Message boards :
Number crunching :
3nd: Compiling faster Windows client with Intel C++ compiler (and fftw?)
Message board moderation
Author | Message |
---|---|
Sir Ulli Send message Joined: 21 Oct 99 Posts: 2246 Credit: 6,136,250 RAC: 0 |
the green Planet has a client who is also working on Athlon64 with SSE-2 original Mesage in German http://www.planet3dnow.de/vbulletin/showthread.php?t=218401 download http://home.arcor.de/jojo4ubin/seti/ only for Info but be patient this is a test Version, if you would try this, First, make a copy of your Boinc Directory, and then copy the Files app_info.xml, and the setiathome_4.11_windows_jo-athlon64-sse2_1105-1.exe to your Boinc Directory xxx/Boinc/projects/setiathome.berkeley.edu ... and dont forgett, it is a test Version, i will report WU times later when the First Wus are done. Greetings from Germany NRW Ulli S@h Berkeley's Staff Friends Club m7 © |
Ulrich Metzner Send message Joined: 3 Jul 02 Posts: 1256 Credit: 13,565,513 RAC: 13 |
Just want to report: Running seti-p3 on Pentium III M Notebook without a hitch and damn' it's fast :)) Same applies to the seti-plainvanilla on an Athlon TB 1400. @Tetsuji Maverick Rai: You made my day! Thank you very much :) Aloha, Uli |
tekwyzrd Send message Joined: 21 Nov 01 Posts: 767 Credit: 30,009 RAC: 0 |
I installed the p3 optimized seti v 4.11 that has recently become available and was able to get it to resume the work unit started by v 4.09 with no problem. Everything was going fine until I decided to pause the einstein work unit I had running and allow two seti units to run for a while. A few seconds later the unit that had been running terminated with a computation error. Luckily I had a back-up copu of my BOINC directory and only lost an hour of computation time. When the error occurred I shut down BOINC and reverted to the standard version. Can anyone point me to the files that will tell me the cause of the problem? System Description : Operating System: Windows XP Pro Version 5.1 (Build 2600.xpsp2_sp2_gdr.050301-1519 : Service Pack 2) Processors: Dual P3 700 MHz Coppermine (Generation 6 Revision 8 Stepping 3) Package : SECC(2) Slot1 Cartridge Stepping Mask : cB0 Microcode : MU068313 Memory : 256MB Registered ECC PC100 Display : ATI all-in-wonder 7200 AGP Driver version : 6.14.10.6483 BIOS information : 113-73701-101 AIW Samsung/Micron DDR RAID Controller : Promise Technologies FastTrak 66 (PDC20262) Driver: V 1.30 Build 33 (Win 2000 PDC20262 Driver) BIOS version : 1.30 Hard Drives : ATA33 : (1) WDC AC310100B Stripe set : (1) IBM-DJNA-371350 Mirror Set : (2) IBM-DJNA-371350 Nothing travels faster than the speed of light with the possible exception of bad news, which obeys its own special laws. Douglas Adams (1952 - 2001) |
Sterling_Aug Send message Joined: 27 Sep 02 Posts: 54 Credit: 14,105,725 RAC: 0 |
I just started your new BOINC client on my dual 250 Opteron processor system. I have completed 6 WUs in the last 3.6 hours which is about 45% faster than version 4.09 standard client. So far, none of them have verified yet. Great work here and THANKS! |
Tetsuji Maverick Rai Send message Joined: 25 Apr 99 Posts: 518 Credit: 90,863 RAC: 0 |
<blockquote>the green Planet has a client who is also working on Athlon64 with SSE-2 original Mesage in German http://www.planet3dnow.de/vbulletin/showthread.php?t=218401 download http://home.arcor.de/jojo4ubin/seti/ only for Info but be patient this is a test Version, if you would try this, First, make a copy of your Boinc Directory, and then copy the Files app_info.xml, and the setiathome_4.11_windows_jo-athlon64-sse2_1105-1.exe to your Boinc Directory xxx/Boinc/projects/setiathome.berkeley.edu ... and dont forgett, it is a test Version, i will report WU times later when the First Wus are done. Greetings from Germany NRW</blockquote> Nice to hear that!! I've never thought of using /QxW option!! In the optimization window of Visual Studio, there are only K,N,B,P,(and nothing). And I didn't know the difference between N and W options. I still don't know because I cannot understand German :( from green planet (jojo4u's post): QxW: optimiere für P4 inkl. SSE, nicht kompatibel zu älteren, ohne Abfrage but I can understand his guess was correct; he indicated I had used N option and never tried W option. In Intel's document coming with compiler, /QxW and /QxN have the same description: "Intel Pentium4 and compatible Intel Processors" So I'd never thought of trying /QxW option (this is my excuse :) ) But I still had another possibility! And I still don't know the differences. Sorry and thank you!! and at the same time I'm happy to know someone (must be jojo4u in green planet forum) has the license!! But...I'm curious jojo4u here has 0 credit now. Maybe he is using another id. There are so many jojo's here. Luckiest in the world. WMD = Weapon of Mass Distraction. Click this table. |
Spectrum Send message Joined: 14 Jun 99 Posts: 468 Credit: 53,129,336 RAC: 0 |
Hi All Is there a link to download 4.11?, I would like to try it on my Athlon XP3200 running windows |
Tetsuji Maverick Rai Send message Joined: 25 Apr 99 Posts: 518 Credit: 90,863 RAC: 0 |
<blockquote>Hi All Is there a link to download 4.11?, I would like to try it on my Athlon XP3200 running windows</blockquote> See Uill's link [url=http://home.arcor.de/jojo4ubin/seti/ ]http://home.arcor.de/jojo4ubin/seti/ [/url] and http://www.planet3dnow.de/vbulletin/showthread.php?t=218401 but the filename says it's athlon64... Luckiest in the world. WMD = Weapon of Mass Distraction. Click this table. |
Alexander Lazar Send message Joined: 11 Apr 03 Posts: 16 Credit: 448,266 RAC: 0 |
Thank you, Tetsuji! Are you taking PayPal donations? I've switched over my Athlon XP 1800 to the Seti-p3 and it has gone from 4:20 to 3:20 per WU on average. Everything has been very stable for over a day. I've still not switched the P4 3.2HT because I'd like to get the latest opinions on two topics: a) Are the new WUs validating OK? Is there any significant variation in the results being generated? b) At the risk of opening a can of worms, could someone summarize the latest thinking on how credit is claimed in the new scheme - my brain hurt when I tried to understand it the first time. Thanks! Tony Lazar San Carlos, CA |
Spectrum Send message Joined: 14 Jun 99 Posts: 468 Credit: 53,129,336 RAC: 0 |
See Uill's link [url=http://home.arcor.de/jojo4ubin/seti/ ]http://home.arcor.de/jojo4ubin/seti/ [/url] and http://www.planet3dnow.de/vbulletin/showthread.php?t=218401 but the filename says it's athlon64...</blockquote> Thank's Mav, I tried it but when I restarted the manager everything was gone, no seti@home project or work units showing in the manager just einstien! I copied the app_info.xml, and the setiathome_4.11_windows_jo-athlon64-sse2_1105-1.exe to my Boinc Directory xxx/Boinc/projects/setiathome.berkeley.edu and also the work sah file. Also when I reopened the folder they were the only files there, somehow the old client and workunits had been deleted so I just overwrote the whole shabang with the backup and am up on 4.09 again. Any ideas or comments, did I stuff up? |
Mike Send message Joined: 17 Feb 01 Posts: 34263 Credit: 79,922,639 RAC: 80 |
<blockquote>Thank you, Tetsuji! Are you taking PayPal donations? I've switched over my Athlon XP 1800 to the Seti-p3 and it has gone from 4:20 to 3:20 per WU on average. Everything has been very stable for over a day. I've still not switched the P4 3.2HT because I'd like to get the latest opinions on two topics: a) Are the new WUs validating OK? Is there any significant variation in the results being generated? b) At the risk of opening a can of worms, could someone summarize the latest thinking on how credit is claimed in the new scheme - my brain hurt when I tried to understand it the first time. Thanks! Tony Lazar San Carlos, CA </blockquote> Hi I´ve more than 10 results validated ok with TMRs optimized p3 client on my XP 2800+. yesterday. All is running fine. Credits are nearly the same between 28 and 38 cobbles for 9000 secons. greetz from germany Mike With each crime and every kindness we birth our future. |
Spectrum Send message Joined: 14 Jun 99 Posts: 468 Credit: 53,129,336 RAC: 0 |
Hi I´ve more than 10 results validated ok with TMRs optimized p3 client on my XP 2800+. yesterday. All is running fine. Credits are nearly the same between 28 and 38 cobbles for 9000 secons. greetz from germany Mike</blockquote> Hi Mike. Where did you get the client from as I can't get the 64bit one to run on my XP 3200? |
Tetsuji Maverick Rai Send message Joined: 25 Apr 99 Posts: 518 Credit: 90,863 RAC: 0 |
<blockquote>Thank you, Tetsuji! Are you taking PayPal donations? </blockquote> No..please don't send money to me :) I won't buy Intel C++ for Windows. Some others (ex, jojo4u on planet) have the real compiler license and will provide binaries, and I'll also use them. I won't make any further clients because of the evaluation license problem. Since others provide binaries, anyone can use them. I prefer Linux and I compile Linux client with Intel C++ compiler which is free for non-commercial use. That's a major reason I won't continue compiling Windows client. <blockquote> I've switched over my Athlon XP 1800 to the Seti-p3 and it has gone from 4:20 to 3:20 per WU on average. Everything has been very stable for over a day. I've still not switched the P4 3.2HT because I'd like to get the latest opinions on two topics: a) Are the new WUs validating OK? Is there any significant variation in the results being generated? b) At the risk of opening a can of worms, could someone summarize the latest thinking on how credit is claimed in the new scheme - my brain hurt when I tried to understand it the first time. Thanks! Tony Lazar San Carlos, CA </blockquote> First of all if you are going to use the new client on P4, be sure your P4 really supports SSE3 or not, and make a full backup copy of the whole folder, in case you damage/lose your data. a) all my results are fine; valid or pending(initial), sometimes "no concensus yet" but finally turns to valid. So far I have confirmed 20-30 valid results. I've never got "Invalid" for successfully completed WU's for some ten WU's. b) sorry, I don't know much about it....all I can say is....however low credit your client claims, if two others claim high credits, you'll get the middle..yes, higher credit than yours :) My client usually claims somewhere around 15 rarely exceeds 20, but getting 20-40 granted or even 45. Usually I'm the lowest, and others pull up credits for me (my client is to prove the results are valid) :) Happy crunching! I believe I'm luckiest in the world :) Luckiest in the world. WMD = Weapon of Mass Distraction. Click this table. |
genes Send message Joined: 25 May 99 Posts: 117 Credit: 580,187 RAC: 0 |
This program is very useful for identifying what instructions your CPU can handle, and many other things as well... http://www.cpuid.org/index.php |
mikey Send message Joined: 17 Dec 99 Posts: 4215 Credit: 3,474,603 RAC: 0 |
<blockquote>Thank you, Tetsuji! Are you taking PayPal donations? I've switched over my Athlon XP 1800 to the Seti-p3 and it has gone from 4:20 to 3:20 per WU on average. Everything has been very stable for over a day. I've still not switched the P4 3.2HT because I'd like to get the latest opinions on two topics: a) Are the new WUs validating OK? Is there any significant variation in the results being generated? b) At the risk of opening a can of worms, could someone summarize the latest thinking on how credit is claimed in the new scheme - my brain hurt when I tried to understand it the first time. Thanks! Tony Lazar San Carlos, CA </blockquote> I have an AMD64 3200+ and am using Tetsuji's optimized program and it is validating just fine. I went from around 2 hours 20 minutes to 90 minutes, 25 to 30 credits per unit to 20 to 25 per unit. Paul's site gives as good an explanation of how the credits are doneas any I have seen: http://boinc-doc.net/index.php This is is specifically about pending credit: http://boinc-doc.net/site-common/oman-web/pending-credit.php |
jshenry1963 Send message Joined: 17 Nov 04 Posts: 182 Credit: 68,878 RAC: 0 |
I have a question concerning using the optimized seti and boinc. If I understand it correctly, the new seti utilizes less instructions to get the same resultant as the normal seti. And there is also a change in the optimized boinc to match the benchmarks with the optimized seti. How this works, I don't need to understand, just that it gives some sort of credit at the end. But, what I want to know is, will the change (as I understand it) in the boinc computation of the benchmarks affect the other clients credits that I am running? E@H, P@H, CP? As I understand it, benchmarks one of the variables used in computing claimed credits for all projects work units. So if we change how the benchmarks are calculated for the optimized seti, won't that throw off the credits for the other projects? I don't understand all of the credit stuff completely, and I really don't need to, just a simple, yes it will, or no it won't, and maybe the extent of the damage it will cause. (or maybe I don't even need to worry about this at all) Thanks to the pros for getting this working, and for all the help, 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"> |
Mike Send message Joined: 17 Feb 01 Posts: 34263 Credit: 79,922,639 RAC: 80 |
Hi @jshenry1963 Yes it will. @spectrum You can use the version from jojo4 at http://home.arcor.de/jojo4ubin/seti He is actually working on a version for Athlon XP an XP 64 with SSE3. The version that is finnished yet is running fine on XP 64 with SSE2. Its tested by some members of the planet team. But dont forget app_info.xml. greetz from germany Mike With each crime and every kindness we birth our future. |
Spectrum Send message Joined: 14 Jun 99 Posts: 468 Credit: 53,129,336 RAC: 0 |
<blockquote>Hi @jshenry1963 Yes it will. @spectrum You can use the version from jojo4 at http://home.arcor.de/jojo4ubin/seti He is actually working on a version for Athlon XP an XP 64 with SSE3. The version that is finnished yet is running fine on XP 64 with SSE2. Its tested by some members of the planet team. But dont forget app_info.xml. greetz from germany Mike </blockquote> Mike, I have tried both jojo & the P3 version, I think I must be doing something wrong as I lose all my existing work units, and I have downloaded the workunit app info from the previous thread that should allow completion of existing 4.09 workunits but I come up empty, any help or walkthrough of the install of the 4.11 version from anyone reading this would be greatly appreciated. |
jshenry1963 Send message Joined: 17 Nov 04 Posts: 182 Credit: 68,878 RAC: 0 |
Thanks, One more question, and this may not be a good one. The benchmark results for boinc optimized are lower than benchmark results for normal boinc. This means I am now getting less credit for P@H than when I was using the normal boinc. Now, if I switch back to the normal boinc, and use the optimized seti, won't that give me higher numbers for seti? given that I have higher benchmark results wiht the normal boinc? Kinda cheats the system one way or the other dontcha think? 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"> |
Mike Send message Joined: 17 Feb 01 Posts: 34263 Credit: 79,922,639 RAC: 80 |
Hi @jshenry1963 You can use Boinc version you want, there is no need to copy boinc.exe. When you have an better benchmark with the original core client then use it. But the difference is that small maybe 2-5 cobbles per result. @spectrum Did you change the app_info.xml with the one from AndyK ??? You can get it here http://www.andyk.de/app_info.xml This one is needed otherwise you will loose all WUs. First step backup Boinc directory closeing boinc Than copy optimized setihome4.11 exe and the modified app_info.xml in project/seti dir. Copy boinc.exe in parent boinc directory but isn´t necassary. restart boinc thats all. greetz Mike With each crime and every kindness we birth our future. |
Metod, S56RKO Send message Joined: 27 Sep 02 Posts: 309 Credit: 113,221,277 RAC: 9 |
<blockquote> The benchmark results for boinc optimized are lower than benchmark results for normal boinc. This means I am now getting less credit for P@H than when I was using the normal boinc. </blockquote> Just use boinc that gives you the highest benchmark results. As for the credit claimed in other projects: if your benchmarks go say 20% up, your claimed credit for other projects will go 20% up. Claimed credit for seti will go up or (more likely) down, depending on how well the seti is optimized for your CPU. Better optimized seti will end up claiming less credit. Metod ... |
©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.