List of Optimized BOINC Clients

[gamer007]

General Info + Links:
http://boincdenmark.dk/optimize_en.html

BOINC Optimized Clients:
Harold Naparst (http://naparst.name/)
Ned Slider (http://www.pperry.f2s.com/)
Tetsuji Maverick Rai (http://www.marisan.nl/seti/)
http://www.mkx.si/boinc/
TruXoft (http://boinc.truxoft.com/)
For Linux (http://www.godefroy.t.freesurf.fr/seti/)

Einstein@Home Optimized Clients:
Akosf's (http://einstein.phys.uwm.edu/forum_thread.php?id=3898)
Akosf's - Mirror site (http://einstein.terrorfront.info/en/index.htm)

SZTAKI Desktop Grid Opt. Clients:
http://szdg.lpds.sztaki.hu/szdg/forum_thread.php?id=348

SIMAP Opt. Clients:
http://boinc.bio.wzw.tum.de/boincsimap/forum/viewtopic.php?t=96

I'm currently using Crunch3r's. My "Measured integer speed" more than doubled after switching. And I've gained about 10-20 more credits per Rosetta WUs. :)

[NeoGen]

It was Trux's optimized clients. Mitro's the one who told us about it. :)
http://boinc.truxoft.com/

[gamer007]

Hehe. Thanks NG. :)

[Empty_5oul]

so with these, what do they do? and whats best (for a 3200+ running XPpro)?

[daddygeek]

The 3200+ came out on both the 64-bit and the regular Athlon XP
Athlon 64 , use http://www.bm-makler.de/setiathome/boinc/pc/Boinc_5.2.13_P4_Athlon64_SSE2.zip

Athlon XP ,use http://www.bm-makler.de/setiathome/boinc/pc/Boinc_5.2.13_P3_AthlonXP_SSE.zip


I finally just upgraded from optimized 4.x to Cruch3r's 5.2.13 and I've seen quite an improvement over the old 4.x

[Empty_5oul]

ok thanks.
lucky i didnt specify which 3200+ -- as i have both :P i will put them on later. cheers.

[torch]

what is optimized with an optimized client? is it just the credit, or ist it really more work in the same time?
i mean is ist cheating or something like that? :)

[NeoGen]

No, don't worry, it's not cheating. :)

The optimized client is just the same boinc client but specially compiled to take advantage of advanced processor features and instructions, making it work faster and more efficiently.
But we know that the client itself doesn't do any work, so why optimizing it? Because the only time the client actually does hard work is when it is benchmarking our computer. And with an optimized client the benchmark results of our computer are much higher because our computer does more. As the credits claimed values are tied to the benchmark values of our computer, a higher benchmark value means higher credit claimings.
And if you claim more, chances are that you'll get more. (in most cases) In Rosetta it seems that what you claim is what you get, so you see now why we try to claim as much as possible? :)

[Lagu]

Hello

If I want to upgrade Boinc how shall I do? Will I remowe the old client and do an clean install?

Lagu :)

[NeoGen]

If you want to upgrade boinc to a more recent version, you just install it over the previous and it will keep all your projects, workunits and settings.
Even if you're upgrading from v4 to v5 all projects and settings will remain.

[Nflight]

I currently have the version 5.2.7 There is a newer one out there and until now I have been able to successfully improve on my stats just about everyday. Yesterday was 1103 points improvement on Predictor Scripps.
With this insightful forum subject I now know how to install over and save all pertinent information and improve on my improving status.
Oh yeah and my Current Average credit is now 764. I must be doing something right! Check on this link to see how!
http://www.amdusers.com/modules.php?name=Forums&file=viewtopic&t=1780#23102

[Strongbow]

Does anyone know if there is an optimized BOINC client for Opteron? Linux or Windows I don't mind!

I'm currently using Crunch3r's Boinc 5.2.13 Windows SSE2 and I get the following for each Opteron 248 so I am sure I can do better than this;

3003 double precision MIPS (Whetstone) per CPU
9140 integer MIPS (Dhrystone) per CPU

[drezha]

Crunch3r does a Boinc 5.3.6 64 bit SSE2 for linux.

BUT, you gotta have a 64 bit linux OS and it doesn't give any higher benchmarks really (Not when I tried it) for two reasons...one, BOINC on linux is unfairly weighted towards windows. (Windows gets higher benchmarks than a linux machine even if exactly the same) and it only enables it to run at 64 bit.

Oh and most projects wont recognise it as an OS at all and so it wont get work. I know Rosetta, SIMAP, QMC and Predictor are all the same. I tried it the other day.

You get
3/10/2006 6:23:23 AM|rosetta@home|Message from server: platform 'x86_64-unknown-linux-gnu' not found

and those benchmarks seeem fine to me :) I'm running a stock A64 3500+ and my benchmarks are
Measured floating point speed 2935.6 million ops/sec
Measured integer speed 9071.06 million ops/sec

[NeoGen]

In lack of a specific one for opteron, I guess one optimized with SSE2 should do very well too.

[gamer007]

I've reverted back to BOINC v5.2.13 from GridRepublic.

CPU: AMD Athlon XP 1700+ 1.46GHz

Benchmarks before Cruch3r's optimized:

11/03/2006 11:18:30 PM||Benchmark results:
11/03/2006 11:18:30 PM|| Number of CPUs: 1
11/03/2006 11:18:30 PM|| 1088 double precision MIPS (Whetstone) per CPU
11/03/2006 11:18:30 PM|| 1779 integer MIPS (Dhrystone) per CPU


After:

11/03/2006 11:22:27 PM||Benchmark results:
11/03/2006 11:22:27 PM|| Number of CPUs: 1
11/03/2006 11:22:27 PM|| 1205 double precision MIPS (Whetstone) per CPU
11/03/2006 11:22:27 PM|| 3335 integer MIPS (Dhrystone) per CPU

[drezha]

Was gonna say that looked very much like a linux box but then noticed it was only 1.46Ghz.

Bit of a shame that Linux users seem to be penalised for using Linux to crunch BOINC :mad: (Even though I only play with linux, not seriously use it :P )

[NeoGen]

That I don't understand either.

I don't even play with Linux, only try to figure it out sometimes on a Virtual Machine. But my idea is that any version of Linux is less resource demanding and more efficient than Windows. How come Boinc doesn't gain from it?

[drezha]

I believe that it's not that it's not gaining, it's just that BOINC is optmized for windows by default.

Therefore it gives Windows higher benchmarks.
Downloading the crunch3r client nullifies this and I believe you get fairly simliar benchmarks to a windows machine (mainly because your getting a client that's optimised for linux and for SSE2) :D

[Frederic Brillouet]

hey guy's I tried to use the 5.2.13 optimized client for my athlon 64 3000+. The boinc thing refused to connect to localhost. Is there another optimized client i could use? I'm using the bbc client and it's version 5.3.19

[NeoGen]

Do you have a firewall? Sometimes antivirus/firewall software will block a program that appear to have changed due to possibility of virus. I know that at least mcafee and norton both instantly block boinc if I overwrite it with an optimized client. Then a popup appears asking if I want to allow the connection, I say "Always allow", and boinc starts working normally again.

[Frederic Brillouet]

No that doesn't happen to me. I have Panda Anti-Virus + firewall, but it doesn't give any notification about that.

[NeoGen]

I don't know how Panda works... do you have access to the list of programs that are blocked? Maybe you can check out if Panda has automatically blocked it or something...

[mitro]

hey guy's I tried to use the 5.2.13 optimized client for my athlon 64 3000+. The boinc thing refused to connect to localhost. Is there another optimized client i could use? I'm using the bbc client and it's version 5.3.19You must be using the full BOINC client (5.2.13) to begin with. You cannot just replace the files of the BBC version.

[drezha]

mitro is right.

You have to have the original BOINC client to be enabled to use the optimised client.

Though if your only using it for the BBC one, it'll have no effect on your score.

I think it MIGHT be safe to uninstall it (might wanna get a backup of the folder first...zip it and move out of harms way..) then install the proper BOINC client from http://boinc.berkeley.edu/download.php then move the files (minus the .dll's and .exe) back into to it.

I've noticed when I've installed the current BOINC one for some reason, it leaves all the .xml files and work unit file sin the BOINC folder...a reinstall activates it all again and it starts where it left off...that might be the same.. I've not tried with the BBC client.

Then once you've installed the "real" BOINC client, you can add the crunch3r client ;)

HTH

[Steve Lux]

mitro is right.

Now did you really have to go and say that?

[drezha]

Cant hide the truth this time :P

[mitro]

ok... then I'm wrong. :P

[Frederic Brillouet]

Thanks guys, i installed the normal and the optimized client. The qmc thing seems to be running faster. Thanks for the tips, now I hope we will climb a little more in the rankings

[gamer007]

Updated 1st post with another BOINC client + 1 Einstein@Home Optimzed app. Also added link to general page of info + links (to somewhat the same stuff I already linked)

[Evil-Dragon]

Amazing! I just put on Crunchers BOINC on my BBC BOINC and my climate change model time went from 1800 hrs remaining to 1200 hours remaining... thats 600 hrs less work time!

Before:

2006-04-24 00:11:02 [---] Running CPU benchmarks
2006-04-24 00:12:01 [---] Benchmark results:
2006-04-24 00:12:01 [---] Number of CPUs: 1
2006-04-24 00:12:01 [---] 2102 floating point MIPS (Whetstone) per CPU
2006-04-24 00:12:01 [---] 3908 integer MIPS (Dhrystone) per CPU
2006-04-24 00:12:01 [---] Finished CPU benchmarks


After:

2006-04-24 00:13:04 [---] Running CPU benchmarks
2006-04-24 00:14:03 [---] Benchmark results:
2006-04-24 00:14:03 [---] Number of CPUs: 1
2006-04-24 00:14:03 [---] 3042 double precision MIPS (Whetstone) per CPU
2006-04-24 00:14:03 [---] 9235 integer MIPS (Dhrystone) per CPU
2006-04-24 00:14:03 [---] Finished CPU benchmarks

[PcManiac]

hey guys, is there a good client for a Turion Processor?

I wasn't sure if I should just use the one for the Athlon 64 or not ...

thanks!

[Electabots]

You can use any athlon 64 client with either sse1, 2, or 3(3 would be fastest)

[PcManiac]

my Turion has sse3 so I guess I'll search these links and try to find the best one ...

:Edit: hmm, I can't seem to find any sse3 clients for amd's ... :(

[NeoGen]

I haven't checked in a while, but I also don't remember seeing SSE3 enabled boinc clients... only SSE/SSE2.

[Electabots]

Err, yea there isn't any sse3 clients(yet) but theres sse3 applications for einstein/seti@home. Sorry for the confusion :oops:

[Steve Lux]

It was suggested that I use the "optimized" client application for Predictor for my AMD 3800+. After installation my RAC dropped from 615 per day to 545.

I bumpped up to 2 gig of faster ram and my RAC only increased about 5-7 per day. Not much of a gain.

I have tried to re-install the origional client application and have been unable to. I may have to completely wipe BOINC from my system and start over again from scratch. Any idea how to do this and still be able to retain your computer number, record and account with BOINC?

[drezha]

An uninstall should be fine. It doesn't delete the project XML's etc that do this ;)

Though why you cant use the installer to repair. Try downloading the installer (5.2.13 I imagine) then run it and it'll give you the option to repair or remove. Just click repair and it'll restate everything to a default BOINC install (ie any optimised BOINC stuff will be back to normal)

[Steve Lux]

I've run the repair twice and am still using the same Athlon 64 version.

[Keith75]

I have tried the crunchr and trux version. The Trux version didn't help much at all where as the crunchr version about doubled my integer score.

[gamer007]

I've updated my 1st post with Crunch3r's new site (http://calbe.dw70.de/). Also a new version as well for Windows. For some reason it's 5.5.0 when the latest Windows is 5.4.9. :?

[drezha]

Believe the 5.5.0 reports results as soon as they have finished according to the site...(thats how I interpret it)

[Evil-Dragon]

Yup thats correct.

I gained about 2000 extra whetstones on both my CPU's from the last released version, so all is good :)

[NeoGen]

There's a version 5.5.0 already, only compiled for Linux, and only accessible if you go to the downloads root folder...
http://boinc.berkeley.edu/dl/

Scroll down until you see it.
I'm guessing that Crunch3r picked up the source code, got ahead and compiled it for windows, adding the tweak of returning results immediatly.

[mitro]

I have tried the crunchr and trux version. The Trux version didn't help much at all where as the crunchr version about doubled my integer score. The trux version takes as long as a week to gradually adjust your amount of credit claimed. So its not instant like Crunchers.

[mitro]

I highly reccommend upgrading to the latest version of boinc and using the latest Crunch3r's version... I got a 25% increase in claimed credit over the last version.

I generally recommend the Trux calibrating client but with the BOINC race it doesn't give enough time for the client to calibrate since yu are switching projects every week. (or atleast thats my theory)

[gamer007]

Great.. w/ immediate reporting, how am i gonna get more Predictor WUs? I don't want my credits to be added before the start. I can't get more than 20 WUs... stupid P@H scheduler. :evil:

[NeoGen]

Doesn't the option to not allow connection override the immediate reporting?

[gamer007]

I just tried it. I'm getting more WUs now. Looks like some of the uploaded WUs got reported, but some didn't.

But it's not good seeing as how several WUs will be continously be reported before the start-date everything I want more WUs. :-(

EDIT: Guess I'll have to live w/ my lack of WUs for the time. I'll leave it suspended like I did last night until the start.

[dAVE]

Neogen Won't that stop it downloading more WUs?

[NeoGen]

Yea... I was hoping he could queue up a good amount of workunits, and then disable the network connection. So that boinc wouldn't report back the finished workunits before the contest began.

[AMD-USR_JL]

I really like the new BOINC. My whetstone went up 800 and my dhrystone went up 400. I love that crunch3r guy.

[dAVE]

It’s a bit cheeky in the middle of a competition I know but some one might take pity on me and put me right. I upgraded to BOINC 5.4.9 from the Berkley site and was impressed. Simap WUs down from 1hr 9m to 54min. I then tried to install CRUNCH3R 5.5 over the top but it will not install because it can’t find LIBEAY32.dll (upper case). I have searched my files and found that I have libeay32.dll (lower case) installed by BOINC5.4.9 and from the files it looks to be the same library. Any ideas how I can make the CRUNCH3R program see it as there is no option to direct the installation to it, it just fails and gives me an error message.
Grandpa dAVE, or Grumpy as my granddaughter calls me!

[NeoGen]

That's an odd one... I too got the file in my boinc folder in lower case and it never complained.

Maybe a reinstall would set things right?

[dAVE]

Thanks NeoGen, I’ve tried a reinstall/repair but still have the same problem. I’m thinking along the lines of uninstalling and a fresh install now but I’m not sure of exactly what files I need to back up so that I don’t have to re-attach to all the projects again and be given new computer IDs. Any ideas? dAVE

[Nflight]

The new Dynamic Link Library file may be the attempt of the programmers to do away with Optimized clients. In order to thwart the effort and be able to install and run the Optimized program you would need to be able to have all other access to the capitalized file. This means that while it seems irrevalant to most of us capitalizing the file actually redefines the process of Boinc by eliminating the use of Optimizers. That is of course till the optimizers re-configure the script to adapt to the new hazards put in to keep everything the same routine proposed by the Leaders of Boinc.

I was searching for a link to show what I am talking about but I am still reading into the depth of this frustration, will edit with link shortly, I hope!

EDIT: Seems we are the first to bring this to the lime light, Other forums have some mention of a few frustrated members unable to load Optimizers since installing the new Boinc, but not enough to find an answer! YET ! ! I am still working on this issue.

[dAVE]

Thanks Nflight. She who must be obeyed says that I have to go out now to choose tiles for the bathroom. Such is the fun of old age, what happened to clubbing all evening and spending a night on the beach together under the stars! :? Will look back this evening. dAVE.

[mitro]

So are we saying there is a newer version of 5.4.9? If so I've got the version that works with crunch3r's 5.5.0 and could host it if people need it.

[Empty_5oul]

mitro it could be hosted here under the "downloads" section.

it is something to do with the BOINC version. when i did it with boinc 5.2.13 it got a new dll file. but on my 64 i didnt :S.
both have increased the benchmarks significantly though.

[dAVE]

There has been a big fall out on SETI and Crunch3R has left after allegations of cheating. The thread was deleted but CRUNCH3R had copied it and posted it on his site. I wish I had read more of it at the time, there seemed to be allegations that some apps were looking for intel processors and then implimenting routines that gave them a speed advantage. There were also claims that everytime he brought out a new client then a new version of BOINC would come out to thwart him.I bookmarked it to read later but when I went back his whole site had gone. If any of you have his 5.5 version then I should hang on to it. He may have gone to Einstien or Seti Beta so hopefully we shall see him soon. dAVE

[Evil-Dragon]

His site is now redirecting to Einstien@Home.

It's a shame, his optimised BOINC clients were good.

[gamer007]

Crun3r's gone? Crap, anyone got a backup of his latest version? :-(

[daddygeek]

OK
http://cookngeek.com/Daddygeek/BOINC_5.5.0_SSE2.zip

More to come when I find them.

[daddygeek]

From Crunch3r
June 02, 2006


Hello,

It's not quite easy for me to write this, to honest to you all i'm writing this now with a tear running from my eye.

It's been a fantastic time here for quite a while, since things have changed for the bad.

That's why i'm taking the consequences and have to say god bye to you all.

Take care of yourself and keep on crunching.

Good by SETI.

Man i'm really p****d of ...
Can't take this crap any more ...

This whole c++ bug, timer crap could have been resolved a long time ago ... but ignoring people doesn't help getting things solved..

[gamer007]

Thanks daddygeek. Will be useful when I get my new computer in a month. :)

It's too bad he's gone.

[AMD-USR_JL]

Wow! Daddygeek, you may have in your posession one of the greatest boinc clients ever made. Thanks a bunch for sharing this with us. I think you should charge people for it, if that's possible. :lol: I found a thread (http://setiathome.berkeley.edu/forum_thread.php?id=31485) on seti about what happend.

[NVM]

what drama over at Seti! Crunch3r's taking his ball and going home. all because of one guy who really doesn't crunch that much.

[Strongbow]

It's all blown up over there! ...what a shame, especially that Crunch3r has given up trying to help them (I totally understand his reasons though!)! :roll:

This is what I've just scraped from SETI's forum...

"CRUNCHER's MESSAGE

Hi folks,

I am now jolly well fed up with the manure that happens @ Berklely.

That's why my page is offline for now...

I just can't bare and do not need Ignorant arXXXXXX like those guys over there...

Example:
Anybody using the official setiathome_enhanced application has got plenty of it...
C++ exceptions, WUs whose progress time is stuck etc...

Just as a side note... there has long been a solution for those problems now...

Me, foolish as I was, have sent Eric Korpela (the ignorant responsible for all those problems!) a version of the SETI@HOME enhanced application that is FREE of all these problems...

He could have distributed that one unheeded by anyone and all would have been at peace... none of you would have noticed anything of it... except for the new version number, e.g. 5.17.

BUT ignorance wins and the problems persist... thanks to Eric Korpela!!!

Furthermore I am tired of any of those iXXXXX who know a damn about how BOINC or setiathome functions but will still give their opinion on the credits...

That's why my page is offline for now...

But instead of it you'll find some funny reading from those idiots who blame me of "cheating"...
(and a great cartoon from seti.usa on that topic)

I hope for your support to fight that ignorance from seti@home's side...

Regards,
Crunch3r"

[taters]

WOAH! :shock:
I am not exactly sure... but my task time went down from 13K-14K to about 2.2K! Amazing! :D
I do ask for an explanation: Some of you say that the client is not crunching, but I can clearly see in the task manager that that's what's crunching "albert_4.37_windows_intelx86.exe" (it's at 99% CPU usage). Then why do people call it cheating, since it really isn't - you just utilize a more efficient way of crunching specific to your CPU, nothing more?

Now for something else - I now have a funny "issue", which really isn't that big of a deal, since I still get my regular 40-45 credits/task from the server, but my client currently requests 6 points / job. :\

By the way, what do benchmarks matter? My CPU benchmark results did not change at all (still ~2000 floating/3700 integer), but as I said my time decreased by about 620%!

And one more thing, reading that forum at SETI I come to a conclusion that mmciastro guy is full of it. Claimed credit does not matter anything, as shown by my above results, where I claim 6 credits, even though I've done the full 40 in less time. For all I know, I can claim 0.05 credits or 500,000 credits and it would still give me about 40-45. -_-

[drezha]

"albert_4.37_windows_intelx86.exe"

That was done by the user askof and there's been no people accusing him of cheating ;)

[taters]

Sorry, may be I'm missing something, wasn't that what Crunch3r modified? >_>

[gamer007]

Crunch3r modified the boinc.exe and the boinc.dll.

He modified it to use things like SSE2 resulting in higher benchmarks, meaning more credits.

[taters]

Meh, that debate just sucks.

There doesn't seem to be anything wrong:
a) if you use SSE2 to crunch, it just makes sense to benchmark with SSE2
b) if you don't use SSE2 to crunch, it doesn't really make sense to bench with SSE2, but hey... everyone still gets the same, even if claimed is higher or lower than that figure. (Then again, I haven't been around long enough to know if it has always been the case).

[gamer007]

He made different version for each thing (Don't know what SSE, SSE2, SEE3, etc. are specifically called). Meaning each CPU will have it's own optimized client.

I wish I kept a back up of each version. If I knew of this argument, I would've.

[Empty_5oul]

i have the standard 32bit optimised client which was posted above.

on my other machine i think i have the .zip for 64bit (using SSE2) if its still there i will post it later.

[gamer007]

I've updated my 1st post. Crunch3r's link has been removed due to the incident @ the SETI forums. Added SZTAKI and SIMAP optimized clients, untested by me.

[taters]

Mm... which modified client works best for you people?
I assume mine is the Crunch3r's v 5.5.0, with SSE2. (Not sure it's not really labeled in any way). It did increase my benchmarks, but it still requests less than normal credit, which is I believe either 33 (I read it somewhere on boinc forums I think) or 40 (what it used to req before I got the modified cruncher program). It now requests 22, so I was wondering if anyone had luck making it req one of those numbers? :P

[Evil-Dragon]

For the most up to date Einstein@Home aksof optimised applications have a look here:
http://einstein.phys.uwm.edu/forum_thread.php?id=3898

http://eclient.tvn.hu/alb_U4105.zip = SSE3 compatible windows executable (AMD: from E3 stepping, Intel: Prescott? and newer) (AMD X2, would not work with my AMD64 3200+ due to lack of SSE2)
Normal Runtime per unit = 4-5 hrs
Using this application = 35 mins

http://eclient.tvn.hu/alb_S4108.zip = SSE/SSE 2 compatible windows executable (Pentium-III,Pentium4,AthlonXP,Duron(1GHz+),...) (AMD64 will work on this one)
Normal Runtime per unit = 5-6 hrs
Using this application = 43 mins

Not bad eh?

[gamer007]

Ya, that's better. I just have it linked to somebody's (?) site who mirror's Akosf's clients. It's a great client, using it now. :)

Updated.

[taters]

Yup, I'm using the SSE2 one on an Athlon 64 3000+ clocked at 50Mhz higher than 3500+ (so, practically, it's a 3600+, hehe). Doing about 35-36 minutes/WU. Good stuff :) Just wish I had two cores - double the work in the same time. I'm waiting for July 23rd (Conroe) to hopefully get a dual-core Opteron or Athlon, since imo they will go down in price significantly. (Yah, I'm bad... but you gotta ride the market)

[Evil-Dragon]

Using these optimised applications are really beneficial from a credit point of view. Take this for example:

Unoptimised: 1 WU takes 4 hrs 30 mins to complete. For that time you would claim 124.93 credit. After other clients have validated the result, you are granted 54.73 credit which means you have wasted CPU time unnecessarily thinking you would gain more credit.

Optimised: 1 WU takes 35 minutes to complete. For that time you would claim 19.39 credits. After other clients have validated the result, you are granted 51.13 credits. In other words you have been given extra credit because the other clients earned similar amounts of credit. You can see an example of this here: http://einstein.phys.uwm.edu/workunit.php?wuid=9543319

Even if you only gain the amount of credit for the time spent on the workunit then there is no CPU time lost or credits lost. There are times when you'll actually get more credit than the time spent on a WU. Free credit anyone? :)

[vaughan]

That looks great but remember the artificial limit of 32 tasks / CPU / day so at 35' each you can only report about 16 hours of work.

[Evil-Dragon]

I'll just have to run some other project alongside it to stop it from running out of work. Either way i'll end up earning more credit than just running the official unoptimised application.

I red a post on the message board from Crunch3r talking about faking the number of CPU's in the BOINC client to allow it to get more work. Apprently it's already being done in BOINCStudio (although i have not had a look at it yet)

Is NeoGen gonna start this weeks competition or should I just stop crunching and do some work on another project...