Author

Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 803. (Read 2347664 times)

sr. member
Activity: 506
Merit: 252
Look, I'd actually like to improve Spread (as well as X11 itself) for Nvidia, but that means I have to go over 5,375 lines of AMD GCN assembly by paw (yes, I just checked) and understand it perfectly in order to not only write it in OpenCL, but then port it to CUDA. You can't imagine what a pain in the ass this is.

why not do a pledge with escrow?

eg. 5 BTC for x11

if 5BTCs are reached you start doing your magic ...

I prefer not to promise something I dont have yet - especially because people are going to ask what the hashrate is going to be, and the answer is "fuck if I know" until I've done it.

therefore escrow ...

Bro, are you high?

you are not?
hero member
Activity: 1974
Merit: 502
Vave.com - Crypto Casino
Look, I'd actually like to improve Spread (as well as X11 itself) for Nvidia, but that means I have to go over 5,375 lines of AMD GCN assembly by paw (yes, I just checked) and understand it perfectly in order to not only write it in OpenCL, but then port it to CUDA. You can't imagine what a pain in the ass this is.

why not do a pledge with escrow?

eg. 5 BTC for x11

if 5BTCs are reached you start doing your magic ...

I prefer not to promise something I dont have yet - especially because people are going to ask what the hashrate is going to be, and the answer is "fuck if I know" until I've done it.

therefore escrow ...

Bro, are you high?
sr. member
Activity: 506
Merit: 252
Look, I'd actually like to improve Spread (as well as X11 itself) for Nvidia, but that means I have to go over 5,375 lines of AMD GCN assembly by paw (yes, I just checked) and understand it perfectly in order to not only write it in OpenCL, but then port it to CUDA. You can't imagine what a pain in the ass this is.

why not do a pledge with escrow?

eg. 5 BTC for x11

if 5BTCs are reached you start doing your magic ...

I prefer not to promise something I dont have yet - especially because people are going to ask what the hashrate is going to be, and the answer is "fuck if I know" until I've done it.

therefore escrow ...
sr. member
Activity: 506
Merit: 252
Look, I'd actually like to improve Spread (as well as X11 itself) for Nvidia, but that means I have to go over 5,375 lines of AMD GCN assembly by paw (yes, I just checked) and understand it perfectly in order to not only write it in OpenCL, but then port it to CUDA. You can't imagine what a pain in the ass this is.

why not do a pledge with escrow?

eg. 5 BTC for x11

if 5BTCs are reached you start doing your magic ...
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
full member
Activity: 180
Merit: 100
Sp
sent ya a pm to get your donation addy as the site don't like addy's posted in forums
full member
Activity: 180
Merit: 100
If you guys can improve x11 or Ethereum that'd be worth some coin. Please consider targeting the big algos.


Hm. As you wish.

It's not nice to play with a young boys heart. Cry

Spread is also a good one worth targeting right now.

Spread? I hadn't thought about that, but I probably could do something with that...


To give you and ideal of how hot it is......
I've been mining it on 1 750ti for about 5 months and have gotten ~0.6 btc @ today's pricing

ive been mining spreadcoin since it first started - then stopped for a long while due to the laborious changeover of miners and reboots involved with thefarm whenever i needed to mine it ...

btc is NOT always the driving factor here - and i admit - it is a nice project ... i have been with it from the beginning ...

but unless the kernel will be incorporated into a universal miner ... there will ALWAYS be a limited fraction of the worlds miners mining this coin ...

i know - im one of them ... and used to be one of the larger of the farms ...

#crysx

Spread crapped itself for awhile and occasionally really huge miners hop on it that adds like 25% network hash and tanks the income from it. Ethereum is more stable if you have large hashrate.

I've had problems with Ethereum.   The only wallet I could find was a contract type. The pools I've used when they would send out my coin, it would not get into my wallet. If I mined directly to an exchange, I'd get the coin. 
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
SP-mod private version 2 is soon ready. Here are som stats:

(faster than version private 1)

qubit:

gtx 970: +2.2%   12070 / 11810 (+260khash) +4.7% 12370 / 11810
gtx 960: +2.6%  7900 / 7730 (+170khash)
gtx 750ti:+2.5%   4860 / 4740 (+120khash)

x11

750ti: 3090/3045  +1.42% (+55KHASH)
970: 8130/7950  +2.22% (+180KHASH)
960: 5170/5119  +1%    (+51KHASH)


x13

750ti: 2426/2400 +1.08% (+26KHASH)
970: 6550/6400 +2.3%  (+150KHASH)
960: 4096/4080 +0.4% (+16KHASH)

x15

750ti: 2084/2060 +1.17% (+24KHASH)
970: 5645/5520 +2.22% (+75KHASH)
960: 3540 /3525 +0.42% (+15KHASH)

lyra2v2
750ti: 4630/4610 +0.43% (+20khash)
960,970 slower than private #1

I have sendt release 2 to the supporters..

If you want to support my code, please donate 0.1BTC to recieve the private mod.

release 1 sp-mod private (bether than release 78):

X11 is up 30khash (750ti) (+1%)
x13 is up 100khash(750ti) (+4.3%)
x15 is up 66Khash(750ti) (+3.3%)
quark is up 200khash(750ti) (+3.2%)
lyra2v2 is up 250khash(750ti)(+6%)

release 2 sp-mod private (bether than sp-mod private 1):

x11: +1.42% (+55KHASH(750ti))
x13: +1.08% (+26KHASH(750ti))
x15: +1.17% (+24KHASH(750ti))
lyra2v2: +0.43% (+20khash(750ti))
qubit:+2.5%  (+120khash(750ti))

legendary
Activity: 2940
Merit: 1091
--- ChainWorks Industries ---
Wolf0 was coding miners for nvidia back when maxwell came out.
Some of that inline PTX is mine, in the various ccminer forks... I had 5x750Ti.

Did you ever do any CPU miners? I merged three CPU miners, see my thread for details, and
if you ever did any significant work on CPU mining I could add you the long credits.

Release is impending by the way.

https://bitcointalksearch.org/topic/ann-cpuminer-opt-v3881-open-source-optimized-multi-algo-cpu-miner-1326803

im heading to the office now joblo - a few personal issue out the way - i can focus a little on the work now ...

so will be compiling / testing in the next few hours ...

tanx ...

#crysx
legendary
Activity: 2940
Merit: 1091
--- ChainWorks Industries ---
If you guys can improve x11 or Ethereum that'd be worth some coin. Please consider targeting the big algos.


Hm. As you wish.

It's not nice to play with a young boys heart. Cry

Spread is also a good one worth targeting right now.

Spread? I hadn't thought about that, but I probably could do something with that...

To give you and ideal of how hot it is......
I've been mining it on 1 750ti for about 5 months and have gotten ~0.6 btc @ today's pricing

ive been mining spreadcoin since it first started - then stopped for a long while due to the laborious changeover of miners and reboots involved with thefarm whenever i needed to mine it ...

btc is NOT always the driving factor here - and i admit - it is a nice project ... i have been with it from the beginning ...

but unless the kernel will be incorporated into a universal miner ... there will ALWAYS be a limited fraction of the worlds miners mining this coin ...

i know - im one of them ... and used to be one of the larger of the farms ...

#crysx

Spread crapped itself for awhile and occasionally really huge miners hop on it that adds like 25% network hash and tanks the income from it. Ethereum is more stable if you have large hashrate.

i cant seem to get the ethereum miner compiled ... which is the reason i havent tested mining with it - let alone mined with it ...

ive also been focussing on the development of granite - and helping out a few other coins also - so i havent really too hard at eth ... it seems more trouble than its worth - but the eth / btc value seems quite good ... may have a look again soon ... i have no idea what hashrate thefarm can pull with hashimoto though - so i dunno if its a large sum or not ...

#crysx
legendary
Activity: 1470
Merit: 1114
Wolf0 was coding miners for nvidia back when maxwell came out.
Some of that inline PTX is mine, in the various ccminer forks... I had 5x750Ti.

Did you ever do any CPU miners? I merged three CPU miners, see my thread for details, and
if you ever did any significant work on CPU mining I could add you the long credits.

Release is impending by the way.

https://bitcointalksearch.org/topic/ann-cpuminer-opt-v3881-open-source-optimized-multi-algo-cpu-miner-1326803
legendary
Activity: 1764
Merit: 1024
If you guys can improve x11 or Ethereum that'd be worth some coin. Please consider targeting the big algos.


Hm. As you wish.

It's not nice to play with a young boys heart. Cry

Spread is also a good one worth targeting right now.

Spread? I hadn't thought about that, but I probably could do something with that...

To give you and ideal of how hot it is......
I've been mining it on 1 750ti for about 5 months and have gotten ~0.6 btc @ today's pricing

ive been mining spreadcoin since it first started - then stopped for a long while due to the laborious changeover of miners and reboots involved with thefarm whenever i needed to mine it ...

btc is NOT always the driving factor here - and i admit - it is a nice project ... i have been with it from the beginning ...

but unless the kernel will be incorporated into a universal miner ... there will ALWAYS be a limited fraction of the worlds miners mining this coin ...

i know - im one of them ... and used to be one of the larger of the farms ...

#crysx

Spread crapped itself for awhile and occasionally really huge miners hop on it that adds like 25% network hash and tanks the income from it. Ethereum is more stable if you have large hashrate.
legendary
Activity: 2940
Merit: 1091
--- ChainWorks Industries ---
If you guys can improve x11 or Ethereum that'd be worth some coin. Please consider targeting the big algos.


Hm. As you wish.

It's not nice to play with a young boys heart. Cry

Spread is also a good one worth targeting right now.

Spread? I hadn't thought about that, but I probably could do something with that...

To give you and ideal of how hot it is......
I've been mining it on 1 750ti for about 5 months and have gotten ~0.6 btc @ today's pricing

ive been mining spreadcoin since it first started - then stopped for a long while due to the laborious changeover of miners and reboots involved with thefarm whenever i needed to mine it ...

btc is NOT always the driving factor here - and i admit - it is a nice project ... i have been with it from the beginning ...

but unless the kernel will be incorporated into a universal miner ... there will ALWAYS be a limited fraction of the worlds miners mining this coin ...

i know - im one of them ... and used to be one of the larger of the farms ...

#crysx
full member
Activity: 180
Merit: 100
If you guys can improve x11 or Ethereum that'd be worth some coin. Please consider targeting the big algos.


Hm. As you wish.

It's not nice to play with a young boys heart. Cry

Spread is also a good one worth targeting right now.

Spread? I hadn't thought about that, but I probably could do something with that...

To give you and ideal of how hot it is......
I've been mining it on 1 750ti for about 5 months and have gotten ~0.6 btc @ today's pricing
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
SP-mod private version 2 is soon ready. Here are som stats:
(faster than version private 1)
qubit:

gtx 970: +2.2%   12070 / 11810 (+260khash) +4.7% 12370 / 11810
gtx 960: +2.6%  7900 / 7730 (+170khash)
gtx 750ti:+2.5%   4860 / 4740 (+120khash)

x11

750ti: 3090/3045  +1.42% (+55KHASH)
970: 8130/7950  +2.22% (+180KHASH)
960: 5170/5119  +1%    (+51KHASH)


x13

750ti: 2426/2400 +1.08% (+26KHASH)
970: 6550/6400 +2.3%  (+150KHASH)
960: 4096/4080 +0.4% (+16KHASH)

x15

750ti: 2084/2060 +1.17% (+24KHASH)
970: 5645/5520 +2.22% (+75KHASH)
960: 3540 /3525 +0.42% (+15KHASH)

lyra2v2
750ti: 4630/4610 +0.43% (+20khash)
960,970 slower than private #1
reporting 0.5% increases is rather meaningless (It is within error measurement...)

I have optimized SIMD. one of the 15 algos in x15.
In lyra2v2 I have merged two kernals.  Faster on compute 5.0 but slower on 5.2

 
The X series is not like neoscrypt or lyra2v2 where 90% of the work is done in the memory kernal.

You need to add the improvements in private #1 to get the total increase over my opensource.

release 1 sp-mod private (bether than release 78):

X11 is up 30khash (750ti) (+1%)
x13 is up 100khash(750ti) (+4.3%)
x15 is up 66Khash(750ti) (+3.3%)
quark is up 200khash(750ti) (+3.2%)
lyra2v2 is up 250khash(750ti)(+6%)

release 2 sp-mod private (bether than sp-mod private 1):

x11: +1.42% (+55KHASH(750ti))
x13: +1.08% (+26KHASH(750ti))
x15: +1.17% (+24KHASH(750ti))
lyra2v2: +0.43% (+20khash(750ti))
qubit:+2.5%  (+120khash(750ti))


legendary
Activity: 1400
Merit: 1050
SP-mod private version 2 is soon ready. Here are som stats:

(faster than version private 1)

qubit:

gtx 970: +2.2%   12070 / 11810 (+260khash) +4.7% 12370 / 11810
gtx 960: +2.6%  7900 / 7730 (+170khash)
gtx 750ti:+2.5%   4860 / 4740 (+120khash)

x11

750ti: 3090/3045  +1.42% (+55KHASH)
970: 8130/7950  +2.22% (+180KHASH)
960: 5170/5119  +1%    (+51KHASH)


x13

750ti: 2426/2400 +1.08% (+26KHASH)
970: 6550/6400 +2.3%  (+150KHASH)
960: 4096/4080 +0.4% (+16KHASH)

x15

750ti: 2084/2060 +1.17% (+24KHASH)
970: 5645/5520 +2.22% (+75KHASH)
960: 3540 /3525 +0.42% (+15KHASH)

lyra2v2
750ti: 4630/4610 +0.43% (+20khash)
960,970 slower than private #1


reporting 0.5% increases is rather meaningless (It is within error measurement...)
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
SP-mod private version 2 is soon ready. Here are som stats:

(faster than version private 1)

qubit:

gtx 970: +2.2%   12070 / 11810 (+260khash) +4.7% 12370 / 11810
gtx 960: +2.6%  7900 / 7730 (+170khash)
gtx 750ti:+2.5%   4860 / 4740 (+120khash)

x11

750ti: 3090/3045  +1.42% (+55KHASH)
970: 8130/7950  +2.22% (+180KHASH)
960: 5170/5119  +1%    (+51KHASH)


x13

750ti: 2426/2400 +1.08% (+26KHASH)
970: 6550/6400 +2.3%  (+150KHASH)
960: 4096/4080 +0.4% (+16KHASH)

x15

750ti: 2084/2060 +1.17% (+24KHASH)
970: 5645/5520 +2.22% (+75KHASH)
960: 3540 /3525 +0.42% (+15KHASH)

lyra2v2
750ti: 4630/4610 +0.43% (+20khash)
960,970 slower than private #1

legendary
Activity: 1764
Merit: 1024
If you guys can improve x11 or Ethereum that'd be worth some coin. Please consider targeting the big algos.


Hm. As you wish.

It's not nice to play with a young boys heart. Cry

Spread is also a good one worth targeting right now.

Spread? I hadn't thought about that, but I probably could do something with that...

gigabyte 980ti extreme is sitting on the nvidia test machine now wolf ...

wanna play? ... its all yours Wink ...

remember to stop it mining first before any testing on it please ...

thats IS you are interested in dabbling with nvidia again mate ...

im off - back tomorrow ...

#crysx

I was thinking AMD, though.

That doesn't help anyone. Why do you even exist?

Anyone? Just because you don't have AMD cards doesn't mean no one does - Christ, you're self centered.
I shouldn't exist because I do not benefit you. Now that's some logic.

You would understand if you had Nvidia cards.

Oh god... that's insane.
Wolf0 was coding miners for nvidia back when maxwell came out.

Yeah, sorta expected a catchall when he said 'ok'... XD


Not sure if you guys have seen the ROI on video cards lately. ASICs seem to have a much more stable ROI and sooner. Actually debating on selling some cards off, taking the hit and buying some ASICs. I honestly don't know who has been purchasing cards in the last six or so months. Things really fell apart after Quark ended.
legendary
Activity: 1400
Merit: 1050
btw an early Pascal chip could be available soon (rumored in next 2 months)


14nm finfet and new chip design optimized for perf/watt
http://videocardz.com/58069/amd-demonstrates-revolutionary-14nm-finfet-polaris-gpu-architecture


same strategy which nvidia did with their maxwell 1.0 chips aka 750 / ti ...
to bring a low core count (probably less than 1024 cores) to the market early on.

not sure it has much interest beside learning how to work with the Pascal chip
sr. member
Activity: 506
Merit: 252
btw an early Polaris chip from amd could be available soon (rumored in next 2 months)


14nm finfet and new chip design optimized for perf/watt
http://videocardz.com/58069/amd-demonstrates-revolutionary-14nm-finfet-polaris-gpu-architecture


same strategy which nvidia did with their maxwell 1.0 chips aka 750 / ti ...
to bring a low core count (probably less than 1024 cores) to the market early on.
Jump to: