| 1 |
On Tue, 21 Sep 2010 13:56:49 +0200 |
| 2 |
Claes Gyllenswärd <letharion@×××××.com> wrote: |
| 3 |
|
| 4 |
> |
| 5 |
> Regardless, I add that when graphite was still the coolest things |
| 6 |
> since sliced bread (when 4.4 first came out, IIRC), I tried it out |
| 7 |
> locally. |
| 8 |
> |
| 9 |
|
| 10 |
Well for me it is only partly a matter of being "cool." I do a fair |
| 11 |
amount of image processing and anything that offers the possibility |
| 12 |
of more speed would be welcome. Also, I burn a lot of DVDs and for |
| 13 |
data protection I usually include some sort of data-recovery capability. |
| 14 |
Producing par2 archives, for example, is a very slow process and more |
| 15 |
speed or efficiency produced by the compiler would be welcome for this |
| 16 |
task as well. |
| 17 |
|
| 18 |
From time to time I get the urge to write my own routines in pure |
| 19 |
assembly language. Hand-crafted assembly code would be much faster |
| 20 |
than compiled code. But I usually quickly come to my senses -- it |
| 21 |
would be an enormous job. For now, the gcc compiler is the only option. |
| 22 |
|
| 23 |
I also thought trying the Intel ICC compiler, but no version is |
| 24 |
available for a pure 64-bit system (i.e. no multilib). It seems |
| 25 |
that those highly paid Intel programmers, even after many years, |
| 26 |
can't produce a pure 64-bit Linux version. |
| 27 |
|
| 28 |
Frank Peters |
Archives