Rosetta: Top RAC Systems on the TeAm as of 3 Feb 2007

[Smoke]

I would like to welcome jpeyton to TeAm AnandTech - Rosetta.

I agree with you, Rudy Toody.

IMHO, "siglines" should not carry political statements, especially statements that would be considered inflammatory by others. Putting such messages in one's sigline seems to me to be a "dodge" on the OFF TOPIC rule here on the AnandTech Forums. However, control over what is allowed in one's sigline is strictly under the control of the AnandTech Moderators. If someone has a sigline that "you" think is objectionable you should write the AnandTech Moderator and make a complaint.

You should never respond to a "sigline".

We, as TeAm members, need to be tolerant of other's political views. We definitely should not get into political debates here on the Distributed Computing Forum. Such discussions should only take place in the NEWS AND POLITICS FORUM or in the OFF TOPIC FORUM. Personally, I don't think TeAm members should even get into these discussions at all because they have led in the past to hard feelings and departures from the TeAm.

 

[jpeyton]

Another really quick question.

I'm running the stock BOINC 5.8.8 client, but are there 'optimized' clients available that I should use instead (SMP/AMD K8/Windows x64 optimized)?

 

[petrusbroder]

Originally posted by: Smoke
I would like to welcome jpeyton to TeAm AnandTech - Rosetta.

I agree with you, Rudy Toody.

IMHO, "siglines" should not carry political statements, especially statements that would be considered inflammatory by others. Putting such messages in one's sigline seems to me to be a "dodge" on the OFF TOPIC rule here on the AnandTech Forums. However, control over what is allowed in one's sigline is strictly under the control of the AnandTech Moderators. If someone has a sigline that "you" think is objectionable you should write the AnandTech Moderator and make a complaint.

You should never respond to a "sigline".

We, as TeAm members, need to be tolerant of other's political views. We definitely should not get into political debates here on the Distributed Computing Forum. Such discussions should only take place in the NEWS AND POLITICS FORUM or in the OFF TOPIC FORUM. Personally, I don't think TeAm members should even get into these discussions at all because they have led in the past to hard feelings and departures from the TeAm.

QFT

I agree fully!

 

[petrusbroder]

Originally posted by: jpeyton
Another really quick question.

I'm running the stock BOINC 5.8.8 client, but are there 'optimized' clients available that I should use instead (SMP/AMD K8/Windows x64 optimized)?

There are optimised "clients" but they do not optimise the crunching just the submitted credits by taking the different processor's FP-properties into account.
I think that this issue has become out of date since the BOINC-client version 5.2.XX has changed the way the credits are reported and calculated.

If you wanted to use your processors more efficiently you could (before the current crop of e.g. seti- and einstin-applications) download optimized crunching-applications. They were optimised by compiling them with more efficient compilers, which used different hard-coded functiuons in the CPU better than the "stock" application.

Now however, this issue is not an issue anymore since quite a few of the applications have been optimised by the developers themselves: e.g. einstein@home has had help from one of the most successful optimising programmers ...

I have a limited understanding of this matter - so if I have made an error in the post above, please let me know...

 

[jpeyton]

Originally posted by: petrusbroder

Originally posted by: jpeyton
Another really quick question.

I'm running the stock BOINC 5.8.8 client, but are there 'optimized' clients available that I should use instead (SMP/AMD K8/Windows x64 optimized)?

There are optimised "clients" but they do not optimise the crunching just the submitted credits by taking the different processor's FP-properties into account.
I think that this issue has become out of date since the BOINC-client version 5.2.XX has changed the way the credits are reported and calculated.

If you wanted to use your processors more efficiently you could (before the current crop of e.g. seti- and einstin-applications) download optimized crunching-applications. They were optimised by compiling them with more efficient compilers, which used different hard-coded functiuons in the CPU better than the "stock" application.

Now however, this issue is not an issue anymore since quite a few of the applications have been optimised by the developers themselves: e.g. einstein@home has had help from one of the most successful optimising programmers ...

I have a limited understanding of this matter - so if I have made an error in the post above, please let me know...

Thanks for your help! The client seems to be running efficiently, in line with results from other people. This system looks like it'll put out ~1100 credits/day, which isn't too bad (especially since it is almost silent at full load). Unfortunately I had to remove BOINC from my SFF because the fan at 100% was driving me nuts

 

[petrusbroder]

Unfortunately I had to remove BOINC from my SFF because the fan at 100% was driving me nuts

I have exactly the same experience .
I had to put the old SFF-computer in the basement after I purchased an other one (which was much more silent).

 

[Canai]

Alright #5!!

I've got my c2d running at 2.8, so I'm happy!

 

[petrusbroder]

Congrats, Canai! :sun:

 

[Markfw]

I don;t get how this list was compiled... My box listed has 742.65 RAC right now. Far cry from 485. I also don;t get how the AMD cpus are do high. My 170@2.5 is half the speed on almost everything compared to my C2D's@3.4, but the one is coming up with the same number as my 6300@3.4 ! My Opteron 248's are less than half.

 

[jpeyton]

Originally posted by: Markfw900
I don;t get how this list was compiled... My box listed has 742.65 RAC right now. Far cry from 485. I also don;t get how the AMD cpus are do high. My 170@2.5 is half the speed on almost everything compared to my C2D's@3.4, but the one is coming up with the same number as my 6300@3.4 ! My Opteron 248's are less than half.

I'm a little curious why your Opteron 248s have almost double the benchmarked integer speed as my Opteron 2214s.

I don't think benchmark speeds correlate directly to WU speeds, but it's a curious anomaly since they are running at the same core clock speed. Does the fact that I'm running 64-bit windows have anything to do with it?

 

[Smoke]

Originally posted by: Markfw900
I don;t get how this list was compiled... My box listed has 742.65 RAC right now. Far cry from 485. I also don;t get how the AMD cpus are do high. My 170@2.5 is half the speed on almost everything compared to my C2D's@3.4, but the one is coming up with the same number as my 6300@3.4 ! My Opteron 248's are less than half.

Hello Mark

The list was compiled on 3 February 2007 as was so stated in the first line. At that time your fastest system clocked in at the listed "485" RAC mark. The actual method of calculating the RAC was referenced earlier in this thread. Though not the actual calculation methodology I have found that if you think of BOINC as using a divisor of 30 (days) you will come pretty close to the RAC number. In my rule of thumb way of looking at this your system's indicated RAC will always be below its potential high number until you have a full 30 day history. For the report I only use the official numbers listed by BOINC on each user's stats page. I have stated that I intend to recompile this list once a week. This will be the page I look at for your stats on the next update (only 2 days away).

Your system will definitely climb quickly up the list. :beer:

 

[Markfw]

Originally posted by: jpeyton

Originally posted by: Markfw900
I don;t get how this list was compiled... My box listed has 742.65 RAC right now. Far cry from 485. I also don;t get how the AMD cpus are do high. My 170@2.5 is half the speed on almost everything compared to my C2D's@3.4, but the one is coming up with the same number as my 6300@3.4 ! My Opteron 248's are less than half.

I'm a little curious why your Opteron 248s have almost double the benchmarked integer speed as my Opteron 2214s.

I don't think benchmark speeds correlate directly to WU speeds, but it's a curious anomaly since they are running at the same core clock speed. Does the fact that I'm running 64-bit windows have anything to do with it?

Do you have 2 cpu's ? and 4 sticks of memory ? Numa aware ? 64 bit windows also may be a factor. They still stink compared to my C2D's@3.4

 

[jpeyton]

Originally posted by: Markfw900

Originally posted by: jpeyton

Originally posted by: Markfw900
I don;t get how this list was compiled... My box listed has 742.65 RAC right now. Far cry from 485. I also don;t get how the AMD cpus are do high. My 170@2.5 is half the speed on almost everything compared to my C2D's@3.4, but the one is coming up with the same number as my 6300@3.4 ! My Opteron 248's are less than half.

I'm a little curious why your Opteron 248s have almost double the benchmarked integer speed as my Opteron 2214s.

I don't think benchmark speeds correlate directly to WU speeds, but it's a curious anomaly since they are running at the same core clock speed. Does the fact that I'm running 64-bit windows have anything to do with it?

Do you have 2 cpu's ? and 4 sticks of memory ? Numa aware ? 64 bit windows also may be a factor. They still stink compared to my C2D's@3.4

Yes, 2 CPUs, 8 DIMMs (4 per CPU). How do I find out if it is NUMA aware?

EDIT: NM, SS Sandra says "Yes" to NUMA.

 

[drkicker]

I need to upgrade my CPU. My Athlon 64 3200+ has an RAC of only 262.20. Maybe the Opty 170's will get cheap enough that I can afford one soon.

 

[Assimilator1]

How about overclocking your 3200 in the meantime? .... I've just seen in your sig link that you have it at 2.2GHz ,depending upon the core many 3200s hit 2.4-2.6+GHz