[VM] NCM performance report

Dan Ingalls DanI at wdi.disney.com
Tue Apr 6 15:17:57 UTC 1999


Peter -

>For interest, I tried bodging the SystemTracer to write out an image with
>3-word headers for every object in the distribution 2.3 image.  It took a
>3.5Mb image to a shade over 5Mb if I made the correct changes --- a large
>penalty on a PDA.

To argue the agressive side, the space penalty is not that bad.  It is much easier to compute than running the ST, but if you want to do that, then compare it with a normally written ST image.  The reason is that Squeak shrinks the display before snapshotting, but the ST does not.

>The flip side is that some of us [hi Adrian] trying to
>run multi-user systems on Squeak are willing to pay that space for a large
>speedup if we could get one.  I've had my head in the object allocation and
>GC stuff, so may have a biased view, but it looks like some considerable
>speed could be gained by doing away with the twisty turny maze of tests
>about object formats.  Has anyone done any experiments in this area?

I wouldn't call it twisty-turny, but it IS an extra memory cycle.  As I say, we have been withholding judgement until Ian has a nearly final version of Jitter running.  This is because

	1.  He may be able to circumvent the most common tests
		for compact classes, and
	2.  The faster JII goes, the more any(!) wasted cycles will matter.

>If not, it looks like I'll have to do this anyway at some point, as adding an
>optional fourth security word to the header causes all kinds of merry hell
>if you try to keep the existing variable-length headers.
>
>All thoughts welcome!
>
>		- Peter





More information about the Squeak-dev mailing list