LinX - A simple Linpack interface

[xpower]

thanks @ stasio

ver 0.6.9-2017.0.011 still crashing ......

[stasio]

thanks @ stasio

ver 0.6.9-2017.0.011 still crashing ......

Look bellow (0.6.9K English Hotfix is Chinese version...):
http://hwtips.tistory.com/1611

[xpower]

Thanks for sharing

[stasio]

Intel Linpack 2017.1.014

http://registrationcenter-download.i...2017.1.014.zip

[stasio]

LinX v0.7.0 (2017.1)

http://www.mediafire.com/file/g0evl5...17.1.014%29.7z

[xpower]

^^^ LinX v0.7.0 same problem crash after 1min
but Alternate works fine....

[stasio]

Intel Linpack 2017.3.019

http://registrationcenter-download.i...2017.3.019.zip

[Fritz_the_germ]

What about linpack 2018.0.00.8?
This forum seams to be dead

[stasio]

What about linpack 2018.0.00.8?
This forum seams to be dead

LinX v0.8.0 (Intel/AMD Edition) :

http://hwtips.tistory.com/1611

http://registrationcenter-download.i...2018.0.008.zip

[Barr3l Rid3r]

here Linpack 0.8.0 AMD_AVX International

https://drive.google.com/file/d/0B-g...RpVzJtdm8/view

Please ignore the 0.6.5 tittle on the box, I'm lazy, libraries and dll running are all the latest same for performance and stress.

[Barr3l Rid3r]

LinX v0.8.0 (Intel/AMD Edition) :

http://hwtips.tistory.com/1611

http://registrationcenter-download.i...2018.0.008.zip

thanks bud!

[stasio]

LinX v0.8.0 Update 1

http://hwtips.tistory.com/1611

Intel Math Kernel Library (MKL) 2018.1.009 -Win

http://registrationcenter-download.i...2018.1.009.zip

[NEOAethyr]

Cool, thanks for the heads up.
It works in linux under wine just fine, also intel provides a native linpack binary somewhere, I messed with it once.

Does anyone know if Linx still uses the hard drive as temp storage?
It would be awesome if it got an update to stop that.
I killed my corsair ssd that way, had a hardlock and had to power off mid test, and because I had put it on my desktop for ease of access, it was a testing os...
That partition was on the ssd, totally killed it.
I think my samsung is safe in this respect but I still don't doit because I'm scared lol.
I always make sure to put this program on my normal hardrives just in case.
If I got a heads up sometime that it no longer did this it would be awesome news.

[xpower]

Cool, thanks for the heads up.
It works in linux under wine just fine, also intel provides a native linpack binary somewhere, I messed with it once.

Does anyone know if Linx still uses the hard drive as temp storage?
It would be awesome if it got an update to stop that.
I killed my corsair ssd that way, had a hardlock and had to power off mid test, and because I had put it on my desktop for ease of access, it was a testing os...
That partition was on the ssd, totally killed it.
I think my samsung is safe in this respect but I still don't doit because I'm scared lol.
I always make sure to put this program on my normal hardrives just in case.
If I got a heads up sometime that it no longer did this it would be awesome news.

thats very scary if it does uses ssd as temp. but I doubt. someone might have any more info on that?

[NEOAethyr]

It happened to my corsair in my sig, it's toast, it won't show up even in the bios after that.
I haven't had any probs when using it on actual platter hard drive.
Intel ssd's have a capacitor on them to prevent that, it ensures it writes everything it needs to.
It must have something todo with how it maps out the chips on the controller and it's cache.
I read something about my samsung 850 evo not having this prob, but I haven't taken the chance anyways.
But yeah last I've checked (been a while) Linx still writes temp files next to the exe, kinda like superpi.
Which reminds me holy crap, I better take that off my ssd lol.
If you wanted the best pi times, well..., since it uses temp files next to it, an ssd would be the way to go.
Better would be ram drive actually.

Anyways not to scare you but, I'de be careful putting any testing program that uses temp files on a ssd.
Perhaps newer ssd's are unaffected, but the 1st few gen's, bad idea.
Thing is I had it for years, and I've done tons of testing.
Just one time it screwed up during a hard lock, disappeared completely after forcing it to power off.
I usually hit the reset button, but there's times when that doesn't work and that was one of those times.
I still have the drive on top of my dresser, I tried it again months after it happened, same, it's gone.
No smart data issues, though tbh smart data is crap, it doesn't mean a darn thing other then if you got a bad drive from the get go, what I mean it is it should show all good, and it won't really tell you when a drive is bad.

Good thing I checked this thread, I totally forgot about superpi's temp files.
Not that I expect that to mess up in this day and age, but still.

[xpower]

I have used 850 evo too with previous ver of linx in win8.1 and testing right now with no temp file creates while test runs. that might be linux that create superpi temp? but this is interesting to find out though...

[stasio]

LinX v0.9.1 out....

http://hwtips.tistory.com/1611

[Fritz_the_germ]

Far out

[stasio]

Intel Math Kernel Library (MKL) 2018.2.010 -Win

http://registrationcenter-download.i...2018.2.010.zip

[xpower]

^^^thanks

anyone knows if OCCT linpack be updated with this latest update???

[stasio]

LinX v0.9.2 out....(chinese atm...)

http://hwtips.tistory.com/1611

[Fritz_the_germ]

Is there any english atm?
chinese sucks

[NEOAethyr]

Just a heads up, the Linux ver is here:
https://software.intel.com/en-us/art...nchmarks-suite

Readme stuff:
https://software.intel.com/en-us/mkl...pack-benchmark

Extract and browse to the dir (ex.):
l_mklb_p_2018.1.009.tgz\l_mklb_p_2018.1.009\benchm arks_2018\linux\mkl\benchmarks\linpack
Then run the command from the terminal:
./runme_xeon64

Make sure to be in the dir, otherwise it fails.

lininput_bench I believe is the example file. (Actually... that's probably the test run file I made..., that's what I'm thinking, edit: ah yep it's the one I created)
lininput_xeon64 I believe is the test run, ran with the above script "runme_xeon64".

You can customize your own runtime file...

Just some notes:
./xlinpack_xeon64 -i lininput_xeon64 25000 25000 1 4
1000 1000
2000 2000
5000 5008
10000 10000
18000 18008
20000 20016
22000 22008
25000 25000
26000 26000
27000 27000
30000 30000
35000 35000
40000 40000
45000 45000

5000 5008
18000 18008
20000 20016
22000 22008

If you make your own test script thing.
Start from at least 20000, the higher the better it seems.

./xlinpack_xeon64 -i lininput_xeon64 10000 10008 1 4
./xlinpack_xeon64 -i ./lininput_bench (<--- this cmd works, this is the custom run stuff, cat'ed file contents below for example)

The above is the default Linx setting.
I have to give you a heads up, the above cmd is no good though.
I swore it had to of worked at one time, I don't know...
Anyways just make the test file thing for it and it's all good.

cat ./lininput_bench
Shared-memory version of Intel(R) Distribution for LINPACK* Benchmark. *Other names and brands may be claimed as the property of others.
Sample data file lininput_xeon64.
1 # number of tests
10000 # problem sizes
10000 # leading dimensions
1 # times to run a test
4 # alignment values (in KBytes)

./xlinpack_xeon64 -i ./lininput_bench
OMP_NUM_THREADS=12 ./xlinpack_xeon64 -i ./lininput_bench
KMP_AFFINITY=nowarnings,compact,1,0,granularity=fi ne

./xlinpack_xeon64 -i ./lininput_bench
The threads setting slows it down, the affinity doesn't seem to do anything if much at all.

Some results to compare diff os'es:
wine
143.9650
142.8114

linux
159.1263

windows
problem size 10000
memory 772
run 20 times

size 10000
lda 10008
align 4
time 4.948
gflops 134.7674
residual ..
residual norm ..

In linux, cmd line only, no x, I get a whopping 180gflops ! (at this given setup for now, it's been slacked because of mem probs this last year, like I said, thinking I put the 4x sticks of ram in the wrong order)
Which is a pretty impressive boost, esp. compared to windows...

I also have the original superpi for linux, the site is long gone for superpi, but I managed to find it on my drive, thought I had lost them years ago.
If anyone wants it I can up it, just ask.
Some info on that.

./super_pi

16k
32k
64k
128k
256k
512k
1m
2m
4m
8m
16m
32m

7 128
8 256
9 512
10 1024 1k
11 2048 2k
12 4096 4k
13 8192 8k
14 16384 16k
15 32768 32k
16 64k
17 128k
18 256k
19 512k
20 1M
21 2
22 4
23 8
24 16
25 32

Superpi for linux reached ver 2.0, while the ver for windows is 1.9 I believe.
It's also a tad bit faster in linux then in windows, I had read this years ago but only tested it recently setting up some testing stuff for linux.

I have linux essentially ready togo as a testing platform, gaming is another issue... (it can be done but..., native linux games would be the best bet, like dying light for ex may be a good 0.5-2hrs test).
Main issue is I don't have a spare system to let this one go do it's thing...
It would be an extremely boring week without a pc ^^ :\.
And I know I'm unstable, I think I put my ram sticks in the wrong order when I was fooling around with 32gb just before I got the 1080ti.
Only have 16gigs in, I need to work on that 32gb thing eventually .
It's annoying to say the least...

I want to build a pi3b+ setup for casual internet browsing and dl'ing.
But it looks from my budget at the moment I have to wait until sept until I can do that.

Another issue is some of these windows only games I need windows for, I have yet to dl some of them, or even figure out gta5... (d2mn game won't run).
Looks like I need a reinstall for Bloodstained: ritual of the night, can't install the 2015 vc runtime .
Major pain in the bum...
Need part space for ubuntu 18.04...
Need a 2nd and 3rd part for winblows...
I can't even stand windows...
It would take me around 2 weeks to dl the updates for gta5 alone..., if I can even get it to doit.
Definitely need an alt system setup.

But anyways..., sooner or later I'll have a bootable usb setup for linux that is able to run all the needed stress tests.
You can do prime too, but who cares about that useless prog these days.
Y-Cruncher... and linpack is all I need.
Then gaming tests, which unfortunately windows is king in that department.

I wanted to make this post short, but also wanted to try to explain as much as possible so other's don't rush in blindly like I did trying to make it equ to the windows ver LinX.
What else was I gonna say?...
Oh I read something about setting up linpack not to use temp space, or to change it's location, I dn.
Anyways if you had to, which is pretty easy, setup a ram drive and just throw it in there.
That's what I'll do in the future when I'm ready for some serious testing.
No more hardlocks in the middle of testing and having to worry about my drive, going back and forth between cmos backups and etc (once I setup it for a ramdrive that is, already have one so np, it's just I need to setup scripting for it when I'm ready to do that).

Sorry for the random non-organized post.
I just thought of it, spur of the moment type of deal.

Edit:
Oh one last thing before I go, example superpi linux run, 1m from the cmd line..
Note I'm browsing and dl'ing right now..., don't care about the times even though it's at least 2 secs faster then windows doing nothing...

./super_pi 20
Version 2.0 of the super_pi for Linux OS
Fortran source program was translated into C program with version 19981204 of
f2c, then generated C source program was optimized manually.
pgcc 3.2-3 with compile option of "-fast -tp px -Mbuiltin -Minline=size:1000 -Mnoframe -Mnobounds -Mcache_align -Mdalign -Mnoreentrant" was used for the
compilation.
------ Started super_pi run : Mon Apr 2 05:44:27 EDT 2018
Start of PI calculation up to 1048576 decimal digits
End of initialization. Time= 0.117 Sec.
I= 1 L= 0 Time= 0.334 Sec.
I= 2 L= 0 Time= 0.381 Sec.
I= 3 L= 1 Time= 0.381 Sec.
I= 4 L= 2 Time= 0.381 Sec.
I= 5 L= 5 Time= 0.381 Sec.
I= 6 L= 10 Time= 0.381 Sec.
I= 7 L= 21 Time= 0.381 Sec.
I= 8 L= 43 Time= 0.382 Sec.
I= 9 L= 87 Time= 0.381 Sec.
I=10 L= 174 Time= 0.381 Sec.
I=11 L= 349 Time= 0.381 Sec.
I=12 L= 698 Time= 0.380 Sec.
I=13 L= 1396 Time= 0.381 Sec.
I=14 L= 2794 Time= 0.379 Sec.
I=15 L= 5588 Time= 0.379 Sec.
I=16 L= 11176 Time= 0.376 Sec.
I=17 L= 22353 Time= 0.369 Sec.
I=18 L= 44707 Time= 0.356 Sec.
I=19 L= 89415 Time= 0.329 Sec.
End of main loop
End of calculation. Time= 7.497 Sec.
End of data output. Time= 0.044 Sec.
Total calculation(I/O) time= 7.541( 0.172) Sec.
------ Ended super_pi run : Mon Apr 2 05:44:34 EDT 2018

[stasio]

Intel Math Kernel Library (MKL) 2018.3.011 -Win

http://registrationcenter-download.i...2018.3.011.zip

[stasio]

LinX v0.9.3 out....(chinese atm...)

http://hwtips.tistory.com/1611