Meriken's Tripcode Engine No.8

(Continued from )

This is a support thread for Meriken's Tripcode Engine.

3.2.2 (May 21, 2016)
Improve OpenCL/CUDA detection on POSIX systems, including Mac OS X.

"Meriken's Tripcode Engine English" is a cross-platform application
designed to generate custom/vanity tripcodes at maximum speed.
It is arguably the fastest and most powerful program of its kind.
(This program actually runs faster than the original mty or
MTY_CL on the same hardware!) It makes effecitive use of
available computing power of CPUs and GPUs, and the user
can specify flexible regex patterns for tripcodes. It features highly
optimized, extensively parallelized implementations of bitslice
DES and SHA-1 for OpenCL, AMD GCN, NVIDIA CUDA,
and Intel SSE2/AVX/AVX2.

"Meriken's Tripcode Engine" is free software, licensed
under GPLv3 and available for free download at the following link:

github.com/meriken/merikens-tripcode-engine-v3

Other urls found in this thread:

support.amd.com/en-us/download/desktop?os=Linux x86_64
comments.gmane.org/gmane.comp.programming.tools.cmake.devel/10070
cmake.org/files/v3.5/cmake-3.5.2-Darwin-x86_64.tar.gz
github.com/meriken/merikens-tripcode-engine-v3'
pastebin.com/GN700Vgw
reddit.com/r/archlinux/comments/4iczja/a_work_around_if_you_are_using_cuda_and_have/
wiki.archlinux.org/index.php/GPGPU#Using_CUDA_with_an_older_GCC
twitter.com/SFWRedditVideos

you should put the Linux version in the aur

Hope that guy with a hackintosh and GTX 970 is still around...

I dunno how exactly Linux packaging would work with OpenCL and CUDA. We will see.

I guess I came here too late today. Oh well.

it's midnight - 4am in the us right now lol

...

I know... I actually live in the states. It took me a while to fix build issues on OS X.

That is really a dinosaur fossil. I cannot believe people are still using it.

That guy is French, and it's almost noon right now.

oh cool you fixed it?
i'm that guy from yesterday. i'll pull and give it another try

wait, did you add the caveat about 10.10 (as opposed to OS X in general)? is 10.11 not supported?

There shouldn't be any problems.

very weird. last time it didn't have trouble but now it exits out with...
Testing MerikensTripcodeEngine...
./BuildAllWithGCC.sh: line 31: ./MerikensTripcodeEngine: No such file or directory
"MerikensTripcodeEngine -l 10" failed.


lots of issues like
/System/Library/Frameworks/OpenCL.framework/Headers/gcl.h(138): error: expected a ")"

>tfw too poor to use this

I have a hd7750 on Arch Linux.
Can I use this with open cl?

You could try this:

cd CMakeBuild
cmake -DENABLE_OPENCL=OFF ../SourceFiles

Probably not...

support.amd.com/en-us/download/desktop?os=Linux x86_64

You also need to run make after cmake, obviously.

just noticed that the last meaningful error seems to be
Error generating file
/users/myshortname/code/merikens-tripcode-engine-v3/CMakeBuild/CMakeFiles/cuda_compile.dir//./cuda_compile_generated_CUDA12.cu.o


it still makes the CMakeBuild directory but when i cd in there and try to do it with your command it also fails.

Yo meriken what gives I lost about 10% hashing speed with crimson drivers vs non crimson ones.
Love the program though.

That makes sense. Let me see...

Hmm... That's weird. Which video card are you using?

Reported for endorsing tripfaggetory.

Why is it this slow?

Add "^" at the beginning of the pattern.

I'm using a hackintosh, albeit with a 960. I'm not familiar with this tool but let me know if I can help.

This is a bug in CMake.

comments.gmane.org/gmane.comp.programming.tools.cmake.devel/10070

You just need to upgrade CMake. See the updated build instructions on GitHub.

didn't notice this, i'm maybe you're thinking of someone else. just to be clear, this is me

cool, thanks (and sorry for not doing this legwork myself - i don't know where i would begin to be honest)

looking at the link's comments, i'm a little lost...

thanks.
hmm so this adds ~15M, but shouldnt this GPU get at least 200M ?

No prob. These commands should fix the issue.

curl -sSL cmake.org/files/v3.5/cmake-3.5.2-Darwin-x86_64.tar.gz | sudo tar -xzC /opt
export PATH=/opt/cmake-3.5.2-Darwin-x86_64/CMake.app/Contents/bin:$PATH

oh hmm, i saw those in the readme about an hour ago and (thought i) ran it. i'll try it again just in case, but /opt/.../Cmake.app/Contents/bin is in my path now

oh i guess the homebrew version i had installed was showing up in the path first. i uninstalled that one to avoid dealing with path orders for now. the cmake from curling shows up now when i check where it is, so it's building now (and should (hopefully) work)

Non-Maxwell NVIDIA video cards are not particularly suitable for tripcode generation. You definitely want Compute Capability of 5.0 or higher.

Oh, good!

still crashing on ./BuildAllWithGCC.sh:

[...]/SourceFiles/MerikensTripcodeEngine.h:124:10: fatal error: 'cuda.h' file not
found
#include
^
3 warnings and 1 error generated.
make[2]: *** [CMakeFiles/MerikensTripcodeEngine.dir/Main.cpp.o] Error 1
make[1]: *** [CMakeFiles/MerikensTripcodeEngine.dir/all] Error 2
make: *** [all] Error 2

That's pretty odd...

install cuda devel package

i did that yesterday. i can do it again, but... seems like kind of a waste of time (unless there's good reason to think it'd help to do it again).

You simply add OGL/CUDA as package dependency.

What does this say?

ls /usr/local/cuda/include

I really should get a hackintosh myself.

>ls /usr/local/cuda/include

it's nice except for this kind of stuff

Is it this simple? I dunno...

When you make a build package yes,

>>ls /usr/local/cuda/include

cuda.h is right there! You can try:

cd CMakeBuild
cmake -DCUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda ../SourceFiles
make

I am trying to build the app with clang on OS X now. Let's see how that goes.

Really looking forward to that AUR PKGBUILD buddy :^).

Tripfags are the cancer of this site.

6:58:18 @ hck ~/code/tripping/CMakeBuild $ cmake -DCUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda ../SourceFiles
-- ARCH_64BIT = TRUE
-- ARCH_X86 = TRUE
-- Configuring done
-- Generating done
-- Build files have been written to: /Users/[...]/code/tripping/CMakeBuild
6:58:24 @ hck ~/code/tripping/CMakeBuild $ make
In file included from /Users/[...]/code/tripping/SourceFiles/Main.cpp:39:
/Users/[...]/code/tripping/SourceFiles/MerikensTripcodeEngine.h:124:10: fatal error: 'cuda.h' file not
found
#include
^
3 warnings and 1 error generated.

>regex
the hell would you expect

>tripfags

The cancer that was and is always killing this site

we discussed this in the last thread, newfag. cancer is cancer regardless of the name field. this just makes it easier to track and filter for autistic people who have tantrums when they see tripcodes.

This is *really* strange. You could try rebuilding everything with BuildAll.sh instead. If this doesn't work, I really need to get a hackintosh myself.

Thanks a lot for helping me with debugging, though. I really appreciate it.

I like your 2ch Browser.

no, thank you - i hate feeling like i'm just blindly pasting shit in, but i don't know enough about this stuff to really troubleshoot more than high level stuff

i'll try and see what comes up

I'll look into packaging after ironing out build issues.

It could be so much better, though. I just don't have enough time for my ideas.

That would be really great!

Thanks, really appreciated.

Let's see if I can create a Debian package first...

AUR WHEN

Crashed my display drivers when I run it

You could try "-b 2" option.

I mean, "-x 2" option.

Here's a bad beginning of the PKGBUILD

# This is an example PKGBUILD file. Use this as a start to creating your own,
# and remove these comments. For more information, see 'man PKGBUILD'.
# NOTE: Please fill out the license field for your package! If it is unknown,
# then please put 'unknown'.

# Maintainer: Your Name
pkgname=merikens-tripcode-engine-v3-git
_gitname=merikens-tripcode-engine-v3
pkgver=VERSION
pkgrel=1
pkgdesc='"Meriken''s Tripcode Engine" is a cross-platform custom tripcode generator.'
arch=('i686' 'x86_64')
url='github.com/meriken/merikens-tripcode-engine-v3'
license=('GPLv3')

# Find a way to say either/or
depends=() #('cuda' 'amdapp-sdk')
makedepends=('p7zip' 'cmake')
source=("git+$url")
md5sums=('SKIP')


pkgver() {
cd "$srcdir/$_gitname"
echo "$(git rev-list --count HEAD).g$(git rev-parse --short HEAD)"
}


build() {
cd "$srcdir/$_gitname"
./buildAll.sh
}


package() {
cd "$srcdir/$_gitname"
make DESTDIR="$pkgdir/" install
}

I'll try around a bit more. I commented out the deps for now because cuda is 800mb.

Nice!

Make a gui and add an stress test option.

AVX2+Opencl is the ultimate complete system stress test but you need to run 2 separate programs on Windows because none exist that features both.

The Japanese version already has these features. I will translate the GUI soon.

Why don't you use real regex?
Would it be hard to save the 'expanded' stuff so it doesn't have to build it every time? It takes a while...


Also,
RAR your *chan folder and upload it somewhere like mega, would be eternally grateful.

Hey Meriken:
# Maintainer: meriken

pkgname=merikens-tripcode-engine-v3-git
_gitname=merikens-tripcode-engine-v3
pkgver=110.g2d9cd3b
pkgrel=1
pkgdesc='"Meriken''s Tripcode Engine" is a cross-platform custom tripcode generator.'
arch=('x86_64')
url='github.com/meriken/merikens-tripcode-engine-v3'
license=('GPLv3')
optdepends=(
'cuda: Support for NVIDIA GPUs'
'amdapp-sdk: Support for AMD GPUs'
)
makedepends=('p7zip' 'cmake')
source=("git+$url")
md5sums=('SKIP')


pkgver() {
cd "$srcdir/$_gitname"
echo "$(git rev-list --count HEAD).g$(git rev-parse --short HEAD)"
}


build() {
cd "$srcdir/$_gitname"
./BuildAll.sh
}


package() {
cd "$srcdir/$_gitname/CMakeBuild"
make DESTDIR="$pkgdir/" install
}

This seems to work. I'm not sure if BuildAll.sh automatically picks up on available optional dependencies (I think it does). If it does, this should nearly be completely -- there should be a way to easily specify DENABLE_CUDA_DES_MULTIPLE_KERNELS_MODE=OFF, I think one would have to modify the make call in build() with all available flags then -- not sure.

What do you mean by "real regex"?
The regex engine of this program is optimized for tripcode generation.
With the GUI version, you can pause tripcode generation. Let me see what I can do with the CUI version.

Thanks! I will try it later and submit it to the AUR if it works. BuildAll.sh does detect optional dependencies automatically.

arch=('i686' 'x86_64') Use this instead, I think building should work transparently for both architectures.

Something to consider is using sed and some regex magic to enable/disable the testing step (this imho shouldn't be in BuildAll.sh) and enabling the CUDA_DES thing optionally by modifying the cmake line in the buildAll script.

So for instance if I wanted to search for tripcodes that contained the word 'Gentoo' what would patterns.txt be?

I agree. Testing is mostly for Travis-CI. I will modify BuildAll.sh.

testing. Temps holding at 82c

#noregex
Gentoo

or

#regex
^Gentoo

Don't forget to leave the last line blank.

how many trips/s are you guys getting?

im at ~550M using a R9 390, temps holding steady around 75°C

Temps up to 86.

I remember that old Japanese one that I used years ago used to smash the cpu to pieces and never get any codes over 6 characters.

Is ^gentoo faster than gentoo?

That's a pretty good speed.It should be OK as long as the temp stays below 90.

This program is very different from other tripcode generators as it has self-morphing codes everywhere.

The two patterns in are exactly the same.

Will a benchmarking aspect be added to the GUI once it gets added in?

Hey I'm the PKGBUILD gui. Just installed CUDA and tried again, small issue: pastebin.com/GN700Vgw

Without CUDA it works.

You could try g++-5. There seems to be a problem with a combination of g++-6, Boost, and CUDA.

I think so. Hope Mono works well with the GUI.

I'll try that, if it works I'll update the PKGBUILD accordingly.

Ugh still compiling gcc5 from the AUR. This is painful...

Ok, time to look stupid. What is a tripcode?

the thing this guy for example:

is using. It uniquely identifies you on 2ch-style boards. Just google around.

You could try clang. Just replace gcc with clang in BuildAllWithGCC.sh.

# Maintainer: meriken

pkgname=merikens-tripcode-engine-v3-git
_gitname=merikens-tripcode-engine-v3
pkgver=110.g2d9cd3b
pkgrel=1
pkgdesc='"Meriken''s Tripcode Engine" is a cross-platform custom tripcode generator.'
arch=('i686' 'x86_64')
url='github.com/meriken/merikens-tripcode-engine-v3'
license=('GPLv3')
optdepends=(
'cuda: Support for NVIDIA GPUs'
'amdapp-sdk: Support for AMD GPUs'
)
makedepends=('p7zip' 'cmake')
source=("git+$url")
md5sums=('SKIP')


pkgver() {
cd "$srcdir/$_gitname"
echo "$(git rev-list --count HEAD).g$(git rev-parse --short HEAD)"
}


build() {
cd "$srcdir/$_gitname"
./BuildAll.sh
}


package() {
cd "$srcdir/$_gitname/CMakeBuild"
make DESTDIR="$pkgdir/" install
}

Nope, I give up. Fix it to work with gcc6 please. When using gcc5 from AUR it doesn't even pass tiny compiler tests and when using clang it can't figure out where CUDA Is even when exporting the cuda root dir.

> Fix it to work with gcc6 please.

I suppose I need to find a workaround for bugs in libraries... Thanks for the hard work!

that would be neat. And of course, I'm interested in seeing this work on Arch.

I'll work on it!

It turned out this problem was more serious than I thought:

reddit.com/r/archlinux/comments/4iczja/a_work_around_if_you_are_using_cuda_and_have/

I'd wait for CUDA 8 for full CUDA support. We could still use NVIDIA cards as OpenCL devices, you know.

I mean, I'd wait for CUDA 8 for full CUDA support if you cannot use gcc-4.

Well that's disappointing :<

Might be a while before they release it.
There's also this: wiki.archlinux.org/index.php/GPGPU#Using_CUDA_with_an_older_GCC

I might muck around with this tomorrow.