[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Bug-apl] OpenMP patch, test harness and benchmark

From: Juergen Sauermann
Subject: Re: [Bug-apl] OpenMP patch, test harness and benchmark
Date: Mon, 17 Mar 2014 18:56:31 +0100
User-agent: Mozilla/5.0 (X11; Linux i686; rv:17.0) Gecko/20130330 Thunderbird/17.0.5


I have added a workspace "Performance.apl" that gnuplots performance data nicely.

According to that the sequential time (interpretation + malloc) for V+V is pretty much 5 uS on my machine,
plus the length-dependant computation time

/// J├╝rgen

On 03/15/2014 08:23 PM, David B. Lamkins wrote:
See attached files.

Configured using ASSERT_LEVEL_WANTED=0, the execution ratio improved
slightly to about 2:1.

Note that this test uses a wider range of ravel lengths. The execution
time ratio is pretty much constant, falling off only for the smallest
(100 element) ravels.

I did monitor the system this time. All 8 cores were busy at near to
100% utilization. Memory utilization remained below the swap threshold
during the entire test.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]