RSA-576 Project

[magnav0x]

For those looking for another project to help out on DC Repository rankings I'd like to point out RSA-576. Unlike a lot of math projects these work units don't take long. My Athalon XP 1800 completes a packet in about 55 seconds and my Xbox (celeron 733) completes a packet in about 133 seconds. They have windows and linux clients and both are very user friendly as long as you register your accound ahead of time at the website.

http://rsa576.d2g.biz:8080/en/index_html

Currently we are ranked 14th, but US Distributed has a big production output :scared:


Stats pimpage:

http://www.us-distributed.com/stats/rsa576/rsa_all.php
http://stats.dbestern.net/rsa576/index.html

[magnav0x]

Welcome aboard rsbriggs :thumbs:

Btw, I think those US-D stats are realtime

[rsbriggs]

The packets seem to move along pretty quickly. Shouldn't take long to move past the roadkill at the bottom of the list. :D Might be a while before the two of us make much of a dent here though.....

Real time statage is ....GOOD...

I'm at 578... ooops. Make that 563...

[willy1]

Added an XP2600 to the pile.

It looks like the GUI can be bypassed by running client-gui.exe as a CLI client under Windoze.

[PY 222]

I've registered and installed the client on one 1.3GHz machine.

So far, I like what I see.

Lets go Free-DC! :cheers:

[magnav0x]

Realtime stats + fast wu's = fun :|party|:

dang willy that thing did 100 already?! :scared:


Anyone else notice the typo in the GUI client? Let's see who can find it first :p

[PY 222]

I saw two.

Both are under the Options tab. Do you want me to say it out or not?? :smoking:

[rsbriggs]

You mean "DOS consol"? "Hight"? Credit to PY though - he mentioned that there were two. I suspect the author isn't an English-as-first-language person.

Climbing quickly. Expect to catch up to you, eventually.....

/me 509 /you 463 changes every few minutes :D

12 more blocks to finish in my SoB test, then my production should double.

[PY 222]

Also one more under the Cruncher Priority. "Hight". Maybe that is how they spell it from where they come from. :confused:

This project is getting interesting. Last one is a rotten egg. :bonk:

[magnav0x]

I didn't even check the options tab for typos :D

The first that I saw was on the main GUI at the bottom where it should read "Active Users", but reads "Activs Users". Hey we could almost make stats out of this alone. Let's see....PY is leading. Just wait until the next version, I'll find them ALL before you PY :D


Maybe that is how they spell it from where they come from
Probably used an online translator...


Afraid my P4 is unavailable until later tonight, but I'll get it going on RSA soon as I can. We may not be able to catch US Distributed unless we get more support but as mentioned earlier, we CAN get the roadkill. Oddly enough I'm liking the project. I was expecting to try it and dump it like I did with GoT, but it's rather nice :thumbs:


BTW, what que sizes are you guys using? I have a 50 WU que on my Athalon XP 1800 and 25 WU que on my Xbox

[rsbriggs]

I think I just left it at the default 10, or something. You get stats back faster that way(?) :)


Yes - it's a fun project, and the realtime stats make it quite interesting.....

[PY 222]

Since you all are being such nice competitors, I've just moved another box from DF to RSA-576.

My current work queue is 10 on one machine and 5 on a slower machine. You guys think this is enough or should I bump it up a little higher?


The first that I saw was on the main GUI at the bottom where it should read "Active Users", but reads "Activs Users". Hey we could almost make stats out of this alone. Let's see....PY is leading. Just wait until the next version, I'll find them ALL before you PY

Oh yeah, I see you try that! :Pokes:

[magnav0x]

You guys think this is enough or should I bump it up a little higher?

The only reason I set mine higher is so it communicates less with the server, I figure it takes a little less away from processing time, but rsbriggs is right, the lower the faster the stats roll in :cool: I have my ques set up for about 1 hour of work each.

Oh yeah and I noticed MrTRS has put a box on :thumbs:

[PCZ]

No not another Project !

[PY 222]

Originally posted by PCZ
No not another Project !

I believe this has been around for sometime now.

Don't worry PCZ, its all in the name of fun. :|party|:

[PCZ]

Whats a good crunch time ? is 35 seconds ok ?

[rsbriggs]

That would be an insanely fast crunch time. I'm doing them in about a couple of minutes each....
What the heck are you using?

[Bok]

PCZ, 35 seconds is good, I'd guess this is on a fast athlon?

Fastest I've seen is on a 64bit optimized compile on my opteron 1.8 running Suse linux 8.2, that was around the 29sec mark. An athlon running at 2.3Ghz would be 34-35 secs. Just this weekend the author, Bruno (who is french) got a sun sparc version compiled on one of my Ultra 10's, but it's pretty slow... 400secs per unit :P

Welcome to the project.

I guess you guys are stalking us now :Pokes:

Bok :)

[PY 222]

Guys, I was wondering how do you all come up with the benchmark numbers?

[PCZ]

That time is from one of my PXE boot nodes.
It is one of the slower ones only having SDRAM not DDR
clock speed 145 x 15.

I expect the faster ones 155x15 with DDR to beat that time

[magnav0x]

PY 222, in the GUI client look under the progress bar in the section called "Performances". You will see a Last packet crunched in followed by the time in seconds it took to complete the last packet. Then below it you will see key/s benchmark.

I get about 500,000 key/s and each packet has a crunch time of 55 seconds on my Athalon XP 1800 (@1.63ghz w/Windows XP)

I get about 220,000 key/s and each packet has a crunch time of 130 seconds on my Xbox (Celeron 733 w/Gentoo Linux)

[magnav0x]

Ack I knew PCZ would find out! He's going to bury us all :notworthy

[PY 222]

Darn it.

I am only managing 122 241 keys per second on a Pentium 4 1.3GHz. Damn Pentium machines. :bang:

[PCZ]

The linux version gives you a display showing the crunching time and the progress.

Seems the linux client is faster than the windows version

[magnav0x]

Hrm tonight I'll put it on my P4 1.6 and see what kind of results I get PY. Is that in Windows or Linux? I'll test in both.

[PY 222]

Originally posted by magnav0x
Hrm tonight I'll put it on my P4 1.6 and see what kind of results I get PY. Is that in Windows or Linux? I'll test in both.

I'm running Win2k on this machine.

I guess I will also throw in my AXP 1800+ when I get home tonight to see if it matches your numbers.

[PCZ]

Ack I knew PCZ would find out! He's going to bury us all

How long do you think it is going to be before IB and Fobot roadkill you ?

[rsbriggs]

It appears that AMDs are faster than P4's, and Linux is faster than Windows. I'm getting 45 seconds per on a Barton 2500 running Windows.

I can't seem to get the client to work correctly on my diskless nodes. It starts, but once it tries to display anything, it just hangs.


How long do you think it is going to be before IB and Fobot roadkill you ? Oh, not them, too :mad: This project sure got crowded in a hurry :D Hey - give us a break, we've only been running this stuff for a couple of hours now....

[pfb]

Originally posted by rsbriggs
It appears that AMDs are faster than P4's, and Linux is faster than Windows. I'm getting 45 seconds per on a Barton 2500 running Windows.

Yep - AMD seems to be a lot faster: my P4 @ 2.7 and Win XP does a packet in ~90 seconds, my MP 2400+ and Win XP is ~60 seconds, and my Barton 2800+ and Mandrake 9.2 is ~40 seconds...

[rsbriggs]

D'oh - :notworthy Ironbits is already up in 146th place overall. No sign of FoBot so far. Maybe this project will sneak in under his radar..... :thumbs:

[magnav0x]

PCZ, Ironbits is already way up there :p He had an output of 5,000+ yesterday.


Oh yeah and I was getting a bit curious as to why the P4's where going so slow, so I threw the client on a Celeron 333 to see what it's numbers would look like compared to a P4.

Celeron 333 (w/Gentoo Linux) ran the first packet at 104,914 keys/s in 285 seconds

[willy1]

Arrrgh!! Another client I can't run at work - uses some non-standard port (not 80 or 23) to get work.

Bloody hell...:swear:

[magnav0x]

I'm a bit confused as how the hell your P4 1.3 only gets 122,241 keys/s PY 222. Seeing how this 333 is getting 104,914 keys/s :scratch:

I'm thinking that the Linux client is quite a bit faster than the Windows clients (btw I'm using the i586 optimized client on both of my Linux boxes for those interested.).

[FoBoT]

Originally posted by rsbriggs
D'oh - :notworthy Ironbits is already up in 146th place overall. No sign of FoBot so far. Maybe this project will sneak in under his radar..... :thumbs:


Originally posted by PCZ
How long do you think it is going to be before IB and Fobot roadkill you ?

what is the most you can cache/queue up? or is this a strictly online project?

i don't generally participate in the "guess a number" projects, but stats ARE stats :rolleyes:

[PCZ]

rsbriggs

try copying the usr/share/terminfo directory to the ltsp tree

[magnav0x]

FoBoT, as far as I know, you can get as many as you want.

[FoBoT]

Originally posted by magnav0x
FoBoT, as far as I know, you can get as many as you want.

so if a slower box takes 2-3 mintues per, then grabbing 5000 should last a couple of days offline

hmmm...

[rsbriggs]

rsbriggs
try copying the usr/share/terminfo directory to the ltsp treeAlready did that previously to be able to see the ANSII type output from DF.

[PY 222]

Originally posted by magnav0x
I'm a bit confused as how the hell your P4 1.3 only gets 122,241 keys/s PY 222. Seeing how this 333 is getting 104,914 keys/s :scratch:

I'm thinking that the Linux client is quite a bit faster than the Windows clients (btw I'm using the i586 optimized client on both of my Linux boxes for those interested.).

Dude, I don;t know why my Win2k machine is as slow as a snail but it could be that I am using it right now.

Update: The client now says 141k keys per second. Now I am hoping that it will go higher.

WARNING: FoBoT in the picture. Run people. :scared:

[PCZ]

willy1

You can run it at work by sneakernetting.

You startup a client, download say 5000 WU's, put the down_packets file on a floppy and transfer it to an offline machine for crunching.

[magnav0x]

Well glad to see it's gone up some. Still a ways from out crunching my Xbox....you know those GPU's are good for distributed computing lol. I'm not sure what to think. I would figure if P4's were slow on it that Celeron's would be a LOT slower than the P4's, especialy at near half the clock speed. We'll get it figured out :D

[rsbriggs]

Might want to move this over to the new RSA576 forum :D

[magnav0x]

Ok PY222, I put the client on my P4 1.6 and here are the results:

Windows 2000: 232,000 keys/s in 129 seconds
Gentoo Linux (P4 Optimized Client): 398,000 keys/s in 75 seconds

[willy1]

The XP2600 is averaging about 650000 keys/sec.

Still hassling with the corporate firewall issue - don't know the host name it's trying to connect to to validate the userid and get work (may be 2 different hosts?). It seems to work on port 1710 - got that much info from the built-in proxy support. Trying HTTPort, and looking for other solutions.:confused:

[magnav0x]

Here is a little info about the proxy authentication I dug up over at their site



Client is using 'desproxy' http://desproxy.sourceforge.net to connect proxy server... to pass basic HTTP proxy server authentification, desproxy's using an environnement variable where login and password are stocked.

You must to create manually this environement vriable:

underWin9X/ME
Edit C:\autoexec.bat file. Add(at end of file for example) this line: SET PROXY_USER=proxylogin:proxypass
where proxylogin is the username and poxypass the password to connect to proxy...
You must to reboot.

underWinNT/2000/XP/2003
Control panel > System > Advanced > Environnements variables. you must create now the new environement variable : PROXY_USER=proxylogin:proxypass
where proxylogin is the username and poxypass the password to connect to proxy...


Now when desproxy is running, it's using PROXY_USER variable to connect server...

http://desproxy.sourceforge.net/manual-en.html for more help



Not sure if it helps at all, but I figured I might as well throw it out there. Apparently the client runs on port 1708 only by default.

[magnav0x]

willy1, I was curious where the authentication was going as well. After monitoring the packets going in and out upon sending/recieving packets and the authentication it turns out that all communications are going to the following host:

rocques01.ikoula.com

and the client only utilizes port 1710 (by default)

[willy1]

OK - I was just about there - I think I spelled rocques01 wrong :(

I'll give it another shot tomorrow.