Eric Haszlakiewicz <erh@swapsimple.com> writes:

> On Sun, Feb 18, 2007 at 06:50:38PM +0000, David Laight wrote:
>> On Sun, Feb 18, 2007 at 10:29:42AM -0800, Bucky Katz wrote:
>> > 
>> > It has to be the later. The context switch overhead of 1:1 on ARM will
>> > cause a severe performance degradation.
>> 
>> That shouldn't matter.
>> Siince the threads share the VM space, the cache flushing needn't be done.
>> So there isn't any expensive context switch to do when one LWP wakes
>> another in the ame process.
>
> This sounds like something begging for a before-and-after benchmark.
> Does anyone have time (and hardware) to do such a benchmark?


Um, since the theory behind newlock2 was performance improvements,
shouldn't benchmarks have already been done with the results
available?