<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
I don't think PCJACOBI is working on the GPU yet.<br>
<br>
I have some local fixes to make it work efficiently (without extra
copies), but changes won't be pushed too soon because the design is
not sufficiently general.<br>
<br>
-Paul<br>
<blockquote
cite="mid:CAM9tzSmUXOih3E9k6daP+_=WpSiTJK9tYjkxheyU_d8D+77r8Q@mail.gmail.com"
type="cite">
<div class="gmail_quote">2012/3/13 Xiangze Zeng <span dir="ltr"><<a
moz-do-not-send="true" href="mailto:zengshixiangze@163.com">zengshixiangze@163.com</a>></span><br>
<blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt
0.8ex; border-left: 1px solid rgb(204, 204, 204);
padding-left: 1ex;">
<div>I do set preallocation after setting the matrix type.</div>
</blockquote>
<div><br>
</div>
<div>Run with -info to make sure it is used (i.e. that the
matrix type isn't changed later). Note that PCSOR does not run
on the GPU, so it will do lots of copying (run with
-log_summary to see). You should start by running PCJACOBI on
the GPU.</div>
<div> </div>
<blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt
0.8ex; border-left: 1px solid rgb(204, 204, 204);
padding-left: 1ex;">
<div> <br>
Zeng</div>
<div><br>
在 2012-03-13 22:14:12，"Jed Brown" <<a
moz-do-not-send="true" href="mailto:jedbrown@mcs.anl.gov"
target="_blank">jedbrown@mcs.anl.gov</a>> 写道：
<div class="im">
<br>
<blockquote style="padding-left: 1ex; margin: 0px 0px 0px
0.8ex; border-left: 1px solid rgb(204, 204, 204);">
<div class="gmail_quote">2012/3/13 Xiangze Zeng <span
dir="ltr"><<a moz-do-not-send="true"
href="mailto:zengshixiangze@163.com"
target="_blank">zengshixiangze@163.com</a>></span><br>
<blockquote class="gmail_quote" style="margin: 0pt 0pt
0pt 0.8ex; border-left: 1px solid rgb(204, 204,
204); padding-left: 1ex;">
At the beginning and end of the codes for setting
the matrices values, I add "printf", and compute the
time of this period. It is much longer than that
when I don't use the GPU. I just guess the time is
used for copping data. My PCTYPE is sor. And 2000
iterations. Do you have any suggestion about this?</blockquote>
</div>
</blockquote>
</div>
</div>
</blockquote>
</div>
<br>
</blockquote>
<br>
</body>
</html>