CCminer(SP-MOD) Modded GPU kernels.

[tbearhere]

still very new at compiling, where can i get curl to 7.38.0 windows 8.1 from a safe site please.

from the site of the author

thanks djm34  going to fetch it 

[tbearhere]

still very new at compiling, where can i get curl to 7.38.0 windows 8.1 from a safe site please.

from the site of the author

thanks djm34  going to fetch it  

i cant find it

[djm34]

still very new at compiling, where can i get curl to 7.38.0 windows 8.1 from a safe site please.

from the site of the author

thanks djm34  going to fetch it 

i cant find it

google libcurl (no it isn't on microsoft page) it is open source

[Epsylon3]

I should probobly merge the latest changes in the main branch. (1.4.9) but I'm too lazy. My focus is on the kernals, and not the rest.

The work on the echo is not done. There is more to remove.

I almost didnt changed the .cu files recently, you can maybe refork my project

[tbearhere]

still very new at compiling, where can i get curl to 7.38.0 windows 8.1 from a safe site please.

from the site of the author

thanks djm34  going to fetch it  

i cant find it

google libcurl (no it isn't on microsoft page) it is open source

i got it but it needs to be compiled im use to doing it bigjme way. so i cant do it. 6 hrs to try 1.4.9 no luck

[djm34]

still very new at compiling, where can i get curl to 7.38.0 windows 8.1 from a safe site please.

from the site of the author

thanks djm34  going to fetch it 

i cant find it

google libcurl (no it isn't on microsoft page) it is open source

i got it but it needs to be compiled im use to doing it bigjme way. so i cant do it. 6 hrs to try 1.4.9 no luck

hu ? still haven't compile it ?
You know that you haven't anything to do (as epsilon told you... since libcurl has been put into the compat)

[sp_]

You need to install the latest cuda 6.5 and visual studio 2013.

[tbearhere]

You need to install the latest cuda 6.5 and visual studio 2013.

i got it thank you  djm34  and  sp  and all  yes curl was build into it.. it was open zip to folder then  sln..opens  vs 2013 automatically  then 64x release and done...3 minutes  

[sp_]

On windows, ccminer runs faster when compiled for x86

[tbearhere]

On windows, ccminer runs faster when compiled for x86

i get less hash on x11 750ti and no improvement on quark.
EDIT: I should say no improvement on x11

[djm34]

On windows, ccminer runs faster when compiled for x86

i get less hash on x11 750ti and no improvement on quark.

compiled with x64 or x86 ?

[tbearhere]

On windows, ccminer runs faster when compiled for x86

i get less hash on x11 750ti and no improvement on quark.

compiled with x64 or x86 ?

x86
EDIT: I should say no improvement on x11

[sp_]

the problem is that the throughput is set to be fast on the 980 in the source version. download the 1.4.9 source and replace the file:

cuda_x11_echo.cu from my fork

you should get a small boost in x11. On quark I have optimized bmw and blake a tinybit

1.4.9 with the intesity parameter is found here:

 https://github.com/tpruvot/ccminer

[tbearhere]

sp your older ccminer  and im really pushing it. looking forward to your new one.  

the fastest hashing is 1.4.6  

[Epsylon3]

the problem is that the throughput is set to be fast on the 980 in the source version. download the 1.4.9 source and replace the file:

cuda_x11_echo.cu from my fork

you should get a small boost in x11. On quark I have optimized bmw and blake a tinybit

1.4.9 with the intesity parameter is found here:

 https://github.com/tpruvot/ccminer

Indeed, +15kH on the 750 Ti (2ms improvement on your repo, its the biggest optimisation you have made on a single algo, was 0.5 before, i will pick it for the 1.5.0)

on mine, i get +9 KH (2791 vs 2800KH in benchmark mode) but i didnt take the launch bounds change for the moment...

39.171ms before, 38.522ms before = 0.65ms on mine, enough for me (but not fully comparable)

EDIT: but on windows :// seems to be lowered, investigating...

[Schleicher]

Possible small optimization at the end of cuda_echo_round:

Code:

	for (int i = 0; i<15; i += 4)
	{
		W[i] ^= W[32 + i] ^ 512;
		W[i + 1] ^= W[32 + i + 1];
		W[i + 2] ^= W[32 + i + 2];
		W[i + 3] ^= W[32 + i + 3];
	}
	W[15] ^= W[47] ^ 512;

(we don't need more than 16)

[sp_]

Indeed, +15kH on the 750 Ti (2ms improvement on your repo, its the biggest optimisation you have made on a single algo, was 0.5 before, i will pick it for the 1.5.0)
on mine, i get +9 KH (2791 vs 2800KH in benchmark mode) but i didnt take the launch bounds change for the moment...
39.171ms before, 38.522ms before = 0.65ms on mine, enough for me (but not fully comparable)
EDIT: but on windows :// seems to be lowered, investigating...

In addition to the launchbound change, did you remember to go from 256 to 320 threads when calling the kernal?. The launchbound will force the compiler to use 64 registers. We get more spills to memory, but it seems to run faster.

[sp_]

Possible small optimization at the end of cuda_echo_round:

Code:

	for (int i = 0; i<15; i += 4)
	{
		W[i] ^= W[32 + i] ^ 512;
		W[i + 1] ^= W[32 + i + 1];
		W[i + 2] ^= W[32 + i + 2];
		W[i + 3] ^= W[32 + i + 3];
	}
	W[15] ^= W[47] ^ 512;

(we don't need more than 16)

Thanks, it works.

   for (int i = 0; i<15; i += 4)
   {
      W ^= W[32 + i] ^ 512;
      W[i + 1] ^= W[32 + i + 1];
      W[i + 2] ^= W[32 + i + 2];
      W[i + 3] ^= W[32 + i + 3];
   }

is enough.

[sp_]

I have checked in some more performance improvements. I moved the precalc table in echo from constmem to the instruction cache. Improved registers/launchbounds on shavite.

The 980 is now around 400KHASH faster than the release 6. on stock clocks.(x11).

Here is the link:

http://www.filedropper.com/release7

The sourcecode is available here:

https://github.com/sp-hash/ccminer

[jpouza]

I have checked in some more performance improvements. I moved the precalc table in echo from constmem to the instruction cache. Improved registers/launchbounds on shavite.

The 980 is now around 400KHASH faster than the release 6. on stock clocks.(x11).

Here is the link:

http://www.filedropper.com/release7

Nice, 9MH/s with 185+ on 980 GPUs.
10MH/s with extreme overclock GPU at 300+ and overvolted.

750Ti boost to 2.9MH/s with 135+ GPU 460+ MEM.

Cheers