CCminer(SP-MOD) Modded GPU kernels.

[Epsylon3]

make && nvprof ./ccminer -a x11 -o stratum+tcp://mine.xpool.ca:8888 -u XeVrkPrWB7pDbdFLfKhF1Z3xpqhsx6wkH3 -p password

on windows it doesnt work well, but i prefer linux to do that much more faster to compile

on windows, you have nsight, which also gives you the timing of the different algo (an average is however missing)

Yes i use it for another reason, there are additional graphes where you can see what is "limiting" the perfs (often sign conversions)

[sp_]

aes.cu also do small improvements on echo (was the remaining diff of 0.2ms), but not on shavite. Seems ok
for groestl, i just didnt pick it (there is a real improvement too (0.5ms), but .... i dont understand it yet

int andmask1 = ((threadIdx.x & 0x03) - 1) >> 16;

This creates a mask of eighter 0x0000ffff or 0.

The code should create an andmask of 0x0000ffff only when if ((threadIdx.x & 0x03) == 0)

For the given inputs:

(3 & 0x03)-1 >>16=0
(2 & 0x03)-1 >>16=0
(1 & 0x03)-1 >>16=0
(0 & 0x03)-1>>16=0x0000ffff

The purpose of the shift is to remove the low bits in the andmask when the number is positive, and also set  the 16 low bits to 1 only when (threadIdx.x & 0x03) == 0 .

You also need to know that:

number^0=number
number&0=null

[sp_]

The SIMD change should give a couple of 100 KHASH on the 970. At least it did it for me when I tested it. Strange that you don't get any difference.

[th00ber]

from ccminer-tpruvot
I get this error at runtime "double free or corruption (out): 0x00007fecf00096c0 ***"

running Quark algo, but benchmark mode running fine

Same from ccminer-sp

[Epsylon3]

seems to run fine here, which os / release ?

Code:

[2014-11-07 07:39:08] qrk.suprnova.cc:6666 quark block 1428192
[2014-11-07 07:39:08] GPU #1: GeForce GTX 750 Ti, 5127 kH/s
[2014-11-07 07:39:12] GPU #0: GeForce GTX 750 Ti, 4478 kH/s
[2014-11-07 07:39:12] accepted: 558/560 (99.64%), 9604 khash/s yay!!!
[2014-11-07 07:39:15] GPU #0: GeForce GTX 750 Ti, 4363 kH/s
[2014-11-07 07:39:15] accepted: 559/561 (99.64%), 9489 khash/s yay!!!
[2014-11-07 07:39:19] qrk.suprnova.cc:6666 quark block 1428195

[th00ber]

git clone from yesterday, on Linux Kopiemtu 2 (cuda 6.5)
The previous version of the ccminer-tpruvot run well, but the one with sp_ optimizations do not work on my rig

[SS2006]

is adding neoscrypt to this possible/ Jealous of those AMD guys mining away at FTC. Cudaminer can't do neoscrypt and cgminer + NVIDIA is aweful.
I guess a better question is, is there anyone working on cudaminer just like you gentleman are taking ccminer to new heights?

vertcion is coming soon too to lyra2, and nvidia peeps wanna be ready!

"Vertans,

Firstly I’d like to introduce myself. Many of you will know me from Reddit as the developer of Greenpool. I have joined the development team recently to aid with the implementation of Lyra2RE and I plan to stay for the foreseeable future to see Vertcoin grow and ultimately succeed in this oversaturated crypto coin market.

Currently, development is well under way for the fork to Lyra2RE. The algorithms that Lyra2 will be chained with to form Lyra2RE have been decided and many essential pieces of software have been ported. A Python module for Lyra2RE has been created which means that Stratum-Mining and P2Pool are ready for the fork. Sgminer has also been modified to make it ready for the completion of a working OpenCL kernel. At this time, wallet integration needs to be completed as well as OpenCL and CUDA kernels created, the latter two of which we are looking for members of the community who are experienced with these systems to help us complete this task. Contact a432511@vertcoin.org if you have OpenCL or CUDA experience and would like to help out.

Regards,

Jamesl22"

[bathrobehero]

While mostly everything is about the X-series nowadays, I personally don't see them being viable long term and I never liked them due to some interesting things going on with them in the background like the massive, 3.5GH/s "rig" rentals, the occasional new x11 coins being dumped way below cost of GPU production and there are upcoming ASICs with cleverhash - however far that might be.

Anyway, my point is that I'd also be more interested in neoscryt, and I'm sure we could crowdfund it.

[sp_]

I think djm34 has made a cuda implementation of neoscrypt.

[Epsylon3]

A tip for windows users using Chrome :

type chrome://flags/ and disable WebGL

Chrome will be faster if you mine on the GPU (and the miner too

[djm34]

I think djm34 has made a cuda implementation of neoscrypt.

no, I haven't tried yet

[jpouza]

Watercooled 980 max hash X11:


subir fotos a internet

[sp_]

Checked in some more modded kernals:

x11 +300KHASH (980) NIST5 boost, faster x13,x15

https://github.com/sp-hash/ccminer

[sp_]

windows binaries here:

http://www.filedropper.com/release6

[sp_]

[italeffect]

A tip for windows users using Chrome :

type chrome://flags/ and disable WebGL

Chrome will be faster if you mine on the GPU (and the miner too

This made no difference to hash rate on my win 8 mining rigs.

[jpouza]

9MH/s +


subir fotos a internet

[Epsylon3]

A tip for windows users using Chrome :

type chrome://flags/ and disable WebGL

Chrome will be faster if you mine on the GPU (and the miner too

This made no difference to hash rate on my win 8 mining rigs.

Its not for rigs, i guess you dont use chrome on them, its more for normal users which like me have seen decreased perfs recently on all algos (when chrome is open)...

[Epsylon3]

Checked in some more modded kernals:

x11 +300KHASH (980) NIST5 boost, faster x13,x15

https://github.com/sp-hash/ccminer

perf is reduced on the 750Ti (linux), else i found the way to enhance a bit your groestl change (commited)
EDIT: hmm in fact not exactly, but... its hard to compare

my current version for x11 on a 750Ti / linux (2800kH) :

Code:

Time(%)      Time     Calls       Avg       Min       Max  Name
 20.75%  3.64387s        93  39.181ms  39.064ms  41.831ms  x11_echo512_gpu_hash_64(int, unsigned int, unsigned long*, unsigned int*)
 18.75%  3.29341s        94  35.036ms  34.963ms  39.033ms  quark_groestl512_gpu_hash_64_quad(int, unsigned int, unsigned int*, unsigned int*)
 12.87%  2.26079s        93  24.310ms  24.180ms  27.077ms  x11_shavite512_gpu_hash_64(int, unsigned int, unsigned long*, unsigned int*)
 11.11%  1.95157s        93  20.985ms  20.926ms  23.382ms  x11_simd512_gpu_expand_64(int, unsigned int, unsigned long*, unsigned int*, uint4*)
  7.24%  1.27073s        94  13.518ms  13.483ms  15.056ms  x11_cubehash512_gpu_hash_64(int, unsigned int, unsigned long*, unsigned int*)
  5.32%  933.86ms        94  9.9347ms  9.8739ms  11.096ms  quark_jh512_gpu_hash_64(int, unsigned int, unsigned long*, unsigned int*)
  5.04%  884.74ms        94  9.4122ms  9.2574ms  10.502ms  x11_luffa512_gpu_hash_64(int, unsigned int, unsigned long*, unsigned int*)
  3.08%  540.45ms        94  5.7494ms  5.7279ms  6.3724ms  quark_bmw512_gpu_hash_64(int, unsigned int, unsigned long*, unsigned int*)
  3.07%  539.01ms        93  5.7958ms  5.7538ms  5.8993ms  x11_simd512_gpu_compress2_64(int, unsigned int, unsigned long*, unsigned int*, uint4*, int*)
  2.80%  491.53ms        93  5.2852ms  5.1886ms  5.4446ms  x11_simd512_gpu_compress1_64(int, unsigned int, unsigned long*, unsigned int*, uint4*, int*)
  2.76%  484.47ms        94  5.1540ms  5.1358ms  5.7421ms  quark_blake512_gpu_hash_80(int, unsigned int, void*)
  2.71%  475.78ms        94  5.0615ms  5.0070ms  5.6117ms  quark_skein512_gpu_hash_64(int, unsigned int, unsigned long*, unsigned int*)
  2.60%  456.75ms        94  4.8591ms  4.8225ms  5.4034ms  quark_keccak512_gpu_hash_64(int, unsigned int, unsigned long*, unsigned int*)
  1.61%  283.04ms        93  3.0434ms  3.0159ms  3.3809ms  x11_simd512_gpu_final_64(int, unsigned int, unsigned long*, unsigned int*, uint4*, int*)
  0.28%  49.941ms        93  537.00us  534.29us  543.03us  cuda_check_gpu_hash_64(int, unsigned int, unsigned int*, unsigned int*, unsigned int*)

[Travis9x]

Code:

@ECHO off
setx GPU_MAX_ALLOC_PERCENT 100
ccminer.exe -a x11 -o stratum+tcp://us1.coinking.io:6666 -u Travis9x.ASRockX11 -p x -D
PAUSE

Code:

SUCCESS: Specified value was saved.
*** ccMiner for nVidia GPUs by Christian Buchner and Christian H. ***
         This is the forked version 1.4.7.SP (sp-hash@github)
          Built with VC++ 2013 and nVidia CUDA SDK 6.5

          based on pooler-cpuminer 2.3.2 (c) 2010 Jeff Garzik, 2012 pooler
            and HVC extension from http://hvc.1gh.com/

        Cuda additions Copyright 2014 Christian Buchner, Christian H.

        Include some of djm34 additions, cleaned by Tanguy Pruvot
                  Optimized Kernals By SP^Cryptoburnes

[2014-11-08 19:18:16] 2 miner threads started, using 'x11' algorithm.
[2014-11-08 19:18:16] Starting Stratum on stratum+tcp://us1.coinking.io:6666
[2014-11-08 19:18:16] Binding thread 0 to cpu 0
[2014-11-08 19:18:16] Binding thread 1 to cpu 1
[2014-11-08 19:18:17] Failed to get Stratum session id
[2014-11-08 19:18:17] Stratum difficulty set to 0.004
[2014-11-08 19:18:17] DEBUG: job_id=45ecd6d dc92 xnonce2=00000000 time=19:11:57
[2014-11-08 19:18:17] us1.coinking.io:6666 sent x11 block 2849
[2014-11-08 19:18:17] sleeptime: 500 ms
[2014-11-08 19:18:17] job 45ecd6d dc92 target change: f9ff060000 (1.0)
[2014-11-08 19:18:17] sleeptime: 500 ms
[2014-11-08 19:18:17] job 45ecd6d dc92 00000000
[2014-11-08 19:18:17] job 45ecd6d dc92 target change: f9ff060000 (1.0)
[2014-11-08 19:18:17] job 45ecd6d dc92 7fffffff
Cuda error in file 'C:/code/ccminer-sp/x11/cuda_x11_simd512.cu' in line 647 : in
valid texture reference.
Press any key to continue . . .

Any thoughts about this texture reference error?