[MAGEEC] BEEB features analysis

Kerstin Eder Kerstin.Eder at bristol.ac.uk
Thu Aug 28 09:51:48 BST 2014


H Stoil,

These look interesting - thank you. We need to analyse/evaluate/discuss 
them. We need to agree on what we define to be a "good spread". ;)

Wrt the feature vector that is fed into the learning, do you also have a 
graph that shows the "population" of the feature vector for the 
benchmarks? (This might already exist, but I've not seen it as I've been 
away for most meetings.)

Best wishes,
Kerstin

On 27/08/2014 17:51, Stoil Ganev wrote:
> As some of you know I have been working on extracting data from the 
> benchmarks that could help us determine if they provide a good spread 
> of features.
>
> This graph shows us the spread of gimple statements in the gimple tree 
> right after the 'ssa' pass:
>
>
>
> This graph shows us the spread of basic blocks with different number 
> of edges leading from them:
>
>
>>
> This graph shows us the spread of instructions in the binary when the 
> benchmark has been compiled for atmega328p:
>
>
>> ​This graph shows us the the spread of instructions executed during 
> the runtime of the program. Note that some benchmarks are not present 
> because they didn't execute properly in the simulator I was using.:
>
>
>>
>
> _______________________________________________
> mageec mailing list
> mageec at mageec.org
> http://mageec.org/cgi-bin/mailman/listinfo/mageec

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mageec.org/pipermail/mageec/attachments/20140828/0a2b10b7/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/png
Size: 160182 bytes
Desc: not available
URL: <http://mageec.org/pipermail/mageec/attachments/20140828/0a2b10b7/attachment-0004.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/png
Size: 131687 bytes
Desc: not available
URL: <http://mageec.org/pipermail/mageec/attachments/20140828/0a2b10b7/attachment-0005.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/png
Size: 140489 bytes
Desc: not available
URL: <http://mageec.org/pipermail/mageec/attachments/20140828/0a2b10b7/attachment-0006.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/png
Size: 117907 bytes
Desc: not available
URL: <http://mageec.org/pipermail/mageec/attachments/20140828/0a2b10b7/attachment-0007.png>


More information about the mageec mailing list