|
Posted new commit: "Reduced GPU RAM usage, code cleanup"
|
|
|
|
Win 8.1, gtx1060
Good hashrate! It looks like the 1070 should do more, as the 980Ti @180W is doing 16.8 Mh/s with the lastest commits. |
|
|
|
Another 0.02 although I don't use the miner yet ...  It's just a way to show appreciation of your work. Just done the same. If you would release win binaries I would appreciate that pallas. Keep up the good work, and thanks for your efforts. Let it rain donations guys. Thanks! The links on your signature have some wrong characters in them. Do you own the xtremehardware site? |
|
|
|
|
Someone donated 0.05 btc: thanks to whoever you are :-)
|
|
|
|
|
I've posted another update that fixes hashrate calculation, now it's much more stable.
|
|
|
|
Posted an hashrate increase, see the quoted thread. |
|
|
|
don't think, I'll be able to get my coins out.... the wallet shows zero as well as all the transaction it doesn't count...
did you have coins on a "long lost" wallet? I'm sure you did not leave many, didn't you? :-) |
|
|
|
|
Price is steady, I didn't expect that! :-D
BTW, pushed a bunch of optimisations.
Now the 1070 does 15.5 Mh/s from 15 which it did on the first version.
|
|
|
|
Nah kinda doubt SP has +50% available right now, just gloating about what he's already done.
Most of the kernals in this algorithm have already been optimized in the sp-mod private all nicehashalgos optimized. I just need to do some copy and pasting, and then I have an optimized M7 kernal. Yesterday I worked on the spreadcoin miner. M7 is not my priority right now.. look at what algos... they are all fast (except whirlpool and tiger, which I improved but lead to very little hashrate increase). the speed is limited by the muls: if you can optimize those, then you have some chance. |
|
|
|
pallas
And you can do it? you are welcome)
I tend to avoid windows when I can. I prefer to use my time working on cuda or opencl code. |
|
|
|
I could make the windows bin for a 0.1BTC donation. :-)
Someone will do it for free. |
|
|
|
Nice work, thanks. What was the error which lead to the buffer overflow ? github commit history doesn't really help  I know it's a mess because my editor fixed a ton of weird spacing ;-) There was a throughput coefficient initialised to -1 which broke the gpu threads which in turn had random hashrate which in turn didn't fit into a string buffer LOL! lol  so what the price of m7 and its name, actually I don't even remember the name of the coin Good to see you back :-) |
|
|
|
can you tell already if it is possible to push the hashrate for a 1070 to 30MH?
Unlikely, but I still have to look deeply into the algo. I have some ideas for optimizations but lacking the time to try them. |
|
|
|
|
Let's hope the exchanges will burn automatically and convert the open orders.
|
|
|
|
Nice work, thanks. What was the error which lead to the buffer overflow ? github commit history doesn't really help I know it's a mess because my editor fixed a ton of weird spacing ;-) There was a throughput coefficient initialised to -1 which broke the gpu threads which in turn had random hashrate which in turn didn't fit into a string buffer LOL! |
|
|
|
Here is my forked ccminer for M7 algo (Cryptonite coin), fixed to compile and mine on cuda 8 (required) and newer cards. It includes optimisations, and more I will add. Tested on Linux (no visual studio support, sorry) and 970, 980Ti, 1070, 1080 and 1080Ti. Here is the github repo: https://github.com/pallas1/ccminer-m7-branch |
|
|
|
|
Show Posts
