Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.programming.threads > #4974 > unrolled thread

Read again, i correct a last typo

Back to article view | Back to comp.programming.threads

Contents

  Read again, i correct a last typo Horizon68 <horizon@horizon.com> - 2019-03-28 11:32 -0700

#4974 — Read again, i correct a last typo

Hello...


Read again, i correct a last typo:

More explanation about scalability..

As you have noticed i said below the following:

"The scalability of my Parallel Sort Library in the mergesort mode will 
go to around 12X"

You have to distinguish between the scalability that is the combination 
of scalability of hardware and software, and the scalability that is 
only software or only hardware.

So you have to know that i am speaking below about the "total" 
scalability of both the hardware(from one channel memory to 8 channels
memory) and software on actual Intel and AMD processors, but 3D stacking
of memory will give much more scalability.


Read again:

About Multi-channel memory architecture and my Parallel Sort Library..

On actual Intel and AMD processors data is moved from and to main memory 
only in 64 bytes chunks, so in a 8 channels of a Multi-channel memory 
architecture in a ganged mode, i think that the memory speed of
my Parallel Sort library in the mergesort mode will go to 8X speed,
and the actual scalability of my Parallel Sort Library in the mergesort 
mode will go to around 12X. So i will advice you to use the ganged mode 
of the Multi-channel memory architecture because it works great.

My Parallel Sort Library was updated to version 3.64

I have enhanced it more, and i think it is stable and fast
and it scales more.

You can download it from:

https://sites.google.com/site/scalable68/parallel-sort-library


Thank you,
Amine Moulay Ramdane.

[toc] | [standalone]

Back to top | Article view | comp.programming.threads

csiph-web