Fundamentals of Quantitative Design and Analysis - Computer Architecture: A Quantitative Approach - page 57

The overall ssj_ops/wat of the three servers is 3034 for the Intel R710, 2357 for the AMD
dual-socket R815, and 2696 for the AMD quad-socket R815. Hence, the Intel R710 has the best
power-performance. Dividing by the price of the servers, the ssj_ops/wat/$1000 is 324 for the
Intel R710, 254 for the dual-socket AMD R815, and 213 for the quad-socket MD R815. Thus,
adding power reverses the results of the price-performance competition, and the price-power-
performance trophy goes to Intel R710; the 48-core R815 comes in last place.
1.11 Fallacies and Pitfalls
The purpose of this section, which will be found in every chapter, is to explain some com-
monly held misbeliefs or misconceptions that you should avoid. We call such misbeliefs falla-
cies . When discussing a fallacy, we try to give a counterexample. We also discuss pitfalls —eas-
ily made mistakes. Often pitfalls are generalizations of principles that are true in a limited con-
text. The purpose of these sections is to help you avoid making these errors in computers that
you design.
Fallacy Multiprocessors Are A Silver Bullet
The switch to multiple processors per chip around 2005 did not come from some breakthrough
that dramatically simplified parallel programming or made it easy to build multicore com-
puters. The change occurred because there was no other option due to the ILP walls and power
walls. Multiple processors per chip do not guarantee lower power; it's certainly possible to
design a multicore chip that uses more power. The potential is just that it's possible to contin-
ue to improve performance by replacing a high-clock-rate, inefficient core with several lower-
clock-rate, efficient cores. As technology improves to shrink transistors, this can shrink both
capacitance and the supply voltage a bit so that we can get a modest increase in the number of
cores per generation. For example, for the last few years Intel has been adding two cores per
generation.
As we shall see in Chapters 4 and 5 , performance is now a programmer's burden. The La-
Z-Boy programmer era of relying on hardware designers to make their programs go faster
without lifting a finger is officially over. If programmers want their programs to go faster with
each generation, they must make their programs more parallel.
The popular version of Moore's law—increasing performance with each generation of tech-
nology—is now up to programmers.
Pitfall Falling Prey To Amdahl's Heartbreaking Law
Virtually every practicing computer architect knows Amdahl's law. Despite this, we almost all
occasionally expend tremendous effort optimizing some feature before we measure its usage.
Only when the overall speedup is disappointing do we recall that we should have measured
irst before we spent so much effort enhancing it!