Topic: Unfinished CryptoNight OpenCL (AMD) miner - page 4. (Read 20750 times)

nice 

don't tease me with a new post!!  watching this with interest and waiting for the more technical among us to unveil something new and revolutionary!

Did that. While the code is not the bottleneck, I'll tell you - that code is SLOW. Also, a little painful to look at. Just to prove it, I'll give you a little bit of code that is a drop-in replacement for his ccminer that shows what I'm talking about: http://pastebin.com/UqUJ7WKu

Like I said, it won't make it much faster, because it appears the code is not the bottleneck - but if you just look at it, it makes you think, "WTF was he smoking?"

Another example - in the older versions of ccminer, he had this:


That was used four times in a loop that executes 252,144 times. Now, the compiler can optimize some of this away - AES_BLOCK_SIZE and MEMORY are both constants, 16 and 1 << 21 (0x200000, off the top of my head.) Divides are OUCH slow, but to be fair the compiler should fix that. Now, he has since replaced it with this, showing he's getting better:


But he still doesn't seem to see the fail here - the result of E2I is multiplied by AES_BLOCK_SIZE without exception. Therefore...


... can replace the whole E2I macro PLUS the multiply. Why? The first divide by AES_BLOCK_SIZE and the multiply by the same that is always done after cancel. By the way, idk if 32-bit ops are any faster or slower on the GPU, but you don't need 64-bit width - you're masking out everything anyway. So if 32-bit is faster:


The 0 on the end is to mask out four bits - the bit shift down by four, then up by four would clear them, so an AND by zero does the same.

Bump 

Just a quick fist bump from me, just to say I'm eager to see this working!
Been supporting the guys over at PiMP lately as I thing they're doing an awesome work with their platform. Heard they had been in touch with you, hope you can build something great! 

test under pimp

wget http://ftp://ftp.gnu.org/gnu/autoconf/autoconf-2.69.tar.gz

tar -xvzf autoconf-2.69.tar.gz

cd autoconf-2.69/

./configure

make

make install

autoconf --version

cd ..

ln -s /opt/AMDAPP/lib/x86/libOpenCL.so /usr/lib/libOpenCL.so

wget https://github.com/wolf9466/opencl-cryptonight/archive/master.zip

unzip master.zip

cd opencl-cryptonight-master/

cp -R /opt/sgminer/winbuild/dist/include/CL CL

CFLAGS="-O2 -Wall -march=native" ./configure --prefix=/usr

make

minerd -a cryptonight -o stratum+tcp://mine.moneropool.org:80 -u 463tWEBn5XZJSxLU6uLQnQ2iY9xuNcDbjLSjkn3XAXHCbLrTTErJrBWYgHJQyrCwkNgYvyV3z8zctJL PCZy24jvb3NiTcTJ -p x -t 2
[2014-07-02 20:38:35] Using JSON-RPC 2.0
[2014-07-02 20:38:35] 1 miner threads started, using 'cryptonight' algorithm.
[2014-07-02 20:38:35] Starting Stratum on stratum+tcp://mine.moneropool.org:80
[2014-07-02 20:38:35] Pool set diff to 10000
[2014-07-02 20:38:35] Stratum detected new block
[2014-07-02 20:38:36] thread 0: 1 hashes, 0.96 H/s
[2014-07-02 20:38:40] thread 0: 59 hashes, 14.90 H/s


it will not support the gpu yet?
or I forget something?

Wolf0,
Do not worry, you are supported by those of us who can see clearly.  Even if we require corrective lenses to do so.
 

Is there any chance for Windows binary?

I support that.

How do you know Claymore's is faster? This doesn't even work yet - it uses ONE GPU work-item and ONE thread.

I do not claim authorship or reward. I want everyone to know - programmer vehre - dishonest

Then tell your friend he needs to stop ripping off MY open source work. This is simple - your friend vehre or whoever cloned my repo and did some OpenCL. I used my repo to do the OpenCL. So yes, they look similar.

It doesn't mean it will stay the fastest.
If more people get involved in the open GPU miner, we're probably going to see better optimisations.

Speaking of open-source gpu-miners... Wolf0, you might want to take a look at tsiv's latest release for cuda-miner... made a binary to mine on compute 2.1 and compute 2.0 (fermi) cards.

Maybe that can help?

My 1GB 560ti's get 200h/s each. Just saying.

It really shouldn't matter who made it first or claim come with it first, main focus is to get it working. Then credit can be given to all involved.

anyhow, I fail to see the interest claymore algo is way much faster... (if it is just to get the bounty.... at 50USD/hour, it would cost you more)

This topic has appeared on the following day after vehre refused to quit. This is also a chance?

I did not pay because they do not get the finished product.
Most likely, vehre sold his unfinished work

cryptonight_lobotomized.c - file name coincided by chance?