| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Help Cure Muscular Dystrophy - Phase 2 Forum
Thread: Continual error 193 |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 42
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The best thing to do for now is if u really want to crunch HCDM2 on P3 an older machines, just use windows instead of linux.
I tried HCDM2 the other day on my old 1.3 tualatin celeron and so far have 1 PV ,7 valid results and 0 errors. |
||
|
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1671 Status: Offline Project Badges: 
|
Hello,
----------------------------------------Thank you Uplinger for your precise answers and comments. Indeed "compiler science" is not so simple as many people could suppose it would be. Especially since the last 15 years, looking for efficiency stays in many cases fully in contradiction with the compliance need with so many processors as possible. Additionally, because all WCG projects are not using the same programming language, the number of combination between processor, operating system, language and finally boinc version grows dramatically. Regarding the mentioned problem, I can add one "killing" question: could the problem be caused by a processor bug ? We know that some steps of P2 and P3 processor are buggy (probably P4 and Core processors too). Because Linux uses a processor differently than Windows, maybe a processor bug can cause this particular problem (especially by using SSEx commands) ! ... Cheers, Yves |
||
|
|
TXR13
Cruncher Canada Joined: Dec 5, 2005 Post Count: 36 Status: Offline Project Badges: 
|
The best thing to do for now is if u really want to crunch HCDM2 on P3 an older machines, just use windows instead of linux. You do realize, Haris, that some of us don't have Windows licenses for the machines we use, yes? Or that the Windows licenses are stupidly old, such as the P3-800 I have that only came licensed for NT4 Workstation. Using Linux is sometimes less a matter of choice and sometimes just a matter of economic reality. It's not as simple as "just use Windows instead of Linux."  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
True. Given that the only option for now is to run anything else beside HCMD2 on these old machines.
----------------------------------------Question from me though: How can it be the windows version of the science does work on a P3, and the linux version doesn't. I understand they both use different compilers, but if SSE2 functions are needed to achieve acceptable performance under Linux, I assume these would also be needed under Windows... Anyway I'm sure the techs who understand this stuff a lot better than are breaking their brains about this as well, so lets hope they figure out how to tweak the linux compiler so it will produce P3 compatible software which is as quick as its windows counterpart  [Edit 2 times, last edit by Former Member at May 20, 2009 6:33:42 PM] |
||
|
|
Dadinck
Cruncher Joined: May 14, 2009 Post Count: 6 Status: Offline Project Badges: 
|
Anyway I'm sure the techs who understand this stuff a lot better than are breaking their brains about this as well, so lets hope they figure out how to tweak the linux compiler so it will produce P3 compatible software which is as quick as its windows counterpart That is the problem. I remeber reading in this thread that Fortran is still used. The intel fortran compiler (windows) is superior to then g77 (gnu fortran) compiler (linux). I had experience with fortran models on both comilers, and some models on g77 did not converge when they converged for the Intel fortran.   |
||
|
|
TXR13
Cruncher Canada Joined: Dec 5, 2005 Post Count: 36 Status: Offline Project Badges:
|
Good news!
----------------------------------------The new version of HCMD2 (6.13) now in Beta testing seems to fix the error 193 issue. At least, I haven't had that error pop up on any of the Beta units my P3 Linux machines have seen. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
If MAXDo runs on AIX, then all we would need is an AIX BOINC client and I would have access to quite the array of additional machines. I haven't heard of one, but is there an AIX BOINC client?
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
World Community Grid does not support AIX.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The best thing to do for now is if u really want to crunch HCDM2 on P3 an older machines, just use windows instead of linux. This is the wrong way around, I'm afraid. "just use windows" is equivalent to "just throw the machine out". I have some useful machines (a file server, a firewall machine, etc) that run reliably for months headless (no mouse, no monitor) and I ssh into them occasionally. Their primary tasks are simple and take up only about 1% of their CPU time, but Windows would do the same tasks inefficiently and/or insecurely and/or unreliably. (Vista or Win7 might manage to be secure and reliable, but are so bloated as to not be feasible for a P3 with 128MB or 256MB of RAM.) Anyhow, as TXR13 mentioned, excellent to see the new betas working fine on P3 and Athlon machines. I hope the performance is okay and they end up in production really soon. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
A bit over a day ago, a new version which works on older CPUs was released into the world. Sadly, it is marking some WU results as invalid, though.
|
||
|
|
|