No Subject

[lowry@xxxxxxxxxxxxxxxxx (Dave Lowry)]

Eugene Leitl writes:

>I know you are not trying to do wild claims on F21 for the sake of
>seriosity, Jeff, but I think your estimate from the 3d package you
>wrote (i486/40, you said?) is too low. Using 3- and even 4-dim matrix
>maths with lots of multiplications is not the only way to do fast 3d.
>Multiplication is the weak side of F21, and 3d matrixmult makes it
>show more than the average code mix would. In fact I am amazed
>you still have the i486/40 performance in spite of above handicap.

I wrote the 3D package and made the 486DX40 (P21 running in SRAM) claim.

Eugene, the intent of this admittedly non-scientific benchmark was to show
the relative performance of P21 and x86 *running the same code*.  That
is the point of all benchmarks.  The intent was not to "showcase" P21,
but to provide a "fair" comparison for a particular application domain.

The benchmark was posted in response to a question about how x21 would
perform scientific computations.  In the absence of the more traditional
SPECmarks, the 3D application was offered.

Eugene makes a very good point, though.  To equal an x86 in an application
with lots of multiplications, and many divisions, P21 must be doing some
things very, very fast indeed!

Jeff Fox has posted, numerous times, benchmarks (i.e. CORDIC algorithms)
where P21 blows other processors out of the water.

-Dave