Quote:
Originally Posted by CRACING
Ok got it.
I use spb benchmark. Before converting modules to files ( Untouched SYS ), The overall performance in benchmark report was 265 but the ezinput caused problems so I had convert modules inside IE packages to files and in benchmark report overall performance gone 1 step down thats 264.
And also as I'm using yours source now, in benchmark report overall performance is 263. 
So that mean converting modules to files, makes the OS slower. 
However, I'm low in these slots, (Re)allocation and etc. 
Best Regards |
Well, I would say that the performance difference is practically NIL and having more free VM space in Slot 0 would be more important. What good does a slight increase in performance do if the application you want to run does not run due to the VM contraints? (i.e. Slot 0 is to full and therefore only has a few MB's to run apps instead of the approximate 29MB instead if it were to be entireley empty minus some of the XIP stuff). Of course this all mainly applies to the 6.1 kernel as WMReloc and platform rebuilder will fill Slot 1, then 0. But the 6.5 kernel allows them to fill up slot s 1,60,61 then 0. So you have an extra 64MB of module allocation space. Basically with 6.1, we only have 32MB of allocation since we really shouldn't be using 0.
The performance COULD be related to the fact that MORE of the modules have to be dynamically loaded from the flash into ram, as opposed to eXecuting In Place (XIP)?