With the release of 1.20, is there any chance for us poor Linux users to get the latest client? There seem to have been quite a few nice fixes since 1.02, and it appears we are suffering with much older client.
Printable View
With the release of 1.20, is there any chance for us poor Linux users to get the latest client? There seem to have been quite a few nice fixes since 1.02, and it appears we are suffering with much older client.
Well, with the issues of the (now old) Windows version, these fixes of course had top priority (Windows machines do more than 90% of the PRP testing).Quote:
Originally posted by Exdaix
With the release of 1.20, is there any chance for us poor Linux users to get the latest client? There seem to have been quite a few nice fixes since 1.02, and it appears we are suffering with much older client.
Seeing Louie on a much more regular basis the last days (plus the fixed Windows client) indicates that study left him some time lately (maybe holiday season? ;)). A Linux version could be available soon, but I won't expect it within the next two weeks.
If I was Louie, I'd have other things to do right now. :|party|:
btw. thanks for that christmas gift, Louie! :cheers:
I have a new linux client compiled from v1.20 source.
I'm trying to fix alignment issues though. Windows knows better than to misalign SSE2 instructions but Linux doesn't.
If I can't fix it this afternoon, I may just release a non-SSE2 linux version that at least works. I think I fixed a couple of the other major bugs (ie the client now runs on machines that would seg fault on any previous version). I'll let you know when it's ready.
Cheers,
Louie
ok, so it's online now. go to the download page on the site to try it out.
please, everyone give feedback on how it works as quickly as possible. if there's something seriously wrong, i may have time to update it once more before i go out of town for a few weeks.
i can't wait to hear all the positive results :)
Cheers,
Louie
Works perfectly. This is much more fun!
Thanks!
Thank you very much Louie, we have needed this for a long time.
Could we have a version that only depends on glibc 2.2? I'm not the administrator of all the machines I use for SoB, so I can't just upgrade. Furthermore I have a couple of machines (which does GIMPS work at the moment, but I have consider changing that) for which I require the stability of Debian stable, which also has glibc 2.2.
That also means that I not able to comment on the speed.
The good news is that the new client doesn't segfault immediately on my P4 laptop, like 1.0.2 did. (But it's not always online so I do P-1 factoring on it).
It doesn't behave very nicely when killed:
.HenrikCode:grove@julia:17_or_bust% kill 3512
grove@julia:17_or_bust% caught signal 15. dumping unfreed memory:
unfreed ptr 0x822e048, size 36, in file client.c:349 at time 1072389002
[1] + exit 15 ./sb
grove@julia:17_or_bust%
Thank you very very much for the new client version!
Same with me:Quote:
Originally posted by hc_grove
It doesn't behave very nicely when killed:
This box is a P3 700 with Debian testing.Code:~/sb$ killall sb
~/sb$ caught signal 15. dumping unfreed memory:
unfreed ptr 0x822dd68, size 36, in file client.c:349 at time 1072390182
free(): invalid pointer 0x822dd90!
[1]+ Exit 15 ./sb sclient.conf
There seems to be a problem connecting to the server, too:
I'm using my old sclient.conf without any changes.Code:[Thu Dec 25 22:49:17 2003] iteration: 4690000/5242646 (89.46%) k = 55459 n = 5242630
[Thu Dec 25 22:57:31 2003] logging into server
[Thu Dec 25 22:57:32 2003] login successful
[Thu Dec 25 22:57:32 2003] n.high = 4693020 . 61 blocks left in test
[Thu Dec 25 23:16:32 2003] iteration: 4700000/5242646 (89.65%) k = 55459 n = 5242630
[Thu Dec 25 23:22:53 2003] client process [v1.2.1] invoked
[Thu Dec 25 23:22:53 2003] priority set to idle
[Thu Dec 25 23:22:53 2003] got k and n from cache
[Thu Dec 25 23:22:53 2003] Intel(R) Pentium(R) III processor detected. Enabling cpu specific optimizations.
[Thu Dec 25 23:22:54 2003] restarting proth test from cache (k=55459, n=5242630) [89.7%]
[Thu Dec 25 23:27:47 2003] resolving hostname
[Thu Dec 25 23:27:47 2003] opening connection
[Thu Dec 25 23:30:56 2003] temporarily unable to connect to server -- block added to submit queue
Greetings,
Othello
I get the memory dump, but not the server connection problem. The memory dump doesn't seem to affect SB at all.
Here is my output with a new config file and everything all setup, the only thing I did was copy my cache and such files to the new folder, so I would not have to get a new data packet. (I pressed Ctrl-C after the 4th output line, it didn't catch signal 2 by itself, I killed it manually.)
Code:exdaix@Exdaix sb $ ./sb
[Thu Dec 25 17:43:39 2003] client process [v1.2.1] invoked
[Thu Dec 25 17:43:39 2003] priority set to idle
[Thu Dec 25 17:43:39 2003] got k and n from cache
[Thu Dec 25 17:43:39 2003] AMD Athlon(tm) XP 1800+ detected. Enabling cpu specific optimizations.
caught signal 2. dumping unfreed memory:
unfreed ptr 0x822dd68, size 36, in file client.c:349 at time 1072392219
exdaix@Exdaix sb $
Same connect problem on all three boxes here.
Maybe I didn't update the right way?
That's what I did:
- killed the client
- moved sclient.conf to sclient.conf.backup
- untared the archive of 121
- moved sclient.conf.backup to sclient.conf
- started the client
Anything wrong?
Greetings,
Othello
After letting it run for a while, I am getting the same network connection error othello is.
I'm going to try putting in the server's IP address instead or the DNS name in the config file.
EDIT - The IP hack didn't work, it still won't upload results.
SB v1.2.1 for linux with glibc 2.2 is [removed link because of bug]
Will a 1.2.2 release be put out to fix the server connection error?
I have the same communications problem with the new 1.2.1 client. I running Linux on a dual Athlon system. I upgraded to the new 1.2.1 client without changing the .conf file since the one in the tar file seemed similar. Both processes could not communicate. I switched one back to version 1.0.2 and that process communicates with no problem. So on a dual processor system, one processor running 1.2.1, will not communicate, while the other processor running 1.0.2 has no problem communicating. I've tried using the sb ip address and the sb.pns.net for the ServerAddress in the configuration file, and 1.2.1 will not communicate with either configuration.
Also, I noticed that the new client reports an AMD Athlon MP detected... but sometimes reports an AMD Athlon MP (2000+) detected. I'm curious why it doesn't consistently report one or the other.
Okay, yes, the linux v1.2.0 and .1 are broken, i've removed the download links for now until we can fix the problem and compile a new client.
I heard the Windows version has communication problems too.
As far as I know the windows version 1.2.0 is fine, except for the exit problem (but that's rare..)
Well let's keep our ears open for anyone else that cannot connect to the server with Windows v2.21.
Thank you.Quote:
This seems to have the network problem too, though, so I hope you'll make a version of 1.2.2, when that comes.