[cairo] New tessellator work and thoughts

[M Joonas Pihlaja]

On Tue, 21 Nov 2006, Behdad Esfahbod wrote:

> On Tue, 2006-11-21 at 00:03 -0500, M Joonas Pihlaja wrote:
>> (Although, that's getting a
>> little too low level for testing on just one machine, and the
>> profile's likely different on newer processors.)
>
> Is callgrind's timing in any way related to the processor you are
> running on at all?

No, I don't believe so[1], but the numbers can be misleading just 
because of that.  I've been bitten one time optimising an image 
scaler on my machine using cachegrind as a guide, and then left 
wondering why there was no speedup at all on the newer machines 
it was meant to run on.

Joonas

[1] IIRC cachegrind tries to match the native CPU's cache sizes 
though.  Not that that makes a difference for instruction counts.