Re: [PATCH/RFC] Shared page tables

[Brian Twichell]

Ray Bryant wrote:

> On Friday 27 January 2006 16:50, Brian Twichell wrote:
> <snip>
>
>  
>
> > Hi,
> >
> > We collected more granular performance data for the ppc64/hugepage case.
> >
> > CPI decreased by 3% when shared pagetables were used.  Underlying this was
> > a 7% decrease in the overall TLB miss rate.  The TLB miss rate for
> > hugepages decreased 39%.  TLB miss rates are calculated per instruction
> > executed.
> >
> >    
>
> Interesting.
>
> Do you know if Dave's patch supports sharing of pte's for 2 MB pages on 
> X86_64?
>  
I believe it does.  Dave, can you confirm ?

> Was there a corresponding improvement in overall transaction throughput for 
> the hugetlb, shared pte case?    That is, did the 3% improvement in CPI 
> translate to a measurable improvement in the overall OLTP benchmark score?
>  
Yes.  My original post with performance data described a 3% improvement
in the ppc64/hugepage case.  This is a transaction throughput statement.

> (I'm assuming your 25-50% improvement measurements, as mentioned in a previous 
> note, was for small pages.)
>
>  
That's correct.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/