CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 812.

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: sp_ on January 11, 2016, 02:19:49 PM

I just compared the original lyra2v2 from djm34 on my asus strix 750ti

1. https://github.com/djm34/ccminer-lyra/releases/tag/v0.5
2. https://github.com/sp-hash/ccminer/releases/tag/1.5.78 (opensource)
3. Buyable private miner (0.1BTC) pm for details.

asus strix 750ti:

1. djm34: 4260
2. 78-sp: 4470khash (+4.9%)
3. priv-sp: 4700khash. (+10.3%)

windforce 970 g1:

1. djm34: 9600
2. 78-sp: 9800 (+2.08%)
3. priv-sp: 10200 (+6.25%)

Asus strix 960:

1. djm34: 5750
2. 78-sp: 6100 (+6%)
3. priv-sp: 6330(+10.08%)

ok... so finally it isn't 5% more on every release, but 5% on the total of the 50 releases or so, since I released the private miner.
which, by the way, can already do the number you quote with a little tuning and OC (if my gainward 750ti can do 4.7MH/s there is no reason that
the more powerful 750ti can't...)

Might do a pledge for my fast lyrav2 version which does about 20% faster than your private version Grin

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

x13

750ti asus strix:

sp-mod 74: 2375
sp-mod 78: 2315 (-2.6%)
sp-mod priv: 2420 (+1.8%)

970 gigabyte g1

sp-mod 74: 6410
sp-mod 78: 6170
sp-mod priv: 6410 0%

960 strix

sp-mod 74: 4070
sp-mod 78: 3921
sp-mod priv: 4070 0%

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

I just compared the original lyra2v2 from djm34 on my asus strix 750ti

1. https://github.com/djm34/ccminer-lyra/releases/tag/v0.5
2. https://github.com/sp-hash/ccminer/releases/tag/1.5.78 (opensource)
3. Buyable private miner (0.1BTC) pm for details.

asus strix 750ti:
core: 1266
mem: 1350

1. djm34: 4260
2. 78-sp: 4470khash (+4.9%)
3. priv-sp: 4700khash. (+10.3%)

windforce 970 g1:
core: 1366
mem: 1500

1. djm34: 9600
2. 78-sp: 9800 (+2.08%)
3. priv-sp: 10200 (+6.25%)

Asus strix 960:
core: 1380
mem: 1500

1. djm34: 5750
2. 78-sp: 6100 (+6%)
3. priv-sp: 6330(+10.08%)

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

Quote from: joblo on January 11, 2016, 01:50:48 PM

Again I question your math, You can only mine one algo at a time so a miner only gets the increase of one algo.
Also most miners are only interested in a couple of algos and don't care about improvements in the others.
If you increase an algo by 10% you should expect only 10% of what the design of the original algo is worth.

The most profitable algo is switching from day to day.

Today the x15 is the most profitable. my +2% in the x15 algo might be worth more than +10% in lyra2v2 or +25% in neoscypt.

You need all the most profitable algos to be optimized.

Right now these algos are the most profitable in random order for NVIDIA maxwell cards: x11,x13,x15,quark,qubit,lyra2v2 and neoscrypt

Results vary from card to card.

the gtx 950,960,970,980,980ti are similar, but a small kernal change can boost one model and slow down another.

joblo

legendary

Activity: 1470

Merit: 1114

Quote from: sp_ on January 11, 2016, 01:19:07 PM

Quote from: joblo on January 11, 2016, 01:04:46 PM

You deliver marginal increases in performance. These improvements are simply not worth it for smalll miners, it requires
scaling before the increases become substantial. If you released less often and with bigger gains it would look better
from a sales perspective

From release 74 to release 78 private. +12% in the lyra2v2 algo is significant. This algo is one of the most profitable.

These algos can be mined at nicehash:

gtx 960:

keccak: +20%
lyra2v2: +11,64%
quark: +3.07
Qubit: +3.03%
Nist5: +2.2%
x11: +2%
x13: +1.8%
Whirlpoolx: +1.2%
x15: 1.14%
Neoscrypt: 0%

Sum them up and you get (46% increase divided on 10 algos) (An average of 4.6% increase per algo)
I have optimized all of the algos in the nicehashminer. Some of my work is opensource, and some is closed source.

Again I question your math, You can only mine one algo at a time so a miner only gets the increase of one algo.
Also most miners are only interested in a couple of algos and don't care about improvements in the others.

If you increase an algo by 10% you should expect only 10% of what the design of the original algo is worth.

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

Quote from: joblo on January 11, 2016, 01:04:46 PM

You deliver marginal increases in performance. These improvements are simply not worth it for smalll miners, it requires
scaling before the increases become substantial. If you released less often and with bigger gains it would look better
from a sales perspective

From release 74 to release 78 private. +12% in the lyra2v2 algo is significant. This algo is one of the most profitable.

These algos can be mined at nicehash:

gtx 960:

keccak: +20%
lyra2v2: +11,64%
quark: +3.07
Qubit: +3.03%
Nist5: +2.2%
x11: +2%
x13: +1.8%
Whirlpoolx: +1.2%
x15: 1.14%
Neoscrypt: 0%

Sum them up and you get (46% increase divided on 10 algos) (An average of 4.6% increase per algo)
I have optimized all of the algos in the nicehashminer. Some of my work is opensource, and some is closed source.

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

The x86 and the x64 build is different. I get 2070khash with release 78 and 2040 in release 74 (x15)

joblo

legendary

Activity: 1470

Merit: 1114

Quote from: sp_ on January 11, 2016, 02:59:04 AM

Quote from: chrysophylax on January 10, 2016, 08:40:51 PM

im putting some btc for the quark kernel - some btc for the neoscrypt pledge - some more btc for the distribution to other devs ... all i need now is to find a way of proper distribution to you all ... if willing community members want to come forward and help with advice on how - or with ideas - im all ears ...
#crysx

You should buy my private kernal and run it on your 6x980ti rig.The 980ti is doing more than 16MHASH in the lyra2v2 algo on the standard clocks.
100MHASH@lyra2v2 = 0.1BTC in just 7 days.

I'm not so sure about your math. Only the delivered increase in performance counts toward ROI, the existing performance
has already been figuratively paid for.

You deliver marginal increases in performance. These improvements are simply not worth it for smalll miners, it requires
scaling before the increases become substantial. If you released less often and with bigger gains it would look better
from a sales perspective

To appeal to small miners you need a market edge like being the first to market with a new algo, as you did with spread
and DJM did with neoscrypt, or the early days of Maxwell optimization when larger gains were realized. This is the kind of work that
will motivate small miners to contribute. I think DJM's approach is just that, hold back until market conditions improve.
I don't think the razor blade marketing approach will work in this case.

Razor blade marketing, for those who don't know, is a technique where the big money is made by frequent replacement
of wearable items. Rasors are cheap, blades are expensive. Printer ink use the same technique, printers are cheap, ink
is expensive.

I'm not saying I agree or disagree with anyone's chosen approach, just saying if it was me and I was trying to make
money writing mining software.

scryptr

legendary

Activity: 1797

Merit: 1028

Quote from: sp_ on January 11, 2016, 12:50:14 PM

Quote from: s7icky on January 11, 2016, 12:23:07 PM

Quote from: sp_ on January 11, 2016, 12:06:19 PM

X15?
Wich version are you running?

Im running ur FREE ONE cuda 7.5 .
for shits and giggles just cause i wanted to see.
and yeaup x15

Nice find.

The private version is faster than release 74 but abit slower than release 78 on the 750ti when mining x15. I will correct this and add more hash later.

X15 HASH RATE--

I am getting 13MH/s with a 6x 750ti FTW rig running self-compiled release dot 74 with CUDA 6.5, on Ubuntu. --scryptr

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

Quote from: s7icky on January 11, 2016, 12:23:07 PM

Quote from: sp_ on January 11, 2016, 12:06:19 PM

X15?
Wich version are you running?

Im running ur FREE ONE cuda 7.5 .
for shits and giggles just cause i wanted to see.
and yeaup x15

Nice find.

The private version is faster than release 74 but abit slower than release 78 on the 750ti when mining x15. I will correct this and add more hash later.

joblo

legendary

Activity: 1470

Merit: 1114

Quote from: pallas on January 11, 2016, 11:14:55 AM

Quote from: skunk on January 11, 2016, 11:05:09 AM

Quote from: chrysophylax on January 11, 2016, 04:09:08 AM

it is - but it would MUCH more difficult for the devs to supply a compiled version ( even as a static build ) to work will most linux distributions ...

im very specific with what i have int he farm - fedora 23 x64 cuda 7.5 with all the latest updates ...

so if a miner can be compiled by the devs - that would be great AND i would pay another 50% more for their trouble ...

#crysx

it's not if you staticize the binary with ermine, the trial version works good enough... Wink

Nice, but it makes linux more like windows Cheesy

(i.e. distributing the shared libraries with every program makes them... well... no longer shared)

Three points that all can deliver a linux binary while keeping the source private.

1. Static builds can be compiled, as previously mentioned. Depending on the amount of extra work this may be viable.

2. The dev could use a VM with your OS to build a standard dynamically linked version, extra work for the dev but pure linux.

3. Run the windows build using Wine with no performance loss, a bastard but easiest to implement.

bensam1231

legendary

Activity: 1764

Merit: 1024

Quote from: Liquid71 on January 10, 2016, 07:10:40 PM

Quote from: chrysophylax on January 10, 2016, 06:47:08 PM

hi all ...

i see a lot has happened with the testing and miner release ...

sp - ill be in the office later today ( and will be there for the next few days ) to get a LOT organized - which includes the 'big' machine ... so testing will begin with what i have currently on your latest ccminer-spmod78 c75 ...

djm34 - i would really like to see the neoscrypt miner running on my system - but i cant put 2.5btc forward to see it ... if there is no other way but to wait to see it - then that is what ill do ... i will ( and have ) donated to you in the past - and will continue to do so regardless of whether i advertise i do or not ...

you ( as are other devs ) are one hell of a coder - but this btc grabbing system that ALL the ~~devs~~ MINERS are participating in at the moment is actually making me cringe ...

NOT because you are all asking for payment / donation for the work you are diligently doing ( as you ALL deserve to be paid for this work ) - but for the MASSES of miners out there that are NOT donating in one way or another ... the handful of donors are those that are usually the ones that put the work into the threads and help support devs for the work - but im becoming increasingly distasteful of those miners ( the majority that use the products ) that do NOT donate ...

its actually quite disgusting ... there are MANY methods and MANY avenues to donate to these wonderful devs ( regardless of whether they bicker at each other or not ) and it seems that even a few btc is 'too much' to distribute amongst them ... that really is pathetic ... and yes - im aiming this directly at the miners that USE the miner software and oss kernels and private kernels ...

donation by direct btc address - donation by private kernel purchase - donation by pledge - donation by mining ( whether mining using the granite dbm system or direct address mining ) ... there are SO many ways to donate ... where is the FULL community support? ... it seems only a few handful of donors pitch in with as much as they can ...

come on mining community - sp djm34 pallas wolf0 ( just to name a few ) ALL need our help to continue ... lets get behind them ...

im off to the office now to get some of the masses of work i have piled up - done ... and to think about more ways to open avenues to donate ...

this needs to change ... it really does ... open your wallets people - and GIVE as much as the devs are GIVING ...

#crysx

Fixed it.
Also no need to double space every sentence

This is towards the rest of the miners bitching, not you #crysx,
GREED seems to be on part of the miners not the devs. It's a donation not an investment, in return for the donation you get a faster miner and support the development of future mining software. If djm34 doesn't work on Nvidia kernals then AMD kernals will dominate neoscrypt. You're keeping the ecosystem healthy by supporting devs, if you stomp your feet like that self entitled benslam character stomping your feet demanding to profit from the work of others without contributing...well you're a parasite and the reason it's so hard to get good devs to put the work in on open source software that benefits everyone.
end rant

I donated to the last development of neoscrypt, then also bought the miner after that before it was released publicly...and never used it. I was just trying to support the development of ccminer and figured if FTC took a moon shot it would be nice to have around. I bought SP cryptonite miner and tested it but never used it again, just wanted to support dev and the miner is an nice bonus if xmr ever takes a moonshot lol..

Without devs releasing miners to the public it's all over. Most small miners are bagholders and support the alts they mine, and devs going to all private to farms will basically push out all miners pretty much destroying the ecosystem we have going now. bad for everyone..can't believe how greedy people are.

Hey if anyone wants a guarantee of ROI why not try treasury bonds Huh

Guaranteed ROI is different from 'never going to ROI'. You can throw your money away. Both sides need to benefit from the relationship or it's pointless. Right now miners need Neo and Spread kernels and are willing to pay for them, devs optimized everything except that (with the exception of the DJM drive happening right now). I know developers a lot of times like just working on whatever they feel like, but that's not going to encourage people to donate or purchase a miner. You can't have a 'ecosystem' when you're basically employing someones hobby.

If you want people to donate, make something that miners need.

Most small miners that actually attempt to make money aren't bag holders and after getting burned the first half dozen times stop doing it because they're pissing electricity down the drain.

Private farms mine profitable coins. They aren't bag holders. They know better. Being a bag holder doesn't help anyone, not even the coin you're 'supporting', which is just going to spiral down the drain anyway unless it's something extremely unique. I've been there, done that. I stopped after Wolf0 ran me out of the AMD card range with his private kernels I didn't even know existed because none of the developers advertised they were selling them.

s7icky

member

Activity: 106

Merit: 10

Quote from: sp_ on January 11, 2016, 12:06:19 PM

X15?

Wich version are you running?

Im running ur FREE ONE cuda 7.5 .
for shits and giggles just cause i wanted to see.

and yeaup x15

skunk

sr. member

Activity: 329

Merit: 250

Quote from: pallas on January 11, 2016, 11:14:55 AM

Quote from: skunk on January 11, 2016, 11:05:09 AM

Quote from: chrysophylax on January 11, 2016, 04:09:08 AM

it is - but it would MUCH more difficult for the devs to supply a compiled version ( even as a static build ) to work will most linux distributions ...

im very specific with what i have int he farm - fedora 23 x64 cuda 7.5 with all the latest updates ...

so if a miner can be compiled by the devs - that would be great AND i would pay another 50% more for their trouble ...

#crysx

it's not if you staticize the binary with ermine, the trial version works good enough... Wink

Nice, but it makes linux more like windows Cheesy

(i.e. distributing the shared libraries with every program makes them... well... no longer shared)

it's the only way to be sure binary blobs will work regardless dependencies and local versions of shared libs...
btw, you shouldn't install binary blobs anyway if you don't trust the source or if you are an open source evangelist...

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

X15?

Wich version are you running?

s7icky

member

Activity: 106

Merit: 10

Stock clocks.

So sp_ i can pay you for a slower version?

pallas

legendary

Activity: 2716

Merit: 1094

Black Belt Developer

Quote from: skunk on January 11, 2016, 11:05:09 AM

Quote from: chrysophylax on January 11, 2016, 04:09:08 AM

it is - but it would MUCH more difficult for the devs to supply a compiled version ( even as a static build ) to work will most linux distributions ...

im very specific with what i have int he farm - fedora 23 x64 cuda 7.5 with all the latest updates ...

so if a miner can be compiled by the devs - that would be great AND i would pay another 50% more for their trouble ...

#crysx

it's not if you staticize the binary with ermine, the trial version works good enough... Wink

Nice, but it makes linux more like windows Cheesy

(i.e. distributing the shared libraries with every program makes them... well... no longer shared)

skunk

sr. member

Activity: 329

Merit: 250

Quote from: chrysophylax on January 11, 2016, 04:09:08 AM

it is - but it would MUCH more difficult for the devs to supply a compiled version ( even as a static build ) to work will most linux distributions ...

im very specific with what i have int he farm - fedora 23 x64 cuda 7.5 with all the latest updates ...

so if a miner can be compiled by the devs - that would be great AND i would pay another 50% more for their trouble ...

#crysx

it's not if you staticize the binary with ermine, the trial version works good enough... Wink

scryptr

legendary

Activity: 1797

Merit: 1028

QUARK HASH RATES--

Following are Quark results for the Private Miner on light computer load:

GTX 960 SC mining Quark at startup

GTX 960 SC mining Quark with light computer load

With release dot 74, the same card achieved 10.9-11.1MH/s mining Quark. --scryptr

scryptr

legendary

Activity: 1797

Merit: 1028

Quote from: sp_ on January 11, 2016, 06:36:39 AM

Quote from: pokeytex on January 11, 2016, 06:32:13 AM

@sp - what is your private kernel doing on x15? 750ti, 960, 970?

I have posted the results earlier:

the +% is compared to release 78

neoscrypt (same as release 74)
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%) (asus strix 5150Khash (with +150 oc))

x15 does 2055KHASH on the asus strix 750ti standard clocks.

results gtx 960:

scryptr did a test on his gtx 960 card and compared it with the nicehash miner(release 74) here are the results:

keccak: +20%
lyra2v2: +11,64%
quark: +3.07
Qubit: +3.03%
Nist5: +2.2%
x11: +2%
x13: +1.8%
Whirlpoolx: +1.2%
Neoscrypt: 0% (after reducing virtual mem)

The private miner is for sale for a 0.1BTC donation.

NEOSCRYPT CODE IS A WIN AT ZERO PERCENT--

The private miner based on release dot 78 is compiled with CUDA 7.5, and the popular algos are now mining at equal or better hash rates than earlier versions of CCminer compiled with CUDA 6.5. This brings CCminer current with what is now considered the standard CUDA environment, CUDA 7.5.

When I reduced the size of my RAMDisk, my memory load dropped to about 30% from about 50%. Neoscrypt performance increased. The 16GB Win 7 x64 system became less "laggy" in general. I run CCminer under NiceHash Miner in the RAMDisk, about 500MB total disk used. I simply renamed the private miner to match the "ccminer_sp.exe" executable (release dot 74) in the NiceHash Miner "bin" subdirectory. I then benchmarked the new binary executable.

--scryptr

Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 812. (Read 2347664 times)