[tremor] Ogg/Vorbis report, FFT optimizations

[Segher Boessenkool]

On 12-mrt-04, at 17:50, timmy brolin wrote:

>>> On x86 you'll get a speed-penalty for misaligned accesses.
>>
>> Not on the implementations I used...  But that is _long_ ago.
>
> Of course you get a speed penalty. Consider a misaligned 32bit read 
> from a memory with a 32bit databus. Two 32bit read cycles must be 
> performed from the memory in order to piece together the misaligned 
> 32bit word.

Except that you're not reading from memory.  And cache-line sizes
are more than 4 bytes ;-)  Yes, sometimes you get a hit, and a quite
big one too, if you're crossing memory pages, etc.

Let's stop on this subject?  It's not too useful to discuss here...

<p>Segher

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'tremor-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.