Good news :o)

I'm in final testing of an SSE-optimized Linux cruncher. No numbers yet, but it seems to run fine across platforms. As soon as I can get some independent test and verification results, I'll put it up for download.

Regards,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

Sorry I wasn't clear, the -nographics argument is only available when testing the setiathome app standalone (without BOINC):
path\\setiathome_5.15_windows_intelx86.exe -nographics[/pre]

To me that sounds like a usefull boinc feature: to be able to spawn the app with the -nographics option (or any other, maybe project-specific, options). Would such a feature give everybody who uses it (I personally never use graphics) a 5-10% speedup?

My testing indicates no change is needed. If you don't turn on the graphics in BOINC the graphics thread doesn't use enough CPU time to matter, probably much less than 1%.

For this project, there's also control of the graphics from the project preference settings. Included in that is a setting for the maximum percent of CPU time the graphics can use. The default is 50%, but it could be set to zero.

                                                          Joe

Ok. So for someone who knows absolutely nothing about programming, is this something that I could do? Where would I find everything I need to download to begin doing this? I'm very interested in tryin this out.

Also, I know that you have test WU's, do you run your cache dry of SETI WU's, plug these in, disable network access? I'm pretty illiterate when it comes down to this stuff. But, I'm interested in learning. Thanks for all the work you've done Simon. :-)

Jeremy

---

Hi KB7RZF,

I made two How-To's, one for Windows, one for Linux, that have both been stickied. They have links to the necessary stuff and information how to install and use it. Those instructions are geared towards people with little to no experience as much as I could make it so.

Linux How-To

Windows How-To

Hope that helped,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

Hi KB7RZF,

I made two How-To's, one for Windows, one for Linux, that have both been stickied. They have links to the necessary stuff and information how to install and use it. Those instructions are geared towards people with little to no experience as much as I could make it so.

Linux How-To

Windows How-To

Hope that helped,
Simon.

Thanks for those, forgot about um. :-) I have no clue where to find C++ 2003 at all. Where would I come across this?

---

It's a commercial product. eBay or the like should have used version for sale pretty cheaply.

Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

Oh yeah,

just saw this among the other hosts on one of my WUs -

Windows optimized S@H Enhanced application by Daniel Schaalma -SETI.USA
Developed for Team SETI.USA(www.setiusa.net) JOIN Team SETI.USA!!!
Special Thanks to Crunch3r, Hans Dorn, TMR, Harold Naparst, and Financier!
Version info: Windows Intel P4 SSE2 V5.15c by Daniel Schaalma -SETI.USA

Work Unit Info:
...............
WU true angle range is : 0.5692

Flopcounter: 12903131849229.531000

So I guess team Seti.Usa has already gotten their own optimized windows client. kevint, you haven't answered whether you want to release anything to the public. Judging by what I'm seeing, it doesn't look likely.

Also, if someone's already testing this, why do it as "anonymous"? Seems a bit fishy to me - also, why test it online? Grab a few reference WUs and do it offline, where problems don't affect the project as a whole.

All in all, I'm moderately surprised you a) didn't announce anything b) are doing it in this sort-of furtive way (forgive my choice of words, but that's what it seems like to me). Oh, and thanks for deciding for me whether I wanted to include a team ad into my clients...you managed to put it once on each line on average.

Anyway, props to Daniel for getting his build to work :o)

Regards,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

6,000 seconds slower on a CPU that is only 200 MHz slower? At least it looks like he hasn't managed to beat your times yet :)

---

A member of The Knights Who Say NI!
For rankings, history graphs and more, check out:
My BOINC stats site

@ KWSN

Sorry. Another question. If you remeber i asked about threading. Anyway i had another thought on that. (you know all this already so might as well scip to the second paragraph) On a HT machine you can run two units -will take longer but because doing two they're both done quicker. I remember on the old SAH app when running optimised some people got more efficient crunching with HT off.

My question is would you get better results (time wise) from running a optimized app two units with HT on - or limit SAH to 1cpu and running a threaded optimised app? i know the compiler can only do so much to thread the app because it wasn't writen threaded.

Still currently testing the SSE3 app on my HT machine and i got my first result in (hybrid) and is valid. few other shorts. All valid and granted.
http://setiathome.berkeley.edu/result.php?resultid=345128248
http://setiathome.berkeley.edu/result.php?resultid=348661796
http://setiathome.berkeley.edu/result.php?resultid=348661792
http://setiathome.berkeley.edu/result.php?resultid=348661807
Need to rmember to add a personal bit to the version report.

Oi,

Pepperammi, you seem to be picking up a lot in a short time :o) In my experience, HT has been quicker than disabling it. Yes, you can crunch a WU in a shorter amount of time if you switch it off, say by 33% (estimate). So then you switch on HT, and you have two virtual CPUs crunching 2 results at the same time, each 33% slower than before.

So say the time taken without HT is 1 time unit. The time taken with HT is 1.33. Let's do 10 WUs. This will take 10 time units without HT, and 5x1.33 with HT, which is 6.65. So you see, even with HT taking longer per WU, you will crank out about 50% more results (33% less time taken).

Hope that helped :)

Oh, and right now (as this was set before) all apps are set to use multi-threaded libs in Windows - I'm doing some testing whether anything changes if I set it to use single-threaded stuff instead.

Regards,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

Also, if someone's already testing this, why do it as "anonymous"? Seems a bit fishy to me - also, why test it online? Grab a few reference WUs and do it offline, where problems don't affect the project as a whole.

All in all, I'm moderately surprised you a) didn't announce anything b) are doing it in this sort-of furtive way (forgive my choice of words, but that's what it seems like to me). Oh, and thanks for deciding for me whether I wanted to include a team ad into my clients...you managed to put it once on each line on average.

Anyway, props to Daniel for getting his build to work :o)

Regards,
Simon.

You are making some wayward assumptions there Simon!! First, you assume that it was a test when actually all of the testing took place using reference WUs and on the beta site as is proper. Your assumption that something was fishy and or furtive is also way off base since the WU you saw was from one of our members who has always elected to be anonymous. But by far the worse assumption you made was that it was "your client". The client was made from scratch using the nightly tarballs and Daniel's own programming knowledge with input from the people listed. As a matter of fact it was being developed at about the same time you were doing yours and your assumption that it was your work that was changed is false. The Team voted to keep this within the Team since it was funded by one of our members. So please in the future have your facts straight before making cloaked accusations.

---

Sarge,

nowhere did I assume it was my client - you misunderstood me there.
And no, I don't assume it was my work.

When I make accusations, I don't cloak them. When I ask about things, I do so in the form of a question. If you understood my questions as an accusation, you inferred that meaning.
Also, I honestly salute Daniel for getting his build to work - there was no sarcasm included (nor is there now).

So no, that's not what I'm on about, and please don't try to infer I am :o)

Instead, seeing that message, I was already thinking about whether to include one line with a link where to get updates for the client and including my team name. That's what I'm talking about.

And again - when I assume something, I say "it seems", not "these are the facts" or something. So thanks for explaining - but please don't put words in my mouth.

Also, I had asked kevint to post whether he planned to release anything to the public or not. As it seems, that question has been answered (by a vote, as you say), but not by anyone from Seti.USA until right now. Again, doing things behind closed doors is something I cannot understand. Even if you wanted to keep it to yourself, why not even tell anyone? Seti.USA has always been most vociferous in other regards.

What I find worthy of accusation is this: Seti.USA caring only for Seti.USA, not for SETI. Of course, the vote was a team decision, but it just goes to strengthen that point. I'm willing to be persuaded my opinion here is wrong (in a reasonable discussion), but so far, I have been less than impressed.

I recognize it was funded by one of your members (seems it was kevint) and that you are in no way required to share. That's why you're Seti.USA and I'm a KWSN ;o)

Regards,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

nowhere did I assume it was my client - you misunderstood me there.
And no, I don't assume it was my work.

My apologies for apparently doing what I accused you of. Yes I did misunderstand that part of your statement.

However, comments like:
Seems a bit fishy to me
and
are doing it in this sort-of furtive way
Would have been better made to us in private rather than in this public forum. By making them publicly they became inflammatory and will be seen as accusatory by many people. So I elected to answer them as accusations.

---

Oh, and right now (as this was set before) all apps are set to use multi-threaded libs in Windows - I'm doing some testing whether anything changes if I set it to use single-threaded stuff instead.

Regards,
Simon.

Thanks. I think you answered it more in your last paragraph. :)
I meant if HT-enabled and limit SAH to 1cpu it useto on the old app (for me) only use the 1 cpu and the other logical was free and report only using 50%cpu. Not sure if still the case, Havn't messed with it for a long time.
so you could have;

2WUs running two optimized apps together -one each physical/logical cpu and as you say they'l be done faster overall than with HT off running 1WU.
Or (?) 1WU HT-Enabled and SAH set to 1cpu but the app optimized-with-Threading and this i think should utilise both physical and logical cpu? I just wasn't sure if this setup might utilize the cpu and its threading abilities more intelligently/efficiently than two opti-apps fighting over it. I suppose this would only come into play (if at all) if boinc set affinity but windows probly does all this already.

I think you've already answered all this though because as you say its compiled with mutithreading already. :) sorry if i've wasted your time.

P.S. learning most of this from the forums and your how-to pages :)

nowhere did I assume it was my client - you misunderstood me there.
And no, I don't assume it was my work.

My apologies for apparently doing what I accused you of. Yes I did misunderstand that part of your statement.

However, comments like:
Seems a bit fishy to me
and
are doing it in this sort-of furtive way
Would have been better made to us in private rather than in this public forum. By making them publicly they became inflammatory and will be seen as accusatory by many people. So I elected to answer them as accusations.

That's okay, maybe I used the wrong words. If so, I apologize.

Private forums are exactly what I'm trying to avoid here, like I said, I aim for maximum transparency.

Thanks for answering!
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

Oh, and right now (as this was set before) all apps are set to use multi-threaded libs in Windows - I'm doing some testing whether anything changes if I set it to use single-threaded stuff instead.

Regards,
Simon.

Thanks. I think you answered it more in your last paragraph. :)
I meant if HT-enabled and limit SAH to 1cpu it useto on the old app (for me) only use the 1 cpu and the other logical was free and report only using 50%cpu. Not sure if still the case, Havn't messed with it for a long time.
so you could have;
...

Pepperammi,
be aware that the applications using multi-threaded libs does not mean they are specially optimized to take advantage of multithreading, as well! That's a separate setting you can find in "C/C++ - Optimization - Parallelization" (scroll down all the way, it's the last). I haven't played with it much yet, so I'm very interested in your results.

As to it using 50% of available CPU - no matter whether it's a HT system or a real SMP system, on a 2-CPU machine 100% usage on one CPU shows as 50% overall. I'm not sure whether setting CPU affinity will change anything there. Optimizing for Parallelization may cause both logical/physical CPUs to share the load and overall system usage still be 50%, maybe. Check your CPU graphs in task manager when you do benchmarking :o)

Regards,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

Pepperammi,
be aware that the applications using multi-threaded libs does not mean they are specially optimized to take advantage of multithreading, as well! That's a separate setting you can find in "C/C++ - Optimization - Parallelization" (scroll down all the way, it's the last). I haven't played with it much yet, so I'm very interested in your results.

As to it using 50% of available CPU - no matter whether it's a HT system or a real SMP system, on a 2-CPU machine 100% usage on one CPU shows as 50% overall. I'm not sure whether setting CPU affinity will change anything there. Optimizing for Parallelization may cause both logical/physical CPUs to share the load and overall system usage still be 50%, maybe. Check your CPU graphs in task manager when you do benchmarking :o)

Regards,
Simon.

Ah. Thanks. I'm still just checking that the app is making valid results on the HT machine but yea I will try the different single-threaded and Parallelization options you mention. Will be interesting to see if any difference. I'll try the parallelization first if your already trying single-threaded.

Hi Simon,

First, let me start out by offering my congratulations on releasing your Linux apps. I have not had the time to spend here reading these boards for a while. I first started working on an optimized app just after Crunch3r resigned from the Seti Community. My team has supplied me with a fully licensed version of ICC and IPP, to produce optimized apps for our team. It was decided that after the turmoil that led to Crunch3r's loss to our community, that we would build apps for team members only, to avoid a repeat of such events. I would like to reassure you that my apps were very thoroughly tested in "standalone" mode, and all results checked with TMR's "rescmp", and ALL completed results were graded "Strongly Similar". After the "standalone" testing phase, these apps were then thoroughly tested on Seti Beta. In that phase, there were NO INVALID results, except through installation errors by one or two testers. It was only after very thorough testing that the apps were released here for production. I would NEVER release an app for production without thorough testing.

I wish you the best of luck in your app development, Simon, and I just hope that by your releasing your apps to everyone, that you don't suffer the same misfortunes as Crunch3r. Take care, and again, great job on getting your builds released!

[edit]We also decided to remain silent about our apps, to attempt to avoid the flaming that will unfortunately, enevitably appear, for as long as physically possible. I am sorry if you took this to mean that we had some sort of nefarious plans, but that is not the case. We have been keeping our comments mostly among team members, in order that people on all sides can cool off for a while, and we can all get back to crunching.[/edit]

Regards, Daniel.

---

Around $900 AFAIK, but I haven't checked closely. Sadly, the licenses are time-limited, I believe that's for one year and includes ICC, IPP and MKL (the latter two being library packages).

Intel pricing page

Also don't forget both Linux and Windows (and maybe even OS X, for X86-based Macs) are required, so that doubles/triples normal license costs.

--edit
Intel Compiler for Linux - $399
Intel Compiler for Windows - $399
Intel Compiler for OS X - $399

IPP - $199
MKL - $399 (maybe not necessary, but probably faster than fftw).

From the pricing page, it seems that all supported OS flavours are included when you buy one license for the library packages. The compiler has to be licensed once per OS.

There's a promotional MacOS package available that includes Compiler, IPP and MKL and costs $549. Since the libraries seem to license cross-OS, that might be the cheapest path overall.

Regards,
Simon.

I find it hard to believe that Crunch3r was paying $900 US every year just so he could make BOINC and s@h optimizations!

Erik,
Why do you find it so difficult to believe that Crunch3r was not paying this, it is not so hard to understand. Why do you think he got so upset and left the project ?

ACTUALLY, I believe Crunch3r's copy was donated by another Seti user.
It has been awhile, but I do remember this from A LONG TIME AGO now, a user had a copy and was not using it.

---

Thanks for your answer, Daniel!

I was aware that you were already working on an optimized build when I started doing my stuff, so that's why I was sure you had gotten there on your own.

Thanks for explaining the testing regimen you put your clients through, I commend you for taking extra care.

I do have to disagree on your decision to cut down on the flaming by keeping things private. I do believe that (even had I not noticed it) this will cause more turmoil in the long run exactly because you didn't tell anyone beforehand.

Also best of luck in your further app development :o) I'm pretty sure I won't let things like that happen - hence my policy of total transparency. That's really the best way to stop rumours - never give anyone anything to wonder about.

Just my 2 cents...
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information