Slide 2
    The 75% claim might be better stated as "Better overall throughput,
    as high as 75% less power" or something like that.  It all depends on
    the confluence of multiple factors, including existing utilization
    rate, application workload footprint, existing configuration, and
    so on.  Maybe something like "50-75% less power".  The 75% number
    comes from specific bids prepared for prospective customers after
    carefully their application and analyzing all those, and possibly
    a couple more, factors.

Slide 5/6
    These are the same slide.  I guess I like #6 a little better.

    Probably just delete the #.

    Also, we could add 20 and 24 core servers.

    The "Lopoco Servers" bar should go all the way up to the bottom of
    the HPC block.

Slide 7
    TDP Software, maybe change to Accurate Total Design Power measurement
    software or something similar.

Slide 8
    Probably a no-go on this slide.  I eventually came to the conclusion
    that the SPECpower benchmark was just a CPU benchmark, and we only
    do slightly better, or slightly worse when going up against the
    Microsoft tuned Java stack, than existing servers.  The last thing I
    was working on was a whole server power benchmark, which I basically
    have, but the science to translate (normalize) the results of the
    benchmark required some serious scientific/mathematical thought,
    and I haven't been able to do that for a while.  So perhaps just
    do a slide with that benchmark results on a Lopoco machine and the
    similar Dell machine that we bought.