| 1 |
Am 17.08.2012 19:57, schrieb Jorge Almeida: |
| 2 |
> On Fri, Aug 17, 2012 at 2:53 PM, Florian Philipp <lists@×××××××××××.net> wrote: |
| 3 |
>> Am 17.08.2012 10:58, schrieb Jorge Almeida: |
| 4 |
>>> |
| 5 |
>>> 1) Is this strategy right? If so, any other flags to add? (or any |
| 6 |
>>> flags to remove from the list?) |
| 7 |
>>> |
| 8 |
>>> 2) The --param flags are the ones of the computer that will do the |
| 9 |
>>> compiling. I'm guessing the produced binaries are compatible with cpu |
| 10 |
>>> with different --param flags. Is this right? |
| 11 |
>>> |
| 12 |
>>> TIA |
| 13 |
>>> |
| 14 |
>>> Jorge Almeida |
| 15 |
>>> |
| 16 |
>> |
| 17 |
>> 1) Yes. But as you can see, -march=prescott is basically a subset of |
| 18 |
>> atom. In fact, before there was a -march=atom option, prescott was the |
| 19 |
>> best flag for atoms. I think you can avoid some hassle by simply |
| 20 |
>> enabling "-march=prescott --param l1-cache-line-size=64 --param |
| 21 |
>> l2-cache-size=512". |
| 22 |
>> |
| 23 |
>> 2) Yes, the param flags do not affect compatibility. Using the lower |
| 24 |
>> value will probably be better but this is just an educated guess. |
| 25 |
>> |
| 26 |
> |
| 27 |
> What about: |
| 28 |
> |
| 29 |
> CFLAGS="-O2 -pipe -fomit-frame-pointer -march=prescott -mtune=atom --param |
| 30 |
> l1-cache-size=16 --param l1-cache-line-size=64 --param l2-cache-size=1024" |
| 31 |
> |
| 32 |
> If prescott were exactly a subset of atom, this would yield the best of both |
| 33 |
> worlds. Can it still be safe? |
| 34 |
> |
| 35 |
> I read in http://forums.gentoo.org/viewtopic-t-895104.html : |
| 36 |
> atom |
| 37 |
> Intel Atom CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3 |
| 38 |
> instruction set support. |
| 39 |
> |
| 40 |
> Does this mean that these flags are pulled by -mtune=atom, or do we need to |
| 41 |
> ask for them explicitly? The WiKi shows how to find which flags are pulled by |
| 42 |
> -march=native, but not the other cases. |
| 43 |
> |
| 44 |
> Thanks |
| 45 |
> |
| 46 |
> Jorge Almeida |
| 47 |
> |
| 48 |
|
| 49 |
Oh, sorry, I didn't think of the second generation atoms. I guess your |
| 50 |
newer atom is a bit more different from prescott than the one I talked |
| 51 |
about. Anyway, using -march=prescott is still viable. It just means you |
| 52 |
lose a bit more in terms of usable SSE extensions on your atom. |
| 53 |
|
| 54 |
Your CFLAGS look good to me. They won't pull in anything that your |
| 55 |
prescott cannot handle. |
| 56 |
|
| 57 |
Regards, |
| 58 |
Florian Philipp |
Archives