Performance Analysis and Benchmarks

Skip to first unread message

Adam Preble

Feb 26, 2005, 4:02:20 AM2/26/05
I do a lot of performance analysis work at my job at Advanced Micro
Devices, particularly with Just-In-Time compilers. We do a lot of
automation in Perl and I've been impressed by what I've read so far on
the pending Perl 6 features. We're hoping to use it for future
projects--as opposed to moving on to Python for some OOP. I'm trying to
come on board as best as I can right now to make sure it happens.

I saw in the TODO that there is a need for "Mo' betta benchmarks," and
I'd like to know what's currently out there along with what is lacking.
Writing some benchmarks looks like a good way to transition into Perl
6 since it'll force me to study the new semantics.

There are some architectural things I'd need to know about the virtual

1. Does it optimize dynamically?
2. Are there different optimization levels? What are they?

This would affect how I'd approach things. As of now, I'm assuming #2
is "kind of" and #1 is "no," but I'll try to write some stuff assuming
it will dynamically optimize someday.

I'm also curious if somebody is already working on an AMD64 code
generator, and if they need code quality assessments. That's
essentially my bread and butter right now, and I can try to help as best
I can.

Reply all
Reply to author
0 new messages