Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > comp.programming > #4625 > unrolled thread
| Started by | Ramine <ramine@1.1> |
|---|---|
| First post | 2014-06-16 20:46 -0700 |
| Last post | 2014-06-16 21:34 -0700 |
| Articles | 2 — 1 participant |
Back to article view | Back to comp.programming
More about scalability... Ramine <ramine@1.1> - 2014-06-16 20:46 -0700
Re: More about scalability... Ramine <ramine@1.1> - 2014-06-16 21:34 -0700
| From | Ramine <ramine@1.1> |
|---|---|
| Date | 2014-06-16 20:46 -0700 |
| Subject | More about scalability... |
| Message-ID | <lno353$oo5$1@dont-email.me> |
Hello, You have seen me doing some scalability projections and calculations for my parallel Dual Simple system solver, and i have come to the conclusion that on DDR3 memory systems you will get around 4X scalability for larger problems, but i have forgot to tell you that the future stacked memory at 1TB/s will allow my parallel Dual Simplex to scale much more than 4X, cause the serial part that constitutes the memory transfers of my parallel algorithm on stacked memory will become more and more smaller compared to the parallel part that constitutes the Divive and Multiply and Add asm instructions on two doubles of my parallel algorithm , and that will get more and more exciting on stacked memory, so i think that i will write my parallel Dual Simplex algorithm. Read about the stacked memory here: http://www.theinquirer.net/inquirer/news/2256401/nvidia-says-voltas-stacked-dram-will-increase-memory-capacity Thank you, Amine Moulay Ramdane.
[toc] | [next] | [standalone]
| From | Ramine <ramine@1.1> |
|---|---|
| Date | 2014-06-16 21:34 -0700 |
| Message-ID | <lno5v7$7c6$1@dont-email.me> |
| In reply to | #4625 |
On 6/16/2014 8:46 PM, Ramine wrote: > Hello, > > > You have seen me doing some scalability projections and > calculations for my parallel Dual Simple system solver, I mean here parallel Dual Simplex solver, not parallel Dual simple solver. > and i have come to the conclusion that on DDR3 memory systems > you will get around 4X scalability for larger problems, > but i have forgot to tell you that the future stacked memory > at 1TB/s will allow my parallel Dual Simplex to scale much more than 4X, > cause the serial part that constitutes the memory transfers of > my parallel algorithm on stacked memory will become more and more > smaller compared to the parallel part that constitutes the Divive and > Multiply and Add asm instructions on two doubles of my parallel > algorithm , and that will get more and more exciting on stacked memory, > so i think that i will write my parallel Dual Simplex algorithm. > > Read about the stacked memory here: > > http://www.theinquirer.net/inquirer/news/2256401/nvidia-says-voltas-stacked-dram-will-increase-memory-capacity > > > > > Thank you, > Amine Moulay Ramdane. > > > >
[toc] | [prev] | [standalone]
Back to top | Article view | comp.programming
csiph-web