General :: Command Returns Too Much Of System Time
Aug 20, 2009
I use the time command to measure the wall-clock time of a GPU implementation of an algorithm. When I time the CPU execution of the algorithm time returns a negligible sys time. However, when I time the GPU execution time returns a sys time that is around 20-30% of the total time. If that time was comparable with the negligible sys time of the CPU I would achieve a speedup of a few times higher.
I suspect that the increased sys time is because of the GPU usage, which, I assume, takes some time for the OS because of the drivers etc. I am not sure though, and it is important to figure this out because it will improve my results a lot if I can ignore the sys time and use just the user time for speedup calculations. Also, is there a way to see, in detail, what is the sys running and takes so much time. I am thinking that I might be able to see if it is the driver indeed that causes this delay.
First time here at LQ, so if the post is in the wrong place, this is a bit programming related. I've been trying to find the maximum resident memory of a process in a memory constraint situation.
Instead of using top, or ps, which gives me real-time snapshot at the memory usage of a process, is there a way to determine the maximum resident memory used by a process? /usr/bin/time seem to provide this functionality within the format string
[Code]...
The minor page-fault is suspicious, does that mean memory is being used.. but not recorded? if so, how can I know how much memory is being mapped?
I want to use ssh to execute a command and to wait endlessly to log everything (in file) that comes as a stream of the connected server. But unfortunately, in the manual its written "If command is specified, it is executed on the remote host instead of a login shell"
So what happens is that when I specify my command: ssh user@server "my_command"
It executed the command and the flow of execution returns to bash shell. So basically my session ends right after the command is executed. This happens only in case I specify command in the command line. If I login into ssh manually and then type "my_command", then the session doesn't end. I want the ssh not to exit, because after "my_command" executes, I want to capture everything in the session.
logging in a server through putty in the same network when i executed last command its showing system ip logged in time and logged out time the output as followsthis is my system oot pts1 xx.xx.xx day month date time in time out timeand similarly am geeting other than this likeroot :0day month date time still logged in this is from more than 3 days its logged in
i have a debian 5 vps system.. it reports the time as beeing one hour behind, i have tried to change this by setting the time to GMT+1 and setting the time to my local region (Europe-Brittian) using the tzselect command but none get the time to the correct time, one hour ahead of the current time.
I'm just wondering what the limits for time are. I have a program that always takes exactly 20 ms, so I assume this is the lowest it can measure, but I want to see if there's some sort of documentation of this.
get the values for the user time and system time for a process.i have tried getrusage to get values of ru_utime and ru_stimebut these don't seem to be correct
I have a service on my Suse 11 server which runs an ssh command (using openSSH) on another box.The output below is returned, but legitimate output is also returned as if the command had executed successfully. For example, the service executes an ls command through ssh, gets the error below, but also the contents of the remote directory as output. This is a problem because the service retries the command until it receives no error. I have been unable to replicate this manually from the command line. Does anyone know what might cause this or what this error really means?
I have the following function, and it appears to set the var correctly, but then tries and executes the line as a command. Anyone know how to keep it from doing that?
I have a test socket on a server. If I connect to it using telnet, I get exactly the response I expect. So I know the socket works, and the script on the server works:
Code: my-desktop:~$ telnet 192.168.1.1 3333 Trying 192.168.1.1... Connected to 192.168.1.1. Escape character is '^]'. RAM: 90 % Free ( 2793 M free, 3082 M total)
Connection closed by foreign host. And if I script it, I get almost the same response from Telnet, but still a valid response from the socket. So I know that my script works:
I'm using Ubuntu 9.10 with kernel 2.6.31 and gcc 4.4.1. I began getting the following errors from gcc after installing bison, g++ and g++-multilib and also after compiling 2.6.34-rc3 as a test. I've tried removing the packages but the error still shows up.
I'm timing how long it takes to run a command foo. I'm looking to append the results from the time command to a file, and discard the results from the foo command. I tried the following, but it didn't do what I want:
$ time ./foo > /dev/null >> output_from_time_command.txt
I am trying to use the time command to measure the execution time of a small program. The problem is that the command has three outputs. They look like this:
Code: $ time ./a.out real0m51.935s user0m51.060s sys0m0.040s
Should the execution time be the sum of the user and the sys time? sys time is really small.
I want to know how to set time format (12 hrs or 24 hrs) using command.I tried thisode:date +%T -s "2011-02-23 14:00"But it only displays 24 hrs format on TERMINAL but it does not SET 24 hrs format on the system
I am running a script with nohup and this generates a lot of logs.
In order to view the log I use tail -f nohup.out
The problem is that the info supplied by this command is not always the latest//sometimes I need to use the command again order to view the latest info added to the nohup.out file.
I would appreciate help with how to extract the date and time from at command jobs. From what I can tell, the date and time is embedded in the file name (/var/spool/atjobs).I'd be using this information in a (bash) shell script.
I was reading that if I want to do a one time scheduled command, I should use at, which I've never done, as opposed to cron, which i'm kinda familiar with. But what I want to do is reboot my server at 3am tomorrow and force it to check the file systems with a shutdown -rF. For this do I even need to use "at" or could I just say shutdown -rF 3:00.Will that also know that I mean 3am tomorrow and not say in 3 minutes from now or 3pm?
I am running shell command through C program using system() routine.
I am executing "opcontrol --status" an executable using the this routine and I get the following error. access: unix error (2) No such file or directory
But when I give the complete path to the executable it runs perfectly.
The executable is installed in "/usr/local/bin/" And the path variable has this path.
When I use the `date' command on RHEL5, I can get the time shown in 24 hr format. I wonder if the time can be shown in 12 hr format. But I don't want to use the `date +FORMAT' to do that, neither `alias date='date +FORMAT''. I just want a simple `date' to show time in 12hr format.Is there any configuration file about the 12/24 hr format? Is the format related to the value of the environment variable 'TIMEFORMAT'? I can't find its default value by 'set | grep -i timeformat'.How can I know what the current format is? I mean, when I use `date', I can get "Sun Sep 19 13:22:50 CST 2010", which seems like 24hr format; but when I use `hwclock', I can get "Sun 19 Sep 2010 01:23:05 PM CST -0.174299 seconds", which seems like 12hr format.
I am troubleshooting file copy time issues between 3 servers. I need to copy the same file from server A to both server B and server C and compare the elapsed times for the copy. Is there an easy way to do this?
When booting Fedora 11, my system hangs for a very long time on starting udev. Sometimes I get an I/O error. However, my hardware is fine. I do eventually get in to the system.
I am running my Ubuntu 32 bit server on top of Windows 7 64 bit with VirualBox. It's a 2 core Atom. It's been working good for about half a year. But the last about 6 weeks the system time only in Ubuntu is going slow. About -8 per 24 hours! I can only guess because I have more things running in my Windows 7 and Ubuntu.
I can set it right by coping the hareware time to system time with this command:
Code: hwclock --hctosys
I want to run a crontab to have that command run every minute. But it don't seem to run.
I want to get the system idle time till a mouse move or a key press. How is it possible to get it from a char terminal running through ssh/telnet as well as a from an X-terminal session?