CCminer(SP-MOD) Modded GPU kernels.

[chrysophylax]

Read and replied

You've got my skype and facebook name (check in my signature for confirmation eventually). Write me on fb, I'm on the move

I don't use face book

nikolaj is a good guy mate ...

if he says he can help - you can rest assured that he means what he says ...

#crysx

[1714874227]

1714874227

[1714874227]

1714874227

[giagge]

Nvidia miner pascalcoin for pool when ? .

[sp_]

My version of Vorksholk/PascalCoin-CUDA, with fixes and speedups.

BTC donations: 1H7qC5uHuGX2d5s9Kuw3k7Wm7xMQzL16SN

https://github.com/pallas1/Pascalcoin-CUDA

Work in progress...

Thanks. 8% Faster than mine with 2 threads on the 1070.

I get 970 vs 900

I am running a test now and compare the klaus_t kernel with the same input data as yours.

edit: compiled with cuda52 instead of 61 I get 1050
edit: With npt=16 gives 1100

[reb0rn21]

In last week I wasted 36h+ with 25G, so far no block and I am stubborn as hell

[pallas]

My version of Vorksholk/PascalCoin-CUDA, with fixes and speedups.

BTC donations:

https://github.com/pallas1/Pascalcoin-CUDA

Work in progress...

Thanks. 8% Faster than mine with 2 threads on the 1070.

I get 970 vs 900

I am running a test now and compare the klaus_t kernel with the same input data as yours.

edit: compiled with cuda52 instead of 61 I get 1050
edit: With npt=16 gives 1100

If you increase npt, it will likely overflow the nonce range.

[sp_]

If you increase npt, it will likely overflow the nonce range.

The original source had a limit on the nonce range of 2^20

With your block config. npt of 16 should be the maximum possible value.

8192*8*16=2^20

But not sure. I seem to get less solutions with npt=16 than npt=1

[pallas]

If you increase npt, it will likely overflow the nonce range.

The original source had a limit on the nonce range of 2^20

With your block config. npt of 16 should be the maximum possible value.

8192*8*16=2^20

But not sure. I seem to get less solutions with npt=16 than npt=1

You forgot to multiply by the block size...

[sp_]

You forgot to multiply by the block size...

Yes but you have done something with the code.

//const uint32_t id = (headerIn[16] << 20) | (blockDim.x * blockIdx.x + threadIdx.x) * npt; (old code)
const uint32_t id = headerIn[16] + (blockDim.x * blockIdx.x + threadIdx.x) * npt; (your code)


In the old code (blockDim.x * blockIdx.x + threadIdx.x) * npt had to be less than 2^20.

[AzzAz]

ocminer is starting pool, also have some miner for windows and linux
https://bitcointalk.org/index.php?topic=1583719.4400

[Amph]

ocminer is starting pool, also have some miner for windows and linux
https://bitcointalk.org/index.php?topic=1583719.4400

we need ccminer, opencl is not good with nvidia, in fact my rig freezed and slowed down when i started it...

[sp_]

This is what I get.

gtx 1070 (366MHASH)
3x gtx 970 (270-300 MHASH)

100% cpu load. Unusable.

[ioglnx]

Well CPU load is very high yes.. but for me it runs stable


If you help KlausT to fix the cpu verify issue in ccminer of his implementation then we have something to go...
Eventho lets combine our nvidia power to supernova lets beep these AMD morons where the sun never shines ;-D

[ioglnx]

Suprnova mined the first block of Pascal ..dancing like a queen :-D

And the next one:

[00:02:41] Found block for pasc.suprnova.cc!
[00:02:41] Accepted a4dedb2f Diff 397/8.222 BLOCK! GPU 1

[du44]

Suprnova mined the first block of Pascal ..dancing like a queen :-D

And the next one:

[00:02:41] Found block for pasc.suprnova.cc!
[00:02:41] Accepted a4dedb2f Diff 397/8.222 BLOCK! GPU 1

Now waiting for ccminer...

[ioglnx]

Well there is just the cpu verification issue..and you can't disable it sadly.

[reb0rn21]

Well there is just the cpu verification issue..and you can't disable it sadly.

I think Klaus have a solution soon...

anyway sgminer with -w 1024 and at 2Ghz is doing 1.05Gh/s, 2x 1070 are doing 61% CPU usage on i5, so I did not try it on celeron with 5 GPUs

I think sp_ made mistake and tried it with -g 2 which is not good in NV with pascal 

[ioglnx]

Let rock guys we need to beat nanopool and these AMD nerds.
Even the cpu usuage is high it performs good. I throw all my cards into it :-D damn hell yeah eat that AMD fanboys. NV rocks you again.

[bensam1231]

There are a decent number of other coins earning as much as Pascal at 900Mhs without all the issues. Waiting for all of this to get ironed out before I decide to burn an entire days worth of energy again.

[ioglnx]

I tried KlausT last build but it doesn't work yet. But looks like he is just 1 step behind the solution.
Getting the error result dosn't verify on cpu..

[tbearhere]

Read and replied

You've got my skype and facebook name (check in my signature for confirmation eventually). Write me on fb, I'm on the move

I don't use face book

nikolaj is a good guy mate ...

if he says he can help - you can rest assured that he means what he says ...

#crysx

Thx crysx
Remember how sick you were? I'm still that way and only now I can get my treatments for mercury and lead chronic toxicity... but I'm thrown out of my house bank stole it ..so now I can't get them.
With the treatments I'm fine but now at extreme poverty.
The banks won't let me get a loan ether.
6 days at most of mining left and in my house.
Yes nikolaj is a great guy .. and he is trying to help..but I'm way to sick.
Sorry people for messy up the thread.
I'll say goodbye to everyone now because I don't know how exactly when I'll be shutting down my 2 rigs.. but within 6 days.
Don't know if I'll be back. Been a very educational time and thanks to all that helped me.
The downing of America.

Tbear