Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 27 Jun 2013 18:50:20 +0400
From: Solar Designer <>
Subject: Re: Daniel's weekly report #2

On Mon, Jun 24, 2013 at 03:26:47PM +0200, D?niel Bali wrote:
> Accomplishments for last week:
> 0. Finally added the longer description for my project on google-melange

I've just reviewed it.  Looks mostly OK to me.  This is an ambitious
project, which means it's somewhat likely to end up achieving only a
subset of its stated goals - but it's fine to try anyway!

For others to read:

> 1. I worked on the elf patching utility and learned a lot about the elf
> format/libelf
> (It doesn't work on OpenCL binaries yet, but I have more important tasks
> for now)


> 2. Removed the code duplication from the keccak256 format with the help of
> Lukas

Is it working reliably, at least on both GPUs in bull?

> 3. Started working on the assembler/reading documentation


> Priorities:
> 1. Get the assembler working on a small subset of op codes
> 2. Try patching the "dummy kernel" that contains NOPs with working
> microcode that changes the behavior of the kernel.
> 3. Try to figure out how the kernel's initialization works
> 4. Expand the subset of supported instructions

How much progress at these do you expect to make this week?

How are you going to produce that dummy kernel specifically with NOPs?
Request some insane code alignment via an OpenCL directive maybe?  Or
generate a kernel with other than NOPs (with some dummy non-NOP code)
and overwrite its code with NOPs using your would-be assembler?


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.