-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://reviews.m5sim.org/r/654/#review1142
-----------------------------------------------------------



src/mem/cache/base.hh
<http://reviews.m5sim.org/r/654/#comment1564>

    This can pretty easily be done with a template:
    
    template <class STAT>
    void
    incThreadVectorStat(PacketPtr pkt, STAT &stat)
    {
    }
    
    I'm pretty sure that you could even do this:
    
    template <class STAT>
    STAT::Proxy &
    getThreadVectorStat(PacketPtr pkt, STAT &stat[])
    {
        if (pkt->req->hasContextId()) {
            return stat[pkt->cmdToIndex()][pkt->req->contextId() % 
_numSharingContexts];
        } else {
            assert(pkt->req->contextId() == -1);
            assert(FULL_SYSTEM);
            return stat[pkt->cmdToIndex()][_numSharingContexts];
    }
    
    And do this:
    getThreadVectorStat(pkt, misses)++;
    getThreadVectorStat(pkt, misses) += 1;
    
    Actually, it may just be cleaner to have a function that returns the index. 
 Then you don't need all of the template mubmo jumbo.



src/mem/cache/base.hh
<http://reviews.m5sim.org/r/654/#comment1566>

    _numSharingContexts has the + 1 in it for devices.  Is that a problem?  
Seems like you need a parameter to know that.



src/mem/cache/base.hh
<http://reviews.m5sim.org/r/654/#comment1565>

    Maybe you can use a constant to indicate the -1 is a device.  Also, let's 
make sure that you initialize to something other than -1.



src/mem/cache/cache_impl.hh
<http://reviews.m5sim.org/r/654/#comment1568>

    This makes me think that some sort of index function is the right way to go.



src/mem/cache/cache_impl.hh
<http://reviews.m5sim.org/r/654/#comment1569>

    Is it not possible to figure out which thread caused the wb?



src/mem/cache/cache_impl.hh
<http://reviews.m5sim.org/r/654/#comment1567>

    I think we should force context IDs to be set even by devices.



src/mem/cache/cache_impl.hh
<http://reviews.m5sim.org/r/654/#comment1570>

    Index function.


- Nathan


On 2011-04-22 17:16:05, Lisa Hsu wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://reviews.m5sim.org/r/654/
> -----------------------------------------------------------
> 
> (Updated 2011-04-22 17:16:05)
> 
> 
> Review request for Default, Ali Saidi, Gabe Black, Steve Reinhardt, and 
> Nathan Binkert.
> 
> 
> Summary
> -------
> 
> Cache: fix vector stats in classic cache to have matching lengths
> 
> 
> Diffs
> -----
> 
>   configs/common/CacheConfig.py 914389024c33 
>   src/cpu/base.cc 914389024c33 
>   src/mem/cache/BaseCache.py 914389024c33 
>   src/mem/cache/base.hh 914389024c33 
>   src/mem/cache/base.cc 914389024c33 
>   src/mem/cache/blk.hh 914389024c33 
>   src/mem/cache/cache_impl.hh 914389024c33 
>   src/mem/cache/tags/base.cc 914389024c33 
>   src/mem/cache/tags/lru.cc 914389024c33 
> 
> Diff: http://reviews.m5sim.org/r/654/diff
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Lisa
> 
>

_______________________________________________
m5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/m5-dev

Reply via email to