[uae] Re: segmentation fault with latest

  • From: Richard Drummond <evilrich@xxxxxxxxxxxxxx>
  • To: uae@xxxxxxxxxxxxx
  • Date: Tue, 30 Nov 2004 20:11:00 -0500

Hi Lee

On Tuesday 30 November 2004 06:36 pm, general questions wrote:
> Now I can get both demos working, but using 68000 and
> higher accuracy, which wasn't necessary under
> uae-0.8.22.
> I did a direct comparison with uae-0.8.22:
> Whereas on uae it only required about 50% to 60%
> processing power to work perfectly smoothly, with the
> minimum to get it to work under e-uae-0.8.27-RC2, it
> doesn't quite pull it off with 100% processing power
> in most places.

Try RC3. Here the 'Technological Death' demo works with normal 68000 CPU 
emulation with cpu_speed=real. top reports 20%-30% CPU utilization on my 
1800MHz Athlon.

It's probably worth pointing out that on Linux hosts unless "Don't busy wait 
when idle" is checked in the GUI, CPU usage won't reflect the actual work 
that UAE is doing, since E-UAE will use busy waiting for delays. On 2.4 
kernels (unless its a kernel with low latency patches), this'll be off by 
default and UAE will busy wait. On 2.6 kernels task scheduling has a much 
lower latency, and so UAE can use system-friendly methods for waiting.

> On 'Technological Death' demo.  Using Sprite
> Collisions: 'Sprites & Playfields' only, then under
> uae-0.8.22 the red demon head shrinks fine.  Under
> e-uae it gets green blobs.
> Similarly happens under different software.
> Don't if uae was really doing the full emulation or
> e-uae is miss doing it...

I don't see this. Try with RC3 and see if this still occurs.

> Also, don't know if mentioned previously but there
> seems to be an issue with swapping the disk.
> try swapping, pciking new disk, then resetting, and
> occasionally the previous disk appears when I reset.
> Seems to happen in succession or not at all
> intermittently.

I'll investigate this.


Other related posts: