> On May 20, 2012, 3:34 p.m., Steve Reinhardt wrote:
> > src/cpu/decode_cache.hh, line 188
> > <http://reviews.gem5.org/r/1193/diff/1/?file=26343#file26343line188>
> >
> >     Why the change from the if/else structure in the original function?  
> > That seemed reasonable to me, and avoided duplication of a couple of 
> > (admittedly simple) lines of code.

The idea was to isolate the different stages of the lookup. The first chunk 
looks in the address based cache, the second looks in the ExtMachInst based 
cache, and the third actually decodes the instruction. Each is responsible for 
fixing up the earlier caches if there's a hit. With the old structure, things 
were mixed together which, I thought, would make it more difficult to mix and 
match components, or at least see how this particular caching organization is 
structured.


- Gabe


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://reviews.gem5.org/r/1193/#review2757
-----------------------------------------------------------


On May 15, 2012, 5:56 a.m., Gabe Black wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://reviews.gem5.org/r/1193/
> -----------------------------------------------------------
> 
> (Updated May 15, 2012, 5:56 a.m.)
> 
> 
> Review request for Default.
> 
> 
> Description
> -------
> 
> Changeset 9007:906c07bc6917
> ---------------------------
> CPU: Simplify the implementation of the decode cache.
> 
> Also reorganize it to make it more amenable to being rearranged later.
> 
> 
> Diffs
> -----
> 
>   src/cpu/decode_cache.hh f681719e2e99 
> 
> Diff: http://reviews.gem5.org/r/1193/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Gabe Black
> 
>

_______________________________________________
gem5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev

Reply via email to