[self-interest] Re: speed

[David Ungar]

Actually, I used Self for a while in a 266Mhz machine.
It was not really that slow, even without the SIC.
I did a lot of programming within Self, then moved up to a 400MHz machine
and have done  a lot more hacking.

I'm not sure what Steve is seeing.

- Dave





At 8:03 PM -0200 12/17/99, Jecel Assumpcao Jr. wrote:
>Steve Dekorte wrote:
>>  I got a chance to run Self on a G3 with 128M.
>>  It was cool to see it running on a Mac, but it was very slow.
>>  Is this because it's an early version or did my machine not have
>>  enough memory or something?
>
>Self includes two compilers:
>
>   - NIC, the Non Inlining Compiler, is very fast but generates very poor code
>that doesn't run much faster than an interpreter would. All methods are
>initially compiled with it
>
>   - SIC, the Simple Inlining Compiler, is not as fast as NIC but it generates
>very good code. Instead of doing extensive type analysis like the compiler in
>Self 1.0 and 2.0 did, it uses type information generated by actually running
>the code created by the NIC for a while. So the SIC is used to recompile
>methods that are determined to be critical to overall performance (the "hot
>spots" of the application, hence the name of the HotSpot technology 
>in Java 2.0)
>
>Well, this is how things work on the Sparc. For the PowerPC port, David Ungar
>only converted the NIC, so the lack of performance you are seeing is to be
>expected. The Linux Intel port that Gordon Cichon started to write also only
>included the NIC ("simple" is such a relative term ;-)
>
>In short: the NIC is enough to get Self running on a machine, but the SIC is
>needed to make it fast.
>
>-- Jecel
>
>------------------------------------------------------------------------
>Want to send money instantly to anyone, anywhere, anytime?
>You can today at X.com - and we'll give you $20 to try it!  Sign
>up today at http://clickhere.egroups.com/click/2227.  It's quick,
>free, & there's no obligation!
>
>
>-- Check out your group's private Chat room
>-- http://www.egroups.com/ChatPage?listName=self-interest&m=1



     David Ungar
     Sun Microsystems Laboratories
     (650) 336-2618