Flask e nuove cpu

Flask e nuove cpu

di pubblicata il , alle 08:56 nel canale Uncategorized
 
Anche BX Boards ha pubblicato un breve articolo, a questo indirizzo, con informazioni e benchmark sulle prestazioni che i nuovi processori Intel Pentium IV e AMD Athlon Thunderbird possono ottenere utilizzando versioni ottimizzate di Flask, il software di conversione MPEG-4 di filmati MPEG-2:
For those of you who do not know what Flask MPEG is, let me explain. Flask is a program created to encode files using the new MPEG-4 standard. This standard allows near-DVD quality movies to be created in less than 1GB of space. Previous compression techniques, such as MPEG-1, were able to compress the data sufficiently to fit on a single CD but the quality and resolution were very poor. MPEG-4 is a much more CPU intensive encoding algorithm which expands on the MPEG-2 standard used for DVDs while allowing much higher compression rates. Its one major downfall is the CPU requirements. MPEG-4 encoding is very processor intensive but this makes it a perfect subject for benchmarking. Since a rather marginal 5% improvement in performance can mean HOURS less encoding time, the results from this test have very significant real-world implications.

While I don't think I'll ever heard the end of why Flask might not be the best MPEG encoder on the market, the fact is that it is the most popular MPEG-4 encoder available. Since it is available via the GPL, open source license, the source code is freely available as well which allows anyone the ability to modify and try to improve the program.

Back to the optimizations at hand. In an effort to fairly compare the Pentium 4 with other CPUs, the code must be heavily optimized for SSE2. After this optimization, it is only fair that the competitors also have their optimizations made available. Thus, after Intel submited their optimized binaries, some AMD fans decided to do the same. After only about an hour of coding, they had produced a program that revealed almost triple the performance of the original un-optimized binaries. Merely compiling with a new compiler resulted in a 133% gain and additional 3dNow! optimizations resulted in an additional 10%.
0 Commenti
Gli autori dei commenti, e non la redazione, sono responsabili dei contenuti da loro inseriti - info

Devi effettuare il login per poter commentare
Se non sei ancora registrato, puoi farlo attraverso questo form.
Se sei già registrato e loggato nel sito, puoi inserire il tuo commento.
Si tenga presente quanto letto nel regolamento, nel rispetto del "quieto vivere".

La discussione è consultabile anche qui, sul forum.
 
^