[ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX]

[NeuroticFish]

nice looking ccminer splitscreen

+1

Shame it doesn't work with my card

[Newwsr]

Just for interest here is what I am getting at the moment. On a 750 GTX Ti running djm34's latest fork off ccminer (https://github.com/djm34/ccminer) compiled with CUDA 6.0.
Per card of course, and rounded down.

X11    2,300k/s
X13    1,800k/s (about 50w)
x15    1,200k/s
Nist5  7,500k/s (about 54w)

I have tried various versions with CUDA 6.0 and 5.5 but its really hard to see any difference.

I am sticking on TradeMyBit at the moment with a homemade switcher that is flipping between X13 and Nist5 most of the time

I'm using this Mod v1.02-X13-version - 12/06/2014 she works for x11 x13 and x11 doing more in 2.200k / s

I can't recall if v1.02 contains the improved groestl implementation. Give v1.2 a try and you might see an improvement with both x11 and x13.

Got 2,700 k / s with a bit of overclocking, using ccminer v1.1

[DougB62]

@cayars: I have been using nvminer since a day or so after you released it, and everything has been fine. However, yesterday, it suddenly started incorrectly reporting all my cards as GTX 660's (I have 2x EVGA GTX 660's and 2x EVGA GTX 750Ti's). It had been reporting correctly all the time before it started this. All of the other versions that I have (I have most of them) are still reporting correctly. Re-booting the system, and disabling/re-enabling the cards have had no effect on this behavior. Any ideas???

That is strange indeed.  My gut feeling is that it's a driver issue/glitch.

What does GPU-Z report as the card type?

Reporting correctly:

[pinoyminer1977]

I heard so much good with 750 Ti, is it still worth buying this card despite of the profitability of mining right now?

[Parana]

Following to know about the news.

 

[djm34]

if your mining with nvidia then you should ALWAYS be using ccminer or cudaminer

Again a notice to the following people

• Amph
• Bombadil
• Cbuchner1
• tsiv
• zelante

Message me for your login details to cudamining.cc
So far cayars has his and has been using it, i like the new description for your release by the way
djm34 has his logins but has yet to submit anything

sorry I didn't have time to far...

[bigjme]

if your mining with nvidia then you should ALWAYS be using ccminer or cudaminer

Again a notice to the following people

• Amph
• Bombadil
• Cbuchner1
• tsiv
• zelante

Message me for your login details to cudamining.cc
So far cayars has his and has been using it, i like the new description for your release by the way
djm34 has his logins but has yet to submit anything

sorry I didn't have time to far...

you lazy.... only joking. Dont worry i'm not expecting you to. And it isn't a rant for anyone wondering, just wanted to get everyone's attention save messaging them directly and forcing the details on you 

i had to put something in the message to explain why you weren't in the list
zelante has his details for the profitCalc and ive added the option for the program on the site so he can add that.

Basically anything that can help with mining that is from this community or very helpful will be added now

[Amph]

8 days from miner start
nice to see this stability in windows 8.1  

that's cool, source is available?

if your mining with nvidia then you should ALWAYS be using ccminer or cudaminer

Again a notice to the following people

• Amph
• Bombadil
• Cbuchner1
• tsiv
• zelante

Message me for your login details to cudamining.cc
So far cayars has his and has been using it, i like the new description for your release by the way
djm34 has his logins but has yet to submit anything

sorry I didn't have time to far...

you lazy.... only joking. Dont worry i'm not expecting you to. And it isn't a rant for anyone wondering, just wanted to get everyone's attention save messaging them directly and forcing the details on you 

i had to put something in the message to explain why you weren't in the list
zelante has his details for the profitCalc and ive added the option for the program on the site so he can add that.

Basically anything that can help with mining that is from this community or very helpful will be added now

i've only compiled the x15 source of djm and x14, but they are pointless now, because of the last fresh algo release

[Bombadil]

if your mining with nvidia then you should ALWAYS be using ccminer or cudaminer

Again a notice to the following people

• Amph
• Bombadil
• Cbuchner1
• tsiv
• zelante

Message me for your login details to cudamining.cc
So far cayars has his and has been using it, i like the new description for your release by the way
djm34 has his logins but has yet to submit anything

sorry I didn't have time to far...

you lazy.... only joking. Dont worry i'm not expecting you to. And it isn't a rant for anyone wondering, just wanted to get everyone's attention save messaging them directly and forcing the details on you  

i had to put something in the message to explain why you weren't in the list
zelante has his details for the profitCalc and ive added the option for the program on the site so he can add that.

Basically anything that can help with mining that is from this community or very helpful will be added now

*cough* bombadil has his... *cough* I'm very busy adding new features, and once my code is cleaned up again, I'll do another release-round, probably this weekend.

[bigjme]

*cough* bombadil has his... *cough* I'm very busy adding new features, and once my code is cleaned up again, I'll do another release-round, probably this weekend.

Look forward to seeing it lmao, i didn't forget about you, honest 

i've only compiled the x15 source of djm and x14, but they are pointless now, because of the last fresh algo release

Well the login is there for anything else. I will just allow us all to keep track of what everyone has done and when

[cayars]

if your mining with nvidia then you should ALWAYS be using ccminer or cudaminer

Again a notice to the following people

• Amph
• Bombadil
• Cbuchner1
• tsiv
• zelante

Message me for your login details to cudamining.cc
So far cayars has his and has been using it, i like the new description for your release by the way
djm34 has his logins but has yet to submit anything

Check out the description for the version I just uploaded.

http://cudamining.cc/url/releases

v1.2U-D3 - 15-07-2014

[StuffOfInterest]

Check out the description for the version I just uploaded.

http://cudamining.cc/url/releases

v1.2U-D3 - 15-07-2014

When you say this....

Compiled to support both Compute 3 and Compute 3.5 devices.

I presume you also mean Compute 5 being that is what the GTX 750 Ti is using.

The chart of hash speeds is definitely nice.

[cayars]

Check out the description for the version I just uploaded.

http://cudamining.cc/url/releases

v1.2U-D3 - 15-07-2014

When you say this....

Compiled to support both Compute 3 and Compute 3.5 devices.

I presume you also mean Compute 5 being that is what the GTX 750 Ti is using.

The chart of hash speeds is definitely nice.

No. It's compiled just like ccminer for Compute 3 and Compute 3.5 only.  It of course also works on Compute 5.

I did however download/install Cuda 6.0 and did compile it also for Compute 5.0 but did not get any speed upgrades.  However on Compute 5.0 FRESH algo would crash.
Since there was nothing to gain from Compute 5.0 I just compiled it as usual.

Hope that's not to confusing.

BTW, the hash speeds posted are from the "averaging" included in this version.  While some of these hash rates might seem a bit lower then some people are used to seeing quoted.  I'm sure these are much closer to reality and better used for calculations.

[bigjme]

Check out the description for the version I just uploaded.

http://cudamining.cc/url/releases

v1.2U-D3 - 15-07-2014

your enjoying the editing system then?

[atc1]

Back after a longish hiatus. Have C&C stopped working on the original cudaminer? No more improvements to the scrypt algo? What algo's trending now? What are you guys mining? Afaik,the 750 Ti's still the best option for mining,yes?

Nice to see so many independent shoot-off's now. Seeing some new names as well. 
Could some please tell me about any thing important that has happened from the beginning of last month?
Thanks a lot.

[StuffOfInterest]

Check out the description for the version I just uploaded.

http://cudamining.cc/url/releases

v1.2U-D3 - 15-07-2014

When you say this....

Compiled to support both Compute 3 and Compute 3.5 devices.

I presume you also mean Compute 5 being that is what the GTX 750 Ti is using.

The chart of hash speeds is definitely nice.

No. It's compiled just like ccminer for Compute 3 and Compute 3.5 only.  It of course also works on Compute 5.

I did however download/install Cuda 6.0 and did compile it also for Compute 5.0 but did not get any speed upgrades.  However on Compute 5.0 FRESH algo would crash.
Since there was nothing to gain from Compute 5.0 I just compiled it as usual.

Hope that's not to confusing.

BTW, the hash speeds posted are from the "averaging" included in this version.  While some of these hash rates might seem a bit lower then some people are used to seeing quoted.  I'm sure these are much closer to reality and better used for calculations.

Interesting.  I would have thought the 750 Ti would benefit from the higher CUDA level.  Wonder if there are any features not being utilized which could benefit from it.

Regarding the averaging, I haven't looked at the code but I did think about how this could be calculated beyond just the last 50.  If you keep a long int with a sum of the share hashrates (in khash) and an int with the count of shares all you have to do is divide the two numbers to get a hashrate across the entire run.

[cayars]

Check out the description for the version I just uploaded.

http://cudamining.cc/url/releases

v1.2U-D3 - 15-07-2014

When you say this....

Compiled to support both Compute 3 and Compute 3.5 devices.

I presume you also mean Compute 5 being that is what the GTX 750 Ti is using.

The chart of hash speeds is definitely nice.

No. It's compiled just like ccminer for Compute 3 and Compute 3.5 only.  It of course also works on Compute 5.

I did however download/install Cuda 6.0 and did compile it also for Compute 5.0 but did not get any speed upgrades.  However on Compute 5.0 FRESH algo would crash.
Since there was nothing to gain from Compute 5.0 I just compiled it as usual.

Hope that's not to confusing.

BTW, the hash speeds posted are from the "averaging" included in this version.  While some of these hash rates might seem a bit lower then some people are used to seeing quoted.  I'm sure these are much closer to reality and better used for calculations.

Interesting.  I would have thought the 750 Ti would benefit from the higher CUDA level.  Wonder if there are any features not being utilized which could benefit from it.

Regarding the averaging, I haven't looked at the code but I did think about how this could be calculated beyond just the last 50.  If you keep a long int with a sum of the share hashrates (in khash) and an int with the count of shares all you have to do is divide the two numbers to get a hashrate across the entire run.

This new routine uses from 50 to 250 rounds depending on the algo.  Faster running stuff like Jackpot will use 250 while slower moving algo like CryptoNight will only use 50 with the others in the middle at 100 or 150.

If you just used a long int how would you know what to roll off the "end" to only keep say 50 or 100 values?  What I did was use an array to track up to 250 positions (depends on algo) with the total as another variable.  I then just implement a counter and roll it back to 0 when I've hit the top.  For each new share I know what position in the array I'm about to populate or replace.  I subtract the value currently in the array from my variable and add the new value to the variable and also to the proper spot in the array.  EASY PEASY.

[StuffOfInterest]

This new routine uses from 50 to 250 rounds depending on the algo.  Faster running stuff like Jackpot will use 250 while slower moving algo like CryptoNight will only use 50 with the others in the middle at 100 or 150.

If you just used a long int how would you know what to roll off the "end" to only keep say 50 or 100 values?  What I did was use an array to track up to 250 positions (depends on algo) with the total as another variable.  I then just implement a counter and roll it back to 0 when I've hit the top.  For each new share I know what position in the array I'm about to populate or replace.  I subtract the value currently in the array from my variable and add the new value to the variable and also to the proper spot in the array.  EASY PEASY.

Why do you need to roll values off?  Why not just track average since the miner started?  If you are seeing the rate swing up or down throughout the course of a run then it would make sense to only use a set number of points but if the number should be essentially stable then you might as well use the entire history.  Alternatively, you could use the array method along side the accumulator method so you can see the short term trend against the entire session average which could help someone to see if their card is speeding up or slowing down over a long run.

[sp_]

The latest sgminer can be run on 750ti as well. Install catalyst 14.6. And the latest nvidia driver. Nvidia support opencl as well.

X11 performs at around around 2000 khash.

[StuffOfInterest]

The latest sgminer can be run on 750ti as well. Install catalyst 14.6. And the latest nvidia driver. Nvidia support opencl as well.

X11 performs at around around 2000 khash.

I don't think you need to install the Catalyst driver as the nVIDIA driver already has OpenCL support.  I've run sgminer against an nVIDIA card a couple of times to test for algorithms that were not supported by ccminer yet.  Still, doesn't make much sense.  Consider that you are getting 2000 khash on X11 while cayars' recent update to nvminer (a fork of ccminer) is pulling over 2600 khash.