CCminer(SP-MOD) Modded GPU kernels.

[dominuspro]

You can boost the core clock as well if you put the gpu in the p1 state.

p2 state: adjust memory
p1 state: adjust gpu boost clock

I think I will make code that does this automaticly with an option.. --Max-Boost --Max-memspeed

1. List all possible modes, parse the information and select the fastest for each gpu in the rig
2. Set p1 mode and set the max possible gpu-boost clock
3. set p2 mode and set the max possible memclock clock

The NVIDIA api is broken in x86 builds, so I will call the commandline from the c++ code in the same way I have done before. (windows only)

For 970 cards with 4pin+3pin connector, boosting is important to reach full speed. (but this costs more power)

Other cards will probobly trottle and perform worse..

(f.ex all the gtx 970 mini cards, and the 970 cards that are made small to fit in a small box.)

I can't get my 980ti to boost at all on mem....

Sometimes I'm using nvidia inspector(from guru3d) to find out the actual state and boost the memory itself.
Maybe it can help You if You are using windows. It works on my 970 and 980.

[tbearhere]

You can boost the core clock as well if you put the gpu in the p1 state.

p2 state: adjust memory
p1 state: adjust gpu boost clock

I think I will make code that does this automaticly with an option.. --Max-Boost --Max-memspeed

1. List all possible modes, parse the information and select the fastest for each gpu in the rig
2. Set p1 mode and set the max possible gpu-boost clock
3. set p2 mode and set the max possible memclock clock

The NVIDIA api is broken in x86 builds, so I will call the commandline from the c++ code in the same way I have done before. (windows only)

For 970 cards with 4pin+3pin connector, boosting is important to reach full speed. (but this costs more power)

Other cards will probobly trottle and perform worse..

(f.ex all the gtx 970 mini cards, and the 970 cards that are made small to fit in a small box.)

I can't get my 980ti to boost at all on mem....

Sometimes I'm using nvidia inspector(from guru3d) to find out the actual state and boost the memory itself.
Maybe it can help You if You are using windows. It works on my 970 and 980.

Thx....I'll give it a try.

[sp_]

I looked into this a for a bit. I rewrote it like this:
block size == 128
lower byte of FNV_PRIME is 147, not 193. (0x01000193 & 0xFF = 0x93 == 147)

Then the precalc table would have to be 0x193 * 4 bytes big.

   c.x = sharedmemprecalc[a.x & 0x1ff] ^ b.x;
   c.y = sharedmemprecalc[a.y & 0x1ff] ^ b.y;
   c.z = sharedmemprecalc[a.z & 0x1ff] ^ b.z;
   c.w = sharedmemprecalc[a.w & 0x1ff] ^ b.w;

But there is a bether way.

It doesn't work. But more importantly, your suggestion of only requiring the lower byte from the DAG entry seems wrong, since that's param b in the fnv function...

I haven't started on my mod yet. But it wlll come later.. I can remove many isntructions, but I cannot remove the memory latency..

[bensam1231]

Hi all, could this thread be relevant for Maxwell based mining? Supposedly there is a trick to boost memclock safely using nvidia-smi.

https://devtalk.nvidia.com/default/topic/892842/cuda-programming-and-performance/one-weird-trick-to-get-a-maxwell-v2-gpu-to-reach-its-max-memory-clock-/

Christian

I tried it some months ago on a 970 and it worked, but you loose the auto clock tuning based on temperature and fan speed, which you may or may not like ;-)
I personally like it and ended up with better hashrate using the default settings than changing it with nvidia-smi, but YMMV.

Keep in mind there are only a handful of algos that get a boost out of memory speed... Juicing your memory on a 970 adds like a extra 20w~.

And the altocoins are up, and the mining profit meassured in BTC is up..

I have twice the hash power now and making the same amount. 

Yeah they're still playing with BTC... probably will be a week or two before altcoins rebound, depending on how long they mess around with BTC. All hands are on deck for BTC right now.

[scryptr]

You can boost the core clock as well if you put the gpu in the p1 state.

p2 state: adjust memory
p1 state: adjust gpu boost clock

I think I will make code that does this automaticly with an option.. --Max-Boost --Max-memspeed

1. List all possible modes, parse the information and select the fastest for each gpu in the rig
2. Set p1 mode and set the max possible gpu-boost clock
3. set p2 mode and set the max possible memclock clock

The NVIDIA api is broken in x86 builds, so I will call the commandline from the c++ code in the same way I have done before. (windows only)

For 970 cards with 4pin+3pin connector, boosting is important to reach full speed. (but this costs more power)

Other cards will probobly trottle and perform worse..

(f.ex all the gtx 970 mini cards, and the 970 cards that are made small to fit in a small box.)

I can't get my 980ti to boost at all on mem....

Sometimes I'm using nvidia inspector(from guru3d) to find out the actual state and boost the memory itself.
Maybe it can help You if You are using windows. It works on my 970 and 980.

Thx....I'll give it a try.

NVIDIA INSPECTOR--

It works on 6 cards, and allows setting clocks and fans in Windows.  It was written by Orbmu2k, if you google his name, you will find his programming.       --scryptr

[scryptr]

Hi all, could this thread be relevant for Maxwell based mining? Supposedly there is a trick to boost memclock safely using nvidia-smi.

https://devtalk.nvidia.com/default/topic/892842/cuda-programming-and-performance/one-weird-trick-to-get-a-maxwell-v2-gpu-to-reach-its-max-memory-clock-/

Christian

INTEL PROCESSOR BOARD--

Christian, just before you lessened your presence on the CudaMiner thread, you posted about your purchase of an Intel processor board (Xeon Phi), and that you wanted to explore programming for it.  Is there a thread where you posted results?  Did you program any mining application for the board?

Really curious about the results...       --scryptr

[bathrobehero]

You can boost the core clock as well if you put the gpu in the p1 state.

p2 state: adjust memory
p1 state: adjust gpu boost clock

I think I will make code that does this automaticly with an option.. --Max-Boost --Max-memspeed

1. List all possible modes, parse the information and select the fastest for each gpu in the rig
2. Set p1 mode and set the max possible gpu-boost clock
3. set p2 mode and set the max possible memclock clock

The NVIDIA api is broken in x86 builds, so I will call the commandline from the c++ code in the same way I have done before. (windows only)

For 970 cards with 4pin+3pin connector, boosting is important to reach full speed. (but this costs more power)

Other cards will probobly trottle and perform worse..

(f.ex all the gtx 970 mini cards, and the 970 cards that are made small to fit in a small box.)

I can't get my 980ti to boost at all on mem....

Sometimes I'm using nvidia inspector(from guru3d) to find out the actual state and boost the memory itself.
Maybe it can help You if You are using windows. It works on my 970 and 980.

How much hashrate are you getting with that? And for which algos?

Personally I never even OC memory clocks as it never giving me any noticeable speed just more power consumption and instability.

[dominuspro]

You can boost the core clock as well if you put the gpu in the p1 state.

p2 state: adjust memory
p1 state: adjust gpu boost clock

I think I will make code that does this automaticly with an option.. --Max-Boost --Max-memspeed

1. List all possible modes, parse the information and select the fastest for each gpu in the rig
2. Set p1 mode and set the max possible gpu-boost clock
3. set p2 mode and set the max possible memclock clock

The NVIDIA api is broken in x86 builds, so I will call the commandline from the c++ code in the same way I have done before. (windows only)

For 970 cards with 4pin+3pin connector, boosting is important to reach full speed. (but this costs more power)

Other cards will probobly trottle and perform worse..

(f.ex all the gtx 970 mini cards, and the 970 cards that are made small to fit in a small box.)

I can't get my 980ti to boost at all on mem....

Sometimes I'm using nvidia inspector(from guru3d) to find out the actual state and boost the memory itself.
Maybe it can help You if You are using windows. It works on my 970 and 980.

How much hashrate are you getting with that? And for which algos?

Personally I never even OC memory clocks as it never giving me any noticeable speed just more power consumption and instability.

Boost only mining eth for me. From something around 78MH to 87MH.
BTW not talking about memory OC, just reaching max original boost clocks.

[antantti]

Boost only mining eth for me. From something around 78MH to 87MH.
BTW not talking about memory OC, just reaching max original boost clocks.

Are you saying you are mining ETH@87MH? Windows?

[dominuspro]

Boost only mining eth for me. From something around 78MH to 87MH.
BTW not talking about memory OC, just reaching max original boost clocks.

Are you saying you are mining ETH@87MH? Windows?

Yes, with 4 gpus. 3x970+1x980 making 719W at the wall. All cores overclocked and memory of all gpus at 7000MHz.
I'm not tottally offtopic because I'm using the sp-mod version of the genoil's miner  

The same rig makes 76MH in quark algo, consuming 840-850W. 

[antantti]

Boost only mining eth for me. From something around 78MH to 87MH.
BTW not talking about memory OC, just reaching max original boost clocks.

Are you saying you are mining ETH@87MH? Windows?

Yes, with 4 gpus. 3x970+1x980 making 719W at the wall. All cores overclocked and memory of all gpus at 7000MHz.
I'm not tottally offtopic because I'm using the sp-mod version of the genoil's miner  

The same rig makes 76MH in quark algo, consuming 840-850W. 

Aah ok then... good luck with fighting with those amd guys...

[chrysophylax]

Hi all, could this thread be relevant for Maxwell based mining? Supposedly there is a trick to boost memclock safely using nvidia-smi.

https://devtalk.nvidia.com/default/topic/892842/cuda-programming-and-performance/one-weird-trick-to-get-a-maxwell-v2-gpu-to-reach-its-max-memory-clock-/

Christian

Holy shit, you're alive!

hahaha ... ditto that ...

pity we dont see you more often christian ...

#crysx

[bathrobehero]

How much hashrate are you getting with that? And for which algos?

Personally I never even OC memory clocks as it never giving me any noticeable speed just more power consumption and instability.

Boost only mining eth for me. From something around 78MH to 87MH.
BTW not talking about memory OC, just reaching max original boost clocks.

Ah, I see.
I just checked and all of my rigs with 970s are in p0 for some reason even after 25 days of uptime while I remember when I initially played with them they all eventually switched to p2.
The other thing is that EVGA Precision reports the memory clocks being at 3000mhz while everything else (GPU-Z, nvidia inspector) shows 3500 (1750/7000).

The hashrates are fine and with +150Mhz GPU OC they boost to 1428-1491 Mhz az 1212-1225mV.

[bathrobehero]

Hi all, could this thread be relevant for Maxwell based mining? Supposedly there is a trick to boost memclock safely using nvidia-smi.

https://devtalk.nvidia.com/default/topic/892842/cuda-programming-and-performance/one-weird-trick-to-get-a-maxwell-v2-gpu-to-reach-its-max-memory-clock-/

Christian

Holy shit, you're alive!

hahaha ... ditto that ...

pity we dont see you more often christian ...

#crysx

Oh I missed it, the man himself! How are you doing nowadays Christian?

[CoffeeCat]

You can boost the core clock as well if you put the gpu in the p1 state.

p2 state: adjust memory
p1 state: adjust gpu boost clock

I think I will make code that does this automaticly with an option.. --Max-Boost --Max-memspeed

1. List all possible modes, parse the information and select the fastest for each gpu in the rig
2. Set p1 mode and set the max possible gpu-boost clock
3. set p2 mode and set the max possible memclock clock

The NVIDIA api is broken in x86 builds, so I will call the commandline from the c++ code in the same way I have done before. (windows only)

For 970 cards with 4pin+3pin connector, boosting is important to reach full speed. (but this costs more power)

Other cards will probobly trottle and perform worse..

(f.ex all the gtx 970 mini cards, and the 970 cards that are made small to fit in a small box.)

I can't get my 980ti to boost at all on mem....

Sometimes I'm using nvidia inspector(from guru3d) to find out the actual state and boost the memory itself.
Maybe it can help You if You are using windows. It works on my 970 and 980.

How much hashrate are you getting with that? And for which algos?

Personally I never even OC memory clocks as it never giving me any noticeable speed just more power consumption and instability.

Do you increase the voltage though?

[dominuspro]

You can boost the core clock as well if you put the gpu in the p1 state.

p2 state: adjust memory
p1 state: adjust gpu boost clock

I think I will make code that does this automaticly with an option.. --Max-Boost --Max-memspeed

1. List all possible modes, parse the information and select the fastest for each gpu in the rig
2. Set p1 mode and set the max possible gpu-boost clock
3. set p2 mode and set the max possible memclock clock

The NVIDIA api is broken in x86 builds, so I will call the commandline from the c++ code in the same way I have done before. (windows only)

For 970 cards with 4pin+3pin connector, boosting is important to reach full speed. (but this costs more power)

Other cards will probobly trottle and perform worse..

(f.ex all the gtx 970 mini cards, and the 970 cards that are made small to fit in a small box.)

I can't get my 980ti to boost at all on mem....

Sometimes I'm using nvidia inspector(from guru3d) to find out the actual state and boost the memory itself.
Maybe it can help You if You are using windows. It works on my 970 and 980.

How much hashrate are you getting with that? And for which algos?

Personally I never even OC memory clocks as it never giving me any noticeable speed just more power consumption and instability.

Do you increase the voltage though?

I would never do that. It is just useless with maxwells.

[fenomenhaa]

Is there anything any algo or coin worth to mine? Or tyer are all dead? 

[sp_]

joincoin is still profitable, but only with my 120% faster penta miner..

0.3BTC and you will get it.

quark, lyra2v2 and etherum are profitable now. But only with private kernals.

[pallas]

joincoin is still profitable, but only with my 120% faster penta miner..

0.3BTC and you will get it.

quark, lyra2v2 and etherum are profitable now. But only with private kernals.

let me partecipate in the advertising :-D
joincoin is profitable on whirlpool as well with my 7x AMD miner.

[chrysophylax]

joincoin is still profitable, but only with my 120% faster penta miner..

0.3BTC and you will get it.

quark, lyra2v2 and etherum are profitable now. But only with private kernals.

let me partecipate in the advertising :-D
joincoin is profitable on whirlpool as well with my 7x AMD miner.

you guys ... ...

pallas - need to chat ... let me know when you have time ...

tanx ...

#crysx