Author |
Message |
|
netham45
Advanced Newbie
Joined: 25 Mar 2006 Posts: 99
|
Posted: 14 Aug 2009 06:47:53 pm Post subject: |
|
|
The project spazzed out on my laptop earlier, BOINC kept crashing, and it produced what's visible on this page: http://boinc.unsads.com/rsals/results.php?hostid=37
Not sure what happened.
Edit: Detaching from the project, rebooting, then reattaching seems to have fixed the problems.
Last edited by Guest on 14 Aug 2009 07:00:28 pm; edited 1 time in total |
|
Back to top |
|
|
Skittle
Newbie
Joined: 11 Aug 2009 Posts: 23
|
Posted: 15 Aug 2009 02:19:13 pm Post subject: |
|
|
the experimental siever is much faster than old!
~.10s/relation down from .~18
sieving 8m-10m
Last edited by Guest on 15 Aug 2009 02:19:43 pm; edited 1 time in total |
|
Back to top |
|
|
fullmetalcoder
Member
Joined: 01 Aug 2009 Posts: 139
|
Posted: 15 Aug 2009 02:30:06 pm Post subject: |
|
|
45% done on 10M-13M. It should be over by the end (around midnight) of monday (GMT+1). I won't be able to do more for the 0A key because I'll be away from any sort of computing device from wednesday to... well, I'm not quite sure yet but I'll be away for at least 10 days so the 0A key should be fully factorized when I get back. |
|
Back to top |
|
|
FloppusMaximus
Advanced Member
Joined: 22 Aug 2008 Posts: 472
|
Posted: 15 Aug 2009 06:01:27 pm Post subject: |
|
|
Squalyl: When you have a chance, please try out this validator. To compile it, you'll first need to compile Msieve, as well as BOINC itself. I believe the validator will work correctly, but I'm not 100% sure since I can't test it in your exact configuration.
Validation is based on the name of each work unit (<keyname>_<q0>_<qintsize>_... .) There is probably a better way to do this (e.g. by looking up the actual input file the client was given) but I don't really want to delve that deeply into BOINC code. Feel free to improve this if you want.
One annoying detail is that you need to supply the validator with polynomial files (or, as Msieve calls them, "factor base" files) in the format understood by Msieve. I've included a script to convert GGNFS poly files into Msieve fb files. I'm not sure where is the best place to store these files; it will look for them in the directory defined as POLY_FILE_DIR. The name of the file should be "<keyname>.fb".
Once we get this validator working, we should be able to use the 64-bit client as well, and then we'll really be moving. (By the way, has anyone managed to compile the experimental siever for 64-bit Windows?) |
|
Back to top |
|
|
Skittle
Newbie
Joined: 11 Aug 2009 Posts: 23
|
Posted: 15 Aug 2009 06:36:10 pm Post subject: |
|
|
what do we need to compile for windows? I can try it out later tonight. |
|
Back to top |
|
|
FloppusMaximus
Advanced Member
Joined: 22 Aug 2008 Posts: 472
|
Posted: 15 Aug 2009 06:54:07 pm Post subject: |
|
|
If you want to try, please do. I can barely manage to get my own software to compile for 32-bit Windows, most of the time. :P
The experimental code is in the GGNFS Subversion repository, under src/experimental/lasieve4_64. There are instructions provided for compiling with GCC on Unix-like systems; you could try doing the same using MinGW. Cygwin, as far as I know, does not support 64-bit programs at all. I have no idea if it would be possible to compile using MSVC or some other compiler. |
|
Back to top |
|
|
Skittle
Newbie
Joined: 11 Aug 2009 Posts: 23
|
Posted: 15 Aug 2009 07:28:33 pm Post subject: |
|
|
Ok, ill try using MSVC later |
|
Back to top |
|
|
FloppusMaximus
Advanced Member
Joined: 22 Aug 2008 Posts: 472
|
Posted: 16 Aug 2009 12:40:19 pm Post subject: |
|
|
BOINC currently "has no jobs available". Taking 35M-36M. |
|
Back to top |
|
|
Skittle
Newbie
Joined: 11 Aug 2009 Posts: 23
|
|
Back to top |
|
|
critor
Member
Joined: 04 Feb 2009 Posts: 132
|
Posted: 16 Aug 2009 04:50:33 pm Post subject: |
|
|
Are we going to take care of the TI-73 system key too? |
|
Back to top |
|
|
Skittle
Newbie
Joined: 11 Aug 2009 Posts: 23
|
Posted: 16 Aug 2009 07:55:56 pm Post subject: |
|
|
sieving 6m-8m |
|
Back to top |
|
|
FloppusMaximus
Advanced Member
Joined: 22 Aug 2008 Posts: 472
|
Posted: 16 Aug 2009 09:16:17 pm Post subject: |
|
|
OK, thanks.
critor: Eventually, I hope so. And the TI-73, 89 and 92+ FlashApp keys, as well. If you would like to work on that one, then set up GGNFS and/or Msieve, and start searching for polynomials. The factMsieve script seems to do a decent job in this regard; Msieve is supposedly better, but it's never worked for me. |
|
Back to top |
|
|
Skittle
Newbie
Joined: 11 Aug 2009 Posts: 23
|
Posted: 16 Aug 2009 09:39:44 pm Post subject: |
|
|
Ive gotten the non experimental siever to compile with 64bit in VS2008, so thats a good first step i suppose. |
|
Back to top |
|
|
FloppusMaximus
Advanced Member
Joined: 22 Aug 2008 Posts: 472
|
Posted: 17 Aug 2009 01:31:09 pm Post subject: |
|
|
That's a good first step, but the key point of the exercise is the 64-bit assembly code from the experimental version. :)
Critor: Let me amend my previous statement. bobti89 has a set of polynomials for most of the keys of interest at this point: http://j.petitjean.club.fr/bobti89/tikey/tikey.zip . Unfortunately, this zip file contains two different polynomials for each of the 01 and 08 keys (!)
At this point, please do not perform any sieving with 01.poly (the polynomial found by fullmetalcoder.) Use keyt92p.poly instead (this is the polynomial that is currently being used by the BOINC project.)
In the case of the V200, fullmetalcoder's polynomial has a substantially better score than the one we have been using. We should run some tests to determine if switching to the better polynomial would be worth throwing away all the work that has been done so far on the V200 key.
This is a great example of why we shouldn't try to work on all the keys at once. We're bound to make some mistakes, and we're better off if we can learn from those mistakes and do better next time. :)
I don't believe anyone has started on the 0103 key (app signing key for TI-89) yet.
Last edited by Guest on 17 Aug 2009 03:18:11 pm; edited 1 time in total |
|
Back to top |
|
|
Skittle
Newbie
Joined: 11 Aug 2009 Posts: 23
|
Posted: 17 Aug 2009 04:24:02 pm Post subject: |
|
|
FloppusMaximus wrote: That's a good first step, but the key point of the exercise is the 64-bit assembly code from the experimental version.
Yes and from talking with jeff ghilcrest who provides the current windows binaries... says im SOL compiling for VS2008 |
|
Back to top |
|
|
FloppusMaximus
Advanced Member
Joined: 22 Aug 2008 Posts: 472
|
Posted: 17 Aug 2009 04:41:01 pm Post subject: |
|
|
Hmph. What about MinGW? From a bit of Googling, it sounds like there is some experimental support for 64-bit programs. |
|
Back to top |
|
|
Skittle
Newbie
Joined: 11 Aug 2009 Posts: 23
|
Posted: 17 Aug 2009 05:02:57 pm Post subject: |
|
|
funny that you mention that... im in the IRC channel now!
(at this point it might be a bit over my head... but ill look at it anyway!)
Quote: <xenofears> absolutely
<xenofears> gmp is required to build a mingw.org or mingw-w64 compiler toolchain (required by gcc)
<xenofears> and gmp is buildable to mingw-w64 (to a 64-bit target, or to no target generic)
thats excellent news to hear
Last edited by Guest on 17 Aug 2009 05:26:17 pm; edited 1 time in total |
|
Back to top |
|
|
Graphmastur
Advanced Member
Joined: 25 Mar 2009 Posts: 360
|
Posted: 17 Aug 2009 05:26:10 pm Post subject: |
|
|
who knew factoring numbers could beso hard. LOL! |
|
Back to top |
|
|
Skittle
Newbie
Joined: 11 Aug 2009 Posts: 23
|
|
Back to top |
|
|
fullmetalcoder
Member
Joined: 01 Aug 2009 Posts: 139
|
Posted: 17 Aug 2009 05:47:00 pm Post subject: |
|
|
10M-13M will be finished by the end of tomorrow (~85% done so far). It'll yield about 4M relations (3.5M so far) |
|
Back to top |
|
|
|