Re: [Help-glpk] GLPK re-endrant

[xypron]

Hello Andrew,

my understanding is that the ILOG claim of >= 1.7 times faster for
multithreading
is on multicore CPUs. I suppose the majority of PCs nowadays has 2 to 4
cores.

A much more radical possibility for parallelization is offered by using the 
Graphic processor of the computer with its single instruction multiple data
capabilities.

For libraries see
http://www.khronos.org/opencl/
http://www.nvidia.com/object/cuda_home.html

For implementations see:
http://www.jakobbieling.de/university/diploma/thesis.pdf
http://ws3.re.mipt.ru/mediawiki/index.php/Simplex_Method

The idea here was to handle  identification of the next column/row to enter
the base using the CPU and the matrix multiplications using the GPU.

Best regards

Xypron



Andrew Makhorin wrote:
> 
>> I see two fields of applications for threading in GLPK.
> 
>> ILOG describes that many MIPs can be solved >= 1.7 times
>> faster using multiple threads and shared memory.
>> http://www.ilog.com/optimization/the-right-hand-side/1/TA_Parallel_CPLEX_Dong.html
>> http://www.ilog.com/optimization/the-right-hand-side/1/TA_Parallel_CPLEX_Dong.html
> 
> Probably if a machine has more than one cpu. However, it is a brute
> force approach. (It would help if the machine has 2^n processors, where
> n is the number of binary variables :) Try to solve gesa2 or gesa3
> from miplib without and with mir cuts. That what's I mean.
> 
> 

-- 
View this message in context: 
http://www.nabble.com/GLPK-re-endrant-tp24256046p24327716.html
Sent from the Gnu - GLPK - Help mailing list archive at Nabble.com.