CCminer(SP-MOD) Modded GPU kernels.

[Ayers]

Hi guys

I have a GTX 960. What is the expected hash rate? Also, which version should I download for Windows 10 and where can I get it? I have been looking at this thread for a long time, but can only see download links in the first couple of posts.

Thanks.

hashrate for what algo? i think you cna do fine with etheruem and 15mega or decred around 1GB, and windows 10 is not recomended, because even if they fixed the lower hashrate, still it's not on par with windows 7 hashrate, for ethereum i'm talking

[sp_]

I have done some tests. I expect the final decred mod #4 will be able to do 3.1 ghash on the 980ti. The 750ti will gain around +30mhash +37MHASH from release #3

[sp_]

Ok. I have found away to do the optimal decred kernal now.
http://stackoverflow.com/questions/15842507/passing-the-ptx-program-to-the-cuda-driver-directly
So I will generate the ptx assembly with the midstate data included in the instructions. Then for every time the midstate is changing, I recompile the kernal runtime with the API calls described in the article.
To estimate the speedgain you can replace all the constant mem access with contstants in the 1.7.4 code.. Release #4 will be optimal..

Interesting technique.
But I doubt you'll gain even 1% from it, likely less.

you will, because some of the first rounds will be gone.. (instructions are removed since they work on constant data..) You can try it. replace the d_data[0]...d_data[23] with constant data 0x01234567 etc; make sure every constant is different from each other.. Compile,read the ptx, and count the lines before and after.

Then you don't have 14 round blake kernal. but a 12 rounds blake kernal that only works for one midstate. And solves the 14 round blake problem for one given midstate.

I just perfomed the test in cuda 7.5 The self modified code decred kernal is only using 28 registers. (Down from 80). The ptx Assembly linecount is Down 200 lines from sp-mod #3 (10%)

On the gtx960M(release #4 is 8% faster than release #3) Edit:  I think I can tune it bether, because the register Count has dropped. 512Threads per Block instead of 256 --->+25mhash etc

I kinda doubt there's a documented and stable, supported method of doing so...

You can do it safe:

1. Put the compiled cubin in a ramdisk. (virtual memory drive)
2. Poke the constant values with the cpu directly in the file. (the locations can be found with disassembly and the offsets might change from compiler to compiler (cuda versions) )
2. call the cuda api call cuModuleLoad

https://www.cs.cmu.edu/afs/cs/academic/class/15668-s11/www/cuda-doc/html/group__CUDA__MODULE_g366093bd269dafd0af21f1c7d18115d3.html

If the cubin files are Equal on Windows and Linux I could make a Linux Version of release #4

[Velgelm]

Nice news! i`ll be waiting #4 ...

[sp_]

Get ready for decred private miner version #4 (0.1 BTC)

with -i 30.9 and more overclocking

No more buyers meens no more releases:) Release #4 will be tested  for a couple of weeks in my private farm.

Later today, I am releasing version #4 of the Decred private miner to the donators. The gtx960M  is up 8% since release #3. I expect the 980 ti to do +200MHASH, but I haven't tested the last build. There will probobly be a build #5 with fintuning and adjustments for the different models later.

[sp_]

Email sendt. Check your spam filters, and report your results

[sp_]

A refork of 1.7.4 with all kernals replaced is coming soon. I will also release a new version of the "All nicehash kernals" optimized with more quark,lyra2v2 and keccak hashrate and added optimized decred. (decred private miner #2)

[sp_]

Smart miners donate 0.1BTC to get a 20% boost in their decred profit

yiimp.ccminer.org

[sp_]

I have 5 private miners for the people who donate.

1. 0.1BTC: Pentablake +100-120% (3 releases)
2. 0.1BTC: Cryptonight +10% (one release)
3. 0.1BTC: Spreadcoin +10-20% (with full sourcecode / linux compatible) (9 releases)
4. 0.1BTC: All nicehash algos optimized. 0-10% (4 releases)
5. 0.1BTC: decred +18-30% (4 releases)

Donators will recieve updates...

[DragonSlayer]

Someone want to post what decred version 4 is hashing on a 980?

[chrysophylax]

Someone want to post what decred version 4 is hashing on a 980?

i will when i get the #4 version ...

#crysx

[sp_]

sendt it now. What are your hashrates?

[lmaonade80]

Not mining on a 980 but getting about 580-610 Mhs on a 750ti. Thanks SP

[Velgelm]

My results :
Version 3


Version 4


Thx SP, nice work!

[sp_]

Thanks for submitting results

2x 960 was doing 1860Mhash with the vanilla 1.7.4 release

1860---》2200 (+21%)

21% faster with the sp-mod #4

[bensam1231]

Get ready for decred private miner version #4 (0.1 BTC)

with -i 30.9 and more overclocking

No more buyers meens no more releases:) Release #4 will be tested  for a couple of weeks in my private farm.

Later today, I am releasing version #4 of the Decred private miner to the donators. The gtx960M  is up 8% since release #3. I expect the 980 ti to do +200MHASH, but I haven't tested the last build. There will probobly be a build #5 with fintuning and adjustments for the different models later.

#4 is faster then #3, and #3 is faster then #2. However, #2 has steady GPU utilization and 3/4 do not. The GPUs oscillate constantly.

Not certain if this effects hashrate at the pool, but it's definitely noticeable on the rigs. There isn't high cpu utilization either.

Pools are also showing about 75% of the speed I should be getting. In release #2 it's closer to 90-92%.

[sp_]

I have only tested on yiimp.ccminer.org and I get correct values on the pool for both release #3 and #4.

[CrazyRussian]

Later today, I am releasing version #4 of the Decred private miner to the donators. The gtx960M  is up 8% since release #3. I expect the 980 ti to do +200MHASH, but I haven't tested the last build. There will probobly be a build #5 with fintuning and adjustments for the different models later.

version #3
GTX970 G1Gaming - 1600Mh
GTX750Ti OC - 580Mh

version #4
GTX970 G1Gaming - 1760Mh
GTX750Ti OC - 620Mh

_SP , Nice work!

[Ayers]

Later today, I am releasing version #4 of the Decred private miner to the donators. The gtx960M  is up 8% since release #3. I expect the 980 ti to do +200MHASH, but I haven't tested the last build. There will probobly be a build #5 with fintuning and adjustments for the different models later.

version #3
GTX970 G1Gaming - 1600Mh
GTX750Ti OC - 580Mh

version #4
GTX970 G1Gaming - 1760Mh
GTX750Ti OC - 620Mh

_SP , Nice work!

1760 on a single 970 is very nice, but on what core and memory?

[fenomenhaa]

@Sp

rev 4 increase from rev3 612mh/s to -663mh/S great work SP...