[Tfug] Processing/memory throughput and suggestions

[Bexley Hall]

--- On Fri, 5/15/09, Jesse Allen <jesse.carl at gmail.com> wrote:

> > I've found a Gigabyte motherboard (GA-E7AUM-DS2H) that takes
> > a quad-core Intel CPU and up to 16GB of DDR2-800 memory.  I'd put
> > into that an Intel Q9550S
> > 2.83GHz quad-core CPU with 12MB cache.
> > 
> > But, it occurs to me that I might possibly achieve better 
> > throughput if I went with a true dual-processor motherboard
> > and put onto it two dual-core CPUs.
> > More paths to the memory means less contention inside
> > the CPUs, but, of
> > course, there can still only be one thing accessing memory 
> > at a time.
> 
> The quad-core is going to be faster. The fact that at least
> L3 cache is shared makes a tremendous difference. Remember
> that every time the CPU has a cache miss, there is a huge
> delay. The difference between an L2 cache miss and a cache
> miss which requires going off-chip is astronomical. When I
> say astronomical, I'm talking orders of magnitude
> difference. Having dual CPUs compared to a single CPU of
> comparable processing power is like working in North Phoenix
> and living in South Tucson compared to living and working in
> the same town.

But his algorithm *expects* to take lots of misses on the *data*.
OTOH, I suspect (?) that the actual algorithm itself WILL fit
entirely in that 12M of cache -- *if* he can keep it there!