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

[joblo]

I get no validation errors on the gtx 960.
Only on the 970. 25% of the nonces are invalid. And the accepted hashrate on the pool is lower than reported in the miner.

I don't get that many validation erros but the pool hash rate is lower than r74.

[joblo]

Whirlpool compile failed on first attempt on Centos 6 cuda 7.5. trying again
and will also try cuda 6.5.

Update: Failed again

x15/cuda_x15_whirlpool.cu(2783): error: identifier "UINT32_MAX" is undefined

[djm34]

Whirlpool compile failed on first attempt on Centos 6 cuda 7.5. trying again
and will also try cuda 6.5.

Update: Failed again

x15/cuda_x15_whirlpool.cu(2783): error: identifier "UINT32_MAX" is undefined

what exactly are you reporting ?

[joblo]

Whirlpool compile failed on first attempt on Centos 6 cuda 7.5. trying again
and will also try cuda 6.5.

Update: Failed again

x15/cuda_x15_whirlpool.cu(2783): error: identifier "UINT32_MAX" is undefined

what exactly are you reporting ?

Something unrelated to your new kernel. I cloming it again in case of a corrupt download.

[djm34]

ok here the list of the donators I got from my pm box and browsing through that thread,
if I forgot your name (it is your fault   as I said to send me a pm),
let me know, I will add it (please just don't make me crossref everything...) :

(to be added to readme.txt and readme.md)

===== acknowledgment ========

djm34 would like to thank all the people who participated in the pledge and made that release possible:


Nicehash
antantti
crackers (www.zpool.ca)
scryptr
Liquid71
sp_
frazier34567
thefix
chrysophylax
ldw-com
Grout
serg_25
s7icky
Grim
Pivlus
lawrencelyl
sambiohazard
ltc_bilic
crsminer

========================================================

[s7icky]

f5 f5 f5 f5 f5 f5

you know, that I sleep from time to time
actually the source has been put on github yesterday (silently... as I am checking a few things in the mean time)
(I will probably also release the source of my own release, because the release of sp seems to cause several problem in performance)

nope ur not allowed to sleep

[djm34]

could you please try the newest version on github.
This is should use the best set of parameter depending if you are using cuda 6.5 or cuda 7.5 (since it isn't the same parameter)
problem with cpu validation should also be corrected, need some optimization as well as not all card need it...

please also report best intensity and fill up eliovp spreadsheet at http://62.212.74.86/~mining/list/

[joblo]

Whirlpool compile failed on first attempt on Centos 6 cuda 7.5. trying again
and will also try cuda 6.5.

Update: Failed again

x15/cuda_x15_whirlpool.cu(2783): error: identifier "UINT32_MAX" is undefined

what exactly are you reporting ?

Something unrelated to your new kernel. I cloming it again in case of a corrupt download.

I just remembered I've seen this before on the same machine and only on that machine.
It started around v1.5.70. I work around it by ading a #define in the fike that fails to
compile. Probably an issue with my system.

[djm34]

Whirlpool compile failed on first attempt on Centos 6 cuda 7.5. trying again
and will also try cuda 6.5.

Update: Failed again

x15/cuda_x15_whirlpool.cu(2783): error: identifier "UINT32_MAX" is undefined

what exactly are you reporting ?

Something unrelated to your new kernel. I cloming it again in case of a corrupt download.

I just remembered I've seen this before on the same machine and only on that machine.
It started around v1.5.70. I work around it by ading a #define in the fike that fails to
compile. Probably an issue with my system.

can't really help you with centos, try to ask epsylon3 is the expert in all kind of linux

[djm34]

ok, first release for neoscrypt: https://github.com/djm34/ccminer-sp-neoscrypt/releases

It contains exe for both cuda 6.5 and cuda 7.5 for windows (the version will pick up the best set of parameter for cuda 6.5/cuda 7.5 depending if you use cuda 6.5/cuda 7.5

[joblo]

Whirlpool compile failed on first attempt on Centos 6 cuda 7.5. trying again
and will also try cuda 6.5.

Update: Failed again

x15/cuda_x15_whirlpool.cu(2783): error: identifier "UINT32_MAX" is undefined

what exactly are you reporting ?

Something unrelated to your new kernel. I cloming it again in case of a corrupt download.

I just remembered I've seen this before on the same machine and only on that machine.
It started around v1.5.70. I work around it by ading a #define in the fike that fails to
compile. Probably an issue with my system.

can't really help you with centos, try to ask epsylon3 is the expert in all kind of linux

Just an FYI, not asking for help, but thanks.

[s7icky]

ok, first release for neoscrypt: https://github.com/djm34/ccminer-sp-neoscrypt/releases

It contains exe for both cuda 6.5 and cuda 7.5 for windows (the version will pick up the best set of parameter for cuda 6.5/cuda 7.5 depending if you use cuda 6.5/cuda 7.5

testing now. Thank you sir!!!!!

[djm34]

ok, first release for neoscrypt: https://github.com/djm34/ccminer-sp-neoscrypt/releases

It contains exe for both cuda 6.5 and cuda 7.5 for windows (the version will pick up the best set of parameter for cuda 6.5/cuda 7.5 depending if you use cuda 6.5/cuda 7.5

testing now. Thank you sir!!!!!

don't forget to report. also, you'll get better performance when overclocked (obviously)

[sp_]

Compiled with cuda 6.5 on Fedora 20.
EVGA 970 sc 640KH/s up from 580. Seven invalid nonces in 10 minutes.
Awsome work. Thanks to you SP and all the donators, especially crackfoo.

10.4% faster. This is far form the 100% that was promised...

[djm34]

Compiled with cuda 6.5 on Fedora 20.
EVGA 970 sc 640KH/s up from 580. Seven invalid nonces in 10 minutes.
Awsome work. Thanks to you SP and all the donators, especially crackfoo.

10.4% faster. This is far form the 100% that was promised...

100% is from the crap which was sitting in your "special cuda 7.5" release
also the code of this morning was optimized for cuda 7.5 and not cuda 6.5 and using 6.5 was giving lower performance
now this has been solved while you were acting like a moron on your thread.

[s7icky]

[2016-01-24 12:30:10] GPU #1: GeForce GTX 980 Ti, 955
[2016-01-24 12:30:10] GPU #0: GeForce GTX 980 Ti, 948
[2016-01-24 12:30:11] accepted: 215/215 (100.00%), 1895 kH/s yes!
[2016-01-24 12:30:11] accepted: 216/216 (100.00%), 1895 kH/s yes!

[2016-01-24 12:30:36] GPU #2: GeForce GTX 750 Ti, 206
[2016-01-24 12:30:36] accepted: 47/47 (100.00%), 202.73 kH/s yes!

[2016-01-24 12:28:55] GPU #0: GeForce GTX 960, 417
[2016-01-24 12:28:55] accepted: 29/29 (100.00%), 400.06 kH/s yes!

[djm34]

[2016-01-24 12:30:10] GPU #1: GeForce GTX 980 Ti, 955
[2016-01-24 12:30:10] GPU #0: GeForce GTX 980 Ti, 948
[2016-01-24 12:30:11] accepted: 215/215 (100.00%), 1895 kH/s yes!
[2016-01-24 12:30:11] accepted: 216/216 (100.00%), 1895 kH/s yes!

[2016-01-24 12:30:36] GPU #2: GeForce GTX 750 Ti, 206
[2016-01-24 12:30:36] accepted: 47/47 (100.00%), 202.73 kH/s yes!

[2016-01-24 12:28:55] GPU #0: GeForce GTX 960, 417
[2016-01-24 12:28:55] accepted: 29/29 (100.00%), 400.06 kH/s yes!

ok looks good, is it cuda 6.5 or cuda 7.5 ? for the moment cuda 6.5 should do better on the big gun

[s7icky]

[2016-01-24 12:30:10] GPU #1: GeForce GTX 980 Ti, 955
[2016-01-24 12:30:10] GPU #0: GeForce GTX 980 Ti, 948
[2016-01-24 12:30:11] accepted: 215/215 (100.00%), 1895 kH/s yes!
[2016-01-24 12:30:11] accepted: 216/216 (100.00%), 1895 kH/s yes!

[2016-01-24 12:30:36] GPU #2: GeForce GTX 750 Ti, 206
[2016-01-24 12:30:36] accepted: 47/47 (100.00%), 202.73 kH/s yes!

[2016-01-24 12:28:55] GPU #0: GeForce GTX 960, 417
[2016-01-24 12:28:55] accepted: 29/29 (100.00%), 400.06 kH/s yes!

ok looks good, is it cuda 6.5 or cuda 7.5 ? for the moment cuda 6.5 should do better on the big gun

yea 6.5 is a little faster maybe 100 if that


do u think u will be able to get more out of 7.5 with a 980ti?

[maxim000]

ASUS 750Ti 2Gb GPU overclocked to 1396 mgz  - 227mh cuda 7.5 , 231mh cuda 6.5

_sp release 69 get 200mh

thanks !

[djm34]

[2016-01-24 12:30:10] GPU #1: GeForce GTX 980 Ti, 955
[2016-01-24 12:30:10] GPU #0: GeForce GTX 980 Ti, 948
[2016-01-24 12:30:11] accepted: 215/215 (100.00%), 1895 kH/s yes!
[2016-01-24 12:30:11] accepted: 216/216 (100.00%), 1895 kH/s yes!

[2016-01-24 12:30:36] GPU #2: GeForce GTX 750 Ti, 206
[2016-01-24 12:30:36] accepted: 47/47 (100.00%), 202.73 kH/s yes!

[2016-01-24 12:28:55] GPU #0: GeForce GTX 960, 417
[2016-01-24 12:28:55] accepted: 29/29 (100.00%), 400.06 kH/s yes!

ok looks good, is it cuda 6.5 or cuda 7.5 ? for the moment cuda 6.5 should do better on the big gun

yea 6.5 is a little faster maybe 100 if that


do u think u will be able to get more out of 7.5 with a 980ti?

can you compile it yourself ? I had to add a __syncthreads() to prevent a few card from spewing cpu validation problem,
if it is removed, it should give you already some speed increase for both cuda 6.5/cuda 7.5

(for the moment for cuda 7.5, I am still searching a few way of optimizing it).

Also as you have more memory than a 980, you could also try to use the --intensity flag to raise the number of thread, this should also increase a bit your performance.