[ANN]pledge for the release of 100% faster cuda 7.5 nvidia miner for neoscrypt

[djm34]

not entirely sure...
#thread = 256 * 64 * 5 on cuda 7.5 intensity is #thread =  2^intensity (so it is between 16.47 and 16.48)
and
#thread = 256 * 64 * 4 on cuda 6.5 (probably possible to raise it here) ==> intensity = 16

THANK YOU djm34!!!!!  

my 970 now run @ 657 kH/s!!!!!!!!

before i'm using sp release 7.4 and get around 542 kH/s.

i'm currently using your cuda 65 version.

can you give more info the difference between standard and nosync version?

Cmon guys, i see a lot of results here, screenshots, speeds from different cards.


If it's not to much to ask.

Add those results here : Mining Hardware Comparison

 

already submitted

the nosync refers to the command __syncthreads() which should be used when initializing the shared memory. However in some case (and some card) it isn't really needed and as it tends to slow down the kernel and removing it may increase the hashrate. however on some card it may cause some cpu validation issues.  

[kangjooe]

the nosync refers to the command __syncthreads() which should be used when initializing the shared memory. However in some case (and some card) it isn't really needed and as it tends to slow down the kernel and removing it may increase the hashrate. however on some card it may cause some cpu validation issues.  

i notice sir. already try it and it does give about 10 kH/s more than standard cuda65 version but yes it also give many cpu validation in my card.

i think i will stick with cuda 65 default version.

i sincerely thank you for your work and all the donator that make this happen


PS:
sorry i can't afford any donation, i'm doing this as a new hobby to run from the stress of my daily work for the past 8 weeks, i mine only about 4-5 hours and not every day, and only get about 0.001xxx BTC once a week on nicehash. purely to make me feel i can achieve something... even for that small amount.

[sambiohazard]

Ok i tried cuda 6.5 version as DJM said it faster. Here are my result which i will submit now to Elivop's site.

No flags/switches used, no OC. Efficiency is 99.5% for 750ti & 970. These are numbers after 1 hour run

sp rel71
750ti: 172 kh/s
970: 520 kh/s

djm 6.5

750ti: 202 kh/s(17.5%)
970: 605 kh/s(16.5%)

djm6.5nosync

750ti: +2-3kh/s
970: +10kh/s

EDIT: Forgot to thank djm34. Thanks for all your work. Last time you didn't answer my PM about mining donation i thought you really only want to sell to big miners & may be you do. Now i can see that with help of all parties in ecosystem a crowdfund can be successful. If you keep doing such crowdfunds every few months with similar improvements i think this can work.

[Liquid71]

Ok i tried cuda 6.5 version as DJM said it faster. Here are my result which i will submit now to Elivop's site.

No flags/switches used, no OC. Efficiency is 99.5% for 750ti & 970. These are numbers after 1 hour run

sp rel71
750ti: 172 kh/s
970: 520 kh/s

djm 6.5

750ti: 202 kh/s(17.5%)
970: 605 kh/s(16.5%)

djm6.5nosync

750ti: +2-3kh/s
970: +10kh/s

EDIT: Forgot to thank djm34. Thanks for all your work. Last time you didn't answer my PM about mining donation i thought you really only want to sell to big miners & may be you do. Now i can see that with help of all parties in ecosystem a crowdfund can be successful. If you keep doing such crowdfunds every few months with similar improvements i think this can work.

Agreed!!

I donated .1 not because I'll ever earn that back but to support devs, especially djm which has done a lot for ccminer.

P.S. anything worth mining with the faster miner, haven't even downloaded yet. Just mining most profitable algo's at the moment. Planning to fill some bags with FTC but for now gotta earn this months electric costs before the bill comes 

[Liquid71]

Cmon guys, i see a lot of results here, screenshots, speeds from different cards.


If it's not to much to ask.

Add those results here : Mining Hardware Comparison

 

Yes this site is/can be a great tool for all of us, thanks to the creator.

[Eliovp]

Cmon guys, i see a lot of results here, screenshots, speeds from different cards.


If it's not to much to ask.

Add those results here : Mining Hardware Comparison

 

Yes this site is/can be a great tool for all of us, thanks to the creator.

You're very welcome!

Now if everyone would keep contributing!

I'll keep it add free for sure! Nothing more irritating than a website with a bunch of adds..

[serg_25]

What i can say. another good job DJM.
my 750Ti card can make

Code:

GPU #1: GeForce GTX 750 Ti, 240 (T= 47C F= 32% C=1397/2750)

previously it was 200.

[djm34]

What i can say. another good job DJM.
my 750Ti card can make

Code:

GPU #1: GeForce GTX 750 Ti, 240 (T= 47C F= 32% C=1397/2750)

previously it was 200.

wow, that's pretty high... What kind of card is it ?

[Pivlus]

18% increase, djm34_neoscrypt compared to 1.5.74_spmod.
720 kH/s on EVGA GTX980 SC using Cuda 6.5 version. Submitted inferior result in Mining Hardware Comparison chart (540 kH/s) probably due to CPU bottleneck (I was mining argon2 on 12 threads while testing ccminer).
Thanks.

[djm34]

18% increase, djm34_neoscrypt compared to 1.5.74_spmod.
720 kH/s on EVGA GTX980 SC using Cuda 6.5 version. Submitted inferior result in Mining Hardware Comparison chart (540 kH/s) probably due to CPU bottleneck (I was mining argon2 on 12 threads while testing ccminer).
Thanks.

make sure you have enough ram or pagefile (this is required because of the pre-allocation).
which version are you using ? with the 980 you can use the __nosync version which should be faster.

It is based on the sp 78 version, actually my dev version is somewhat faster... I will try to put the binary somewhere, so a few test can be done.

You can also try to increase the intensity

[serg_25]

What i can say. another good job DJM.
my 750Ti card can make

Code:

GPU #1: GeForce GTX 750 Ti, 240 (T= 47C F= 32% C=1397/2750)

previously it was 200.

wow, that's pretty high... What kind of card is it ?

This one http://www.gainward.com/main/vgapro.php?id=969&lang=en
clocks you can see above.

[djm34]

What i can say. another good job DJM.
my 750Ti card can make

Code:

GPU #1: GeForce GTX 750 Ti, 240 (T= 47C F= 32% C=1397/2750)

previously it was 200.

wow, that's pretty high... What kind of card is it ?

This one http://www.gainward.com/main/vgapro.php?id=969&lang=en
clocks you can see above.

same as the one I have, must have lost at the chip lottery

[djm34]

Just thought about that when looking at the mining hardware comparison http://62.212.74.86/~mining/list/nvidia/index.php?algo=neo. ;
but the card showing lower performance are running windows 10.
Forgot about that since I am still runing windows 8.1 but there are major slow down on hard mem algo with windows 10.
So if you rig is for mining only, I would suggest to downgrade to a previous version.

[blzflkn]

Two issues I found so far:

1: The myr-gr algorithm doesn't seem to be working. (ccminer 65) Tried on a pool and in benchmark. Pool was all errors, benchmark didn't do anything at all.

2: With --cpu-mining, using a 2 core cpu masked with 0x3, both cores will run at ~50% on cpu priority 0 and 1... but setting it to 2,3,4, or 5 will max out one core and leave the other at zero. (lyra2v2, if that matters.)

Mining on GT 730 cuda 3.5

[djm34]

Two issues I found so far:

1: The myr-gr algorithm doesn't seem to be working. (ccminer 65) Tried on a pool and in benchmark. Pool was all errors, benchmark didn't do anything at all.

2: With --cpu-mining, using a 2 core cpu masked with 0x3, both cores will run at ~50% on cpu priority 0 and 1... but setting it to 2,3,4, or 5 will max out one core and leave the other at zero. (lyra2v2, if that matters.)

Mining on GT 730 cuda 3.5

I think you should bring the issue to sp (as I didn't touch that part of the software)

[sp_]

Looks ike this kernal is good on some cards and slow on others. Elpida? Hynix?

But not profitable. On both. Djm34 haz been trolling my thread for 6 months. A good coder,but a slow troll..

[djm34]

Looks ike this kernal is good on some cards and slow on others. Elpida? Hynix?

But not profitable. On both. Djm34 haz been trolling my thread for 6 months. A good coder,but a slow troll..

hmmm... are you lonesome tonight ?  that you come to my thread just to say that ?  
 
https://www.youtube.com/watch?v=_cS5aCozhcA

 

more seriously, I don't have enough cards (and both have the same brand and memory) to tell anything about that.
However, what I think has an effect, is the OS, win10 slows down things a lot for hard mem algo from what I remember
(of the last time I installed it... I may try that soon again, but it has to pass both neoscrypt and lyra2rev2 test to stay
on my computer)

[DragonSlayer]

djm34,

What would it take to get a pledge running for lyra2rev2 release?  I keep hoping and dreaming of the day.  Thank you for all your work for the community.

[djm34]

djm34,

What would it take to get a pledge running for lyra2rev2 release?  I keep hoping and dreaming of the day.  Thank you for all your work for the community.

a long time...

[djm34]

djm34,

What would it take to get a pledge running for lyra2rev2 release?  I keep hoping and dreaming of the day.  Thank you for all your work for the community.

a long time...

that doesn't seem to be wise to run a second pledge right after this one... I mean most of the donation were from rather large donators and asking them to donate a second time in a month.... unless there were a much larger participation of the community may-be... in terms of pricing, I guess I would be happy with the same goal...