From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <ober.14@osu.edu>
X-Spam-Checker-Version: SpamAssassin 3.1.3 (2006-06-01) on yquem.inria.fr
X-Spam-Level: 
X-Spam-Status: No, score=0.4 required=5.0 tests=AWL,DNS_FROM_RFC_ABUSE 
	autolearn=disabled version=3.1.3
Received: from mail1-relais-roc.national.inria.fr (mail1-relais-roc.national.inria.fr [192.134.164.82])
	by yquem.inria.fr (Postfix) with ESMTP id D472BBC69
	for <caml-list@yquem.inria.fr>; Thu,  3 Jan 2008 17:28:31 +0100 (CET)
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: Ah4FAGSbfEdDWxLC/2dsb2JhbACBV5AKmFY
X-IronPort-AV: E=Sophos;i="4.24,240,1196636400"; 
   d="scan'208";a="6251412"
Received: from ip67-91-18-194.z18-91-67.customer.algx.net (HELO server1.bertec.net) ([67.91.18.194])
  by mail1-smtp-roc.national.inria.fr with ESMTP; 03 Jan 2008 17:28:31 +0100
Received: from kuba.bertec.net (kuba.bertec.net [192.168.2.16])
	by server1.bertec.net (Postfix) with ESMTP id E94FFCDFBC
	for <caml-list@yquem.inria.fr>; Thu,  3 Jan 2008 11:28:30 -0500 (EST)
From: Kuba Ober <ober.14@osu.edu>
To: caml-list@yquem.inria.fr
Subject: Performance questions, -inline, ...
Date: Thu, 3 Jan 2008 11:28:30 -0500
User-Agent: KMail/1.9.6 (enterprise 0.20071123.740460)
MIME-Version: 1.0
Content-Type: text/plain;
  charset="us-ascii"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200801031128.30183.ober.14@osu.edu>
X-Spam: no; 0.00; -inline:01 ocamlopt:01 -inline:01 -unsafe:01 mips:01 utime:01 vectors:01 inlining:01 cheers:01 mips:01 const:01 accum:01 accum:01 const:01 1.3:98 

I haven't looked at assembly output yet, but I've run into some unexpected
behavior in my benchmarks.

This was compiled by ocamlopt -inline 100 -unsafe, the results and code are 
below (MIPS is obtained by dividing 50 million iterations by (Unix.times 
()) . Unix.tms_utime it took to run). I haven't included the timing etc. code 
(it's part of a larger benchmark).

What I wonder is why vector-to-vector add is so much faster than (constant) 
scalar to vector add. Vectors are preinitialized each time with a 1.0000, 
1.0001, ... sequence.

Also, the very bad performance from generic vector-to-vector *with* inlining 
is another puzzler, whereas generic add of scalar-to-scalar performs 
similarly to straight-coded one.

Cheers, Kuba

* add1: add scalar to scalar   120 MIPS
* add3: add scalar to vector   250 MIPS
* add5: add vector to vector   320 MIPS
* add2: generic add scalar to scalar   100 MIPS
* add4: generic add vector to vector   38 MIPS

let start = 1.3

(* generic scalar operation *)
let op1 op const nloop =
	let accum = ref start in
	for i = 1 to nloop do
		accum := op !accum const
	done

(* generic vector operation *)
let op2 op const a b (nloop : int) =
	let len = Array.length a in
	for j = 0 to len-1 do
		for i = 0 to len-1 do
			b.(i) <- op a.(i) b.(i)
		done;
	done

(** addition **)
let add1 nloop =
	let accum = ref start in
	for i = 1 to nloop do
		accum := !accum +. addconst
	done
let add2 = op1 ( +. ) addconst
let add3 a b nloop =
	let len = Array.length a in
	for j = 0 to len-1 do
		for i = 0 to len-1 do
			b.(i) <- a.(i) +. addconst
		done;
	done
let add4 = op2 ( +. ) addconst
let add5 a b nloop =
	let len = Array.length a in
	for j = 0 to len-1 do
		for i = 0 to len-1 do
			b.(i) <- a.(i) +. b.(i)
		done;
	done