Dpad

[glennpat]

Do you guys download new sample results files every so often from http://stephenbrooks.org/muon1/ ? Helps optimize the results better if you download both curent best sample results and combine them into a new 'results.dat' in teh DPAD folder. I use a little batch script to combine them instead of messing around with notepad.

Thanks riptide. I just gave it a try on one of my computers. Also thanks for an example of a bat file to help me learn some new stuff.

Glenn

[glennpat]

'Cept that script requires that you dump your own results.dat. Thought I read that's not a good idea.

I just saw the above after I did my last post. From the website it says:

How do I use the sample file? Decompress it if necessary and either use it to replace your results.dat, or edit your results.dat in WordPad and append the new results (being careful not to merge lines). The latter option will mean Muon1 can use your old results as well as the downloaded ones, so is a better choice.

Awhile back I had done the append method manually and I think I messed up one of my dat files so I stopped doing it. Looks like the append method is suppose to be better though.

[[XC] riptide]

'Cept that script requires that you dump your own results.dat. Thought I read that's not a good idea.

Yes... but... although Stephens site says the best way is the append way he would still acknowledge that its also useful to get results from the best samples aswell.

Remmeber that his site was written up back in the day when they were running diff simulations. Not everything may be a 100% applicable now.

ANd.... it takes a small amount of time at the start of each simulation to parse and load up teh results file .dat file. How big is yours? Actually I'm not entirely sure it loads the results file at each simulation OR at the start of the muon.exe run...

Either way... fell free to change the script to just append the new files.

[[XC] riptide]

Keeping it real!

[Paladin]

ANd.... it takes a small amount of time at the start of each simulation to parse and load up teh results file .dat file. How big is yours? Actually I'm not entirely sure it loads the results file at each simulation OR at the start of the muon.exe run...

I see your point there. I checked my results.dat and it's 43MB which offsets the value of keeping my own machine's history in favor of using the published global ones at 281KB.

[glennpat]

KaptainBlaZzed hits 100,000,000 on DPAD

[LP_]

When you're running DPAD on a machine that doesn't have internet access, when you copy the results, do you need to copy both results.txt and results.dat to send your results in?

Nevermind, I think the answer is no results.dat. Got confused for second.
Stupid tower won't connect to the net.

[KaptainBlaZzed]

KaptainBlaZzed hits 100,000,000 on DPAD

Thanks,

just got back from vacation and got this little surprise.

[[XC] riptide]

When you're running DPAD on a machine that doesn't have internet access, when you copy the results, do you need to copy both results.txt and results.dat to send your results in?

Nevermind, I think the answer is no results.dat. Got confused for second.
Stupid tower won't connect to the net.

Yep. You got it. Results.txt.

@KB. Well done Troy.

[[XC] riptide]

I see your point there. I checked my results.dat and it's 43MB which offsets the value of keeping my own machine's history in favor of using the published global ones at 281KB.

Also... depending on when you started up that particular DPAD folder, it could contain many, now useless, results form previous simulation designs.

[[XC] riptide]

You know what annoys me about this? Is when you have a good simulation and the result is quarantined and re run 5 or more times. So the average of the middle 3 runs are the results that is submitted. BUT. You only get the points for that ONE run. Whats compounds it is that most of the time the best runs are 5xxx Mpts, as opposed to the majorith that are around 1xxx Mpts. So you run 25XXX worth of simulation, only to be awarded 5XXX Mpts.

I started a thread on teh forum to ask about this. Its far from equitable.

[KaptainBlaZzed]

You know what annoys me about this? Is when you have a good simulation and the result is quarantined and re run 5 or more times. So the average of the middle 3 runs are the results that is submitted. BUT. You only get the points for that ONE run. Whats compounds it is that most of the time the best runs are 5xxx Mpts, as opposed to the majorith that are around 1xxx Mpts. So you run 25XXX worth of simulation, only to be awarded 5XXX Mpts.

I started a thread on teh forum to ask about this. Its far from equitable.

i see your point, but if it affects everybody it is not a problem.

we would complete a lot more work if we did not have to run each simulation 5 times.

[[XC] riptide]

i see your point, but if it affects everybody it is not a problem.

we would complete a lot more work if we did not have to run each simulation 5 times.

Ok. Its sorted. The 5XXX runs are actually 5 times ~1XXX runs. Thread was answered on teh forums. All good. Move along, nothing to see here.

[[XC] riptide]

Norways challenge has collapsed. Well done team.

[glennpat]

Norways challenge has collapsed. Well done team.

We did good. Over the past month we had 7 teammates active on here.

Yesterday I added a bunch more cores on here. I now have 36 cores on it. I was looking for a way to have the most fun. The new ones won't be returning results till later Thursday.

[KaptainBlaZzed]

We did good. Over the past month we had 7 teammates active on here.

Yesterday I added a bunch more cores on here. I now have 36 cores on it. I was looking for a way to have the most fun. The new ones won't be returning results till later Thursday.

Good to hear.

I will be leaving my machines on DPAD for a while.

[glennpat]

Only 4 of the cores that I put on Tuesday have sent in results and one of those was because I did a manual send. Looks like today is the big day for the most of the other cores to have enough built up to send in. I think they each do around 20,000 points per day so they have a lot stored up.

Our team already has 841,258 for today. Looks like it is going to be a super good day .

[glennpat]

I am down to 2 cores on this until we gain some ground on the 2 projects that were added to the vault.

I have noticed that on DPAD for the same clock speed that the AMD Athlon X2 (Brisbane) is more than 10% faster then a Q6600. Most projects I see the Intel a lot faster. Last week I upgraded a Brisbane running at 2.6G to a 5050e Brisbane at 3.1G. That is the one I left on DPAD.

[[XC] riptide]

I am down to 2 cores on this until we gain some ground on the 2 projects that were added to the vault.

I have noticed that on DPAD for the same clock speed that the AMD Athlon X2 (Brisbane) is more than 10% faster then a Q6600. Most projects I see the Intel a lot faster. Last week I upgraded a Brisbane running at 2.6G to a 5050e Brisbane at 3.1G. That is the one I left on DPAD.

Memory Controller. If you look at the benchmarking thread on the Muon forum you will see AMD's more effiecient per clock. However.... i7 has destroyed this.... again... memory controller.

[glennpat]

Memory Controller. If you look at the benchmarking thread on the Muon forum you will see AMD's more effiecient per clock. However.... i7 has destroyed this.... again... memory controller.

Thanks for the info. I see your 3rd in a benchmarking chart from awhile back. Nice going!

[[XC] riptide]

Thanks for the info. I see your 3rd in a benchmarking chart from awhile back. Nice going!

Ha! Ya... thast cos Stephen hasn't update the chart recently.

Whats more however is that Phenom II is now killing teh Q6600's at same clock to the tune of >20%!. Of course i7 stays ahead... but without HT it would be a close run thing in DPAD.