cmsearch 0.11 win32/64

[yoyo]

no more multiple starts of the screensaver

[MIZ]

errors FYI:

Code: Alles auswählen

19.3.2010 19:11:38	RNA World	[error] cms_sRib4[e]_Equus-caballus-(horse)_CM000403.lin.EMBL_Hairpin-s_1268999223_892_0: negative FLOPs left -1038306038635.231000
19.3.2010 19:11:48	RNA World	[error] cms_sRib4[e]_Equus-caballus-(horse)_CM000403.lin.EMBL_Hairpin-s_1268999223_892_0: negative FLOPs left -1061406354440.852800
19.3.2010 19:12:49	RNA World	[error] cms_sRib4[e]_Equus-caballus-(horse)_CM000403.lin.EMBL_Hairpin-s_1268999223_892_0: negative FLOPs left -1191635191629.483600
19.3.2010 19:12:59	RNA World	[error] cms_sRib4[e]_Equus-caballus-(horse)_CM000403.lin.EMBL_Hairpin-s_1268999223_892_0: negative FLOPs left -1212445687547.972200
19.3.2010 19:13:25	RNA World	[error] cms_sRib4[e]_Equus-caballus-(horse)_CM000403.lin.EMBL_Hairpin-s_1268999223_892_0: negative FLOPs left -1268891662174.235100
19.3.2010 19:13:55	RNA World	[error] cms_sRib4[e]_Equus-caballus-(horse)_CM000403.lin.EMBL_Hairpin-s_1268999223_892_0: negative FLOPs left -1332988871801.453100
19.3.2010 19:13:55	RNA World	[error] cms_sRib4[e]_Gallus-gallus-(chicken)_CM000111.lin.EMBL_CPEB3_ribozyme-s_1268999223_427_0: negative FLOPs left -551820160688.174190
19.3.2010 19:14:03	RNA World	[error] cms_sRib4[e]_Equus-caballus-(horse)_CM000403.lin.EMBL_Hairpin-s_1268999223_892_0: negative FLOPs left -1351476231139.555700
19.3.2010 19:14:03	RNA World	[error] cms_sRib4[e]_Gallus-gallus-(chicken)_CM000111.lin.EMBL_CPEB3_ribozyme-s_1268999223_427_0: negative FLOPs left -570307520026.276610
19.3.2010 19:14:05	RNA World	[error] cms_sRib4[e]_Equus-caballus-(horse)_CM000403.lin.EMBL_Hairpin-s_1268999223_892_0: negative FLOPs left -1357153035606.615200
19.3.2010 19:14:05	RNA World	[error] cms_sRib4[e]_Gallus-gallus-(chicken)_CM000111.lin.EMBL_CPEB3_ribozyme-s_1268999223_427_0: negative FLOPs left -575984324493.336300

Vista 64b, 4 GB RAM, BM 6.10.29 64b

[Ananas]

Not an RNAworld error, you find reports about this in several projects.

It seems to be something fairly new (later than v6.5) in the BOINC client scheduler, when it simulates the runtime of all jobs that are left to crunch (I guess).

The message in the code looks like this :

Code: Alles auswählen

                // can be slightly less than 0 due to roundoff
                //
                if (rp->rrsim_flops_left < -1) {
                    msg_printf(rp->project, MSG_INTERNAL_ERROR,
                        "%s: negative FLOPs left %f", rp->name, rp->rrsim_flops_left
                    );
                }

"internal error" (trying to make a bug sound nicer), might have been added when they did something towards GPU support.

I haven't found an explanation what it exactly means and how it can be fixed - maybe play with your cache settings a bit - if you have a huge cache, reduce it a bit, if you have a tiny one, increase it. Another thing I can imagine would be this XML file with the GPU settings for your host, if that one has values that are not plausible, it sure can influence the simulation.

This simulation looks very different to the old version, so it might still have some flaws.


The best place to ask what to do is probably http://boinc.berkeley.edu/dev , someone there might be familiar with that source module (rr_sim.cpp is the module name).


p.s.: ... or a BOINC upgrade?

[MIZ]

Now I'm running the last beta version BOINC Manager 6.10.43 64b and WUs cmsearch 0.13 on my two machines with Vista 64b. I need some more time for watch them... A few units ended [at 100% done] with windows error message "app. crashed", but there was no error message in BM messages tab... we will see...
BTW Newer units are bigger. Now I'm downloading one with 73 MB and one with 179 MB (gray-short-tailed-opossum).