CCminer(SP-MOD) Modded GPU kernels.

[Johnjay06]

Tried it, moved directly over from enemy on x16s. I get cannot validate on cpu every time

[sp_]

Raven x16r/x16s Spmod-git #6

new build released. (sp-mod #6 git)

-faster X16R/X16S +10-15% ( Optimalizations to aes,echo and simd)

https://github.com/sp-hash/suprminer/releases

I get best speeds with 100% tdp and +100 coreclock and +700 on the memory clock.

edit:

spmod-git #6,  1800 accepted vs t-rex 0.5.7 1674 . Will take a new screenshot on 2000 accepted..

[sp_]

t-rex vs sp-mod after around 2000 accepted shares at diff 2.

t-rex rejected: 0.55%
sp-mod rejected: 2.5%

The spmod-git opensource is currently +8.4% faster than t-rex, but this can change with other sequence numbers.
 

[scryptr]

SPMOD #6 RUNS SMOOTHLY ON MY 1080ti--

With a quick 15-minute trial, I got about 50 accepts, no errors, 100% accepts.  I was running with "The Pill", a slight memory overclock, and intensity 23.  I did not try to evaluate speed, but there were no red-lines or errors.  There are some low-difficulty Raven clones out there to harvest...

--scryptr

[jugger1028]

SPMOD #6 RUNS SMOOTHLY ON MY 1080ti--

With a quick 15-minute trial, I got about 50 accepts, no errors, 100% accepts.  I was running with "The Pill", a slight memory overclock, and intensity 23.  I did not try to evaluate speed, but there were no red-lines or errors.  There are some low-difficulty Raven clones out there to harvest...

--scryptr

You run the eth pill on X16R? I'd think overclocking with no pill would get you a better hash rate, no? Just because lots of the algos aren't memory intensive.

[Johnjay06]

Is there a secret to getting the open source one to work on x16s? I get result does not validate on cpu errors everywhere.  Is this only for x16r?

[scryptr]

SPMOD #6 RUNS SMOOTHLY ON MY 1080ti--

With a quick 15-minute trial, I got about 50 accepts, no errors, 100% accepts.  I was running with "The Pill", a slight memory overclock, and intensity 23.  I did not try to evaluate speed, but there were no red-lines or errors.  There are some low-difficulty Raven clones out there to harvest...

--scryptr

You run the eth pill on X16R? I'd think overclocking with no pill would get you a better hash rate, no? Just because lots of the algos aren't memory intensive.

I DID NOT WANT TO CHANGE MY SETTINGS--

The test ran only about 15 minutes.  The setting of "-i 23" was already in my batch file for X16R.  The miner worked without errors for the brief test.       --scryptr

[bensam1231]

SPMOD #6 RUNS SMOOTHLY ON MY 1080ti--

With a quick 15-minute trial, I got about 50 accepts, no errors, 100% accepts.  I was running with "The Pill", a slight memory overclock, and intensity 23.  I did not try to evaluate speed, but there were no red-lines or errors.  There are some low-difficulty Raven clones out there to harvest...

--scryptr

You run the eth pill on X16R? I'd think overclocking with no pill would get you a better hash rate, no? Just because lots of the algos aren't memory intensive.

I DID NOT WANT TO CHANGE MY SETTINGS--

The test ran only about 15 minutes.  The setting of "-i 23" was already in my batch file for X16R.  The miner worked without errors for the brief test.       --scryptr

Unless I'm mistaken the pill modification is just for Ethereum, why would you use it for other algos?


Also we're getting realllllly close to oblivion as far as profit goes.

[ironkid]

It modifies you memory timings so it works for everything relevant to that like Monero or other memory intensive stuff.

[sp_]

Is there a secret to getting the open source one to work on x16s? I get result does not validate on cpu errors everywhere.  Is this only for x16r?

X16S is broken, this is for x16r.

[sp_]

The setting of "-i 23" was already in my batch file for X16R.

You should use the default intensities for more speed. The default intensity for the 1080ti is 24

[sp_]

after around 10 hours of mining

spmod-git opensource
5174 accepted
18.58 RVN

https://eu.ravenminer.com/?address=RCQqCp2Ne7DSKYCEAZwjXqjiuKGAC78Hzr

t-rex 0.5.7

https://eu.ravenminer.com/?address=RAZqaXtN1jhrnW2BV8vc4WkNv3YQBne5GN

4989 accepted
17.96220330 RVN


spmod-git +3.70% faster and 0%fee.

now I should swap the cards and run the test again for another 10 hours. but too lazy. Try yourself..

[scryptr]

The setting of "-i 23" was already in my batch file for X16R.

You should use the default intensities for more speed. The default intensity for the 1080ti is 24

THIS IS A CHANGE--

The last version assigned "-i 21" as default.  I tried higher intensities, and anything greater than "-i 23 "resulted in failure.  This time I started where I left off, and did not go further.  Profitability is low, even if performance is high.  I have better coins to mine.

As for use of "The Pill", it is harmless when not needed, and helps when needed.  I game with it.       --scryptr

[bensam1231]

It modifies you memory timings so it works for everything relevant to that like Monero or other memory intensive stuff.

Interesting, basically all the good algos for Nvidia are GPU based so I never really looked into it, although some of the top earning algos right now are AMD centric (like Dagger), unless you know niches. However, even though I have a dozen or so coins that are off radar, they're all pretty much as fucked. Also majority of my rigs are 1070s.


Going back to a topic that I've brought up a few times over the years, AMD still looks like the right bet, when things get rough it all equalizes with Ethereum. Even though for awhile the profit/$hardware was higher for Nvidia, that gap has basically turned completely negative for like the last six months. So something like a 580 or Vega56 earns quite a bit more per dollar spent then a 1080ti or especially a 1070ti (which has better hash/$ and efficiency).

Although, right now, no one should be buying hardware (or should've been for the last eight months). It's a interesting predicament mining has always circled back around to.

[pallas]

Of course. Making money can't last forever.

[sp_]

Still profitable..

in 2014 When bitcoin tanked from $800 to $200 (Us gov. Silkroad dump) the mining profit was destroyed.

"when they can't control the market they attack the POW"
"when they can't control the POW they attack the market"

[antantti]

1x1080ti 2018 = 5x750ti 2014.

Cost to build a "rig", earnings, power consumption, everything. Well except those free pci-e slots you theoretically have today.

Still too much hope and hype here, is it just me or is BTC about to be programmed to get parabolic again?

[soothaa]

1x1080ti 2018 = 5x750ti 2014.

Cost to build a "rig", earnings, power consumption, everything. Well except those free pci-e slots you theoretically have today.

Still too much hope and hype here, is it just me or is BTC about to be programmed to get parabolic again?

I think we sink until 2019 for a variety of TA reasons - we COULD break out before then though.

[pallas]

1x1080ti 2018 = 5x750ti 2014.

Cost to build a "rig", earnings, power consumption, everything. Well except those free pci-e slots you theoretically have today.

That might be true for some algo (some memory intensive ones), but it certainly isn't true for those that make little use of ram. Pascal is 16 nm and Maxwell is 28, that means more transistors and less power.

[wacko]

any other ideas?

Your bot is stupid, it's copying and pasting other users' messages from/in totally inappropriate topics, so it's obvious that it's a bot posting and not a real user.