[ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 486.

Quote from: cbuchner1 on May 14, 2014, 05:27:52 AM

And Christian, you can release only binaries, with no source code, I think you are credible enough for us to trust your binaries Wink

No I can't. The terms of the GPL prohibit this as long as we're forked off of cpuminer.

it is only allowed if it is for xpm Grin

(when you see a soft which use -xpm instead of -w (for wallet)... you know from where it is coming)

leofur

member

Activity: 117

Merit: 10

And Christian, you can release only binaries, with no source code, I think you are credible enough for us to trust your binaries Wink

No I can't. The terms of the GPL prohibit this as long as we're forked off of cpuminer.

Ohh, well if anybody here wants to rewrite cpuminer, that can be awesome, wish I could help with that.

zelante

full member

Activity: 263

Merit: 100

Quote from: zelante on May 14, 2014, 05:22:17 AM

it and post early -> https://bitcointalksearch.org/topic/m.6721404

Care to share your binaries with us?
And Christian, you can release only binaries, with no source code, I think you are credible enough for us to trust your binaries Wink

do you trust me? maybe i have implement code for searching for you passwords and wallets? example is find all wallet.dat on you pc and kill it.

izum286

newbie

Activity: 2

Merit: 0

Quote from: cbuchner1 on May 14, 2014, 04:13:17 AM

Quote from: izum286 on May 14, 2014, 03:54:29 AM

Quote

A GTX 780 (non Ti) registered at 5 MHash/s.

HOW? I have 780ti - 3 MHash only...

duh, check out github source code and compile. This one does 4 MHash then.

How can I compile this?
I'm not a programmer ))

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: cbuchner1 on May 14, 2014, 05:20:54 AM

Quote from: cbuchner1 on May 14, 2014, 05:20:54 AM

but with "killer groestl" this problem is not problem? Wink

Yes, do a search for "timeline" and look at the one I posted previously.

the problem is that this code is just way too easy to port to OpenCL, and hence we're looking at a way of releasing
it that doesn't involve disclosure of source code.

Hope you will find soon, I must admit I more interested by groestl than x11
I am compiling the new code and give it a try in a few minutes.
Does this update solves the low gpu usage seen in the previous version ?

jk_14

legendary

Activity: 1292

Merit: 1000

but with "killer groestl" this problem is not problem? Wink

Yes, do a search for "timeline" and look at the one I posted previously.

the problem is that this code is just way too easy to port to OpenCL, and hence we're looking at a way of releasing
it that doesn't involve disclosure of source code.

[2014-05-14 12:25:28] thread 1: 2544086 hashes, 2117 khash/s
[2014-05-14 12:25:28] accepted: 116/116 (100.00%), 6150 khash/s (yay!!!)
[2014-05-14 12:25:28] thread 0: 2067342 hashes, 2118 khash/s
[2014-05-14 12:25:28] accepted: 117/117 (100.00%), 6258 khash/s (yay!!!)
[2014-05-14 12:25:29] thread 2: 9229564 hashes, 2100 khash/s
[2014-05-14 12:25:29] accepted: 118/118 (100.00%), 6334 khash/s (yay!!!)
[2014-05-14 12:25:34] thread 1: 11682100 hashes, 2111 khash/s
[2014-05-14 12:25:34] accepted: 119/119 (100.00%), 6329 khash/s (yay!!!)
[2014-05-14 12:25:35] thread 0: 13765326 hashes, 2096 khash/s
[2014-05-14 12:25:35] accepted: 120/120 (100.00%), 6307 khash/s (yay!!!)

Nice indeed, Christian! Cheesy

P.S
If somebody use KopiemTu, here is hint how to install ccminer & monitoring:
https://litecointalk.org/index.php?topic=16800.msg169584#msg169584

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: zelante on May 14, 2014, 05:22:17 AM

And Christian, you can release only binaries, with no source code, I think you are credible enough for us to trust your binaries Wink

No I can't. The terms of the GPL prohibit this as long as we're forked off of cpuminer.

gpuminer

full member

Activity: 210

Merit: 100

Quote from: gpuminer on May 14, 2014, 05:20:14 AM

Quote from: cbuchner1 on May 14, 2014, 05:20:54 AM

Well if it depends on the algo, I guess the new version won't fix it... Now the question is if it worth getting a strong CPU now for my rig...

intel i7 3770k@4500MHz 1 instance of ccminer with x11 use 6,5% of cpu. Using 19,5% now for 660ti + 2x 750ti (660ti take same percents of usage cpu as 750ti)

On an i3 750ti uses 9% of CPU, on celeron 19%.
Maybe Christian can give a better answer what causes the high CPU usage.

the "throughput" value was set pretty low (i.e. the number of threads thrown at the GPUs at once). The current github code has it increased 8-fold.
This should lower CPU use.

need windows binary. I will test it and report finding.

already test it and post early ->  https://bitcointalksearch.org/topic/m.6721404

there is a difference between testing with 1 person and testing with 10 person. I am using 6 x gtx 750 ti using windows 8.1.

zelante

full member

Activity: 263

Merit: 100

Quote from: zelante on May 14, 2014, 05:22:17 AM

but with "killer groestl" this problem is not problem? Wink

Yes, do a search for "timeline" and look at the one I posted previously.

the problem is that this code is just way too easy to port to OpenCL, and hence we're looking at a way of releasing
it that doesn't involve disclosure of source code.

i think most of our community understand that and will waiting for release "killer groestl" as much time as you need to do it. Cool

leofur

member

Activity: 117

Merit: 10

it and post early -> https://bitcointalksearch.org/topic/m.6721404

Care to share your binaries with us?
And Christian, you can release only binaries, with no source code, I think you are credible enough for us to trust your binaries Wink

zelante

full member

Activity: 263

Merit: 100

Quote from: gpuminer on May 14, 2014, 05:20:14 AM

Well if it depends on the algo, I guess the new version won't fix it... Now the question is if it worth getting a strong CPU now for my rig...

intel i7 3770k@4500MHz 1 instance of ccminer with x11 use 6,5% of cpu. Using 19,5% now for 660ti + 2x 750ti (660ti take same percents of usage cpu as 750ti)

On an i3 750ti uses 9% of CPU, on celeron 19%.
Maybe Christian can give a better answer what causes the high CPU usage.

the "throughput" value was set pretty low (i.e. the number of threads thrown at the GPUs at once). The current github code has it increased 8-fold.
This should lower CPU use.

need windows binary. I will test it and report finding.

already test it and post early -> https://bitcointalksearch.org/topic/m.6721404

cbuchner1

hero member

Activity: 756

Merit: 502

but with "killer groestl" this problem is not problem? Wink

Yes, do a search for "timeline" and look at the one I posted previously.

the problem is that this code is just way too easy to port to OpenCL, and hence we're looking at a way of releasing
it that doesn't involve disclosure of source code.

gpuminer

full member

Activity: 210

Merit: 100

Quote from: cbuchner1 on May 14, 2014, 05:12:50 AM

Well if it depends on the algo, I guess the new version won't fix it... Now the question is if it worth getting a strong CPU now for my rig...

intel i7 3770k@4500MHz 1 instance of ccminer with x11 use 6,5% of cpu. Using 19,5% now for 660ti + 2x 750ti (660ti take same percents of usage cpu as 750ti)

On an i3 750ti uses 9% of CPU, on celeron 19%.
Maybe Christian can give a better answer what causes the high CPU usage.

the "throughput" value was set pretty low (i.e. the number of threads thrown at the GPUs at once). The current github code has it increased 8-fold.
This should lower CPU use.

need windows binary. I will test it and report finding.

zelante

full member

Activity: 263

Merit: 100

new timeline for ccMiner v1.0 on X11. Note how the table based Groestl completely dominates the runtime.
SIMD has been cut into four pieces, and shrunk to a runtime shorter than Echo already.

You may want to compare this to the timeline I posted earlier where Groestl was optimized ("Killer" version)
but the SIMD hash was particularly slow and nasty.

but with "killer groestl" this problem is not problem? Wink

cbuchner1

hero member

Activity: 756

Merit: 502

new timeline for ccMiner v1.0 on X11. Note how the table based Groestl completely dominates the runtime.
SIMD has been cut into four pieces, and shrunk to a runtime shorter than Echo already.

You may want to compare this to the timeline I posted earlier where Groestl was optimized ("Killer" version)
but the SIMD hash was particularly slow and nasty.

supercoin434

newbie

Activity: 20

Merit: 0

If a kind soul coul compile for compute 3.0 and 3.5 ,, yourishiku .... Roll Eyes

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: S_tring on May 14, 2014, 04:55:47 AM

Where is this new version, I only see ver 1.0 on the official release (github)?

gitub HEAD revision. No release.

S_tring

full member

Activity: 252

Merit: 102

OPEN Platform - Powering Blockchain Acceptance

Quote from: sp00lin9 on May 14, 2014, 04:41:19 AM

Quote from: zelante on May 14, 2014, 04:29:10 AM