Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 876. (Read 2347659 times)

Whirlpool(x) is better on AMD as well :-)

Nice. I will rewrite the branch pass so that the hashrate will skyrocket, but I don't think I can beat you in this algo even on the 980ti..
The Whirlpool needs to be rewritten.

Hashpower.com is up again. Estimated quark payout. 0.27BTC/GHASH. Actual(24h) 0.15BTC/GHASH. They have alot of coins, and adding new coins every month.  In miner control you can select wich column you want to use. Estimates or actual payouts or both.

Nice modifications, might try it out later

That is because both hetfy1.c and hefty1.cpp is in the makefile. Try to remove one of them.

that doesnt seem to work ...

-------

sph/ccminer-hefty1.o: In function `HEFTY1_Init':
hefty1.cpp:(.text+0x980): multiple definition of `HEFTY1_Init'
ccminer-hefty1.o:/mnt/compile/ccminer-all/sp-hash/ccminer/hefty1.c:295: first defined here
sph/ccminer-hefty1.o: In function `HEFTY1_Update':
hefty1.cpp:(.text+0x9f0): multiple definition of `HEFTY1_Update'
ccminer-hefty1.o:/mnt/compile/ccminer-all/sp-hash/ccminer/hefty1.c:306: first defined here
sph/ccminer-hefty1.o: In function `HEFTY1_Final':
hefty1.cpp:(.text+0xb20): multiple definition of `HEFTY1_Final'
ccminer-hefty1.o:/mnt/compile/ccminer-all/sp-hash/ccminer/hefty1.c:322: first defined here
sph/ccminer-hefty1.o: In function `HEFTY1':
hefty1.cpp:(.text+0xcf0): multiple definition of `HEFTY1'
ccminer-hefty1.o:/mnt/compile/ccminer-all/sp-hash/ccminer/hefty1.c:359: first defined here
collect2: error: ld returned 1 exit status
Makefile:944: recipe for target 'ccminer' failed
make[2]: *** [ccminer] Error 1
make[2]: Leaving directory '/mnt/compile/ccminer-all/sp-hash/ccminer'
Makefile:1638: recipe for target 'all-recursive' failed
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory '/mnt/compile/ccminer-all/sp-hash/ccminer'
Makefile:564: recipe for target 'all' failed
make: *** [all] Error 2

-------

#crysx

Try to add this to the makefile

sph/hefty1.cpp

and rename all axiom files to bastion (in the makefile)

sp ...

the latest git seems to have a compile error ...

-------

mv -f sph/.deps/ccminer-haval.Tpo sph/.deps/ccminer-haval.Po
make[2]: *** No rule to make target 'x15/axiom.cu', needed by 'x15/axiom.o'.  Stop.
make[2]: *** Waiting for unfinished jobs....
ptxas info    : 0 bytes gmem, 1568 bytes cmem[3]
ptxas info    : Compiling entry function '_Z22pentablake_gpu_hash_64jjPm' for 'sm_50'
ptxas info    : Function properties for _Z22pentablake_gpu_hash_64jjPm
    128 bytes stack frame, 0 bytes spill stores, 0 bytes spill loads
ptxas info    : Used 92 registers, 336 bytes cmem[0]
ptxas info    : Compiling entry function '_Z22pentablake_gpu_hash_80jjPv' for 'sm_50'
ptxas info    : Function properties for _Z22pentablake_gpu_hash_80jjPv
    128 bytes stack frame, 0 bytes spill stores, 0 bytes spill loads
ptxas info    : Used 90 registers, 336 bytes cmem[0]
ptxas info    : Compiling entry function '_Z25pentablake_gpu_check_hashjjPjS_' for 'sm_50'
ptxas info    : Function properties for _Z25pentablake_gpu_check_hashjjPjS_
    0 bytes stack frame, 0 bytes spill stores, 0 bytes spill loads
ptxas info    : Used 6 registers, 344 bytes cmem[0]
ptxas info    : 0 bytes gmem
pentablake.cu: In function ‘int scanhash_pentablake(int, uint32_t*, const uint32_t*, uint32_t, long unsigned int*)’:
pentablake.cu:520:57: warning: passing NULL to non-pointer argument 2 of ‘cudaError_t MyStreamSynchronize(cudaStream_t, int, int)’ [-Wconversion-null]
pentablake.cu:535:57: warning: passing NULL to non-pointer argument 2 of ‘cudaError_t MyStreamSynchronize(cudaStream_t, int, int)’ [-Wconversion-null]
ptxas info    : 0 bytes gmem
ptxas info    : 0 bytes gmem
ptxas info    : Compiling entry function '_Z21groestl256_gpu_hash32jjPmPjj' for 'sm_50'
ptxas info    : Function properties for _Z21groestl256_gpu_hash32jjPmPjj
    0 bytes stack frame, 0 bytes spill stores, 0 bytes spill loads
ptxas info    : Used 128 registers, 8192 bytes smem, 380 bytes cmem[0], 4 bytes cmem[2], 8 textures
make[2]: Leaving directory '/mnt/compile/ccminer-all/sp-hash/ccminer'
Makefile:1618: recipe for target 'all-recursive' failed
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory '/mnt/compile/ccminer-all/sp-hash/ccminer'
Makefile:562: recipe for target 'all' failed
make: *** [all] Error 2

-------

what could this be? ...

up until this latest git ( now that im back ) - it had been compiling fine ...

#crysx

If you ever make a API for your miner so it could work with multi-miner, I would love you forever. MM is totally the best auto-switcher/monitoring software. It was never hooked into Nvidia stuff because there wasn't ever a API for it.

https://bitcointalksearch.org/topic/multiminer-any-miner-any-where-on-any-device-free-open-source-cross-platform-248173

Yep, looks like it.

yep. like trying to chase unicorn farts.

Looks like yiimp is on sidechain... mined 6000 ams. On yobit balance is zero... transfers made more then hour ago...

It was probably the case for only a few hours, but now it's more like 0.0016 per 750 Ti if you look at the coin and the market behind it and it doesn't look like it's going to get better (highest buy order is at 100 sat vs. 175 lowest sell).

Although, block halving is incoming in about 10 days.

Thanks. I removed it from the list.

All the algos are bether it seems..


x11 rates

http://nicehash.com: 0.12 BTC/GHASH
http://ffpool.net:0.36 BTC/GHASH
http://yiimp.ccminer.org:0.73 BTC/GHASH

lyra2re rates

http://nicehash.com: 0.3890 BTC/GHASH
http://ffpool.net:1.53 BTC/GHASH
http://yiimp.ccminer.org:15.52 BTC/GHASH


Looks like the Amsterdamcoin has forked and changed the POW algo from Scrypt to the old lyra2 algo.. (16BTC / GHASH !!)

One 750ti@1 MHASH will make 0,016 a day...

yes but hashpower is broken: wrong estimates (counting the diff of a coin and the exchange value of another) and no shares accepted for days.

The quark miners are escaping from nicehash. Maybe because all the other pools are paying much bether..

Quark rates:

http://nicehash.com: 0.17 BTC/GHASH
http://ffpool.net:0.28 BTC/GHASH
http://yiimp.ccminer.org:0.29 BTC/GHASH
http://hashpower.co/: 0.7671 BTC/GHASH


Here is free opensource software that will switch between the best pools:

http://cryptomining-blog.com/5186-updated-miner-control-tool-packages-for-amd-and-nvidia-3/
http://cryptomining-blog.com/5294-new-miner-control-1-6-5-software-kbomba-edition/

Just like the nicehash algo switcher, only that you get payed more.

Time to bitslice then.. Perhaps not worth the effort.
Bether to optimize x11.
x15 is not very profitable at the moment.

whirlpool is slow on nvidia because it's not bitsliced (like groestl).
table lookup is fast on amd, instead.

I think the fury has buldt in AES instructions. On the NVIDIA side the whirlpool algo is the bottleneck.

The fury is showing its muscles in bastion!
And the opencl compiler for tahiti is weird, at the least...
The hawaii one is much smarter and stable.