Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Date: Mon, 29 Jul 2013 17:01:33 +0200
From: Dániel Bali <balijanosdaniel@...il.com>
To: john-dev@...ts.openwall.com
Subject: Daniel's weekly report #7

Hello!

I'm back to work now. I'll reply to your mail in the 6th weekly report
thread here.

2013/7/24 Solar Designer <solar@...nwall.com>
>
>
> Please note that not only we evaluate you as a GSoC student, but also
> you need to evaluate us (your mentor and org).  I hope you will do that
> in time.
>

Yes, it opens today at 7 UTC (I think), I should be able to fill the form
today.


>
> > Accomplishments:
> > 1. Added support for user-defined formats
>
> What are these?  Any example?
>

Uhh, I meant user-defined aliases here. Don't know why I wrote "format", it
makes no sense. An example for aliases would be:

#define result v7
v_mov_b32 result, 2

Update on the DS format: realhet replied to my post on the AMD forums (
http://devgurus.amd.com/thread/167167).

I did get the DS format working using GDS, but the LDS size is still a
mystery to me. The code that realhet linked to (
http://www.multi2sim.org/svn/multi2sim/trunk/src/arch/southern-islands/asm/bin-file.h)
has a reference to ldsSize in the si_bin_compute_pgm_rsrc2_t struct. I have
to look at the code a bit further to find where the lds size is actually
stored.

By the way, this code is from Multi2Sim which is a CPU-GPU simulator. It
looks interesting.

Accomplishments:
0. None, I was away for the week

Priorities:
0. Fill the midterm evaluation form
1. Find out more about LDS size
2. Work on DS instructions and see if we can improve performance that way
3. Figure out an automated way to measure the performance gain/loss of
"optimized" kernels

Regards,
Daniel

Content of type "text/html" skipped

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.