Author: Maciej Fijalkowski <[email protected]> Branch: extradoc Changeset: r3725:a226b2cf5659 Date: 2011-06-17 09:25 +0200 http://bitbucket.org/pypy/extradoc/changeset/a226b2cf5659/
Log: merge diff --git a/talk/ep2011/talk/talk.rst b/talk/ep2011/talk/talk.rst --- a/talk/ep2011/talk/talk.rst +++ b/talk/ep2011/talk/talk.rst @@ -156,8 +156,24 @@ Real world use case (2) ----------------------- -XXX write slide about this, the site is down ATM -http://www.myhdl.org/doku.php/performance +* **MyHDL**: VHDL-like language written in Python + + - |scriptsize| http://www.myhdl.org/doku.php/performance |end_scriptsize| + + - (now) competitive with "real world" VHDL and Verilog simulators + + +|pause| + +|example<| ``myhdl.org`` |>| +|small| + + [...] the results are spectacular. By simply using a different interpreter, + our simulations run 6 to 12 times faster. + +|end_small| +|end_example| + Real world use case (3) diff --git a/talk/iwtc11/benchmarks/sqrt/sqrt_fix16.c b/talk/iwtc11/benchmarks/sqrt/sqrt_fix16.c --- a/talk/iwtc11/benchmarks/sqrt/sqrt_fix16.c +++ b/talk/iwtc11/benchmarks/sqrt/sqrt_fix16.c @@ -1,12 +1,12 @@ #include <stdio.h> int main() { - long y = 1234 << 16; + long y = 123 << 16; long x = y / 2; long n = 100000000; while (n>0) { n -= 1; - x = ((x + (y << 16)/x)) / 2; + x = ((x + (y << 8)/(x >> 8))) / 2; } printf("%f\n", ((double) x) / ((double) (1<<16))); fprintf(stderr, "sqrt(Fix16): "); diff --git a/talk/iwtc11/paper.tex b/talk/iwtc11/paper.tex --- a/talk/iwtc11/paper.tex +++ b/talk/iwtc11/paper.tex @@ -121,7 +121,19 @@ \maketitle \begin{abstract} -This is the text of the abstract. +By introducing loop peeling into the optimization step of a tracing +jit the effect of optimizations already in place will be increased +greatly. Not only will they become able to move loop invariant code +out of loop. They will also become able to reuse results from the +previous iteration. Also, the implementation of excising optimizations +can be left almost intact as they will not have to interact much with +the loop peeling. + +Several benchmarks, with few guard failures, executed on the +PyPy python JIT show over 2 +times increase in speed when loop peeling was introduced. This makes +some of them almost match optimized C performance and become over XXX +times faster than cpython. \end{abstract} \category{D.3.4}{Programming Languages}{Processors}[code generation, _______________________________________________ pypy-commit mailing list [email protected] http://mail.python.org/mailman/listinfo/pypy-commit
