LinX - A simple Linpack interface

Printable View

Show 100 post(s) from this thread on one page

12-15-2008, 08:31 AM
Dua|ist

Thank you guys for your support.

Bad news is, I don't have enough time right now to continue coding (it's really getting hard here in my university at the end of term). :(

But good news is that I've found a way to set the number of threads for Linpack to use. That's something like what koda has suggested. :) I'm not sure if I can make the affinity to be set from within LinX, but right now it is possible to make Linpack use less threads than the number of cores of a CPU (running it in just one thread works well on my rig). When running only on one core the affinity can be set from within Task Manager I think.

Higher values can be set too, but on my dual-core E8500 I haven't found any difference in performance or whatever with number of threads set to anything higher than 2, so not sure about this. It will be interesting to try and set this value to 8 for a Core i7 system though. It just can't be that easy to force Linpack load all 8 cores in one instance, but who knows? Maybe it is. ;) I'll upload this version as soon as I have some time to finish it and "polish" some things. You'll have to be patient. :)
12-15-2008, 09:27 AM
WaterFlex

You do the great work, Alex :up:
Keep doing your job and good luck with your term exams ;)
12-15-2008, 10:16 AM
koda

You're great mate !

Well, i'm not sure you can set affinity on each core but :
If you click on start with an option limited to one thread ... we will have a new process ... something like linpack32_xeon32.exe (can't remember the exact name) ... then just set affinity to it.

Then do the same again for a second LinX with his linpack process and affinity to another core ... and so on ;)

When one crash ... just look each core in task manager ;)

Take your time ... you tool is great ... we will be patient !

Quote:

Originally Posted by Dua|ist

Thank you guys for your support.

Bad news is, I don't have enough time right now to continue coding (it's really getting hard here in my university at the end of term). :(

But good news is that I've found a way to set the number of threads for Linpack to use. That's something like what koda has suggested. :) I'm not sure if I can make the affinity to be set from within LinX, but right now it is possible to make Linpack use less threads than the number of cores of a CPU (running it in just one thread works well on my rig). When running only on one core the affinity can be set from within Task Manager I think.

Higher values can be set too, but on my dual-core E8500 I haven't found any difference in performance or whatever with number of threads set to anything higher than 2, so not sure about this. It will be interesting to try and set this value to 8 for a Core i7 system though. It just can't be that easy to force Linpack load all 8 cores in one instance, but who knows? Maybe it is. ;) I'll upload this version as soon as I have some time to finish it and "polish" some things. You'll have to be patient. :)
12-28-2008, 09:57 AM
Dua|ist

LinX 0.5.4 with home-made HT support

Thank you all for your patience, here it is: LinX 0.5.4

Changes in this version:
- HyperThreading support added. This is done not in a best way probably, as two instances of Linpack are launched at the same time (their performance is then summed up). This was tested on a P4 and an Atom and worked ok in both cases
- Thread control added. You can now set the number of threads Linpack will create and thus control the number of physical cores that will be stressed. This setting is located in the Settings window (who'd have thought? :D )
- Settings window with all existing and a couple of new settings added. No need to edit the ini-file manually now
- Main menu redesigned
- Lots of minor bugs found and fixed (and who knows how much new are added? :rolleyes: )
- Performance optimized a little bit more
- Graphs have now customizable top and bottom limits
- And changed the icon a bit :cool:

Your comments are welcome as usual. I'm eager to hear how it performs on Nehalems now. :)

P.S. Screenshots and updated first post tomorrow, no time right now.
12-28-2008, 11:09 AM
rge

1 Attachment(s)

Dualist, could not get to work on my core i7. Default was set to 4 threads and HT box was checked. When ran, realtemp, loadtester, speedfan all say no load or 1% load. Checked taksmanager and says cpu idle 99%. I changed to 8 threads, and same. see pic. Any ideas? What is weird is that with it running, system seem very slow...like it is running...something is weird.
12-28-2008, 05:19 PM
andressergio

Quote:

Originally Posted by Dua|ist

Thank you all for your patience, here it is: LinX 0.5.4

Changes in this version:
- HyperThreading support added. This is done not in a best way probably, as two instances of Linpack are launched at the same time (their performance is then summed up). This was tested on a P4 and an Atom and worked ok in both cases
- Thread control added. You can now set the number of threads Linpack will create and thus control the number of physical cores that will be stressed. This setting is located in the Settings window (who'd have thought? :D )
- Settings window with all existing and a couple of new settings added. No need to edit the ini-file manually now
- Main menu redesigned
- Lots of minor bugs found and fixed (and who knows how much new are added? :rolleyes: )
- Performance optimized a little bit more
- Graphs have now customizable top and bottom limits
- And changed the icon a bit :cool:

Your comments are welcome as usual. I'm eager to hear how it performs on Nehalems now. :)

P.S. Screenshots and updated first post tomorrow, no time right now.

Great Work mate :up: it became one of my preferred tools for testing stability

Cheers
Sergio
12-28-2008, 09:09 PM
Dua|ist

Quote:

Originally Posted by rge

Dualist, could not get to work on my core i7. Default was set to 4 threads and HT box was checked. When ran, realtemp, loadtester, speedfan all say no load or 1% load. Checked taksmanager and says cpu idle 99%. I changed to 8 threads, and same. see pic. Any ideas? What is weird is that with it running, system seem very slow...like it is running...something is weird.

Well, I haven't tested it with Core i7, and this was somehow expected. If you uncheck the HT box, it'll (it should) run just like before, only on 4 cores.

Could you please check whether 2 linpack_xeon32.exe processes are running with HT box checked or not? But if it says idle 99% - obviously nothing is running at all. :(

Edit: I'm an idiot. :) We're trying to start two linpacks at the same time, each with maximum memory. That must be the problem. If it is (I sincerely hope so) then with lower memory (say, half from max) values it should work.
12-28-2008, 09:34 PM
rge

Quote:

Originally Posted by Dua|ist

Well, I haven't tested it with Core i7, and this was somehow expected. If you uncheck the HT box, it'll (it should) run just like before, only on 4 cores.

Could you please check whether 2 linpack_xeon32.exe processes are running with HT box checked or not? But if it says idle 99% - obviously nothing is running at all. :(

Edit: I'm an idiot. :) We're trying to start two linpacks at the same time, each with maximum memory. That must be the problem. If it is (I sincerely hope so) then with lower memory (say, half from max) values it should work.

:eek: I was the idiot for running it that way...that would explain the BSOD/linx crashing when I clicked on it first time.:rofl:

Yep, works fine using 1/2 memory, load is 99.9%:up:
12-28-2008, 10:27 PM
safan80

Quote:

Originally Posted by Dua|ist

Thank you all for your patience, here it is: LinX 0.5.4

Changes in this version:
- HyperThreading support added. This is done not in a best way probably, as two instances of Linpack are launched at the same time (their performance is then summed up). This was tested on a P4 and an Atom and worked ok in both cases
- Thread control added. You can now set the number of threads Linpack will create and thus control the number of physical cores that will be stressed. This setting is located in the Settings window (who'd have thought? :D )
- Settings window with all existing and a couple of new settings added. No need to edit the ini-file manually now
- Main menu redesigned
- Lots of minor bugs found and fixed (and who knows how much new are added? :rolleyes: )
- Performance optimized a little bit more
- Graphs have now customizable top and bottom limits
- And changed the icon a bit :cool:

Your comments are welcome as usual. I'm eager to hear how it performs on Nehalems now. :)

P.S. Screenshots and updated first post tomorrow, no time right now.

this version will not run the 64bit test for me even when I set x64=1 in the ini file. 5.3 worked fine.
12-29-2008, 05:27 AM
Dua|ist

rge, thanks for testing the HT stuff. Here's the updated version, which should take the HT checkbox into account and use only half of all available memory as maximum with it checked.
I also fixed a bug when 0.5.4 wouldn't run on a machine with 8GB+ RAM.
LinX 0.5.4.1

Quote:

Originally Posted by safan80

this version will not run the 64bit test for me even when I set x64=1 in the ini file. 5.3 worked fine.

Uhm, I don't remember myself changing anything x64-related. Could you please provide more details? The x64 box is disabled?
12-29-2008, 06:01 AM
stasio

Thank's for update.
Btw,any another link,as many of us are not Premium user on RS.
Uncleweb and me using fileden.com with direct download.
12-29-2008, 06:10 AM
Tobit

Quote:

Originally Posted by stasio

Thank's for update.
Btw,any another link,as many of us are not Premium user on RS.
Uncleweb and me using fileden.com with direct download.

I'm not a premium member of RS, I only had to wait 27 secs to download it. What's the beef with having to wait only a few seconds? :down:
12-29-2008, 06:16 AM
safan80

1 Attachment(s)

Quote:

Originally Posted by Dua|ist

rge, thanks for testing the HT stuff. Here's the updated version, which should take the HT checkbox into account and use only half of all available memory as maximum with it checked.
I also fixed a bug when 0.5.4 wouldn't run on a machine with 8GB+ RAM.
LinX 0.5.4.1

Uhm, I don't remember myself changing anything x64-related. Could you please provide more details? The x64 box is disabled?

the box is disabled in 5.4. and I can't get it to enable even when I change the ini.

downloading 5.4.1 now will update when it's done

works great now!!! thank you!
12-29-2008, 07:51 AM
Dua|ist

Quote:

Originally Posted by stasio

Thank's for update.
Btw,any another link,as many of us are not Premium user on RS.
Uncleweb and me using fileden.com with direct download.

Thanks for that, I don't like Rapidshare either. Just registered at fileden, here's the direct link: LinX 0.5.4.1
12-29-2008, 08:46 AM
stasio

Thank's:clap:
12-29-2008, 08:53 AM
slim142

Dualist, Im impressed :clap:

I will finally get to test my 4Gbs of RAM now that Im on Windows XP Pro x64
12-29-2008, 08:58 AM
rge

1 Attachment(s)

New one is working well on i7, using max memory sets it to one half each for 2 instances, and it max loads all 8 threads:up:
12-29-2008, 09:19 AM
Dua|ist

rge, that is simply great! So glad to hear that it WORKS! :D

Quote:

Originally Posted by slim142

Dualist, Im impressed :clap:

I will finally get to test my 4Gbs of RAM now that Im on Windows XP Pro x64

Thanks, you can always share your results here. ;)
12-30-2008, 02:17 AM
safan80

1 Attachment(s)

I've encountered a problem with LinX 5.4.1 and the 25,000 test. whenever it's about to finish it's run it errors out right as it's about to finish the last run. I've had it at stock too. the amount of tests do not matter it's almost as if it has trouble unloading when it is done. here's example with just 3 tests.

edit: I hope somebody can repeat this or else it might be the bios I'm using.
12-30-2008, 04:32 AM
Extelleron

Quote:

Originally Posted by safan80

I've encountered a problem with LinX 5.4.1 and the 25,000 test. whenever it's about to finish it's run it errors out right as it's about to finish the last run. I've had it at stock too. the amount of tests do not matter it's almost as if it has trouble unloading when it is done. here's example with just 3 tests.

edit: I hope somebody can repeat this or else it might be the bios I'm using.

I had the same problem with my setup, ran 199 out of 200 iterations and then suddenly stopped before the last one.

http://img243.imageshack.us/img243/8721/oc38002lf4.jpg
12-30-2008, 06:41 AM
rge

Yep same here. Funny on short runs 3-5 runs all the way through...but on 25 run, it ran 24 than said unexpectedly stopped before getting to 25, always does it on last run on longer runs.
12-30-2008, 08:41 AM
Dua|ist

That's it, my noobness and no Nehalems around to test on. I think I know why this happens, but can't reproduce that on Atom.
I should have somehow said that HT support is... sort of beta. But since the version number is starting with 0 anyway, I hope you can forgive me that.
Here you go, LinX 0.5.4.2, hopefully fixed.
12-30-2008, 08:55 AM
Ghostleader

You do the great work, keep it up :up: :clap:

I just love LinX, it´s so easy to use :yepp:, great interface. :yepp:

It´s my number one tool for check CPU stability, it stress the living daylight out of the my CPUs.
12-30-2008, 09:14 AM
slim142

Quote:

Originally Posted by Ghostleader

You do the great work, keep it up :up: :clap:

I just love LinX, it´s so easy to use :yepp:, great interface. :yepp:

It´s my number one tool for check CPU stability, it stress the living daylight out of the my CPUs.

This :cool:
12-30-2008, 11:01 AM
rge

1 Attachment(s)

0542 fixed the stopping before last one on mine:up:

Show 100 post(s) from this thread on one page