Comments for Touch Dreams

Comment on Concurrency vs Parallelism, Concurrent Programming vs Parallel Programming by tukaram

tukaram — Sat, 01 May 2010 09:03:30 +0000

hey nice distiction between the two dude …!

great job…

keep it up..

Comment on CUDA Thread, Warp and SIMT by Kevin

Kevin — Sat, 10 Apr 2010 14:59:29 +0000

Thanks for such a great explanation of the concept of warps …

Though i have this doubt , that if i launch a warp of 32 threads , and out of them 30 follow branch A while only 2 follow branch B, is there some definite ordering , like first those 30 will be executed , and then the 2 , or vice versa , or is it something totally random ??

Thanx in advance for any help provided

Comment on Sun Studio does CUDA by David

David — Sat, 20 Mar 2010 22:26:13 +0000

Well, I feel a bit silly. Sorry about that.

Comment on Sun Studio does CUDA by Yuan

Yuan — Fri, 19 Mar 2010 04:41:59 +0000

I am sincerely sorry for any productivity loss. The trick to know the details is to know when this entry was posted or to click the link in the last paragraph.

The closest thing outside there is the PGI Accelerator. Seriously this time.

Comment on Sun Studio does CUDA by Satoshi

Satoshi — Wed, 17 Mar 2010 22:36:35 +0000

This sounds great. Could you please tell me more details about it. I tried it to tranform Fortran code to CUDE C by:

f95 filename -xopenmp=parallel -qoption iropt -Apcg:cuda41

But, it only does only compile the file. Thanks!

Comment on Sun Studio does CUDA by David

David — Tue, 16 Mar 2010 19:31:07 +0000

Could you point me towards the documentation on this option? A group of us are working on GPU processing for computational models in economics and it’d be a lot of help if you have more information on it.
Thanks very much
David

Comment on ThreadSanitizer by Kostya

Kostya — Thu, 18 Feb 2010 17:46:40 +0000

You are welcome to try the tool!

Comment on Understanding Data Races 3: Several currently available tools (2/3) by cwillis-PetraVM

cwillis-PetraVM — Fri, 06 Nov 2009 23:00:15 +0000

Hi Yuan,
You might appreciate a post we did on the difference between static analysis race detectors and simulation-based concurrency tools like Jinx. You can find the two part at article at http://petravm.com/blog

Comment on Common Mistake in Using OpenCL (2): Mis-aligned Vector Field by Yuan

Yuan — Wed, 30 Sep 2009 04:34:19 +0000

On most 64-bit systems, the size of C ‘int’ stays 32-bit. C ‘long’ type and pointer type become 64-bit. So you probably still need to pad the structure with 3 ‘int’s. The size of cl_float4 does not change with the host system. The header file defines the alignment requirement of cl_float4 if the host compiler supports alignment directive.

Comment on Common Mistake in Using OpenCL (2): Mis-aligned Vector Field by John

John — Tue, 29 Sep 2009 03:59:55 +0000

just following up.

manually padding for 32 bit architecture as you did with 3 integers is correct. So is it fair to say that for 64 bit architecture, manual padding would be 1 int? This is in consideration of the size of int changing, hence I could see the use of cl_int, cl_float4

Comment on CUDA Thread, Warp and SIMT by Anton

Anton — Sun, 17 May 2009 18:08:00 +0000

Thank you for such a great clarification of branches.

Now I know that if a condition of «if» statement evaluates to different values at a different threads within a warp, then execution time will be as both of branches are executed.

Comment on Thread Cleanup Handlers, exit(), _exit(), atexit(), and pthread_exit() by Niels de Vos

Niels de Vos — Mon, 11 May 2009 09:16:37 +0000

Nice writeup, thanks!

Comment on Thread Cleanup Handlers, exit(), _exit(), atexit(), and pthread_exit() by J.S.

J.S. — Sat, 25 Apr 2009 05:20:10 +0000

Thanks for explaining this! The description is pretty dense. I read it over 5 times before understanding every points you made. But the information is very helpful. The weird random crash problem in our application during shutdown makes much more sense to me now.

Comment on Extending the OpenMP profiling API for OpenMP 3.0 by OpenMP 3.0 Profiling: a paper and an implementation

OpenMP 3.0 Profiling: a paper and an implementation — Mon, 30 Mar 2009 00:08:01 +0000

[...] I wrote a paper describing the techniques used to do the OpenMP 3.0 performance profiling which was mentioned in my previous blog. The paper is going to be presented at this year’s [...]

Comment on Concurrency vs Parallelism, Concurrent Programming vs Parallel Programming by Pages tagged "concurrent"

Pages tagged "concurrent" — Wed, 18 Feb 2009 23:32:20 +0000

[...] bookmarks tagged concurrent Concurrency vs Parallelism, Concurrent Programming… saved by 2 others kimmi1234 bookmarked on 02/18/09 | [...]

Comment on Common Mistakes in Using OpenMP 1: Incorrect Directive Format by Yuan

Yuan — Fri, 06 Feb 2009 05:57:43 +0000

Sure. Of course!

Comment on Common Mistakes in Using OpenMP 1: Incorrect Directive Format by Slessyspoow

Slessyspoow — Fri, 06 Feb 2009 01:08:47 +0000

Hi. Your site displays incorrectly in Mozilla, but content excellent! Thank you for your wise words:)

Comment on Common Mistakes in Using OpenMP 1: Incorrect Directive Format by Timur Alhimenkov

Timur Alhimenkov — Wed, 28 Jan 2009 02:18:58 +0000

Great! Thank you!
I always wanted to write in my site something like that. Can I take part of your post to my site?
Of course, I will add backlink?

Sincerely, Timur I.

Comment on Maximum Automation for Mundane Tasks by Irongenusneme

Irongenusneme — Mon, 19 Jan 2009 21:00:25 +0000

I think you are thinking like sukrat, but I think you should cover the other side of the topic in the post too…

Comment on More on Concurrency vs Parallelism by Concurrency vs Parallelism, Concurrent Programming vs Parallel Programming

Concurrency vs Parallelism, Concurrent Programming vs Parallel Programming — Tue, 30 Dec 2008 18:39:51 +0000

[...] More on Concurrency vs Parallelism Posted by Yuan Parallel Programming Subscribe to RSS [...]