Fyo, <Naturally Speaking, Netshow Encoder, Photoshop,...>
Of course the things with SSE are different just
because the extended instructions are much more
useful, and AMD's 3Dnow! has had this proven, to Intel's
benefits.
However, the benchmarks you mentioned are largely
meaningless. Why?
1. Speech recognition is the utter nonsense by itself
for PC users. At least in current form it is a joke.
Therefore there is nothing to compare yet.
2. Netshow Encoder. How frequently a home PC user
sends "streaming video" clips to the world
(except teen girls doing net striptease)?
3. Photoshop filtering: the filtering takes so
long time anyway, so 2 minutes vs 3 minutes
makes no difference for home photo processing.
For professional processing you better get
some A21264 machine with a $5000 video processor.
Same for sound editing.
On the other note: we might see some feedback cycle here:
First, 3Dnow proved the concept.
Then, SSE technology gets adopted by software designers.
Next, since algorithms are now in place, SSE and 3Dnow
are structurally identical, experience has gained,
so it will take much less effort
to incorporate 3Dnow into the same software. |
