Message boards :
Number crunching :
Are there any sites providing optimized clients? -- PART II
Message board moderation
Previous · 1 . . . 9 · 10 · 11 · 12 · 13 · 14 · 15 . . . 19 · Next
Author | Message |
---|---|
Crunch3r Send message Joined: 15 Apr 99 Posts: 1546 Credit: 3,438,823 RAC: 0 |
Don't know of what troubles your takling about reagarding a staitic build. Hmmm... The one i did from the std. source is a fully static app. ( as previous releases where dynamic builds .. the linux ones where if i rmember corectly S@H enhanced 5.05 code basis). Another thing i've notticed ... read that there are some complaning that the stock source won't compile on gcc 4.X... don't know what your'all doing but it works the way its supposed too...( Doesent' matter if gcc 4.1 or 4.2 is used) P.S. Why do you link MKL ? It's useless and slower than fftw 3.1.1... TIP remove all references to fftw3 ... it's useless and on 90 precent of the tasks slower than ipp. ( in some cases as you've noticed fftw3 is faster than ipp(stock app) but that's a case you've all have to live with... in 95% using the std. angle of 0.42XX the ipp app is way faster than the stock app. and that's the case that matters... ;) Join BOINC United now! |
KWSN - Chicken of Angnor Send message Joined: 9 Jul 99 Posts: 1199 Credit: 6,615,780 RAC: 0 |
Thanks for the hints :) I'll take your word for MKL being slower than fftw, since I really have no basis for comparison. I have an indecent question: is there any chance you would publicize your LDFLAGS/CFLAGS? Everyone doing optimized builds could benefit. Like I said, an indecent question :o) We can hope, anyway... Also, did you use ICC or GCC? I've been using ICC 9.0 (IPP 5.0 of course as well - did you use 9.1/5.1?). I have so far not tested any builds with GCC and IPP, but that's in my plans. Not needing ICC or MKL would drastically cut the required license costs, I could afford the license for just IPP on my own (for Windows, since Linux is doable with the free non-com license). Regards, Simon. Donate to SETI@Home via PayPal! Optimized SETI@Home apps + Information |
Crunch3r Send message Joined: 15 Apr 99 Posts: 1546 Credit: 3,438,823 RAC: 0 |
Thanks for the hints :) You can count on that.
Short answer .. NO... due to some reasons in the past..
Only used ICC...
ICC or ICL on Windows is required to build the app. MKL is useless for S@H... TIP Read that someone claimed that i did/ could not afored buying icc & ipp license for a year... I find it hard to believe that Crunch3r was paying $900 US every year just so he could make BOINC and s@h optimizations! Well there's a half year left until the license expierces... But anyhow why do people care on what i spend MY MONEY ??? Do you know anything of the total costs to build the previous apps. the none enhanced apps. that you're all enjoyed before and downloaded for FREE ??? You really want to know how much and who did a donation to releave the costs from my own ??? Join BOINC United now! |
KWSN - Chicken of Angnor Send message Joined: 9 Jul 99 Posts: 1199 Credit: 6,615,780 RAC: 0 |
TIP Exactly - someone asked me the same, and that's what I told them. Nobody's business but your own how you spend your money... Since I've been doing work on this, I can imagine how much time in total you must have spent on all the optimized apps you made. Massive doesn't begin to describe it. Nobody here could afford that kind of money if they'd have to pay you for it... I can understand your feelings about not releasing anything, I didn't have high hopes there anyway ;o) So I guess it'll have to be ICC and IPP for Windows. Still better than MKL as well. Thanks again! Simon. Donate to SETI@Home via PayPal! Optimized SETI@Home apps + Information |
Hans Dorn Send message Joined: 3 Apr 99 Posts: 2262 Credit: 26,448,570 RAC: 0 |
Hi Chicken, Hi Crunch3r, sorry for cutting in. Just wanted to say Hi :o) Regards Hans |
Crunch3r Send message Joined: 15 Apr 99 Posts: 1546 Credit: 3,438,823 RAC: 0 |
Hi Chicken, Here's a "Hi" from my side... Greetings my fried Hans ;) Join BOINC United now! |
Hans Dorn Send message Joined: 3 Apr 99 Posts: 2262 Credit: 26,448,570 RAC: 0 |
Hi Chicken, I'm still mucking around with the seti source, in case you're wondering. My latest attempts resulted in a remarkably slow app, but you'll know this goes :o). I'll drop you a line if something comes from this. Regards Hans P.S: Not fried yet, but I didn't release any binaries 'till now :o) |
Crunch3r Send message Joined: 15 Apr 99 Posts: 1546 Credit: 3,438,823 RAC: 0 |
Hi Chicken, Hey you know you're not "fried"... ( should have been friend !!!) In German: Brasilien is raus und WIR WERDEN WELTMEISTER !!! Hatte ein paar bier auf den Sieg der Frazosen (hab's den richtig gegönnt!!!) Naja und da kann so ein "fritierter" Hans schon mal passieren .. prost !!! Join BOINC United now! |
Hans Dorn Send message Joined: 3 Apr 99 Posts: 2262 Credit: 26,448,570 RAC: 0 |
Schau'n mer mal :o)
Prost! |
KWSN - Chicken of Angnor Send message Joined: 9 Jul 99 Posts: 1199 Credit: 6,615,780 RAC: 0 |
P.S: Not fried yet, but I didn't release any binaries 'till now :o)
Well, I am, well and truly :o) Fried chicken...yum.
Watn Krimi das Spiel und Elferschießen :o) Sportfreunde Stiller sollten ihr Lied doch nicht umsonst singen...Go Team Germany! (und das als Ösi ;o) ) Wie ihr schon sagtet: Prost! und schau'n mer mal. Viel Glück ;o) Donate to SETI@Home via PayPal! Optimized SETI@Home apps + Information |
Crunch3r Send message Joined: 15 Apr 99 Posts: 1546 Credit: 3,438,823 RAC: 0 |
Och Nöööö.... Kein Bitburger !!! Bei mir ist es Krombacher ! und dazu ein "fried" Hans :D :D :D @chicken I can remember the sentence "chicken good" ;) P.S. Germany - Italy 2:1 ... that's my vote ;) Join BOINC United now! |
Pepperammi Send message Joined: 3 Apr 99 Posts: 200 Credit: 737,775 RAC: 0 |
I was still trying to figure out how to use MKL because of this; From Intel Site; I mistook this to mean it would be faster but you'd know better. Great to see you again Crunch3r. Be interesting to give your tips a try myself. Somewhat funny seeing how many critical errors a can get this time :) |
mikey Send message Joined: 17 Dec 99 Posts: 4215 Credit: 3,474,603 RAC: 0 |
Great to see you again Crunch3r. Be interesting to give your tips a try myself. Somewhat funny seeing how many critical errors a can get this time :) It IS nice seeing Crunch3r on the Seti boards again!!! I think we are ALL just beginning to realise how much time and effort he put into the "optimised" apps of his! I am personally sorry about the way it went down, and think the whole problem could have been solved by Berkeley(Eric) not being on vacation at the time this came down and then the later on/subsequent post of Eric's saying "optimised" apps are okay by Berkeley! If Berkeley had done this earlier on, maybe even from day 1, the whole situation would not be where it is today, IMO! |
Domestos Send message Joined: 31 Dec 00 Posts: 79 Credit: 7,422,705 RAC: 0 |
Oh yeah, Kick all Seti.USA from the BOINC apps ...as they dont share :/ All optimised apps should be available to all crunchers ...otherwise ...those that hold back apps from the rest of us ...have no place here ... We are supposed to be one big happy family are we not ... ? Remember the old Sci Fi TV that shows ET arriving ...only to be greeted with a barrage of weapons > Sounds familiar ? Seti.USA ... release that optimised proggy to all ...you know it is better than hiding stuff like GWB does... |
Pepperammi Send message Joined: 3 Apr 99 Posts: 200 Credit: 737,775 RAC: 0 |
Only thing i'd say its a bit Anti-competetive isn't it? Its well know that optimized apps can increase the credit (i dont mean because someone programed it that way but because more wus can be done in same amount of time). Teams use to have to rely on attracting new team members and new more powerfull machine running for even longer to get that bit more credit than their rivals. Now it seems have to rely on getting programmers on their side and having a battle of the optimized apps too. I'm not a credit junkie but I can see how that is a bit unfair. Only that team can have the opti app and HAHA to the other teams. Its like two F1 car teams in a race but one's got NO2- theres no competition there. Admittedly if a few teams did start competing on the optimized app front it would boost the optiapps performance but they probly wouldn't release it for everyone. I can kind of understand the what happend to crunch3r reason too. Of course its upto the people who provided the tools and the person(s) who created the app what they want to do with it but you can see my point too. Think seti.USA suits well P.S. @ KWSN Is there a way to modify your benchmark script (brilliant bit'o'kit) to run the same unit twice at the same time? For the HT threading stuff I was babbling on about? thought the only really way to test it would be to put under boinc running conditions. Easyest way would be put into boinc after testing the app but then its difficult to be sure becasue of the WU's bieng wildly different at times. Thanks for your added help by the way :) The How-To itself is brilliant but picking your brain too ;) Well you didn't have to answer all the often silly questions i had so thank you for that too |
KWSN - Chicken of Angnor Send message Joined: 9 Jul 99 Posts: 1199 Credit: 6,615,780 RAC: 0 |
In that regard, I stated before that I did not want to put any advertising or donation links on my site for the work I've done to be able to keep offering the Linux clients for free. Now, to my (admittedly limited) understanding of the legal situation, when I read the license, it specifically states that it applies to the materials and programs created using these materials - hence, the Linux Intel products I used to create Linux optimized apps require me not get any sort of compensation for their creation and usage. Now, the only platform they offer these non-commercial versions for is Linux, currently. So the real problem only exists in the Windows world - here, Intel licenses are not free, neither is Visual Studio (although you could use other IDEs). So, the only part requiring any sort of monetary involvement is and will stay Windows (and maybe OS X on X86, in the future). Therefore, again to my limited understanding, I think it would be legal to accept donations geared towards getting a license for ICC and IPP on WINDOWS ONLY. As long as this is a given, it should be okay from a legal standpoint. Anyone who knows better, please educate me. I have so far held off on anything in this regard as I'd really like to be 100% sure this is okay. What changed my mind for me is the fact that someone asked me whether I would accept a donation towards a compiler license for Windows. Given recent events, SargeD's explanation notwithstanding, I believe there would be enough people that feel everyone should have access to an optimized application for Windows. So - that person motivated me to match his donation. I cannot personally afford to pay all the licenses myself currently - my budget is stretched as it is. However, I can state with satisfaction that totalling his and my donation, we are already more than half the way there! The prices I looked up on the Intel site state that the ICC package is 399 USD, IPP is another 199 dollars. This adds up to 598 dollars. We are already at 350 Euros. This translates to 447 USD at current exchange rates. Do you think there are enough interested people to raise the 151 missing bucks? Personally, I think so. I have not registered a donation button or asked for any donations yet, and as said, I'd like to be sure it's okay before I do so. I plan to send any excess funds over the required amount to Seti@Home. Currently, I'm trying to find out if there's any way to register a separate PayPal account for this purpose and have some sort of publicly accessible means of checking the balance. Anyone know? The only way I've seen so far is that you can give multiple users access to an account and can specify what they may do (like just view balance, which would be what's interesting here). I really don't want to get rich doing this, hey :o) Comments please! Simon. Donate to SETI@Home via PayPal! Optimized SETI@Home apps + Information |
Byron Leigh Hatch @ team Carl Sagan Send message Joined: 5 Jul 99 Posts: 4548 Credit: 35,667,570 RAC: 4 |
ACTUALLY, I believe Crunch3r's copy was donated by another Seti user. http://setiathome.berkeley.edu/forum_thread.php?id=23410#199350 http://setiathome.berkeley.edu/forum_thread.php?id=23410#199350 |
Tye W. Botting Send message Joined: 7 Jan 00 Posts: 11 Credit: 53,631,368 RAC: 0 |
I've edited the files online so the file names are now correct. Any new downloads should be okay. Hi Simon, finally got around to testing the SSE version, but the UPX'd one crashes on my 2500+ Athlon XP (mobile Barton) running Mandrake 2005LE. Am about to test the un-UPX'd one, just thought I'd drop an FYI. Thanks for putting them out there! Kung Fu and Powerlifting |
KWSN - Chicken of Angnor Send message Joined: 9 Jul 99 Posts: 1199 Credit: 6,615,780 RAC: 0 |
Thanks for the feedback. I'm playing around with some hints Crunch3r talked about a few posts previously. If I finally manage to compile a real static executable, there should be no more need for UPX and the executable should be much smaller. Until that's so, I'm going to keep offering both versions. Some people have told me that the UPX-compressed versions show up just as "4" as process name instead of their real file name while the uncompressed versions do not. I guess UPX does have some issues there, which is why I'm looking for a way to make compression unnecessary. I've tested the binary on the exact same CPU (mobile AthlonXP 2500+ Barton) and it worked fine for me (Debian Sarge). So I hope the uncompressed version will work for you :o) Regards, Simon. Donate to SETI@Home via PayPal! Optimized SETI@Home apps + Information |
mikey Send message Joined: 17 Dec 99 Posts: 4215 Credit: 3,474,603 RAC: 0 |
WOW, do you bookmark those things or what? That search must have taken awhile otherwise!!! Good job! |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.