Author

Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 363. (Read 3426976 times)

hero member
Activity: 756
Merit: 502
What will be the next step for ccminer going forward ? It seems that there is no news regarding new release or development. Can we expect to see anything new this month ?

I think even before the weekend. Adding that killer groestl should be an easy exercise.
hero member
Activity: 756
Merit: 502
What is killer groestl quad? I mean what quad does?

quad spreads the state of one hash over 4 threads to save registers.
A necessity for Compute 3.0 devices that have a 63 registers per thread limit.

sr. member
Activity: 300
Merit: 250
What makes the GTX780 so much faster than a 750ti, is it the core speed or the increase amount of memory...

Huh
kepler vs maxwell
highend vs lowend

i think for lowend card the GTX750ti does pretty well thanks to the maxwell architecture.
sad the nvidia postponed the highend maxwell cards...
also sad no vendor did launch the dual GM107 card which could save some space for mining rigs.
just hope one day some vendor will create pure mining card with more GPUs, just like asrock did with motherboard.
legendary
Activity: 1154
Merit: 1001
Me and AizenSou exchanged some settings a few pages back. Shouldn't be too hard to locate if you search my post history.
If these are not helpful, let us know what exactly you're looking for (what card & what's the nfactor), so someone else might have the settings for you.

Cheers,
~ Myagui
legendary
Activity: 3248
Merit: 1070
command-line for mining scrypt jane-n factor?

i can't see it on http://cudamining.cc/
sr. member
Activity: 330
Merit: 252
...and get part of the jackpot (which I missed  Angry,
I had the perfect plan, half of the hashrate on miningpool the other hash@me  Angry

I had exactly the same plan - with the exact same success Cool
newbie
Activity: 18
Merit: 0
Is it normal for jackpotcoin mining to consistently show 1/3 my hashrate being reported on pools via their site?

My 660ti is hashing at a reported 3.5Mh, but the pool I mine on shows 1.5ish.

Also, have I correctly assumed that mining JPC on a pool is better than solo with my limited hashrate?
sr. member
Activity: 252
Merit: 250
Uro: 1 URO = 1 metric tonne of Urea N46 fertilizer
What makes the GTX780 so much faster than a 750ti, is it the core speed or the increase amount of memory...

Its got more than double the amount of GPU cores in the chip.
full member
Activity: 196
Merit: 100
Christian, I remember when you started this thread.

Any chance you would join over at Litecointalk.org ??

Congratulations on all your success.
hero member
Activity: 494
Merit: 500
What makes the GTX780 so much faster than a 750ti, is it the core speed or the increase amount of memory...
member
Activity: 61
Merit: 10
What is killer groestl quad? I mean what quad does?
full member
Activity: 210
Merit: 100

Translation: You'll be able to offer binary versions only in the future?


We don't yet have any codebase for binary only versions.

However the other Christian has dropped out of C&C hash factory (leaving in good terms) to concentrate
more on personal business. So I get to decide the strategy for code releases.

I feel inclined to release the Killer Groestl quad version which works best on Compute 3.0 and still decently
on Compute 3.5. This is nontrivial to port to AMD due to its use of warp shuffle instructions.

What will be the next step for ccminer going forward ? It seems that there is no news regarding new release or development. Can we expect to see anything new this month ?
hero member
Activity: 756
Merit: 502
Hi ppl,
   Managed to compile ccminer (linux) on a system without the nvidia card and ran the binary successfully on the rig with 750Ti. The speed is about 2110kh/s and seems to be similar to what others has posted here.

   My qn is that Windows binaries offered for download has various version for Compute 3, 3.5 and 5.0. I didn't see that option when I was compiling on Linux. How can we choose which compute to compile for? Does it solely depend on the CUDA toolkit version (was using v6)?

on Linux we build all Compute versions into one binary. Under Visual Studio the GUI doesn't have the flexibility to configure it in the same way.
sr. member
Activity: 249
Merit: 250
Hi ppl,
   Managed to compile ccminer (linux) on a system without the nvidia card and ran the binary successfully on the rig with 750Ti. The speed is about 2110kh/s and seems to be similar to what others has posted here.

   My qn is that Windows binaries offered for download has various version for Compute 3, 3.5 and 5.0. I didn't see that option when I was compiling on Linux. How can we choose which compute to compile for? Does it solely depend on the CUDA toolkit version (was using v6)?
hero member
Activity: 756
Merit: 502
I have tried to add another algo the the x11.cu, but all I get is
"result for nonce ... does not validate on cpu.

I tried to repeat the experience by adding a new layer of shavite (known to work) and again all I get is
"result for nonce ... does not validate (instead of boo... if for example I remove an algo)

Did I overlook something ? Is there a limit to the number of algo I can add up (and how can I change it ) ?


there is a CPU version of the hash, and a GPU version. Both have to agree of course,
otherwise you get the dreaded validation error.

Christian
legendary
Activity: 1400
Merit: 1050
I have tried to add another algo the the x11.cu, but all I get is
"result for nonce ... does not validate on cpu.

I tried to repeat the experience by adding a new layer of shavite (known to work) and again all I get is
"result for nonce ... does not validate (instead of boo... if for example I remove an algo)

Did I overlook something ? Is there a limit to the number of algo I can add up (and how can I change it ) ?


Djm34 wants to get us X13 before Christian. Hand clap to him. Bravo dude.  Grin
Btw X13 coins are almost craps now. The future is NIST5 Cheesy
It is just for fun and since it does not work...  Undecided
(I don't really mine X13 either...)
I stopped mining nist5 (the payout is a bit small at the moment) today to go to jpc and get part of the jackpot (which I missed  Angry,
I had the perfect plan, half of the hashrate on miningpool the other hash@me  Angry
 Grin Difficult to be more unlucky as I usually mine on hashharder)

sold this morning another million at 64sat, and I am wondering if I should rebuy it at now 48  Grin (would still be a profit of 0.16btc) or just cash out the 0.64  Grin
hero member
Activity: 938
Merit: 1000
I have tried to add another algo the the x11.cu, but all I get is
"result for nonce ... does not validate on cpu.

I tried to repeat the experience by adding a new layer of shavite (known to work) and again all I get is
"result for nonce ... does not validate (instead of boo... if for example I remove an algo)

Did I overlook something ? Is there a limit to the number of algo I can add up (and how can I change it ) ?


Djm34 wants to get us X13 before Christian. Hand clap to him. Bravo dude.  Grin
Btw X13 coins are almost craps now. The future is NIST5 Cheesy
hero member
Activity: 978
Merit: 506
Hi Christian.

I'm a small miner, just got a share of jackpot mining at hashharder pool.
Is your JPC address in read.me file still actual? Please confirm, and I will send you some coins. Thank you.

If this info can be of any help: Along with 750ti's I'm using MSI 750 non-ti 1gb series also. On jackpotcoin card gives me around 3300+ khs/s in win7 64 and around 3500khs/s with win xp
Using ccminer v0.9. Asrock btc81-pro with Intel Pentium G3220 and 4Gb ram. Classic unpowered risers. OC-ed GPU+170Mhz RAM+100Mhz with precisionX. Cards are cooler than ti's because of less cuda cores, generating less heat. Temp.48c with 48% fan speed. Ti's are hotter, 55c at 66% fan speed. I have mixed rig of 6 cards. Works like a charm, now almost a month.
Tried also solo with 4 cards and found 3 blocks in 48hours at diff around 300 last week.
Thank you again for your dedicated work.
legendary
Activity: 1400
Merit: 1050
I have tried to add another algo the the x11.cu, but all I get is
"result for nonce ... does not validate on cpu.

I tried to repeat the experience by adding a new layer of shavite (known to work) and again all I get is
"result for nonce ... does not validate (instead of boo... if for example I remove an algo)

Did I overlook something ? Is there a limit to the number of algo I can add up (and how can I change it ) ?
hero member
Activity: 644
Merit: 500

Hmm, the "C Hash Factory" has a nice ring to it. Smiley


Perhaps just the "Cash factory"
+1 =D

I feel inclined to release the Killer Groestl quad version which works best on Compute 3.0 and still decently
on Compute 3.5. This is nontrivial to port to AMD due to its use of warp shuffle instructions.
Nice, my gtx760 will love the ring of that Cheesy
But it will still give the 2.8MH to a 750ti? Please, say yes  Kiss

Jump to: