Lew Pitcher
1/11/2015 9:36:00 PM
On Sunday January 11 2015 19:11, in comp.programming, "Ramine" <ramine@1.1>
wrote:
>
> Hello,
>
>
> As you have noticed
Nope. I hadn't noticed.
> i have implemented a Parallel Conjugate gradient solver library that is
> NUMA-aware and cache-aware
Congratulations. So what?
> ... and you have to know
Why do I have to know?
> that my parallel library is scalable on NUMA architecture...
OK, but again, so what?
> I have just done a simulation cause i am using a probabilistic
> mechanism to render my algorithm NUMA-aware,
BECAUSE: you are using a probabilistic mechanism to render your algorithm
NUMA-aware
THEN: you have done a simulation
> and this probabilistic mechanism is very efficient cause i have just done
> a simulation
Cause: you have done a simulation
Effect: the probabilistic mechanism is very efficient
Ergo: the probabilistic mechanism is very efficient BECAUSE you are using a
probabilistic mechanism to render your algorithm NUMA-aware
Tautology?
> and my probabilistic mechanism is giving a very good results
As defined against what standard or test?
> and my very efficient probabilistic mechanism
How did you measure your efficiency?
> have also rendered my parallel algorithm into a scalable parallel
> algorithm on NUMA architecture.
How so?
> You can download my Parallel Conjugate gradient solver library from:
[snip]
Thanks, but no thanks
--
Lew Pitcher
"In Skills, We Trust"
PGP public key available upon request