: Oct 2000
Location: Raleigh, NC, USA
Status: Offline
Reply With Quote
Jul 8, 2006, 12:27 PM
 
I don't think it is doing three times the work. I imagine that by increasing the benchmark, you trick the server into believing that your super-dooper fast machine took three times as long to do xyz WU (a faster computer should finish the WU faster). Therefore, you get three times the credit.
     
   
Thread Tools
 
Forum Links
Forum Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Top
Privacy Policy
All times are GMT -4. The time now is 11:14 AM.
All contents of these forums © 1995-2017 MacNN. All rights reserved.
Branding + Design: www.gesamtbild.com
vBulletin v.3.8.8 © 2000-2017, Jelsoft Enterprises Ltd.,