review request (M): 7121140: Allocation paths require explicit memory synchronization operations for RMO systems

[Tom Rodriguez]

On Dec 16, 2011, at 4:12 AM, Roland Westrelin wrote:

>> Could you do a longer set of specjvm98 to confirm the jess difference?  Looking at the data, I don't really think it's real but another, longer set of runs should tell us that.  Try 20.
> 
> See below.

That looks ok.  Go ahead and push.

tom

> 
> Roland.
> 
> ============================================================================
> t1
>  Benchmark         Samples        Mean     Stdev
>  specjvm98              20      666.33      0.00
>    javac                20      392.47      0.01
>    db                   20      438.68      0.00
>    jess                 20      699.73      0.01
>    jack                 20      593.50      0.01
>    compress             20      536.20      0.00
>    mtrt                 20     1786.87      0.02
>    mpegaudio            20      851.56      0.01
> ============================================================================
> t2
>  Benchmark         Samples        Mean     Stdev   %Diff    P   Significant
>  specjvm98              20      666.90      0.00    0.08 0.518            *
>    javac                20      392.72      0.02    0.06 0.895            *
>    db                   20      439.33      0.00    0.15 0.342            *
>    jess                 20      700.23      0.01    0.07 0.760            *
>    jack                 20      590.51      0.01   -0.50 0.047            *
>    compress             20      537.69      0.01    0.28 0.230            *
>    mtrt                 20     1794.64      0.01    0.44 0.333            *
>    mpegaudio            20      852.44      0.01    0.10 0.678            *
> ============================================================================
>