Unix VM and MacIntel

On Oct 16, 2006, at 1:44 PM, John M McIntosh wrote:

> Ah *slower* you mean

1. '229390681 bytecodes/sec; 7759029 sends/sec'
2. '291904218 bytecodes/sec; 8411337 sends/sec'

(1) is Unix 3.9-9 built from source on Mac Book Pro with register  
variables disabled (i.e., the thing I said was slower).

(2) is Unix 3.9-9 built from source on Mac Book Pro with register  
variables enabled (i.e., the thing I said was faster and the thing I  
dumped in the binary archive in consideration of one person's  
hesitation to build from source).

> mac carbon  3.8.13b3u on pb 17 osx 10.4.8  gives

If the carbon VM is faster on PowerBook then maybe you use some  
clever XCode PPC optimisation flags that Apple doesn't bother to tell  
command-line users about?  In any case, it has no relation to either  
of the things I was comparing when I claimed one was faster than the  
other.

> On testing on macIntel the downloaded application, and the build  
> application both would not open the squeak window
> when given a 2.8 image or a 3.8.latest image.  Squeak CPU usage  
> would run at 100%, not userful. No message to the system log or  
> console.
>
>  www.squeakvm.org unix page only shows 3.9-8

I though I explained that 3.9-8 is the latest release.  3.9-9 doesn't  
exist yet.  Anything you might have seen called 3.9-9 is a figment of  
my imagination that I might do someone a favour by making a binary VM  
available to them.  (Maybe I've learned that I should never copy the  
list in such situations.)

> So if someone has a *working* 3.9-9 macintel VM

I have one.  In fact, I have two of them now.  (One of which is  
faster than the other. ;-)

> Lastly the VM application 3.9-9 posted is macintel only, not powerpc.

I know.  I don't make universal binaries.

FWIW, the 'hang' sounds like a problem with a broken lock on i386.  I  
fixed this and posted new 3.9-9 sources (which don't really exist yet  
sice 3.9-8 is the latest release) shortly after someone found it.  
(FWIW #2, the slow refresh some people reported is probably because I  
reverted to the crappy OS lock on .386 instead of writing a real  
spinlock.  The latter is #13489 on today's 'to-do' list.)

On 16-Oct-06, at 2:28 PM, Ian Piumarta wrote:

> If the carbon VM is faster on PowerBook then maybe you use some  
> clever XCode PPC optimisation flags that Apple doesn't bother to  
> tell command-line users about?  In any case, it has no relation to  
> either of the things I was comparing when I claimed one was faster  
> than the other.

XCode just invokes Apple's GCC so nothing up XCode's sleeve .  man  
gcc is your friend.

> I have one.  In fact, I have two of them now.  (One of which is  
> faster than the other. ;-)

Well email me one, not the list so I can see if I can get it to run  
on my MacIntel Mini.

--
========================================================================
===
John M. McIntosh <[hidden email]>
Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
========================================================================
===

On 16-Oct-06, at 2:28 PM, Ian Piumarta wrote:

> If the carbon VM is faster on PowerBook then maybe you use some  
> clever XCode PPC optimisation flags that Apple doesn't bother to  
> tell command-line users about?  In any case, it has no relation to  
> either of the things I was comparing when I claimed one was faster  
> than the other.

K, well I guess it is faster, but it still lags a bit, you can buy me  
a beer at OOPSLA, and I'll tell you know to make it a bit faster.

Mac Carbon VM 3.8.12b7
'259371833 bytecodes/sec; 5864521 sends/sec'
'258324924 bytecodes/sec; 5879195 sends/sec'
'255872063 bytecodes/sec; 5893942 sends/sec'
'259503294 bytecodes/sec; 5884102 sends/sec'
'259503294 bytecodes/sec; 5898874 sends/sec'
'259109311 bytecodes/sec; 5893942 sends/sec'
'258978249 bytecodes/sec; 5854779 sends/sec'
'258847320 bytecodes/sec; 5889018 sends/sec'
'257545271 bytecodes/sec; 5893942 sends/sec'
'259371833 bytecodes/sec; 5893942 sends/sec'

3.9-9 application build by Ian
'203497615 bytecodes/sec; 5594567 sends/sec'
'203659506 bytecodes/sec; 5612384 sends/sec'
'204146730 bytecodes/sec; 5621335 sends/sec'
'204146730 bytecodes/sec; 5612384 sends/sec'
'203821656 bytecodes/sec; 5612384 sends/sec'
'203659506 bytecodes/sec; 5612384 sends/sec'
'203984063 bytecodes/sec; 5607919 sends/sec'
'203984063 bytecodes/sec; 5581278 sends/sec'
'203821656 bytecodes/sec; 5603461 sends/sec'
'203984063 bytecodes/sec; 5612384 sends/sec'


Please note 3.8.13b4 runs sends/sec slower because of Exupery common  
send changes so I compared a 3.8.12b7 to Ian's 3.9-9 which doesn't  
have the Exupery changes


--
========================================================================
===
John M. McIntosh <[hidden email]>
Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
========================================================================
===