Object memory ideas (was Re: Win32 VM Update & Request for help)
Bert Freudenberg
bert at freudenbergs.de
Thu Aug 30 20:48:36 UTC 2007
On Aug 30, 2007, at 10:46 , Andreas Raab wrote:
> tim Rowledge wrote:
>> On 30-Aug-07, at 8:55 AM, Andreas Raab wrote:
>>> You mean loading the image and relocating pointers is no excuse?
>>> Running the startup code, redrawing the screen? No excuses either?
>> It's a *reason* but not an *excuse*. Yes, there is a load of
>> assorted stuff done as the VM starts up and loads the image (which
>> could take several seconds on some machine/setups) and then as the
>> image starts up. That's the reason(s). The excuse really adds up
>> to "we haven't bothered to improve things". As John mentioned, it
>> seems likely that some of the VM related memory scanning could be
>> improved. Looking at the fairly intimidating list of things done
>> as part of the image startup I feel sure we could do a lot better
>> - not least by having a smaller simpler better thought out image
>> in the first place.
>
> Fix it ;-) In the meantime there is at least *some* feedback that
> the application has successfully launched. I find it highly
> disturbing (to say the least) if you don't get any feedback for
> several seconds when an application launches. The splash screen
> may not be an optimal solution but it sure as hell beats seeing
> absolutely nothing.
There is this idea I discussed on and off with people ... which is
having part of the object memory be "static/read-only". Objects in
there wouldn't be garbage-collected (ven by a full GC), and writing
to them would copy them into regular space. Unfortunately this
wouldn't help with start-up time unless we also move to an object
table (so oops wouldn't have to be rewritten on start-up).
- Bert -
More information about the Squeak-dev
mailing list
|