[Q] Status of blocks

Ken Collins kenncoll123 at hotmail.com
Thu Jan 16 03:40:07 UTC 2003


Hi,

Here are the results I got from a couple of tests.

VI = Squeak3.5alpha[VI4] latest update: #5168.
NonVI = Squeak3.2 latest update: #4957

Win98SE - 1.33 Mhz Athlon - 384 RAM

1) Time millisecondsToRun:[
| s  |
s_Set new.
#(1 2 3 4 5 6 7 8 9) permutationsDo: [:e | s add:e copy].
s asArray]

VI - 24918
NonVI - 50173

2) Time millisecondsToRun: [10000 factorial]

VI - 1388
NonVI - 2249

3) Time millisecondsToRun: [100000 factorial]

VI - 393454
NonVI - 324129

I tried this last test a few times, and NonVI beat it every time. Seems 
kinda strange.

Ken


>From: "Swan, Dean" <Dean_Swan at Mitel.COM>
>Reply-To: squeak-dev at lists.squeakfoundation.org
>To: "'squeak-dev at lists.squeakfoundation.org'" 
><squeak-dev at lists.squeakfoundation.org>
>Subject: RE: [Q] Status of blocks
>Date: Wed, 15 Jan 2003 20:00:22 -0500
>
>Please pardon my ignorance, but if Anthony's VI4 work
>is done, includes block closures and is faster than the
>stock VM, why is there reluctance to take it as a whole?
>
>I've read some of the messages related to this, and from
>my point of view, a new bytecode set and stack contexts
>are both great improvements and we should want them as
>well.
>
>Could anyone enlighten me about the concerns?  I understand
>there are emotional reasons (i.e. too much change for comfort),
>but I don't see any technical drawbacks.
>
>					-Dean
>
>
> > -----Original Message-----
> > From: Anthony Hannan [mailto:ajh18 at cornell.edu]
> > Sent: Tuesday, January 14, 2003 7:11 PM
> > To: squeak-dev at lists.squeakfoundation.org
> > Subject: Re: [Q] Status of blocks
> >
> >
> > Avi Bryant <avi at beta4.com> wrote:
> > > It's redundant, but let me add my agreement: Anthony,
> > *please* do give us
> > > block closures on the stock VM, if at all possible.  I seem
> > to constantly
> > > be doing things these days where I'm bitten by the lack of them.
> >
> > Alright, I will.  But beware, it will make Squeak slower, rather than
> > faster like VI4.  I don't know how much slower yet.  I will
> > try to make
> > closures optional, meaning you can set a preference as to
> > which compiler
> > you want to use when methods are saved.
> >
> > Cheers,
> > Anthony
> >


_________________________________________________________________
Protect your PC - get McAfee.com VirusScan Online 
http://clinic.mcafee.com/clinic/ibuy/campaign.asp?cid=3963




More information about the Squeak-dev mailing list