CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 845.

bathrobehero

legendary

Activity: 2002

Merit: 1051

ICO? Not even once.

Quote from: sp_ on December 10, 2015, 06:22:57 AM

Quote from: bathrobehero on December 10, 2015, 06:01:35 AM

Quote from: sp_ on December 09, 2015, 05:48:27 PM

Quote from: dominuspro on December 09, 2015, 01:54:25 PM

Seems some more Hashes for each gpu. But non profitable nowadays Cry

Seems like Etherum is the only profitable algo at the moment on the NVIDIA unreleased gpu's

I'm not sure about that. Anybody know how much a 750 Ti is expected to earn a day (before electricity)?

The 750ti is not good in mining etherum. You need a compute 5.2 or bether.

Period   ETH   USD
Minute   0.000342   $0.00
Hour   0.020540   $0.02
Day   0.492959   $0.39
Week   3.450715   $2.73
Month   14.788778   $11.71
Notes

Thanks. How about 970s?

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

Quote from: dominuspro on December 10, 2015, 06:11:03 AM

Quote from: sp_ on December 09, 2015, 04:23:09 PM

New compiler alot of work..
Please donate some beers

Just sent 0.05061365 BTC so keep up with the good work Wink

Thanks

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

Quote from: bathrobehero on December 10, 2015, 06:01:35 AM

Quote from: sp_ on December 09, 2015, 05:48:27 PM

Quote from: dominuspro on December 09, 2015, 01:54:25 PM

Seems some more Hashes for each gpu. But non profitable nowadays Cry

Seems like Etherum is the only profitable algo at the moment on the NVIDIA unreleased gpu's

I'm not sure about that. Anybody know how much a 750 Ti is expected to earn a day (before electricity)?

The 750ti is not good in mining etherum. You need a compute 5.2 or bether.

Period   ETH   USD
Minute   0.000342   $0.00
Hour   0.020540   $0.02
Day   0.492959   $0.39
Week   3.450715   $2.73
Month   14.788778   $11.71
Notes

bensam1231

legendary

Activity: 1764

Merit: 1024

Quote from: Genoil on December 10, 2015, 12:28:29 AM

Quote from: sp_ on December 09, 2015, 05:48:27 PM

Quote from: dominuspro on December 09, 2015, 01:54:25 PM

Seems some more Hashes for each gpu. But non profitable nowadays Cry

Seems like Etherum is the only profitable algo at the moment on the NVIDIA unreleased gpu's

With TLB trashing now also affecting AMD, it's getting even better. And then there is a potential DAGpocalypse in 60 days which should be fun to watch

The Dagopalypse is nigh! Anyone with cards with a lot of VRAM should have fun. Cheesy

Have you considered simulating a larger dag file to see what effect it will have on performance?

Curiously what is the point of shifting to cuda 7.5 if it sucks so much for cryptos? Newer is just better?

dominuspro

full member

Activity: 201

Merit: 100

Quote from: sp_ on December 09, 2015, 04:23:09 PM

Quote from: chrysophylax on December 09, 2015, 03:45:58 PM

Quote from: sp_ on December 09, 2015, 12:04:55 PM

-Now compiled and adjusted for cuda 7.5
-Faster quark on compute 5.2 devices
-Neoscrypt is 50% slower.
-Most algos are slower or same speed.(0-5%)
1.5.75(sp-MOD) is available here: (9-dec-2015)
https://github.com/sp-hash/ccminer/releases/
The sourcecode is available here:
https://github.com/sp-hash/ccminer
Please post you hashrates guys.

like i was mentioning many times earlier - cuda 7.5 was working and compiling fine on my fedora 22 x64 systems ... and were on par - if not better ( by a little bit ) than cuda 6.5 ...
now that you are tuning the kernels to this compile - this should be a nice update in hashrate - though the rates that a lot of people are getting here are in no way related to what im getting on stock clocks ...
ill test today if there is any chance to ...
tanx for new update sp ...
#crysx

I have just tuned a few kernals in quark and x11 (qubit). Lyra2v2, neoscrypt, x13,x14,x15,x17 etc. will be done later. Optimizing for compute 5.0 is hurting 5.2 performance. I should split the kernals and make seperate versions.

New compiler alot of work..

Please donate some beers

Just sent 0.05061365 BTC so keep up with the good work Wink

bathrobehero

legendary

Activity: 2002

Merit: 1051

ICO? Not even once.

Quote from: sp_ on December 09, 2015, 05:48:27 PM

Quote from: dominuspro on December 09, 2015, 01:54:25 PM

Seems some more Hashes for each gpu. But non profitable nowadays Cry

Seems like Etherum is the only profitable algo at the moment on the NVIDIA unreleased gpu's

I'm not sure about that. Anybody know how much a 750 Ti is expected to earn a day (before electricity)?

chrysophylax

legendary

Activity: 2940

Merit: 1091

--- ChainWorks Industries ---

Quote from: Daleaus on December 09, 2015, 11:56:09 PM

Quote from: chrysophylax on December 09, 2015, 11:16:25 PM

Quote from: techcoin12 on December 09, 2015, 10:28:29 PM

Daleaus

thanks 75 is a little hot for my office

but good to know

and chrysophylax had a good idea with the 25min test's ... thank you for posting the results!!!

tanx ...

ive found that for any test on the algos - 25mins is a prime ( and in most cases the optimum ) time for the algos to settle with hashrates ...

those evga cards are doing pretty though

... id really like to see what the gigabyte 980ti g1 cards do on stock for all those algos ...

#crysx

These cards can be purchased with a stock water cooler as well. I haven't got the need to buy them as 75 degrees isn't bad as far as GPU's go.

nice ...

asus have made one also ( with slightly lower clocks ) called the poseidon - which allows for both fan or water cooling built in on the one card ...

https://www.asus.com/au/Graphics-Cards/ROG_POSEIDONGTX780P3GD5/specifications/ ...

but for the price of that card ( almost 1300AUD ) i would rather take the risk with the gigabyte 980ti g1 higher clocked cards ...

but it also seems that the evga cards are of a higher quality than the gigbayte ones - as their fans dont continuously die on them ...

Daleaus

member

Activity: 103

Merit: 10

1.75 is giving cuda core errors after extended use on quark. It's also causing graphics to run about 7 degrees warmer compared to 1.74.

Genoil

sr. member

Activity: 438

Merit: 250

Quote from: sp_ on December 09, 2015, 05:48:27 PM

Quote from: dominuspro on December 09, 2015, 01:54:25 PM

Seems some more Hashes for each gpu. But non profitable nowadays Cry

Seems like Etherum is the only profitable algo at the moment on the NVIDIA unreleased gpu's

With TLB trashing now also affecting AMD, it's getting even better. And then there is a potential DAGpocalypse in 60 days which should be fun to watch

Daleaus

member

Activity: 103

Merit: 10

Quote from: chrysophylax on December 09, 2015, 11:16:25 PM

Quote from: techcoin12 on December 09, 2015, 10:28:29 PM

Daleaus

thanks 75 is a little hot for my office

but good to know

and chrysophylax had a good idea with the 25min test's ... thank you for posting the results!!!

tanx ...

ive found that for any test on the algos - 25mins is a prime ( and in most cases the optimum ) time for the algos to settle with hashrates ...

those evga cards are doing pretty though

... id really like to see what the gigabyte 980ti g1 cards do on stock for all those algos ...

#crysx

These cards can be purchased with a stock water cooler as well. I haven't got the need to buy them as 75 degrees isn't bad as far as GPU's go.

chrysophylax

legendary

Activity: 2940

Merit: 1091

--- ChainWorks Industries ---

Quote from: techcoin12 on December 09, 2015, 10:28:29 PM

Daleaus

thanks 75 is a little hot for my office

but good to know

and chrysophylax had a good idea with the 25min test's ... thank you for posting the results!!!

tanx ...

ive found that for any test on the algos - 25mins is a prime ( and in most cases the optimum ) time for the algos to settle with hashrates ...

those evga cards are doing pretty though

... id really like to see what the gigabyte 980ti g1 cards do on stock for all those algos ...

#crysx

chrysophylax

legendary

Activity: 2940

Merit: 1091

--- ChainWorks Industries ---

Quote from: Daleaus on December 09, 2015, 09:50:32 PM

Quote from: chrysophylax on December 09, 2015, 08:47:35 PM

Quote from: Daleaus on December 09, 2015, 06:04:07 PM

My tests with 2x evga 980ti's overclocked 1485mhz

Qubit------1.75 - 39100mh 1.74 - 43000mh
Quark------1.75 - 60572mh 1.74 - 59500mh
x13--------1.75 - 20500mh 1.74 - 24000mh
x11--------1.75 - 26100mh 1.74 - 29000mh
Neoscypt---1.75 - 544.83kh 1.74 - 1.65mh
Lyra2rev2---1.75 - 33048mh 1.74 - 33000mh

great hashrates ...

is there anyway - you would reproduce this test for 25mins on each algo using stock clocks please? ...

no oc for this card would still be high - and not an exact comparison to the gigabyte 980ti g1 cards ... but would be a much closer comparison ...

if not - thats fine - as 25min tests on stock would take some time ...

much appreciated ...

#crysx

My tests with 2x evga 980ti's overclocked 1485mhz

Qubit------1.75 - 39.100mh 1.74 - 43.000mh
Quark------1.75 - 60.572mh 1.74 - 59.500mh
x13--------1.75 - 20.500mh 1.74 - 24.000mh
x11--------1.75 - 26.100mh 1.74 - 29.000mh
Neoscypt---1.75 - 544.83kh 1.74 - 1.65mh
Lyra2rev2---1.75 - 33.048mh 1.74 - 33.000mh

Stock clocks 1290mhz 25 min tests

Qubit------1.75 - 35.533mh 1.74 - 38.856mh
Quark------1.75 - 52.569mh 1.74 - 52.226mh
x13--------1.75 - 18.387mh 1.74 - 20.37mh
x11--------1.75 - 23.045mh 1.74 - 25mh
Neoscypt---1.75 - 549.62kh 1.74 - 1.429mh
Lyra2rev2---1.75 - 30.039mh 1.74 - 30.354mh

By the way these evga 980ti's are the SC version so they come factory clocked at 1290mhz i run them at 1485mhz with modded bios

very much appreciated ...

this gives more of an indication to the stock performance of the cards themselves - and they are doing very well without oc ... in fact - those cards are doing very very well ...

the new ccminer-spmod75 seems to be doing much better now that its been tuned to c7.5 ...

tanx again for the results ...

#crysx

techcoin12

newbie

Activity: 23

Merit: 0

Daleaus

thanks 75 is a little hot for my office

but good to know

and chrysophylax had a good idea with the 25min test's ... thank you for posting the results!!!

Daleaus

member

Activity: 103

Merit: 10

Quote from: chrysophylax on December 09, 2015, 08:47:35 PM

Quote from: Daleaus on December 09, 2015, 06:04:07 PM

My tests with 2x evga 980ti's overclocked 1485mhz

Qubit------1.75 - 39100mh 1.74 - 43000mh
Quark------1.75 - 60572mh 1.74 - 59500mh
x13--------1.75 - 20500mh 1.74 - 24000mh
x11--------1.75 - 26100mh 1.74 - 29000mh
Neoscypt---1.75 - 544.83kh 1.74 - 1.65mh
Lyra2rev2---1.75 - 33048mh 1.74 - 33000mh

great hashrates ...

is there anyway - you would reproduce this test for 25mins on each algo using stock clocks please? ...

no oc for this card would still be high - and not an exact comparison to the gigabyte 980ti g1 cards ... but would be a much closer comparison ...

if not - thats fine - as 25min tests on stock would take some time ...

much appreciated ...

#crysx

My tests with 2x evga 980ti's overclocked 1485mhz

Qubit------1.75 - 39.100mh 1.74 - 43.000mh
Quark------1.75 - 60.572mh 1.74 - 59.500mh
x13--------1.75 - 20.500mh 1.74 - 24.000mh
x11--------1.75 - 26.100mh 1.74 - 29.000mh
Neoscypt---1.75 - 544.83kh 1.74 - 1.65mh
Lyra2rev2---1.75 - 33.048mh 1.74 - 33.000mh

Stock clocks 1290mhz 25 min tests

Qubit------1.75 - 35.533mh 1.74 - 38.856mh
Quark------1.75 - 52.569mh 1.74 - 52.226mh
x13--------1.75 - 18.387mh 1.74 - 20.37mh
x11--------1.75 - 23.045mh 1.74 - 25mh
Neoscypt---1.75 - 549.62kh 1.74 - 1.429mh
Lyra2rev2---1.75 - 30.039mh 1.74 - 30.354mh

By the way these evga 980ti's are the SC version so they come factory clocked at 1290mhz i run them at 1485mhz with modded bios

chrysophylax

legendary

Activity: 2940

Merit: 1091

--- ChainWorks Industries ---

Quote from: Daleaus on December 09, 2015, 06:04:07 PM

My tests with 2x evga 980ti's overclocked 1485mhz

Qubit------1.75 - 39100mh 1.74 - 43000mh
Quark------1.75 - 60572mh 1.74 - 59500mh
x13--------1.75 - 20500mh 1.74 - 24000mh
x11--------1.75 - 26100mh 1.74 - 29000mh
Neoscypt---1.75 - 544.83kh 1.74 - 1.65mh
Lyra2rev2---1.75 - 33048mh 1.74 - 33000mh

great hashrates ...

is there anyway - you would reproduce this test for 25mins on each algo using stock clocks please? ...

no oc for this card would still be high - and not an exact comparison to the gigabyte 980ti g1 cards ... but would be a much closer comparison ...

if not - thats fine - as 25min tests on stock would take some time ...

much appreciated ...

#crysx

Daleaus

member

Activity: 103

Merit: 10

Quote from: techcoin12 on December 09, 2015, 06:18:41 PM

Nice work once again SP_

Daleaus

thanks for the numbers ...

what are your temps on those 980Ti's ?
any idea on watts ?

Mmm about 75 degrees on stock fan with window open lol No idea on watts but they honestly don't use much more power then my old 660's, i don't notice a difference on my power bill anyways. Good cards, mine run 24-7 when I'm not gaming.

techcoin12

newbie

Activity: 23

Merit: 0

Nice work once again SP_

Daleaus

thanks for the numbers ...

what are your temps on those 980Ti's ?
any idea on watts ?

Daleaus

member

Activity: 103

Merit: 10

My tests with 2x evga 980ti's overclocked 1485mhz

Qubit------1.75 - 39100mh 1.74 - 43000mh
Quark------1.75 - 60572mh 1.74 - 59500mh
x13--------1.75 - 20500mh 1.74 - 24000mh
x11--------1.75 - 26100mh 1.74 - 29000mh
Neoscypt---1.75 - 544.83kh 1.74 - 1.65mh
Lyra2rev2---1.75 - 33048mh 1.74 - 33000mh

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

Quote from: dominuspro on December 09, 2015, 01:54:25 PM

Seems some more Hashes for each gpu. But non profitable nowadays Cry

Seems like Etherum is the only profitable algo at the moment on the NVIDIA unreleased gpu's

chrysophylax

legendary

Activity: 2940

Merit: 1091

--- ChainWorks Industries ---

Quote from: sp_ on December 09, 2015, 04:23:09 PM

Quote from: chrysophylax on December 09, 2015, 03:45:58 PM

Quote from: sp_ on December 09, 2015, 12:04:55 PM

-Now compiled and adjusted for cuda 7.5
-Faster quark on compute 5.2 devices
-Neoscrypt is 50% slower.
-Most algos are slower or same speed.(0-5%)
1.5.75(sp-MOD) is available here: (9-dec-2015)
https://github.com/sp-hash/ccminer/releases/
The sourcecode is available here:
https://github.com/sp-hash/ccminer
Please post you hashrates guys.

like i was mentioning many times earlier - cuda 7.5 was working and compiling fine on my fedora 22 x64 systems ... and were on par - if not better ( by a little bit ) than cuda 6.5 ...
now that you are tuning the kernels to this compile - this should be a nice update in hashrate - though the rates that a lot of people are getting here are in no way related to what im getting on stock clocks ...
ill test today if there is any chance to ...
tanx for new update sp ...
#crysx

I have just tuned a few kernals in quark and x11 (qubit). Lyra2v2, neoscrypt, x13,x14,x15,x17 etc. will be done later. Optimizing for compute 5.0 is hurting 5.2 performance. I should split the kernals and make seperate versions.

New compiler alot of work..

Please donate some beers

ill set a miner on the donation link when i get back into the office ... let it mine for a while ... http://donate-sp.granitecoin.com/ ...

i think a split would be good - as the kernels are now becoming quite a task to optimize and keep compatible to both compute versions from what i can see ...

tanx sp ...

#crysx

Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 845. (Read 2347659 times)