Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 215. (Read 3426932 times)

ok good to know

mmh the code are the same mine and yours, but i'm getting rejected, something with the pool?

Hey all, just started trying my hand at monero and I'm getting an odd issue.  Using NVMiner 1.2D6 my GTX 750 tis are hashing in the 280s or so, but my GTX titan is only getting around 220 or so.  Am I missing something?

I have started. But progress is slow:)

compacttest isn't used.
mostly likely you had a difference between the cpu validation (the sph_algo stuff) and the the hash calculation on gpu they
need to match...

i did not remove compacttest, maybe that the problem

i've left the other "quark_check"

anyway diff is 20 already i can't test well in solo, need to much time to find a block

I can't play for a few hours until I get home but have you tried this:
1) copy cubit.cu to luffa512.cu
2) add luffa512.cu to VS
3) in luffa512.cu
a) comment out everything from extern void x11_cubehash512_cpu_init to quark_compactTest_cpu_hash_64
4) rename qubithash to luffa512hash
5) comment out
    //sph_cubehash512_context ctx_cubehash;
    //sph_shavite512_context ctx_shavite;
    //sph_simd512_context ctx_simd;
    //sph_echo512_context ctx_echo;
6) also comment out:
    //sph_cubehash512_init(&ctx_cubehash);
    //// ZCUBEHASH;
    //sph_cubehash512 (&ctx_cubehash, (const void*) hash, 64);
    //sph_cubehash512_close(&ctx_cubehash, (void*) hash);

    //sph_shavite512_init(&ctx_shavite);
    //// ZSHAVITE;
    //sph_shavite512 (&ctx_shavite, (const void*) hash, 64);
    //sph_shavite512_close(&ctx_shavite, (void*) hash);

    //sph_simd512_init(&ctx_simd);
    //// ZSIMD
    //sph_simd512 (&ctx_simd, (const void*) hash, 64);
    //sph_simd512_close(&ctx_simd, (void*) hash);

    //sph_echo512_init(&ctx_echo);
    //// ZECHO
    //sph_echo512 (&ctx_echo, (const void*) hash, 64);
    //sph_echo512_close(&ctx_echo, (void*) hash);
7) rename scanhash_qubit to scanhash_luffa512
more generic here but comment out similar routines to above not using luffa
9) rename function calls inside from qubithash to luffa512hash

then setup externs for luffa512hash in main program?

I of course skipped a couple of things as I just quickly looked though it but that should work I'd think if it's just one round of luffa512.

Carlo

https://bitcointalksearch.org/topic/bbr-boolberry-gpu-miner-discussion-amd-nvidia-opencl-cuda-693118

the "finalhash" routine in whirlcoin is just there for increasing the performance (not entirely it works though... but with github it is as painful to add than to remove...), this just to avoid to do part of the validation on another kernel).
On keccak-256 including the hash detection increased the hashrate by 30MHash/s (or rather not doin it was decreasing the perf. by that much...) so doing the same things on luffa would certainly increase the perf. too

dunno why but any attempt at oc my cards result in a crash after some hours... i can't even set +25(core, mem are fine i can rise them), maybe those gpu are power limited in the bios...and i don't want to flash anything


btw boolberry gpu miner is still private?

Yep, you'll be doing something in the region of 50% TDP. For the venturous types: with the current monero mining performance, you can generally afford a much higher overclock when mining monero. Do remember to switch back if you decide to mine something else afterwards, otherwise you're likely in for a crash 

Happy mining,
~ Myagui

i just removed the four algo from qubit but i don't know if the _80 count as one round(tried both, adding the _64 one also), anyway i'm waiting for pool to see if it work, on solo he just keep running(no error or anything) but dunno if it is working or not

beside that qubit luffa is just for the first hash _80 you certainly need it.... maybe it need that finalhash_80 like whirlcoin 


p.s. i'm mining monero right now gpu are really cool, perhaps it consume less?

Okay, I give up. I compiled like 12 different versions, trying to use both qubit_luffa512 and x11_luffa512 and even did some weird mixes but obviously I only found cpu validation errors and crashes, but mostly air. 

Ahh shit, I'm screwed.  I only have nvMiner.

doomcoin.conf



ccminer.exe -a luffa -o 127.0.0.1:rpcport -u rpcuser -p rpcpassword
pause

If you guys can help me setup the QT wallet for GPU access (I don't remember how- to tired to try and remember) I'll see if I can get you a GPU miner for the coin that will work.

Carlo

PS Is there a spec on the coin? How many rounds of luffa512 is it?
Anything else you may know that could help?

DJM34, did you already do this?

Carlo

u on wrong thread ? This is not a coin thread, 1000% percent is not possible. Unless.....

where is sp_ +100% hashrate x11 ? 

Wolf this coin is nothing to do with CryptoNight.  It has a very similar name being CryptoNite but uses different algo.

Nice try, I was all setup to solo. Got tor to connect and just got

http request failed: empty reply from server
json_rpc_call failed, retry after 30 seconds


booo

If only I would know what I'm doing, but I'm afraid i'm just cutting parts out left and right hoping to get lucky