Re: pack operation is thrashing my server

From: Andi Kleen <andi@firstfloor.org>
Date: Aug 14, 2008, 21:30 UTC
Message-ID: <20080814213033.GE13814@one.firstfloor.org>
In-Reply-To: <alpine.LFD.1.10.0808141215520.3324@nehalem.linux-foundation.org>

> Here's a hint: the cost of a cache miss is generally about a hundred times

100 times seems quite optimistic %)

Show 6 quoted lines

> 
> No, zlib isn't perfect, and nope, inflate_fast() is no "memcpy()". And 
> yes, I'm sure a pure memcpy would be much faster. But I seriously suspect 
> that a lot of the cost is literally in bringing in the source data to the 
> CPU. Because we just mmap() the whole pack-file, the first access to the 
> data is going to see the cost of the cache misses.

I would have thought that zlib has a sequential access pattern that the CPU prefetchers have a easy time with hiding latency.

BTW I always wonder why people reason about cache misses in oprofile logs without actually using the cache miss counters.

-Andi