Page 2 of 2 1 2
Reply
Thread Tools
Joorin is offline Joorin
2011-06-26 , 17:09
Posts: 726 | Thanked: 345 times | Joined on Apr 2010 @ Sweden
#11
Originally Posted by epitaph View Post
Can you elaborate what kind of complicated calculations? I can compile Linux with -O3 and -fastmath with no problems.
A mix of signal analyzis and image manipulation.

Being able to compile some other binary with -O3 is, really, irrelevant since it worked fine to compile the binary I was referring to but it didn't behave in the correct way.

Have you verified that all the code in your kernel generated the expected binary?
 
Joorin is offline Joorin
2011-06-26 , 17:59
Posts: 726 | Thanked: 345 times | Joined on Apr 2010 @ Sweden
#12
Originally Posted by epitaph View Post
I don't know much about signal analyzizis (fft?) but most likely image manipulation is bit shifting (hilbert-curve) and such isn't very complicated.
The end result of the bit mashing and calculations is a combination of image recognition and some path prediction with heuristical algorithms on top.

I have no idea about the comparative complexity but it was enough to confuse g++ when using -O3.

I didnt't verify each binary
That was more of a joke. Most people assume that the compiler won't mess up in this way and it's often the right assumption.

and also I'm used to compile my kernel with ICC a compiler with more power then gcc (and more money). You can download our other work at www.linuxdna.com.

BTW. I have ICC linux license to sell. Half the prize?
I'm not involved in purchasing the tools used. But I'll suggest ICC to the developers so that they might have a look at it. A quick look at Wikipedia suggests good results for parallel computation.
 
Reply
Page 2 of 2 1 2

« Previous Thread | Next Thread »
Thread Tools

 
Forum Jump


All times are GMT. The time now is 17:11.

Contact Us - Home - Archive - Top