@Ned --> Because we both use linux, it should be a quick thing for me to just
compile the stuff before giving it to you. But, a question for you:

The changes this time involve replacing bits like

for (i=0;i<n;i++) sum += a[i];

with IPP routines. ACML also offers sum routines, which I could use.
Is there any interest in having an ACML path? Right now, you'll have a choice
between IPP and the really slow code above.

I'm not playing with ACML at the moment - just don't have the spare time to properly investigate what it has to offer. Strictly gcc/fftw here I'm affraid.

Ned

---

*** My Guide to Compiling Optimised BOINC and SETI Clients ***
*** Download Optimised BOINC and SETI Clients for Linux Here ***

@Ned --> Because we both use linux, it should be a quick thing for me to just
compile the stuff before giving it to you. But, a question for you:

The changes this time involve replacing bits like

for (i=0;i<n;i++) sum += a[i];

with IPP routines. ACML also offers sum routines, which I could use.
Is there any interest in having an ACML path? Right now, you'll have a choice
between IPP and the really slow code above.

I'm not playing with ACML at the moment - just don't have the spare time to properly investigate what it has to offer. Strictly gcc/fftw here I'm affraid.

Ned

I'm interested in ACML since i'm playing around with the PGI compilers (amd64) and if time allows it and i dont have to recompile the whole bunch of clients :-)

---

Join BOINC United now!

Ned, what BOINC source are you compiling against? i haven't had success with GCC/FFTW on any of the caching sources yet.

have you changed the optimization flags at all?

You need the BOINC 4.45 source to get i working.

when getting to xml_util i keep getting:
make: *** No rule to make target ../../boinc/lib/acct_mgr_client.o', needed by `seti_boinc'. Stop.

on harold's sources v2.7-3.1
using boinc trees for 4.45 and 4.72

What switches are you using to configure boinc and seti? I use the following for boinc:

./configure --disable-server --disable-dependency-tracking --build=i686-pc-linux-gnu

and for seti:

./configure --disable-server --disable-gui --disable-dynamic-graphics --disable-dependency-tracking

Also, I use the nightly boinc source from 25 May 2005 (just checked), version 4.43 which was, until recently, the latest stable release for Linux.

Ned

--disable-server --disable-gui --disable-dynamic-graphics --without-wx --enable-static=no

your two others did not make a difference. i will try other sources, as that library is not in boinc/lib/ in any of the versions i compiled

ned, 4.43 source you have been using does have that library. the latest releases of the boinc client do not. with the May sources i was able to easily compile harold's 3.1 seti client with fftw3 and gcc-4.0.2

i am currently testing it and will compare it with TBD versions using P2 and K6-2 flags

Just one question ..sorry if I can't ask for it here...
It s just because I m looking for command line for boinc usage...you know you ve got boin_cmd boinmgr and boinc...
Can you launch boinc in a terminal ? or you must launch boincmgr ??

---

Just one question ..sorry if I can't ask for it here...
It s just because I m looking for command line for boinc usage...you know you ve got boin_cmd boinmgr and boinc...
Can you launch boinc in a terminal ? or you must launch boincmgr ??

On Linux, you surely can. Just launch it ./boinc & (./boinc_client & -- Boinc version 5)

On Windows, I have no idea. If you are running windows, you are in the wrong topic.

---

ned, 4.43 source you have been using does have that library. the latest releases of the boinc client do not. with the May sources i was able to easily compile harold's 3.1 seti client with fftw3 and gcc-4.0.2

i am currently testing it and will compare it with TBD versions using P2 and K6-2 flags

Yes, that's correct. Glad you got it sorted :)

Ned

---

*** My Guide to Compiling Optimised BOINC and SETI Clients ***
*** Download Optimised BOINC and SETI Clients for Linux Here ***

... On Linux, you surely can. Just launch it ./boinc & (./boinc_client & -- Boinc version 5)

On Windows, I have no idea.

Is that in bash or English?

And I suppose you could be awkward and try cygwin!

:-O
Martin

(OK, that lot must be English unless you meant "&&" instead of "&" ...)

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

OK, a very small sample size for the latest optimised clients for a lowly AthlonXP:

Ned's v3.1 (8 or 16 MBytes cache?):

141558138 33746156 Over Success Done 7,818.34 24.00 pending
141518340 33535250 Over Success Done 7,591.95 23.30 31.38


Cruncher's v3.1 (8 or 16 MBytes cache?):

140781388 33556842 Over Success Done 8,433.48 25.89 34.47
140369019 33456229 Over Success Done 8,026.67 24.64 pending


Harald's many Megabytes cache monster:

139728870 33299725 Over Success Done 8,204.25 25.18 21.86
139664492 33284017 Over Success Done 7,527.89 23.11 23.11


One of Ned's optimised clients from sone time ago:

131830143 31373256 Over Success Done 10,509.28 31.34 26.94
95144694 22523486 Over Success Done 10,984.40 33.75 32.61


Note that they are listed most recent first. There's a suggestive improvement there :-)

How is strong/weak similarity reported for those?

Happy crunchin',

Regards,
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

OK, a very small sample size for the latest optimised clients for a lowly AthlonXP:

Ned's v3.1 (8 or 16 MBytes cache?):

141558138 33746156 Over Success Done 7,818.34 24.00 pending
141518340 33535250 Over Success Done 7,591.95 23.30 31.38


Cruncher's v3.1 (8 or 16 MBytes cache?):

140781388 33556842 Over Success Done 8,433.48 25.89 34.47
140369019 33456229 Over Success Done 8,026.67 24.64 pending


Harald's many Megabytes cache monster:

139728870 33299725 Over Success Done 8,204.25 25.18 21.86
139664492 33284017 Over Success Done 7,527.89 23.11 23.11


One of Ned's optimised clients from sone time ago:

131830143 31373256 Over Success Done 10,509.28 31.34 26.94
95144694 22523486 Over Success Done 10,984.40 33.75 32.61


Note that they are listed most recent first. There's a suggestive improvement there :-)

How is strong/weak similarity reported for those?

Happy crunchin',

Regards,
Martin

As you seem to be in a test orgy :-)

Could you try this "OLD" boinc version 4.19 for me ?

Just need to know bench. results and if it can connect to berkeley.

---

Join BOINC United now!

As you seem to be in a test orgy :-)

Could you try this "OLD" boinc version 4.19 for me ?

Just need to know bench. results and if it can connect to berkeley.

Phah! Just when I've finally got the app_...xml file sorted!

Downloaded... What's it supposed to do?

Cheers,
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

As you seem to be in a test orgy :-)

Could you try this "OLD" boinc version 4.19 for me ?

Just need to know bench. results and if it can connect to berkeley.

Phah! Just when I've finally got the app_...xml file sorted!

Downloaded... What's it supposed to do?

Cheers,
Martin

It's just a replacement for your old boinc version and should kick benchmarks a little further :-)

(It's fully static that's why i need to know if it works)

---

Join BOINC United now!

It's just a replacement for your old boinc version and should kick benchmarks a little further :-)

(It's fully static that's why i need to know if it works)

Just looked and that's a boinc client rather than a project client...

I'm on phase 3 of a CPDN so, sorry, I'd rather not risk ditching it.

If there's no other Athlons around, then I could go offline and try it as a duplicate user. But operator sleep first! (And this Athlon is my server!!)

Cheers,
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

Could you try this "OLD" boinc version 4.19 for me ?

Hehe, yeah it works. ;)

It benches over 2x in double and 3x in integer than the 5.x one I compiled.
The only thing is these 4.19 clients all left stale sockets in CLOSE_WAIT
state on all my machines except the masq gateway, your client does this as
well. 5.x clients dont do it on any machine here. Darn downgrade did a project
reset too, bummer, but oh well.

Can you give me the compile flags you used please?

On the XP3200+ (with tvtime running fullscreen on a 2nd display)

2005-11-01 23:02:59 [---] 2412 double precision MIPS (Whetstone) per CPU
2005-11-01 23:02:59 [---] 6378 integer MIPS (Dhrystone) per CPU

M

@Crunch3r

The 4.19 client is claiming about double the credit that my out-of-box 5.3
client does. I tried compiling 5.3.1 cvs with flags I saw in an earlier post...

export CFLAGS="-march=athlon-xp -O3 -static -m3dnow -msse -malign-double -maccumulate-outgoing-args -fomit-frame-pointer -falign-functions=4 -finline-limit=1024 -fprefetch-loop-arrays -funroll-loops -fforce-addr -ftree-vectorize -ffast-math -ftracer"

but it really didnt like the -malign-double flag. It benched ok, nothing
thrilling like your 4.19, but it nagged about results being over quota,
giving some really bogus numbers for result size, and I lost a couple wu.

Taking out -malign-double produced a working client, but of course the benchmarks sucked compared to this one you did. If you can paste the flags you used, I will try again.

M

@Crunch3r

The 4.19 client is claiming about double the credit that my out-of-box 5.3
client does. I tried compiling 5.3.1 cvs with flags I saw in an earlier post...

export CFLAGS="-march=athlon-xp -O3 -static -m3dnow -msse -malign-double -maccumulate-outgoing-args -fomit-frame-pointer -falign-functions=4 -finline-limit=1024 -fprefetch-loop-arrays -funroll-loops -fforce-addr -ftree-vectorize -ffast-math -ftracer"

but it really didnt like the -malign-double flag. It benched ok, nothing
thrilling like your 4.19, but it nagged about results being over quota,
giving some really bogus numbers for result size, and I lost a couple wu.

Taking out -malign-double produced a working client, but of course the benchmarks sucked compared to this one you did. If you can paste the flags you used, I will try again.

M

I used the same FLAGS but the benchmark has changed in the new clients. Besides i'm not able to get a working verion of 5.26 or 5.3.1 compiled that can connect to berkeley no matter what i do.

I guess there's something messed up with libcurl.

I don't like it.

Every increase in the version numbers makes it all even worse.

---

Join BOINC United now!

I used the same FLAGS but the benchmark has changed in the new clients. Besides i'm not able to get a working verion of 5.26 or 5.3.1 compiled that can connect to berkeley no matter what i do.

I guess there's something messed up with libcurl.

I don't like it.

Every increase in the version numbers makes it all even worse.

Yeah, I hear ya there. Oh, well, I will live with stale sockets for awhile
and use this one for a bit. I only start messing with stuff when I get bored,
so that gives me lots of opportunity to play.

M

I just asking for it because (of course i m linux user) i don't know how to use the "boinc" (i mean the boinc client with with boinc command and not with boinc_cmd or boinmgr)...I m looking for command line to run it... if someone knows them...thx

---

Actually I use mouse to switch on the boinc client with boincmgr...and i expect there is an other way...like in a terminal...

---

I just asking for it because (of course i m linux user) i don't know how to use the "boinc" (i mean the boinc client with with boinc command and not with boinc_cmd or boinmgr)...I m looking for command line to run it... if someone knows them...thx

ever tried "./boinc_client --help" ? :-)

anyway you can use it like "./boinc_client -return_results_immediately & "

that will kick the pid in background and you dont have to keep the terminal open.

---

Join BOINC United now!