CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 1198.

chrysophylax

legendary

Activity: 2912

Merit: 1091

--- ChainWorks Industries ---

Quote from: djm34 on December 29, 2014, 08:55:22 PM

Quote from: scryptr on December 29, 2014, 08:47:11 PM

Quote from: djm34 on December 29, 2014, 08:26:58 PM

Quote from: scryptr on December 29, 2014, 07:11:10 PM

DJM34 --

Perhaps the original author of neoscrypt code meant for it to be difficult to port to nVidia.

? still not sure why you are saying that...
and actually doesn't seem to be the case, actually the 980 already beats the 290x/public kernel... (which is around 300kh/s)

Wolf0 answered for you. I thought that the code for neoscrypt may have been obfuscated deliberately in order to hinder its porting to Cuda code, perhaps in response to CBuchner's release of difficult-to-port "Killer Groestl". I'm wondering why development for nVidia is behind, but I am not slighting your efforts.

because there is no bounty Grin

(and paid project or project which are profitable have the priority... )

in that case - what would it take to 'work' on this project to make it worth its weight in gold?

i would be happily pointing my miners to 'donate' for such a project ...

or is there something that could be done in private that would benefit the contributors 'first' before a public release of such optimized code?

#crysx

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: scryptr on December 29, 2014, 08:47:11 PM

Quote from: djm34 on December 29, 2014, 08:26:58 PM

Quote from: scryptr on December 29, 2014, 07:11:10 PM

DJM34 --

Perhaps the original author of neoscrypt code meant for it to be difficult to port to nVidia.

? still not sure why you are saying that...
and actually doesn't seem to be the case, actually the 980 already beats the 290x/public kernel... (which is around 300kh/s)

Wolf0 answered for you. I thought that the code for neoscrypt may have been obfuscated deliberately in order to hinder its porting to Cuda code, perhaps in response to CBuchner's release of difficult-to-port "Killer Groestl". I'm wondering why development for nVidia is behind, but I am not slighting your efforts.

because there is no bounty and no deadline Grin

(and paid project or project which are profitable to me have the priority... ).

actually regarding the code, I could have chosen the original code over Wolf0, so there is nothing obfuscated for nvidia, it is just a choice I made (and I was interested too to import those vector type to cuda, as I was already thinking about it since quite some time)

scryptr

legendary

Activity: 1797

Merit: 1028

Quote from: djm34 on December 29, 2014, 08:26:58 PM

Quote from: scryptr on December 29, 2014, 07:11:10 PM

DJM34 --

Perhaps the original author of neoscrypt code meant for it to be difficult to port to nVidia.

? still not sure why you are saying that...
and actually doesn't seem to be the case, actually the 980 already beats the 290x/public kernel... (which is around 300kh/s)

Wolf0 answered for you. I thought that the code for neoscrypt may have been obfuscated deliberately in order to hinder its porting to Cuda code, perhaps in response to CBuchner's release of difficult-to-port "Killer Groestl". I'm wondering why development for nVidia is behind, but I am not slighting your efforts.

What I asked originally, though, was what the "scaling" referred to, a proportional scale of nVidia model kh/s rate / nVidia model kh/s rate, or perhaps nVidia model rate / AMD model rate. Or something else, like iterations within the code. I don't know, I look at the code and wish I understood it better. --scryptr

chrysophylax

legendary

Activity: 2912

Merit: 1091

--- ChainWorks Industries ---

Quote from: scryptr on December 29, 2014, 08:17:09 PM

Quote from: ?? on ??

Quote from: scryptr on December 29, 2014, 07:11:10 PM

DJM34 --

Perhaps the original author of neoscrypt code meant for it to be difficult to port to nVidia.

I guess, since this is ccminer code, x11 would be more appropriate for a comparison. Does the scaling refer to relative hash power of nVidia models? --scryptr

No, I didn't make it hard to port to Nvidia, I optimized it for GCN cards, though.

Thanks for the answer. I look forward to seeing neoscrypt on nVidia. It's difficult to keep my 280x rig in shape, the algo-switching sgminer 5.x crashed on switching and even single algorithm mining is rough. I think one of my 280x cards is ready for warranty service. I get about 6.6mh/s per card at my current tweak for x11. --scryptr

agreed ... i have a about 9 of my cards for warranty and 6 to throw out ( purchased via ebay - which ill never do again as gigabyte have a 3year warranty on their cards ) ...

for some reason it the AMD cards that seem to have the most issues with failing ...

#crysx

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: scryptr on December 29, 2014, 07:11:10 PM

DJM34 --

Perhaps the original author of neoscrypt code meant for it to be difficult to port to nVidia.

? still not sure why you are saying that...
and actually doesn't seem to be the case, actually the 980 already beats the 290x/public kernel... (which is around 300kh/s)

scryptr

legendary

Activity: 1797

Merit: 1028

Quote from: ?? on ??

Quote from: scryptr on December 29, 2014, 07:11:10 PM

DJM34 --

Perhaps the original author of neoscrypt code meant for it to be difficult to port to nVidia.

I guess, since this is ccminer code, x11 would be more appropriate for a comparison. Does the scaling refer to relative hash power of nVidia models? --scryptr

No, I didn't make it hard to port to Nvidia, I optimized it for GCN cards, though.

Thanks for the answer. I look forward to seeing neoscrypt on nVidia. It's difficult to keep my 280x rig in shape, the algo-switching sgminer 5.x crashed on switching and even single algorithm mining is rough. I think one of my 280x cards is ready for warranty service. I get about 6.6mh/s per card at my current tweak for x11. --scryptr

scryptr

legendary

Activity: 1797

Merit: 1028

DJM34 --

Perhaps the original author of neoscrypt code meant for it to be difficult to port to nVidia.

I guess, since this is ccminer code, x11 would be more appropriate for a comparison. Does the scaling refer to relative hash power of nVidia models? --scryptr

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: scryptr on December 29, 2014, 06:46:31 PM

Quote from: djm34 on December 29, 2014, 05:57:01 PM

Quote from: djm34 on December 28, 2014, 12:39:01 PM

Quote from: djm34 on December 28, 2014, 11:20:45 AM

some progress on neoscrypt:
750ti: ~~81kh/s~~ ~~110kh/s~~ 107kh/s
780ti: ~~165kh/s~~ ~~182kh/s~~ 193kh/s
980: ~~290kh/s~~ ~~299kh/s~~ 335kh/s
(still doesn't scale right Huh

)

DJM34 --

Please explain what you mean by scaling? Should it reach the same ratio between card models as scrypt?

I was expecting something different... but actually it looks like sp_ x11 in terms of scaling

Quote from: scryptr on December 29, 2014, 06:46:31 PM

I.E: [scrypt kh/s 750ti] / [scrypt kh/s 780ti] = [neoscrypt kh/s 750ti] / [neoscrypt kh/s 780ti]

and so forth between other model levels.

Actually, CBuchner said his released "Killer Groestl" code was supposed to be deliberately difficult to port to AMD. Perhaps this code is in a similar vein. --scryptr

not sure what you mean here...

scryptr

legendary

Activity: 1797

Merit: 1028

Quote from: djm34 on December 29, 2014, 05:57:01 PM

Quote from: djm34 on December 28, 2014, 12:39:01 PM

Quote from: djm34 on December 28, 2014, 11:20:45 AM

some progress on neoscrypt:
750ti: ~~81kh/s~~ ~~110kh/s~~ 107kh/s
780ti: ~~165kh/s~~ ~~182kh/s~~ 193kh/s
980: ~~290kh/s~~ ~~299kh/s~~ 335kh/s
(still doesn't scale right Huh

)

DJM34 --

Please explain what you mean by scaling? Should it reach the same ratio between card models as scrypt?

I.E: [scrypt kh/s 750ti] / [scrypt kh/s 780ti] = [neoscrypt kh/s 750ti] / [neoscrypt kh/s 780ti]

and so forth between other model levels.

Actually, CBuchner said his released "Killer Groestl" code was supposed to be deliberately difficult to port to AMD. Perhaps this code is in a similar vein. --scryptr

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: djm34 on December 28, 2014, 12:39:01 PM

Quote from: djm34 on December 28, 2014, 11:20:45 AM

some progress on neoscrypt:
750ti: ~~81kh/s~~ ~~110kh/s~~ 107kh/s
780ti: ~~165kh/s~~ ~~182kh/s~~ 193kh/s
980: ~~290kh/s~~ ~~299kh/s~~ 335kh/s
(still doesn't scale right Huh

)

scryptr

legendary

Activity: 1797

Merit: 1028

Quote from: sp_ on December 29, 2014, 01:55:26 PM

Fixed slow cpu. X11,x13,x14,x15,x17, nist, quark. Lyra is still a bit slow but faster than build 22

The 23nd public beta(windows executable) is available here: (29-12-2014)

http://www.filedropper.com/ccminer23

The sourcecode is available here:

https://github.com/sp-hash/ccminer

Post your stats here. Card name/gpu clock/memclock

SP_BUILD23 --

I have some initial results. First, the following photo has tpruvot's v. 1.5.2-git, a build that has recent improvements but is not the absolute bleeding edge, though I consider it stable:

Second, today's sp_ v23, a version I consider bleeding edge and hope Cheesy

is also stable:

Both results were taken from my slower 6-750ti rig, today. I am building ccminer sp_ v23 on my faster rig as I post this.

EDIT: Here is a snap of my faster 6-750ti rig:

--scryptr

Bombadil

hero member

Activity: 644

Merit: 500

Quote from: K1773R on December 29, 2014, 04:14:10 PM

il try it later with mono

if only basic things are used it should be fine.

Here's a test-version compiled, if you need it (note, this is not a release ^^"): https://mega.co.nz/#!wZdETYQa!ct5DpGhJ_ZpqdmFGGRLalEhcy8YsKfufDsxmA2eBOUg
There's nothing in rig overview, I know Tongue

K1773R

legendary

Activity: 1792

Merit: 1008

/dev/null

Quote from: Bombadil on December 29, 2014, 04:01:55 PM

Quote from: jpouza on December 29, 2014, 04:01:05 PM

Just rolled back to release 21, 6% CPU usage against 44% of release 22.

There's release 23 already: https://bitcointalksearch.org/topic/m.9976879

Quote from: K1773R on December 29, 2014, 04:01:29 PM

Quote from: Bombadil on December 29, 2014, 04:00:55 PM

Quote from: K1773R on December 29, 2014, 03:34:55 PM

does it work with mono?

No idea, and I won't get my hopes up for it ^^" But it works over the network, so another windoze pc or virtualbox with windoze should work.

winblows is no option Wink

Aahw. Well seeing you're a Legendary and you've managed to stick with the game for so long, you'll probably have deeprooted ideals in which Windoze doesn't fit ^^"
Well, I'm sorry to say, but I'm just an autodidact when it comes to programming, IRL I'm an ISP technician, so don't expect that much from me Cheesy

(/me should've just sticked with java, but no, C# just had to be more powerful Tongue

)

il try it later with mono

if only basic things are used it should be fine.

jpouza

legendary

Activity: 2716

Merit: 1116

Quote from: Bombadil on December 29, 2014, 04:01:55 PM

Quote from: jpouza on December 29, 2014, 04:01:05 PM

Just rolled back to release 21, 6% CPU usage against 44% of release 22.

There's release 23 already: https://bitcointalksearch.org/topic/m.9976879

Tks!

Bombadil

hero member

Activity: 644

Merit: 500

Quote from: jpouza on December 29, 2014, 04:01:05 PM

Just rolled back to release 21, 6% CPU usage against 44% of release 22.

There's release 23 already: https://bitcointalksearch.org/topic/m.9976879

Quote from: K1773R on December 29, 2014, 04:01:29 PM

Quote from: Bombadil on December 29, 2014, 04:00:55 PM

Quote from: K1773R on December 29, 2014, 03:34:55 PM

does it work with mono?

No idea, and I won't get my hopes up for it ^^" But it works over the network, so another windoze pc or virtualbox with windoze should work.

winblows is no option Wink

Aahw. Well seeing you're a Legendary and you've managed to stick with the game for so long, you'll probably have deeprooted ideals in which Windoze doesn't fit ^^"
Well, I'm sorry to say, but I'm just an autodidact when it comes to programming, IRL I'm an ISP technician, so don't expect that much from me Cheesy

(/me should've just sticked with java, but no, C# just had to be more powerful Tongue

)

K1773R

legendary

Activity: 1792

Merit: 1008

/dev/null

Quote from: Bombadil on December 29, 2014, 04:00:55 PM

Quote from: K1773R on December 29, 2014, 03:34:55 PM

does it work with mono?

No idea, and I won't get my hopes up for it ^^" But it works over the network, so another windoze pc or virtualbox with windoze should work.

winblows is no option Wink

jpouza

legendary

Activity: 2716

Merit: 1116

Just rolled back to release 21, 6% CPU usage against 44% of release 22.

Bombadil

hero member

Activity: 644

Merit: 500

Quote from: K1773R on December 29, 2014, 03:34:55 PM

does it work with mono?

No idea, and I won't get my hopes up for it ^^" But it works over the network, so another windoze pc or virtualbox with its own network connection should work.

tbearhere

legendary

Activity: 3164

Merit: 1003

About 50kh more on quark with no intensity settings. With #23 750ti.

K1773R

legendary

Activity: 1792

Merit: 1008

/dev/null

Quote from: Bombadil on December 29, 2014, 02:20:50 PM

Quote from: sp_ on December 29, 2014, 01:55:26 PM

Fixed slow cpu. X11,x13,x14,x15,x17, nist, quark. Lyra is still a bit slow but faster than build 22

The 23nd public beta(windows executable) is available here: (29-12-2014)

http://www.filedropper.com/ccminer23

The sourcecode is available here:

https://github.com/sp-hash/ccminer

Post your stats here. Card name/gpu clock/memclock

Same hashrate on X11 but only 12% cpu usage now ^^ It was 8% before build22 for 2x 750TIs. Thx.

(Started build23 after the last red area/line, ignore the others Tongue

)

Now let's hope you'll keep the API so I can keep on monitoring.
I'm nearing the first release of it, and it would be a real shame if all was for nothing.
Source of almost-100%-working ccmonitor is already out there: https://github.com/KBomba/ccmonitor

does it work with mono?

Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 1198. (Read 2347597 times)