Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 1107. (Read 2347601 times)

Did you get the last commit (makefile)?

scryptr buildt it yesterday on linux.

Please do a full clean and get the latest sourcecode and try again.

sp - for some unknown reason - the latest git clone is not compiling ...

it just freezes and sits at the same specific place during compile ...

compilation has not changed for any other compiles that have been made before - just this one ...

where it freezes is below ...

-------

ptxas info    : Function properties for _Z25titan_scrypt_core_kernelAILi1EL12MemoryAccess1EEvPKjii
    0 bytes stack frame, 0 bytes spill stores, 0 bytes spill loads
ptxas info    : Used 25 registers, 336 bytes cmem[0]
ptxas info    : Compiling entry function '_Z25titan_scrypt_core_kernelBILi1EL12MemoryAccess1EEvPjii' for 'sm_50'
ptxas info    : Function properties for _Z25titan_scrypt_core_kernelBILi1EL12MemoryAccess1EEvPjii
    0 bytes stack frame, 0 bytes spill stores, 0 bytes spill loads
ptxas info    : Used 28 registers, 336 bytes cmem[0]
scrypt/titan_kernel.cu: In member function ‘virtual bool TitanKernel::run_kernel(dim3, dim3, int, int, cudaStream_t, uint32_t*, uint32_t*, unsigned int, unsigned int, bool, bool, int)’:
scrypt/titan_kernel.cu:682:27: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
  if (N != prev_N[thr_id]) {
                           ^
ptxas info    : 0 bytes gmem, 480 bytes cmem[3]
ptxas info    : Compiling entry function '_Z24x11_luffa512_gpu_hash_64jjPmPj' for 'sm_50'
ptxas info    : Function properties for _Z24x11_luffa512_gpu_hash_64jjPmPj
    0 bytes stack frame, 0 bytes spill stores, 0 bytes spill loads
ptxas info    : Used 80 registers, 344 bytes cmem[0]
ptxas info    : Overriding global maxrregcount 128 with entry-specific value 64 computed using thread count
ptxas info    : Overriding global maxrregcount 128 with entry-specific value 64 computed using thread count
ptxas info    : 0 bytes gmem, 1152 bytes cmem[3]
ptxas info    : Compiling entry function '_Z26x11_shavite512_gpu_hash_64jjPm' for 'sm_50'
ptxas info    : Function properties for _Z26x11_shavite512_gpu_hash_64jjPm
    32 bytes stack frame, 28 bytes spill stores, 28 bytes spill loads
ptxas info    : Used 64 registers, 4096 bytes smem, 336 bytes cmem[0]
ptxas info    : Compiling entry function '_Z26x11_shavite512_gpu_hash_80jjPv' for 'sm_50'
ptxas info    : Function properties for _Z26x11_shavite512_gpu_hash_80jjPv
    128 bytes stack frame, 132 bytes spill stores, 132 bytes spill loads
ptxas info    : Used 64 registers, 4096 bytes smem, 336 bytes cmem[0]

-------

that last line is where it sits and doesnt move ...

any ideas? ...

edit - when ctrl-c stops the compile - this is what is displayed ...

-------

ptxas info    : Used 64 registers, 4096 bytes smem, 336 bytes cmem[0]
^Cnvcc error   : 'cicc' died due to signal 2
make[2]: *** [yescrypt/cuda_yescrypt.o] Error 2
make[1]: *** [all-recursive] Interrupt
make: *** [all] Interrupt

-------

im stumped ...

#crysx

Don't mine scrypt, mine quark.

ccminer -a quark -o stratum+tcp://yaamp.com:4033 -u -p xx

Yaamp is currently paying around 0.4BTC for 1 GHASH of quark


1 750ti OC does 6.5 MHASH in the quark algo

You earn:
0.0065*0.4=0,0026 (btc per day) (0,4355 LTC Per day.) ($0.64 / day)


SCRYPT pays 0.1840 BTC/GHASH

A 750ti is earning as much as a  15MHASH Scrypt ASIC miner with the current prices. (and use 50watt in the wall) (3.33 watt per MHASH (converted to scrypt MHASH))

A used 750ti can be bought for around $100 second hand.

X11 COMMAND LINE--

Windows batch file format:

":loop

ccminer.exe -a x11 -i 20.5 -o stratum+tcp:// URL:PORT -u USERNAME.WORKER -p PASSWORD -b 0.0.0.0:4068

goto loop"

Linux BASH scrypt, or single command line:

"./ccminer -a x11 -i 20.5 -o stratum+tcp:// URL:PORT -u USERNAME.WORKER -p PASSWORD -b 0.0.0.0:4068"

The "-i 20.5" is the "intensity" setting.  It is a performance flag.  The "-b 0.0.0.0:4068" is an optional API flag, if you want to monitor your performance with software like CCmonitor, etc.  Currently, the "gputhreads", or "-g" switch is also being developed.  Read at least the last 10 pages of this thread to learn about it, it is another performance option, and just about ready for stable use.       --scryptr

It is all-in-one.  You could directly replace the NiceHash URLs and api with WestHash equivalents.

My apologies to Skunk, it is currently a hack-n-slash job, removing code no longer necessary for different mining programs and the defunct TMB site.  I can still see portions that need removal, but that function as-is. 

The script finally mined for me today.  It didn't switch, though, when Yaamp started paying more for Quark.  The script kept running, but the miner was killed and did not reboot properly.  It definitely needs work.  I flopped it out there like a flea circus side-show.

--scryptr

scryptr,

Not sure what context that 'concurrent' falls into but... So, to mine just x11 coins, then, I take it that the sp (latest, of course) will give me the better hashrate? Will it mine LTC? Also, would you mind giving me a correct .bat file for the best hash /w my 750Ti?    Thank you very much for the help!

Edit: did a reread & believe I understand the concurrent remark. Sorry. Not sure if it (sp) will mine just any scrypt tho. I've always thought it was for x11 only. I STILL don't know the correct syntax for my .bat tho. Ive got it OCed already. Help?

Thanks!

Nicely done, I am getting 10,300 kh/s with x11 on gtx980.  I am not using -g, as it still not work in conjunction with -d.

scryptr, thanks for that script.  can any instances of nicehash be replaced with westhash for those of us from the U.S.  And is this script an all-in-one solution, i.e. it will start the miner and then run on auto-pilot?

CCMINER FLAVORS--

There are several current developers.  SP_'s release of v49 (1.5.49) will mine most all algos, and at top hashing rates.  Tpruvot's 1.6.2 has a higher release number, but is concurrent, as is KlausT's v3.00.  The developers are numbering differently.        --scryptr

Ok...so plz allow me to say that I haven't been keeping up with this thread lately. Wow...fast changes. I'm currently lost. Using a single ASUS GTX750Ti2G-DF series card (/w an i7 CPU). Using ccminer 1.5 release 44 strictly for x11.  I was getting 3.25Mh (roughly) and was happy. Then, I see that 1.6.2 has come out and I can mine any scrypt now. Gave it a quick spin on the x11 and only got 3Mh. I OC using GPUTweak. I don't set any xxx.bat perimeters,  just the algo and pool. Here's where I'm lost. What version of ccMiner should I be using now and are there any suggestions for single card settings I need (should) add to my xxx.bat file. I use xHash and suprnova pools for mining the x11s.

Thank you for replies.

BASH SCRYPT for ALGO SWITCHING--

This is a rough edit from Skunk's earlier posting.  3 big differences:

1) NiceHash has changed the ID number for some algos, and eliminated x14 and fresh algos.
2) TradeMyBit (TMB) is gone.
3) ALL algos can be mined with just CCminer now.  No need for CudaMiner or NeoGpuMiner.

This script is set up to mine at NiceHash and Yaamp.  I was getting syntax errors.  I execute the script as "sudo ./switch.sh", it runs, it mines, it doesn't switch just ok yet.  I hope some other user(s) might want to work the bugs out.

Right now, the script launches cleanly, and reports the best paying algo between NiceHash and Yaamp every minute, and mines!  My error was using an abreviation for my home directory.  I just corrected it; use the full path from root to your mining directory with "ccminer".  I am still looking for bugs...

For what it is worth:


============================= switch.sh ========================================

#!/bin/bash

# depends on sys-devel/bc, app-misc/jq, app-misc/screen and sys-process/schedtool
# after starting this script you can execute "screen -r miner" to see miner's output

# algo ids (the first 12 must match with nicehash's api: https://nicehash.com/?p=api)
scrypt=0
sha256=1
nscrypt=2
x11=3
x13=4
keccak=5
x15=6
nist5=7
neoscrypt=8
lyra2re=9
whirlpoolx=10
qubit=11
quark=12

# in the following array keys must match with above ids
declare -a ALGO=('scrypt' 'sha256' 'nscrypt' 'x11' 'x13' 'keccak' 'x15' 'nist5' 'neoscrypt' 'lyra2re' 'whirlpoolx' 'qubit' 'quark')

# hashing rates in Mhs (750ti)
# if not minable, leave it at 0!
RATE[$scrypt]=.28
RATE[$sha256]=0
RATE[$nscrypt]=.105
RATE[$x11]=3.2
RATE[$x13]=2.4
RATE[$keccak]=115
RATE[$x15]=2.11
RATE[$nist5]=8.1
RATE[$neoscrypt]=.180
RATE[$lyra2re]=.076
RATE[$whirlpoolx]=165
RATE[$qubit]=5.6
RATE[$quark]=6.35


# https://www.nicehash.com/index.jsp?p=gstarted#seller
NH_URL=stratum+tcp://stratum.nicehash.com
NH_USER=BITCOIN-ADDRESS.worker
NH_PASS=x
NH_FEE=2
NH_PORT[$scrypt]=3333
NH_PORT[$sha256]=3334
NH_PORT[$nscrypt]=3335
NH_PORT[$x11]=3336
NH_PORT[$x13]=3337
NH_PORT[$keccak]=3338
NH_PORT[$x15]=3339
NH_PORT[$nist5]=3340
NH_PORT[$neoscrypt]=3341
NH_PORT[$lyra2re]=3342
NH_PORT[$whirlpoolx]=3343
NH_PORT[$qubit]=3344
NH_PORT[$quark]=3345

# http://yaamp.com/ (ports numbers and fees are delivered by the api)
YA_URL=stratum+tcp://yaamp.com
YA_USER=BITCOIN-ADDRESS
YA_PASS=worker

# hysteresis: by which (absolute) increment of price a pool/algo switch should be triggered
HYST=.1

# miner's paths
CCMINER=/home/yourusername/miningdirectory/ccminer

# start miner screen session if it doesn't exist
if [ "$(screen -ls|grep miner)" == "" ]
then
        screen -dmS miner
fi

MPRICE=0

while [ 1 ]
do
        # start execution timer
        START=$(date +%s)

        # initialize profitability list
        LIST=''

        # fetch data from pool's api
        NH=$(wget -qO - 'https://www.nicehash.com/api?method=stats.global.current')
        YA=$(wget -qO - 'http://yaamp.com/api/status')

        # process yaamp's data
        for i in $(echo "$YA"|jq '.[].name'|tr -d '"')
        do
                eval x='$'$i
                price=$(echo "$YA"|jq ".[\"$i\"].estimate_current"|awk '{ print sprintf("%.9f", $1); }')
                YA_FEE=$(echo "$YA"|jq ".[\"$i\"].fees")
                price=$(echo "$price * 1000 * ${RATE[$x]}"|bc)
                price=$(echo "$price - ($price * $YA_FEE / 100)"|bc)
                # if port is 0 means this channel is inactive...
                YA_PORT[$x]=$(echo "$YA"|jq ".[\"$i\"].port")
                if [ ${YA_PORT[$x]} -gt 0 ]
                then
                        LIST+="$price\t${ALGO[$x]}\tYA\n"
                fi
        done

        # process nicehash's data
        for i in $(echo "$NH"|jq '.result.stats[].algo')
        do
                # ignore algo id 100 (multi-algo id)
                if [ $i -eq 100 ]
                then
                        continue
                fi
                price=$(echo "$NH"|jq ".result.stats[$i].price"|tr -d '"')
                price=$(echo "$price * ${RATE[$i]}"|bc)
                price=$(echo "$price - ($price * $NH_FEE / 100)"|bc)
                LIST+="$price\t${ALGO[$i]}\tNH\n"
        done

        # most profitable channel candidate extraction
        LIST=$(echo -e $LIST|sort -rn)
        MPRICE_CAND=$(echo "$LIST"|head -n1|cut -f1)
        MALGO_CAND=$(echo "$LIST"|head -n1|cut -f2)
        MPOOL_CAND=$(echo "$LIST"|head -n1|cut -f3)

        # actual channel price extraction
        if [ $(echo "$MPRICE > 0"|bc) -eq 1 ]
        then
                MPRICE=$(echo "$LIST"|grep -P "[0-9.]+\t$MALGO\t$MPOOL"|cut -f1)
                if [ -z $MPRICE ]
                then
                        MPRICE=0
                fi
        fi

        # choose candidate pool/algo if candidate's price difference is greater than HYST
        if [ $(echo "$MPRICE_CAND > ($MPRICE + $HYST)"|bc) -eq 1 ]
        then
                MPRICE=$MPRICE_CAND
                MPOOL=$MPOOL_CAND
                MALGO=$MALGO_CAND
        fi

        # set miner parameters
        eval URL='$'${MPOOL}_URL
        eval PORT='$'{${MPOOL}_PORT[$MALGO]}
        eval USER='$'${MPOOL}_USER
        eval PASS='$'${MPOOL}_PASS


        # choose miner and correctly set algo parameter for nscrypt
        if [ "$MALGO" == "scrypt" -o "$MALGO" == "nscrypt" ]
        then
                MINER=$CCMINER
                if [ "$MALGO" == "nscrypt" ]
                then
                        PMALGO="-a scrypt:2048"
                fi
        elif [ "$MALGO" == "neoscrypt" ]
        then
                MINER=$CCMINER
                PMALGO="-a neoscrypt -i 15.5"
        else
                MINER=$CCMINER
        fi

        # start miner if not running or restart it with new parameters if they've changed
        if [ -z "$(pgrep $(basename $CCMINER))" ]
        then
                screen -S miner -X screen schedtool -B -e $MINER $PMALGO -o $URL:$PORT -u $USER -p $PASS
        elif [ -z "$(pgrep -f -- "$PMALGO.*$USER")" ]
        then
                killall -q $(basename $CCMINER)
                screen -S miner -X screen schedtool -B -e $MINER $PMALGO -o $URL:$PORT -u $USER -p $PASS
                echo "========================================================"
        fi

        # current date/tme and status output
        DATE=$(date +"%x %R")
        echo "$DATE - Mining $MALGO@$MPOOL@$MPRICE (best $MALGO_CAND@$MPOOL_CAND@$MPRICE_CAND)"

        # stop execution timer and sleep until next minute
        END=$(date +%s)
        TOOK=$((END-START))
        SLEEP=$((60-TOOK))
        if [ $SLEEP -lt 0 ]
        then
                SLEEP=0
        fi
        sleep $SLEEP
done

==================================END======================================

@SP_ --

I just flipped you two bits!  Thanks for incorporating Neoscrypt.  Now, if I could only pull together a working BASH scrypt, CCminer could seek the highest paying algos Round-Robin.       --scryptr

brief check #49 with my best previous results (don't remember exact version ...)
x11 goes to 2905 from 2890
x13 2287 from 2275
x15 1990 from 1981
qubit 4585 from 4552
skein 83750 from 83620
lyra 884 from 882
Great work, great release!

COMMIT 775 BUILDS! --

OK, that was the final fix.  It now builds and mines.  I am getting 537kh/s on each of my GTX 970 FTW+ cards.  The rig is getting about 1.05Mh/s for accepted shares. Thanks!       --scryptr

2 sp_

Tried your #49 with neoscrypt

I tried djm's version before and for my poor gtx 750 1gb it runs only with -F 1 and gives 175khs.

Default run with your #49
start /realtime ccminer.exe --cpu-priority 5 -a neoscrypt

it takes almost all gpu memory and pushes load to "Bus interface" as it called in GPU-Z.
With such conditions performance is poor - only 45 khs.
But i tried to play with -i parameter and found that it runs with maximum -i 14.5
With such config it shows 810MB memory usage, 0% bus interface load, 99% gpu load, 30-35% memory controller load.
The numbers are 187khs and this is better then old djm's version. But i didn't check on the pool yet .

So I think you should use not so heavy default launch config for this algo for low-level gpu's like mine.

probably need to add that as well:

COMMIT 775--

@SP_ and DJM34--

In Makefile.am, after the last line beginning with sph/... :
                        
                      ..."sph/bmw.c sph/blake.c sph/groestl.c sph/jh.c sph/keccak.c sph/skein.c \
                          sph/cubehash.c sph/echo.c sph/luffa.c sph/sha2.c sph/shavite.c sph/simd.c \
                          sph/hamsi.c sph/hamsi_helper.c sph/sph_hamsi.h \
                          sph/shabal.c sph/whirlpool.c sph/sha2big.c sph/haval.c \
                          sph/sha256_Y.h sph/sha256_Y.c \" ...

I inserted DJM34's suggested line, "sph/sha256_Y.h sph/sha256_Y.c \".  The linux build still errors out, now with:

"gue512.o x15/x14.o x15/x15.o x15/cuda_x14_shabal512.o x15/cuda_x15_whirlpool.o x15/whirlpool.o x17/x17.o x17/cuda_x17_haval512.o x17/cuda_x17_sha512.o x11/s3.o bitcoin.o cuda_bitcoin.o x15/cuda_whirlpoolx.o x15/whirlpoolx.o yescrypt/cuda_yescrypt.o yescrypt/yescrypt.o ccminer-scrypt.o ccminer-scrypt-jane.o scrypt/blake.o scrypt/keccak.o scrypt/sha256.o scrypt/salsa_kernel.o scrypt/test_kernel.o scrypt/fermi_kernel.o scrypt/kepler_kernel.o scrypt/nv_kernel.o scrypt/nv_kernel2.o scrypt/titan_kernel.o  -L/usr/lib/x86_64-linux-gnu -lcurl -ljansson -lpthread  -lcudart -fopenmp -lcrypto -lssl  -ldl -lcrypto -lssl
ccminer-ccminer.o: In function `miner_thread(void*)':
ccminer.cpp:(.text+0x3533): undefined reference to `scanhash_neoscrypt'
collect2: error: ld returned 1 exit status
make[2]: *** [ccminer] Error 1 ..."

--scryptr

tanx i will try and report
my laptop is lenovo z5070 i5, mining quark for hours the max temp is 65(cpu is 63), i also tried both cpu and gpu mining simultaneously for test,in that case cpu temp passed over 80c and raised gpu temp around 75c.

you need to add to Makefile.am