• Forum has been upgraded, all links, images, etc are as they were. Please see Official Announcements for more information

X11 Mining Optimisation Project

Ignition75

Active member
Hi all, I've been in discussions with a couple of very talented OpenCL (AMD) and Cuda (Nvidia) developers who both think there are decent optimisations to be had from spending some time on existing mining software.

Both have done similar projects for other coins, and, lets just say there was a large gap in the reward vs effort spent on these optimisations. There was a bit of reluctance from both developers to spend huge amounts of time on this.

So I thought about a crowd funding exercise, similar to the way we've funded other projects which have been of benefit to the DRK community.

How many of you would be interested in showing a little support with some DRK, along side myself, to give a nice boost to both AMD and Nvidia mining for x11? In a similar fashion to the Kristov Atlas review, the more we can raise, the more time can be spent on this exercise.

Nothing formal has been set-up at this stage, I'm really just gauging people's appetites for improvements on the mining software, if we all think this is a good idea I'll set-up an address and we can get the ball rolling.

smiley.gif
 
Well, according to Wolf from BCT, it can be improved upon by 400%. I don't know how feasible that actually is though. At this point in time, I'd like to see some development towards FPGA's to be honest with you.
 
Does anybody know why we seem to have lost 30GH/s of net hashrate over the last few days?
 
I've done a lot of optimizations for the AMD kernels and am pretty sure that even a 40% improvement is very unlikely.

Can't speak for CUDA, though.

If you've done optimizations and you doubt a 40% improvement, you sir, do not only know how GCN works - you don't know your asshole from your elbow when it comes to GPUs in general. Over most relevant AMD cards, Pitcairn, Tahiti, and Hawaii, I have an average increase of 34.34% without even trying. Amatuer hour compared to what is possible.
 
If you've done optimizations and you doubt a 40% improvement, you sir, do not only know how GCN works - you don't know your asshole from your elbow when it comes to GPUs in general. Over most relevant AMD cards, Pitcairn, Tahiti, and Hawaii, I have an average increase of 34.34% without even trying. Amatuer hour compared to what is possible.
I haven't heard back from my OpenCL guy, so looks like this is a no-go... Shame, would have been nice to give the community optimisations...
 
I'm a supporter of CUDA so offcourse it's in my interest to get the optimization going. I'm in.
 
I'm a supporter of CUDA so offcourse it's in my interest to get the optimization going. I'm in.
IMO, there's less gains to be had in the CUDA than in the OpenCL on AMD, but there's still some good increases available in it.
 
OK some great news, my OpenCL guy is in, I just have to organise sending him some spare hardware, which is going to be time consuming as I am in Australia and he is in France. I just have to confirm, Wolf0, are you still happy to do the Nvidia optimisation?

He's now having a think about how much time he will require, could you do the same please mate, I think we're going to be good to go in a couple of weeks. I'm going to reach out to every x11, x13, x15 and NIST5 coin to ensure we have a really good go at this.
 
OK some great news, my OpenCL guy is in, I just have to organise sending him some spare hardware, which is going to be time consuming as I am in Australia and he is in France. I just have to confirm, Wolf0, are you still happy to do the Nvidia optimisation?

He's now having a think about how much time he will require, could you do the same please mate, I think we're going to be good to go in a couple of weeks. I'm going to reach out to every x11, x13, x15 and NIST5 coin to ensure we have a really good go at this.

No; your earlier post basically said it was off, so I found other work.

One thing, though: note that almost all optimizations performed to X11 can be applied to X13 & X15, but the reverse is not true - and NIST5 uses a subset of hashes in X11, making it rare that optimizations to the X algos will apply.
 
OK, thanks for letting me know... I'll look for another Nvidia guy, I'm still fairly keen on doing this for the community...:smile:
 
I have optimized the x11 miner for NVIDIA. with compute 5.2 and 5.0

Current hashrates with OC is 3MHASH on the 750ti, 7 MHASH on the 970 and 8 MHASH on the 980. Over the weekend I might reach 3.5MHASH on the TI.
 
I have optimized the x11 miner for NVIDIA. with compute 5.2 and 5.0

Current hashrates with OC is 3MHASH on the 750ti, 7 MHASH on the 970 and 8 MHASH on the 980. Over the weekend I might reach 3.5MHASH on the TI.
Would you like to join as the dev doing OpenCL on this one?
 
I have optimized the x11 miner for NVIDIA. with compute 5.2 and 5.0

Current hashrates with OC is 3MHASH on the 750ti, 7 MHASH on the 970 and 8 MHASH on the 980. Over the weekend I might reach 3.5MHASH on the TI.

Is there a download link?
 
Back
Top