Depending on how radical a rework is it, the optimization plan once the base app is declared bug free is just to prove akos's apps are functionally identical and to just drop the new hot loops in.
realized I screwed this up while rephrasing.
Depending on how radical a rework is it, the optimization plan once the base app is declared bug free *MIGHT* just to prove akos's apps are functionally identical and to just drop the new hot loops in.
If there is a concern about validity of results, are akos' results rejected out of hand?
I always figured this was a relative non-issue because of cross-checking among other work units.
Crosschecking won't work to catch systematic bugs (as opposed to random hardware failure) if the two results are computed by the same client app! Any app (official or unoffical) that has a bug but is so widespread that most users will use it will defeat the validation checks because the results will all be wrong, but equal.
Validation can not fully replace a certain quality assurance and certification process before an app is distributed to the entire user community, I guess.
Well, Akos is in a league of his own. It's all for the benefit of the project, so don't feel negative about about it.
CU
Bikeman
Well I do feel negative about it.
yeah, yeah, yeah I know akosf is doing much for the project but the stats are broken and it's rather "funny" that one person can run his own software on this project that I always thought was serious.
Well I do feel negative about it.
yeah, yeah, yeah I know akosf is doing much for the project but the stats are broken and it's rather "funny" that one person can run his own software on this project that I always thought was serious.
Stats are important for any DC project.
If Akosf can not test the software, then how can we get better optimization? Akosf is designing optimizations for the project to use and will eventually come down the pipeline, when they see that the data is compatible.
With Akosf being the Chuck Yeager of E@H its a good thing he is crunching away openly, i for one like to watch his apps pushing the envelope of optimization. And its not like he is getting those credits for free, i wouldnt think for a second that this stuff is easy or was done in a few hours time.
RE: Depending on how
)
realized I screwed this up while rephrasing.
Depending on how radical a rework is it, the optimization plan once the base app is declared bug free *MIGHT* just to prove akos's apps are functionally identical and to just drop the new hot loops in.
If there is a concern about
)
If there is a concern about validity of results, are akos' results rejected out of hand?
I always figured this was a relative non-issue because of cross-checking among other work units.
RE: If there is a concern
)
Crosschecking won't work to catch systematic bugs (as opposed to random hardware failure) if the two results are computed by the same client app! Any app (official or unoffical) that has a bug but is so widespread that most users will use it will defeat the validation checks because the results will all be wrong, but equal.
Validation can not fully replace a certain quality assurance and certification process before an app is distributed to the entire user community, I guess.
CU
Bikeman
I am just beeing overrun by
)
I am just beeing overrun by akosf in the stats and it doesn't feel like a fair gaim.
No sport at all :(
Me restarted GIMPS and Milkyway
RE: I am just beeing
)
Well, Akos is in a league of his own. It's all for the benefit of the project, so don't feel negative about about it.
CU
Bikeman
RE: Well, Akos is in a
)
Well I do feel negative about it.
yeah, yeah, yeah I know akosf is doing much for the project but the stats are broken and it's rather "funny" that one person can run his own software on this project that I always thought was serious.
Stats are important for any DC project.
RE: Well I do feel negative
)
If Akosf can not test the software, then how can we get better optimization? Akosf is designing optimizations for the project to use and will eventually come down the pipeline, when they see that the data is compatible.
RE: it's rather "funny"
)
He actually works for E@H as a consultant! :-)
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
Hehe, yeah I know that but
)
Hehe, yeah I know that but tests are usually done indoors.
With Akosf being the Chuck
)
With Akosf being the Chuck Yeager of E@H its a good thing he is crunching away openly, i for one like to watch his apps pushing the envelope of optimization. And its not like he is getting those credits for free, i wouldnt think for a second that this stuff is easy or was done in a few hours time.
Team Philippines