[Tremor] Notes on Implementing Tremor on an ARM7TDMI CPU

[Nicholas Vinen]

Andrew Lentvorski wrote:
> Nicholas Vinen wrote:
>>> It would be useful to actually *know* where it's spending all its
>>> time.  Can you run this under a simulator?
>> Yes, it would. I'd like to use a simulator, the main problem is I'd
>> likely have to modify the code since I doubt the simulator will simulate
>> the peripherals etc. I don't know where to get a simulator. I assume
>> they exist...
>
> You might want to slow down a bit.
Too much left to accomplish to slow down yet :)
> You're missing the details and implications in some of these posts.
>
> I pointed you at a simulator, but I guess I wasn't clear enough.  It's
> called Skyeye:
> http://www.skyeye.org/index.shtml
>
> It claims to support the specific chip you're using.  You may still
> have to twiddle with some things, but, if you're really taking 20
> minutes per load on real hardware, the cost/benefit is going to be in
> your favor for even quite a bit of time spent on skyeye.
>
> -a

I see. When you said "skyeye" I didn't know what you were talking about.
Now I do, thanks. I'll give it a try.

I've sped up my JTAG interface so now it's more like 10 minute which is
a bit more bearable. A simulator would still be a lot better. I'll give
it a try.

By the way I just finished applying my changes (static allocation, etc.)
to Tremolo and loaded it onto my test board. It's now using about 90%
CPU to decode in real time, but it's still overclocked. My plan is now
to experiment with loading critical functions into RAM - once I work out
how to tell GCC/binutils to do that. I assume I just use an attribute to
move the function into another section. I have a feeling that I can get
it to be at least slightly better than real time without overclocking
and without radical changes, but it's going to be close.

The simulator may help. Thanks.



Nicholas