Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 229. (Read 3426936 times)

damn it, I can't remember either 

not and this is not

Replace SBOX with sbox_pipelined

In the code:

SBOX(hamsi_s00, hamsi_s08, hamsi_s10, hamsi_s18); \
      SBOX(hamsi_s01, hamsi_s09, hamsi_s11, hamsi_s19); \
      SBOX(hamsi_s02, hamsi_s0A, hamsi_s12, hamsi_s1A); \
      SBOX(hamsi_s03, hamsi_s0B, hamsi_s13, hamsi_s1B); \
      SBOX(hamsi_s04, hamsi_s0C, hamsi_s14, hamsi_s1C); \
      SBOX(hamsi_s05, hamsi_s0D, hamsi_s15, hamsi_s1D); \
      SBOX(hamsi_s06, hamsi_s0E, hamsi_s16, hamsi_s1E); \
      SBOX(hamsi_s07, hamsi_s0F, hamsi_s17, hamsi_s1F); \


------>

   sbox_pipelined(hamsi_s00, hamsi_s08, hamsi_s10, hamsi_s18,hamsi_s01, hamsi_s09, hamsi_s11, hamsi_s19); \
   sbox_pipelined(hamsi_s02, hamsi_s0A, hamsi_s12, hamsi_s1A,hamsi_s03, hamsi_s0B, hamsi_s13, hamsi_s1B); \
   sbox_pipelined(hamsi_s04, hamsi_s0C, hamsi_s14, hamsi_s1C,hamsi_s05, hamsi_s0D, hamsi_s15, hamsi_s1D); \
   sbox_pipelined(hamsi_s06, hamsi_s0E, hamsi_s16, hamsi_s1E,hamsi_s07, hamsi_s0F, hamsi_s17, hamsi_s1F); \

The alu in the cudacore needs perhaps more time, so the sbox could be expanded to take 32 input variables. Then there will always be 4 instructions in different registers.

http://en.m.wikipedia.org/wiki/Instruction_pipeline

yes, but I need to "plug it" to something...

edit: never mind, I read the code, I can try... 

DA4AF09FE5377715856BA0B10A29C95867053ECBF4105DBDD8957DA78B4127E49E4717DD667CEEF B

Don't understand why nobody remember it...  

Yes, but sbox is not the whole hash. As you can see the input and output doesn't change.

which the password to download the x15 file - 07/15/2014?

if you help me find source of nvminer, i'll help you release split screen with nvminer core.

there is only one hash per thread...

I'm using windows, it's just that I always have to manually set oc profiles with EVGA or MSI between algos but I usually forgot.

x13 / cuda_x13_hamsi512.cu


try to replace sbox with sbox_pipelined (stallfree)

#define SBOX(a, b, c, d) { \
        uint32_t t; \
        t = (a); \
        (a) &= (c); \
        (a) ^= (d); \
        (c) ^= (b); \
        (c) ^= (a); \
        (d) |= t; \
        (d) ^= (b); \
        t ^= (c); \
        (b) = (d); \
        (d) |= t; \
        (d) ^= (a); \
        (a) &= (b); \
        t ^= (a); \
        (b) ^= (d); \
        (b) ^= t; \
        (a) = (c); \
        (c) = (b); \
        (b) = (d); \
        (d) = SPH_T32(~t); \
    }

--->>

#define sbox_pipelined(a, b, c, d, a1, b1, c1, d1) { \
        uint32_t t,t1; \
        t = (a); \
        t1 = (a1); \
        (a) &= (c); \
        (a1) &= (c1); \
        (a) ^= (d); \
        (a1) ^= (d2); \
        (c) ^= (b); \
        (c1) ^= (b1); \
        (c) ^= (a); \
        (c1) ^= (a1); \
        (d) |= t; \
        (d1) |= t1; \
        (d) ^= (b); \
        (d1) ^= (b1); \
        t ^= (c); \
        t1 ^= (c1); \
        (b) = (d); \
        (b1) = (d1); \
        (d) |= t; \
        (d1) |= t1; \
        (d) ^= (a); \
        (d1) ^= (a1); \
        (a) &= (b); \
        (a1) &= (b1); \
        t ^= (a); \
        t1 ^= (a1); \
        (b) ^= (d); \
        (b1) ^= (d1); \
        (b) ^= t; \
        (b1) ^= t1; \
        (a) = (c); \
        (a1) = (c1); \
        (c) = (b); \
        (c1) = (b1); \
        (b) = (d); \
        (b1) = (d1); \
        (d) = SPH_T32(~t); \
        (d1) = SPH_T32(~t2); \
    }

and

SBOX(hamsi_s00, hamsi_s08, hamsi_s10, hamsi_s18); \
      SBOX(hamsi_s01, hamsi_s09, hamsi_s11, hamsi_s19); \
      SBOX(hamsi_s02, hamsi_s0A, hamsi_s12, hamsi_s1A); \
      SBOX(hamsi_s03, hamsi_s0B, hamsi_s13, hamsi_s1B); \
      SBOX(hamsi_s04, hamsi_s0C, hamsi_s14, hamsi_s1C); \
      SBOX(hamsi_s05, hamsi_s0D, hamsi_s15, hamsi_s1D); \
      SBOX(hamsi_s06, hamsi_s0E, hamsi_s16, hamsi_s1E); \
      SBOX(hamsi_s07, hamsi_s0F, hamsi_s17, hamsi_s1F); \


------>

   sbox_pipelined(hamsi_s00, hamsi_s08, hamsi_s10, hamsi_s18,hamsi_s01, hamsi_s09, hamsi_s11, hamsi_s19); \
   sbox_pipelined(hamsi_s02, hamsi_s0A, hamsi_s12, hamsi_s1A,hamsi_s03, hamsi_s0B, hamsi_s13, hamsi_s1B); \
   sbox_pipelined(hamsi_s04, hamsi_s0C, hamsi_s14, hamsi_s1C,hamsi_s05, hamsi_s0D, hamsi_s15, hamsi_s1D); \
   sbox_pipelined(hamsi_s06, hamsi_s0E, hamsi_s16, hamsi_s1E,hamsi_s07, hamsi_s0F, hamsi_s17, hamsi_s1F); \

you can use the linux nvidia tools to get the infos
/usr/bin/nvidia-smi --query-gpu=index,temperature.gpu,fan.speed --format=csv,noheader,nounits

I don't think so, but I'd really love that function. The reason for that because every algo has different stable overclocks and so I either run without optimal stable OC or I instacrash half my cards if I switch to something like X15 after an overclock-friendly algo. So with that feature I could always include the optimal overclock profiles in the batch file.

Anyone try the 740 4gb with YACoin???

you don't need to add casing on compute version. the rol macro is already casing.

I think that variables that are written to, needs a break to avoid a stall. So by reordering the instructions / reimplementing in assembly language we can avoid some stalls.

x13 / cuda_x13_hamsi512.cu

#define HAMSI_L(a, b, c, d) { \
        (a) = ROTL32(a, 13); \
        (c) = ROTL32(c, 3); \
        (b) ^= (a) ^ (c); \  //stall
        (d) ^= (c) ^ SPH_T32((a) << 3); \
        (b) = ROTL32(b, 1); \
        (d) = ROTL32(d, 7); \
        (a) ^= (b) ^ (d); \ //stall
        (c) ^= (d) ^ SPH_T32((b) << 7); \
        (a) = ROTL32(a, 5); \
        (c) = ROTL32(c, 22); \
    }

BUG

thanks djm34, Im mining stright to your wallet.    ps i hope thats a whirl coin address

EDIT: for 8 hrs getting 4,300mh/s per card x 6

https://github.com/KBomba/ccminer-KBomba/commit/af4a88615e3416be1adf72870f14dcd9eee527f0
sp_, something like that? Gives me only small boost though, around 50kh/s, so it could be just a calc error.

Wow ... just check my uptime on KopiemTu.
Never hit that with sgminer and BAMT
Stable as fuck !!

Is it possible to change the fan speed, GPU engine etc and read the GPU temps with ccMiner (Xubuntu)?