NSGminer v0.9.2 released with my NeoScrypt OpenCL kernel v7 and other enhancements.
1. All GCN based AMD Radeons get a hash rate increase of 20% to 100% depending on driver version. The difference between 14.6 and 15.7 drivers is less than 2% now, so R9 280X @ 1000MHz may deliver 500KH/s.
2. Performance of the older VLIW based AMD Radeons was doubled simply. HD6970 @ 925MHz delivers 255KH/s now.
3. Added support for the very old VLIW based AMD Radeons of HD4000 series. HD4870 @ 750MHz can do 60KH/s. Not very much, but what do you expect of a card 8 years old?
4. Added initial support for the NVIDIA hardware. Thanks to the Feathercoin community for their donation of 0.3 BTC spent on a GTX 750 Ti. Performance improved from 50KH/s to 185KH/s @ 1400MHz shaders. Older GeForce cards down to the very old 8000 series are also known to work.
5. NVIDIA Management Library (NVML) may be used to provide with temperature and fan speed data. Copy nvml.dll from your driver distribution package to the miner's directory.
A good deal of work has been put into this release, so consider a donation. The addresses and download links are in the OP.
|
|
|
Hey ghostlander, any idea's why FTC is so prone to orphans? It the only coin on my pool that makes more orphans then valid blocks. www.zpool.ca/site/block?id=308Maybe you're low on connections? If you add a few permanent ones to other big pools, it should help. I've tried that already. Tried having 64 connections and still no improvements. What version do you use?
|
|
|
Hey ghostlander, any idea's why FTC is so prone to orphans? It the only coin on my pool that makes more orphans then valid blocks. www.zpool.ca/site/block?id=308Maybe you're low on connections? If you add a few permanent ones to other big pools, it should help.
|
|
|
So this is the third pledge for a faster neoscrypt (two by djm34 and now another one, which is even more expensive). I wonder how people, who already donated two times, feel now :-) I released a 10% faster neoscrypt for free, I must have been crazy!!!! LoL!
Considering the amount of work involved (executable size with static linking is just 1Mb now vs. 40Mb before) and the fact that I'm not really interested in development of ccminer, and hasn't ever used it myself until recently, and have no sponsors like Nicehash, here we go. My performance improvement in absolute hash rate is high as djm34's last work (200KH/s -> 240KH/s -> 280KH/s) yet it's alright for him to ask 2.5 BTC and not alright for me to ask even 0.5 BTC? I understood you were asking 1 BTC and with no promise to release the source. Anyway you can do what you want, I was just thinking about the people who paid for a faster neoscrypt miner TWO times and they still do not have the faster one, and someone paying for it only one time now is faster than them :-) I do not mine neoscrypt so I don't care, but this 3-times-pledge looks a bit ridiculous to me. If the donations meet 1 BTC, I release everything I have: both the source code and a batch of Windows binaries, 32-bit/64-bit, CUDA 6.5/7.5, etc. If they exceed 0.5 BTC, but don't reach 1 BTC, I may release something, but make no promises. If they don't meet 0.5 BTC, only those who donate get the miner. Maybe you and djm34 are the same person? :-D
I don't speak French, that's something I'm sure about
|
|
|
So this is the third pledge for a faster neoscrypt (two by djm34 and now another one, which is even more expensive). I wonder how people, who already donated two times, feel now :-) I released a 10% faster neoscrypt for free, I must have been crazy!!!! LoL!
Considering the amount of work involved (executable size with static linking is just 1Mb now vs. 40Mb before) and the fact that I'm not really interested in development of ccminer, and hasn't ever used it myself until recently, and have no sponsors like Nicehash, here we go. My performance improvement in absolute hash rate is high as djm34's last work (200KH/s -> 240KH/s -> 280KH/s) yet it's alright for him to ask 2.5 BTC and not alright for me to ask even 0.5 BTC? watched your progress ... and its paying off ...
ill test with 980ti g1 cards and the extremes also ...
we have a bit to chat about mate ... i need the linux code to compile ...
great work ...
#crysx
I can share the source code with trusted people who can keep it private.
|
|
|
Let's refresh this thread. I have spent some time on djm34's miner. Stripped it down to NeoScrypt only, modified for Visual Studio 2010 compatibility to produce valid binaries for Windows XP and up, added support for UTF-16, replaced some code with more appropriate, a few more tweaks, etc. Long story short, it runs 17% faster now. That's a GTX 750 Ti @ 1400MHz shaders. It could do almost 240KH/s previously. Don't know how much big Maxwells like GTX 980 gain with it since I don't own any, so here we come to the point. Donate towards my purchase of a GTX 980 for the future development and I release my modded miner open source. The grand total is 1 BTC, however I may consider to let it go public if the donation volume exceeds 0.5 BTC. To make it fair, those who donate substantial amounts get miner binaries anyway. 1KqBw6tEWgmzoorx2PQAvr9GrbnkVWFn6M
|
|
|
There is also community voting at Cryptopia. Users can vote for HAL every day. It's definitely doable this way with active community participation.
|
|
|
cl_amd_media_ops is for bitalign/bytealign mostly which are not used in v6 directly. The compiler is supposed to take care of this, but it doesn't do well in the drivers newer than 14.7. It won't be an issue in the next release.
|
|
|
I find it interesting that of 2291 NeoScrypt miners at hashpower.co 2241 use my CPUminer v2.4.x, though their combined hash rate is only 14.5MH/s. That's an average of 6.5KH/s per connection. Maybe a botnet?
|
|
|
I'm really don't expect moon , but i see lack of support and trade platforms. Don't get me wrong but without exchanges this will stop to exists.. Well, there was a new exchange announced on the previous page, but it seems to be offline more time than online. No doubt a more reliable trade platform is needed. I have finally put together a simple web site. It will do the job for a while.
|
|
|
or tell me how?
Maybe at Cryptocointalk? I see this forum engine allows polls for new threads only. Besides many members of our community visit this forum rather occasionally.
|
|
|
Hey,
You are giving bounties for thread translation?
We have a few translated and abandoned threads already.
|
|
|
This is actually still alive project? I see only Cryptsy as marketplace? What is new here? I had not been for a while.. Do we have some progress finally?
What kind of progress? To the moon?
|
|
|
Couldn't even compile his ccminer for Windows XP where I have my 750 Ti running now. NVCC rejects MinGW and insists on M$ Visual Studio. VS2013 doesn't produce valid code even with vs120_xp target. Missing entry points in kernel32.dll like GetTickCount64 or InitializeCriticalSectionEx. So I had to strip ccminer down to NeoScrypt only and patch for VS2010 compatibility.
damn i like your style ... and the resultant strip down and test? ... btw - its one of the many reasons why i dont use windows ... too messy with compilations ... #crysx GTX 750 Ti @ 1400 = 285KH/s with CUDA 6.5 or 280KH/s with CUDA 7.5 Although I've somehow broken Stratum in the process, so it's solo mining through Getwork now.
|
|
|
Any time line on the new HAL wallet project?
Yea my wallet seems to go to sleep after a day or 2, then all good when I restart it. What does it mean "to go to sleep"? My wallets run and stake fine. It looks like everything is fine,shows connections,shows up to date, staking green. If I send something to it after a couple days,sometimes weeks, it doesn't show up. I restart it and it starts catching up on blockchain and I'm good to go again, for awhile anyway. This on windows 7, I may be changing to Linux in a few days so that may fix my problem. Add the core nodes from the OP to your halcyon.conf in order stay synchronised if you have no incoming connections.
|
|
|
I think ORB has such a great community. Looking forward to the ORB wallet update It's a work in progress, though there is nothing of a critical importance. Minor tweaks and improvements here and there. Is there anything special you'd like to see or get fixed? I'm glad you asked (even if you didn't ask me) I'd like to see at least the daemon working on a RasPi B v1... That would be a blast. As far as my knowledge goes, this would mean a lot smaller memory footprint (maybe with some performance penalties?)... Also the ARM platform support would mean wider audience (given the not-so-recent hype from beaglebone and the likes to PINE A64). You may say I'm a dreamer... I think I'll fix those 32-bit ARM issues, though memory reduction is a different question. Some may be achieved by reworking the block index structure. There are redundant parts like storage of block hashes which are easy to calculate with BLAKE2s and cached in addition. However most users won't be happy to redownload or bootstrap the entire block chain due to this optimisation.
|
|
|
More improvements.
GTX 750 Ti @ 1400 = 180KH/s
R9 280X @ 1000 = 500KH/s
Older Radeons boosted up again: HD6970 @ 925 = 255KH/s
thats some nice improvements ... checkout sp thread for ccminer ( which has djm34 neoscrypt included ) and see what you get from that? ... djm34 released his neoscrypt kernel only recently mate ... so a lot can be had ( and maybe even improved - though djm34 does a very thorough job ) from the kernel ... it is cuda based - but im sure the opensource kernel could help in some ways ... really impressed with your improvements ... when you decide you would like to get some work outside of here done - let me know ... i could use your help and optimizations with granite ... tanx ... #crysx Couldn't even compile his ccminer for Windows XP where I have my 750 Ti running now. NVCC rejects MinGW and insists on M$ Visual Studio. VS2013 doesn't produce valid code even with vs120_xp target. Missing entry points in kernel32.dll like GetTickCount64 or InitializeCriticalSectionEx. So I had to strip ccminer down to NeoScrypt only and patch for VS2010 compatibility.
|
|
|
Dudes, considering I have developed my miner to go with MYR I would expect to be at least mentioned by now... albeit I understand some closed source kernels at last managed to get slightly beyond my perf, I'd like to get some more feedback especially on GRS-MYR.
I don't think I've received any MYR donations on my kernel either. The DigiByte folks were more generous, though it didn't reach even 0.05 BTC anyway. It's a bad sign if a community ignores work done for the coin and discourages from any further work. Post me your MYR address and I'll send over a donation. It's all there in the OP. Thanks. New OpenCL Kernel for Myriad-Groestl (DGB, MYR, etc.)
|
|
|
Well, the price was somewhere about 100 sats before Cryptsy started to go crap. People converted their balance into other altcoins which they could withdraw and dump for BTC elsewhere. PXC wasn't traded anywhere else (Cryptopia had no volume basically), so they dumped PXC straight at Cryptsy.
|
|
|
Dudes, considering I have developed my miner to go with MYR I would expect to be at least mentioned by now... albeit I understand some closed source kernels at last managed to get slightly beyond my perf, I'd like to get some more feedback especially on GRS-MYR.
I don't think I've received any MYR donations on my kernel either. The DigiByte folks were more generous, though it didn't reach even 0.05 BTC anyway. It's a bad sign if a community ignores work done for the coin and discourages from any further work.
|
|
|
|