Message boards :
Number crunching :
2nd: Compiling faster Windows client with Intel C++ compiler (and fftw?)
Message board moderation
Author | Message |
---|---|
Tetsuji Maverick Rai Send message Joined: 25 Apr 99 Posts: 518 Credit: 90,863 RAC: 0 |
|
Pascal, K G Send message Joined: 3 Apr 99 Posts: 2343 Credit: 150,491 RAC: 0 |
Thank you again Tetsuji for all your hard work...... Ok finally got it to work, had to shut down the command line then install 4.27 and it is running 4.09s and has downloaded a 4.11. I am getting this error 5/10/2005 10:55:44 AM|SETI@home|Unparsed line in app_info.xml: 5/10/2005 10:55:44 AM|SETI@home|Unparsed line in app_info.xml: 5/10/2005 10:55:44 AM|SETI@home|Unparsed line in app_info.xml: 5/10/2005 10:55:44 AM|SETI@home|Unparsed line in app_info.xml: 5/10/2005 10:55:44 AM|SETI@home|Unparsed line in app_info.xml: 5/10/2005 10:55:44 AM|SETI@home|Unparsed line in app_info.xml: 5/10/2005 10:55:44 AM|SETI@home|Unparsed line in app_info.xml: and my benchmark did not change to much... it is up about 300 on both...... 5/10/2005 10:56:50 AM||Benchmark results: 5/10/2005 10:56:50 AM|| Number of CPUs: 2 5/10/2005 10:56:50 AM|| 1569 double precision MIPS (Whetstone) per CPU 5/10/2005 10:56:50 AM|| 1444 integer MIPS (Dhrystone) per CPU 5/10/2005 10:56:50 AM||Finished CPU benchmarks Semper Eadem So long Paul, it has been a hell of a ride. Park your ego's, fire up the computers, Science YES, Credits No. |
MikeSW17 Send message Joined: 3 Apr 99 Posts: 1603 Credit: 2,700,523 RAC: 0 |
Excellent idea (the new thread). How about starting off with a summary of the versions you've (wondefully) compiled and what processors they are supposed/confirmed to run-on? I'm getting really lost in the old thread trying to trace what is what. To date, I have the PentiumM version running on my laptop; PlainVanilla on my XP6 3000+ and trying to work out what whould go on the XP 2400. Thanks a million for the effort, it's much appreciated. |
AndyK Send message Joined: 3 Apr 99 Posts: 280 Credit: 305,079 RAC: 0 |
Just to summarize: <blockquote>I added another more reliable and faster server... 1. P3 seti-p3.zip (for AMD XP and 64 use this one!) 2. P4 w/o sse3 seti-p4-no-sse3.zip 3. P4 w sse3 seti-p4-sse3.zip 4. And reference work unit and result unit are in a zip referenceunits.zip The usage (benchmark and validation) of reference work unit is in the lower part of Ned's page for gcc/linux and this is the validation limits. In my personal experience with Linux Intel compiler/fftw, I've never got my results invalidated with them (few weeks.) generic SSE2</blockquote> To crunch the "old" workunits from 4.09 with the new optimized 4.11 replace the app_info.xml file in the BOINC\projects\setiathome.berkeley.edu\ folder with this one: app_info.xml (right click and "save link to...") Otherwise you will lose all already downloaded workunits!! Andy PS:<blockquote>How about starting off with a summary of the versions you've (wondefully) compiled and what processors they are supposed/confirmed to run-on?</blockquote> :-))) just done that, while you were posting. Want to know your pending credit? The biggest bug is sitting 10 inch in front of the screen. |
rsisto Send message Joined: 30 Jul 03 Posts: 135 Credit: 729,936 RAC: 0 |
>I am getting this error >5/10/2005 10:55:44 AM|SETI@home|Unparsed line in app_info.xml: Look at AndyK post in the previous thread. You can download the corrected app_info.xml file from here: app_info.xml (right click and "save link to...")<br /> |
Tetsuji Maverick Rai Send message Joined: 25 Apr 99 Posts: 518 Credit: 90,863 RAC: 0 |
summarizing......yes, it will be necessary, but I'm reluctant right now.....but okay, I'll do it soon. But I need more data on processing time of reference workunit. But even if I do it, is this continueing? My license will expire on May 29.... Luckiest in the world. WMD = Weapon of Mass Distraction. Click this table. |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
Hi @Tetsuji Is it possible to compile only boinc.exe from 4.30. As i figured out that the boinc.exe is the problem of no manual update in Boinc 4.38 because i copied it a bit later. Would be very nice. great work @all my WU time going down from 3.08 houres to 2.35 with my Athlon XP 2800+. greetz Mike With each crime and every kindness we birth our future. |
Tetsuji Maverick Rai Send message Joined: 25 Apr 99 Posts: 518 Credit: 90,863 RAC: 0 |
<blockquote>Hi @Tetsuji Is it possible to compile only boinc.exe from 4.30. As i figured out that the boinc.exe is the problem of no manual update in Boinc 4.38 because i copied it a bit later. Would be very nice. great work @all my WU time going down from 3.08 houres to 2.35 with my Athlon XP 2800+. greetz Mike </blockquote> It's possible. Actually I built a core client on Linux based on May-8 nightly build, but it has a problem in downloading new workunits. It claims it cannot download new WU's so many times, and at the same time it doesn't upload processed result files, while 4.27 can do them very quickly. This problem is discussed in other threads, for ex this thread. So I went back to Apr-12 build again. It works perfectly. The problem is I don't know which nightly build I should use. Will you specify which nightly build works fine? Then I can build it. Luckiest in the world. WMD = Weapon of Mass Distraction. Click this table. |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
HI So far i can say, i tested every Boinc release since available on Windows machine for me was the best one 4.30. Stable and solid. greetz Mike Boincing since March 2003 With each crime and every kindness we birth our future. |
Tetsuji Maverick Rai Send message Joined: 25 Apr 99 Posts: 518 Credit: 90,863 RAC: 0 |
<blockquote>HI So far i can say, i tested every Boinc release since available on Windows machine for me was the best one 4.30. Stable and solid. greetz Mike Boincing since March 2003</blockquote> okay..but which nightly build is it? Luckiest in the world. WMD = Weapon of Mass Distraction. Click this table. |
MikeSW17 Send message Joined: 3 Apr 99 Posts: 1603 Credit: 2,700,523 RAC: 0 |
<blockquote><blockquote>Hi @Tetsuji Is it possible to compile only boinc.exe from 4.30. As i figured out that the boinc.exe is the problem of no manual update in Boinc 4.38 because i copied it a bit later. Would be very nice. great work @all my WU time going down from 3.08 houres to 2.35 with my Athlon XP 2800+. greetz Mike </blockquote> It's possible. Actually I built a core client on Linux based on May-8 nightly build, but it has a problem in downloading new workunits. It claims it cannot download new WU's so many times, and at the same time it doesn't upload processed result files, while 4.27 can do them very quickly. This problem is discussed in other threads, for ex this thread. So I went back to Apr-12 build again. It works perfectly. The problem is I don't know which nightly build I should use. </blockquote> Is this an issue? As far as I can tell, you only need to change the seti....exe in BOINC~projects~setiathome.berkeley.edu. You dont seem to need to change the BOINC.EXE? All(?) the benefit of optimization comes from the seti app. Seems the app works with varsions of 4.2x and 4.3x jsu fine. |
Tetsuji Maverick Rai Send message Joined: 25 Apr 99 Posts: 518 Credit: 90,863 RAC: 0 |
<blockquote> Is this an issue? As far as I can tell, you only need to change the seti....exe in BOINC~projects~setiathome.berkeley.edu. You dont seem to need to change the BOINC.EXE? All(?) the benefit of optimization comes from the seti app. Seems the app works with varsions of 4.2x and 4.3x jsu fine. </blockquote> right....I suspect it shoudn't. Luckiest in the world. WMD = Weapon of Mass Distraction. Click this table. |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
<blockquote><blockquote>HI So far i can say, i tested every Boinc release since available on Windows machine for me was the best one 4.30. Stable and solid. greetz Mike Boincing since March 2003</blockquote> okay..but which nightly build is it?</blockquote> Hi 12 April. @Mikesw17 You can manually update with 4.38 greetz and respectful from germany Mike With each crime and every kindness we birth our future. |
Tetsuji Maverick Rai Send message Joined: 25 Apr 99 Posts: 518 Credit: 90,863 RAC: 0 |
<blockquote> Hi 12 April. @Mikesw17 You can manually update with 4.38 greetz and respectful from germany Mike </blockquote> 12 apr is the one I used; 4.27 :) But it's troublesome to build another client with specific options. Sorry for that, I cannot afford time. Luckiest in the world. WMD = Weapon of Mass Distraction. Click this table. |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
Hi No probs, so i will use original 4.30 boinc.exe. Only get a bit less credits. greetz Mike With each crime and every kindness we birth our future. |
jshenry1963 Send message Joined: 17 Nov 04 Posts: 182 Credit: 68,878 RAC: 0 |
Hey All, Super Congrats and Thanks to Tetsuji, wonderful job at proving this could be done. Now, for permanency, is it possible to incorporate this type of building and downloading of the exe into the regular updates? e.g. someone from seti build an offiical one that is downloaded when new seti apps are updated/downloaded? It would seem a waste to have TMR having built this, Then seti comes out with a new version, and TMRs license expiered, and we go through this all over again. IMHO, if this can be done, it should be incoproated by whomever holds the master seti app, and options of the types to download should be asked, (normal, amd optimized, p3 optimized, pm optimized, p4 sse2 optimized, p4 sse3 optimized...), get my drift? Then ALL would benefit, if they so desire. If it does not make it into the mainstream, then who will maintain it after TMR's license expires? I can already see new variants coming up, and TMR doesn't have time to maintain this for the world. Only did it to see if it could be done, and for learning. maybe another "Maverick" will speak up. Thanks, and Keep on crunchin' John Henry KI4JPL Sevierville TN I started with nothing, and I still have some of it left. <img src="http://www.boincstats.com/stats/banner.php?cpid=989478996ebd8eadba8f0809051cdde2"> |
Heaphus Send message Joined: 1 Apr 03 Posts: 96 Credit: 4,148,549 RAC: 0 |
<blockquote>Is this an issue? As far as I can tell, you only need to change the seti....exe in BOINC~projects~setiathome.berkeley.edu. You dont seem to need to change the BOINC.EXE? All(?) the benefit of optimization comes from the seti app. Seems the app works with varsions of 4.2x and 4.3x jsu fine. </blockquote> This appears to be true. The P3 optimized seti app completes wu in the same amount of time on my Athlon 2700+, no matter which boinc client I have used it with. Standard 4.27, optimized 4.27, and even standard 4.19 are all the same. Actually, I guess using 4.19 is the best, since it completes in the same amount of time and claims a little more credit, because the benchmarks are higher. |
Tetsuji Maverick Rai Send message Joined: 25 Apr 99 Posts: 518 Credit: 90,863 RAC: 0 |
<blockquote>Hi No probs, so i will use original 4.30 boinc.exe. Only get a bit less credits. greetz Mike </blockquote> I hope even if your client claims lower credits, other crunchers claim higher credits, and the middle of first three claimed credits will be granted. Hence your claimed credits will be the least one and won't matter in most of the cases. :) That's how the average credit system of boinc works. So you can use the official core client w/o lower credits granted if you want another version. Even if it affects the granted credits, the difference is very small. But I'm not sure. But for example, look at my results here Even my "optimized" client claims the lowest credits in most cases and I get higher credits than I claim :) I call it "nice cheating." (but all of my results have been valid.) Isn't it funny? regards, Luckiest in the world. WMD = Weapon of Mass Distraction. Click this table. |
-= Vyper =- Send message Joined: 5 Sep 99 Posts: 1652 Credit: 1,065,191,981 RAC: 2,537 |
I just need to poke an x86-64 again into folks eyes so i hope someone reads this who has the urge to compile one for Windows x64 64bit version. I'm all go to try and if that person want me to try it out i'm eager to be a testpilot.. Mail me at laracroft@telia.com if that is required.. Kind regards and for you Tetsui.. A really big thanks and a big hug for that you were the first to bring this out in the open.. I've been waiting for an optimized Windows version for half a year now. Keep up the good spirit and keep it crunching.. //Vyper BTW .. I have an overclocked A64 3200+ running at 2500Mhz. The original client returns a WU in about 6000 - 7200 Secs. With "seti-p3" which is the only version that runs on my A64 it completes WU/s in about 4400 - 4800 Secs.. About 33% more effective and my temperature rise from 49 degrees to 51 degrees.. _________________________________________________________________________ Addicted to SETI crunching! Founder of GPU Users Group |
Ned Slider Send message Joined: 12 Oct 01 Posts: 668 Credit: 4,375,315 RAC: 0 |
<blockquote>I just need to poke an x86-64 again into folks eyes so i hope someone reads this who has the urge to compile one for Windows x64 64bit version. </blockquote> Have patience - it's being worked on. Windows X64 was only released the other day. Ned *** My Guide to Compiling Optimised BOINC and SETI Clients *** *** Download Optimised BOINC and SETI Clients for Linux Here *** |
©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.