Re: [RFC PATCH 00/20] Cleanup and optimise the page allocator

[Mel Gorman]

On Mon, Feb 23, 2009 at 12:57:37AM +0100, Andi Kleen wrote:
> Mel Gorman <mel@xxxxxxxxx> writes:
> 
> > The complexity of the page allocator has been increasing for some time
> > and it has now reached the point where the SLUB allocator is doing strange
> > tricks to avoid the page allocator. This is obviously bad as it may encourage
> > other subsystems to try avoiding the page allocator as well.
> 
> Congratulations! That was long overdue. Haven't read the patches yet though.
> 

Thanks

> > Patch 15 reduces the number of times interrupts are disabled by reworking
> > what free_page_mlock() does. However, I notice that the cost of calling
> > TestClearPageMlocked() is still quite high and I'm guessing it's because
> > it's a locked bit operation. It's be nice if it could be established if
> > it's safe to use an unlocked version here. Rik, can you comment?
> 
> What machine was that again?
> 

It's a AMD Phenom 9950 quad core.

> > Patch 16 avoids using the zonelist cache on non-NUMA machines
> 
> My suspicion is that it can be even dropped on most small (all?) NUMA systems.
> 

I'm assuming it should not be dropped for all. My vague memory was that this
was introduced for large IA-64 machines and that they were able to show a
clear gain when scanning large numbers of zones. Patch 16 disables zonelist
caching if there is only one NUMA node but maybe it should be disabled for
more than that.

> > Patch 20 gets rid of hot/cold freeing of pages because it incurs cost for
> > what I believe to be very dubious gain. I'm not sure we currently gain
> > anything by it but it's further discussed in the patch itself.
> 
> Yes the hot/cold thing was always quite dubious.
> 

Andrew mentioned a micro-benchmark so I will be digging that up to see
what it can show.

> > Counters are surprising expensive, we spent a good chuck of our time in
> > functions like __dec_zone_page_state and __dec_zone_state. In a profiled
> > run of kernbench, the time spent in __dec_zone_state was roughly equal to
> > the combined cost of the rest of the page free path. A quick check showed
> > that almost half of the time in that function is spent on line 233 alone
> > which for me is;
> >
> > 	(*p)--;
> >
> > That's worth a separate investigation but it might be a case that
> > manipulating int8_t on the machine I was using for profiling is unusually
> > expensive. 
> 
> What machine was that?
> 

This is the AMD Phenom again but I might be mistaken on the line causing
the problem. A second profile run shows all the cost in the function entry
so it might just be a co-incidence that the sampling happened to trigger on
that particular line. It's high on the profiles simply because it's called
a lot. The assembler doesn't look particularly bad or anything.

> In general I wouldn't expect even on a system with slow char
> operations to be that expensive. It sounds more like a cache miss or a
> cache line bounce. You could possibly confirm by using appropiate
> performance counters.
> 

I'll check for cache line misses.

> > Converting this to an int might be faster but the increased
> > memory consumption and cache footprint might be a problem. Opinions?
> 
> One possibility would be to move the zone statistics to allocated
> per cpu data. Or perhaps just stop counting per zone at all and
> only count per cpu.
> 
> > The downside is that the patches do increase text size because of the
> > splitting of the fast path into one inlined blob and the slow path into a
> > number of other functions. On my test machine, text increased by 1.2K so
> > I might revisit that again and see how much of a difference it really made.
> >
> > That all said, I'm seeing good results on actual benchmarks with these
> > patches.
> >
> > o On many machines, I'm seeing a 0-2% improvement on kernbench. The dominant
> 
> Neat.
> 
> > So, by and large it's an improvement of some sort.
> 
> That seems like an understatement.
> 

It'll all depend on what other peoples machines turn up :)

-- 
Mel Gorman
Part-time Phd Student                          Linux Technology Center
University of Limerick                         IBM Dublin Software Lab
--
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/