|
|
Atari coding BBS
| Re: dsp profiling patch for Hatari |
|---|
|
Posted by: dml
|
Feb,27.2014-15:55
|
Ok I think I have a fix now which makes the (X:Y) field show the *last* cycle measurements at that instruction as (opcode_cycles:extpenalty_cycles) instead of the unreliable figures which currently appear.
The original code was pulling stale cycle counts direct from the execution core instead of updating last encountered values alongside the other values. So they could get completely out of sync and disagree with the averages.
As I mentioned in the last message these two fields aren't terribly helpful for optimizing code, compared with having an accurate average via the other new fields (which work fine with the patch) - but might as well fix it anyway. Will post an updated diff soon.
|
[All messages in this thread] [Start new thread]
What's the anti-troll code?
That's your personal code to be able to add comments and messages on the dhs.nu site.
Don't have a code or forgot it? Fix it here.
|
