Re: Something like "ping" for windows messages.
- From: "Peter Smithson" <Peter_Smithson@xxxxxxxxxxx>
- Date: 10 Dec 2006 01:15:40 -0800
Joseph M. Newcomer wrote:
Thanks for the long replies! I have read them and found it interesting
but I'm only going to deal with a few bits.
I mentioned there not being many processes running and you said it's a
matter of threads. I was trying to say that I'm not intentionaly
running any other apps. I'm aware that there will be many
processes/threads on the system but there's little that I can do about
that. When I run a benchmark, I run it until it produces the same
(ish) result every time. Then I know there's not disc cache issue or
luck involved.
Interesting idea about timing in PreTranslateMessage and the OnIdle -
not sure how that works though. Is there any gaurentee that OnIdle
will run? I thought there were problems with that if things get busy.
I saturated the network by running my test program (which generated lots of TCP/IP...
packets) on six other machines. So the time spent processing a SendMessage call is not
indicative of performance; you may have hit other limitations, such as the number of
kernel network buffers available, or the amount of network bandwidth available.
You don't want to test the message pump, you want to test the throughput of your HTTP
server.
No I don't. I mentioned earlier in a posting that I'm not interested
at all (at this stage) how fast the http server is. I'm only
interested in what negative effect it may have on the "real" work of
the server. The http server part is a secondary function of the
server.
Sticking it on a thread would obviously help but I want to get some
*rough* idea of the amount of benefit I could get. I don't want to do
the work without a reason or, do the work and not know if it's made a
difference.
I think part of my problem is that we have no benchmark test for the
main purpose of the server. Perhaps I should look into emulating real
messages from the HVAC controllers (which normally use another network
card to the corporate network) and get a measure of how fast the server
deals with them. I have a terrible feeling that this will be very
tricky (huge variation in results) due to the use of WM_TIMER and
polling that is deep in the ancient comms stack which I dare not touch
(and have not worked with).
It'd probably take me less time to just create at least one thread for
this server and forget it! But I'd like to get a rough idea of the
benefit if only to make me feel good!
Cheers
Peter
.
- Follow-Ups:
- Re: Something like "ping" for windows messages.
- From: Joseph M . Newcomer
- Re: Something like "ping" for windows messages.
- References:
- Something like "ping" for windows messages.
- From: Peter Smithson
- Re: Something like "ping" for windows messages.
- From: Joseph M . Newcomer
- Re: Something like "ping" for windows messages.
- From: Peter Smithson
- Re: Something like "ping" for windows messages.
- From: Joseph M . Newcomer
- Something like "ping" for windows messages.
- Prev by Date: Re: ot - updating Visual studio 2005
- Next by Date: Re: Something like "ping" for windows messages.
- Previous by thread: Re: Something like "ping" for windows messages.
- Next by thread: Re: Something like "ping" for windows messages.
- Index(es):
Relevant Pages
|