My Initial ProtoMol Findings

[Tobit]

Well, this public beta release of these new ProtoMol based cores have sure been interesting. I've folded enough and have read enough comments to make an assessment of things from where I sit here. Hopefully, the proper ears will listen in and take note.

• All my hardware here is Intel based. I have had no issues whatsoever sans a warning message that appears to be cosmetic in nature. Many of the projects do not complete to 100% but this is expected and not an issue to me. However, this is an issue to many not accustomed to this.
  
  I'm not entirely sure if these are weighted properly. p10000's are worth 84 credits and p10001's are worth 50 credits. However, these p10001's are taking much longer to process than the p10000's that are actually worth more. This is on a 2.66GHz Core2Duo, one instance of the classic client running. p10000's finish much faster than the p10001's.
  
  The majority of problems I am seeing from others appear to be those with AMD processors. Maybe this should have remained in private beta a bit longer if such blatant AMD compatibility issues were known to be expected.
  
  No offense to toTOW and Co. but why are we hearing about these from a third party first instead of Stanford? Some type of warning ahead of time might have been nice.

I dunno. Stanford is always damned if they do, damned if they don't in situations like this. I'm not casting stones, just trying to constructively make observations to what I am seeing here in the real world. Again, I have no real complaints. Everything is working fine here for the most part.

[uncle fuzzy]

Yes, these usually start with an announcement, but PG has some new members who are still feeling their way into how we, on the outside, expect things to proceed. They'll get there, but there will be a few bumps in the road.

[Pette Broad]

I'm not entirely sure if these are weighted properly. p10000's are worth 84 credits and p10001's are worth 50 credits. However, these p10001's are taking much longer to process than the p10000's that are actually worth more. This is on a 2.66GHz Core2Duo, one instance of the classic client running. p10000's finish much faster than the p10001's.

The majority of problems I am seeing from others appear to be those with AMD processors.

I would agree with the first statement entirely, I'm seeing the same thing.

I'm running about 25 CPU cores, some Intel some AMD and have had problems with both. The problems I'm seeing can't be linked to AMD or Intel just simply bad WU's. What I find interesting is that I can't see that much difference in ppd, Intels on average a bit better but nothing like the difference I expected. Also, and I guess this might just be the way my folders are setup, I'm getting wide variations on ppd from unit to unit, probably as big as anything I've seen. For example on P10000 on my 6000+'s I'm seeing 233ppd and 377ppd on the same machine and 332ppd on another.

Fewer problems today, in fact none at all....


Pete

[7im]

News post, mid-day yesterday, stanford time... Release of new Protomol (Core B4) WUs