Please login or register.
This location

Improved NVIDIA mining software

Gingeropolous posted this 9 years ago

Gingeropolous | 9 years ago

Primary Milestones:

1. Overall software architecture (1000-1500 XMR): 2 weeks

A) Open source software that is nice, clean, well written and documented code that will be easy to maintain and work on; a pleasure for others to jump in and contribute going forward.

B) Cross-platform compatible. e.g. works on apple, linux, microsoft. Binaries available for all platforms when code is released.

C) Parameters can be completely tuned. The software should be designed so that miners can change every value with a config file to tweak settings for their particular hardware. This will possible extend the useful life of the software.

D) Limited dependencies for compiling, and required dependencies must be available via package managers. 100 xmr will be deducted for each dependency required, max 500.

E) Works on as much available hardware as possible. All Maxwell architecture is a must.

F) Works as a solo miner - is able to communicate directly with the monero wallet software. EDITED - Solo mining requires working with the daemon. Sorry for the confusion.

2) Mining performance (up to 3500 XMR) 1 year (open ended)

A) Hash rates reach the price point equivalent of Claymore's AMD GPU miner. Currently the highest reported hashrates are 745 h/s for the Sapphire R9 290X, which at the time of writing is a $350 GPU. The NVIDIA pricepoint equivalent is the GTX 970. Existing hashrate of a GTX 970 is ~400 h/s.

So, there was an initial stall due to djm34 not understanding the milestones, so we talked it out. Review the conversation for yourself to determine if you think we are ready to move to the funding stage.

We modified the hashrate funding curve, such that its linear to the first +100 h/s with a payout of 750 xmr, so the calculation is

hashrate / 0.1333333 = payout

then it goes to

hashrate / 0.18 = payout

so,

A) +100 h/s (compared to existing 970 gtx) = 750 xmr
B) +200 h/s = 2025
C) + 300 h/s = 3037
D) + 346 h/s = AMD 290x equivalence = 3500 xmr

These milestones are incremental, so if 2A is met, then payout 750. Then if 2B is met, and 2A has been paid out, its 2025 - 750, etc.

With djm34's permission, I am copying our conversation for your review.

LINK TO PASTEBIN

So please, respond to express whether you think this is ready for moving to funding.

EDITED TO ADD:

FYI - This a non refundable system. If, for whatever reason, things don't work out, the funds are kept for future developers to receive.

Replies: 27
Gingeropolous posted 9 years ago Weight: -249 | Link [ + ]

Okay, for points 1B,C, and D, the conversation with djm34 indicates the following:

B: djm34 does not have any macs to test the software on, therefore, he really can't develop for macs. If someone can come up with a solution, maybe macs can still be worked on?

C: per the conversation, he doesn't think there will be many more parameters that can be tuned, so beyond what the existing software does (threads and blocks), he doesn't see that there could be more tunable parameters.

D: dependencies - he agrees that some things can be included as libraries or whatever, but the whole cuda development kit is still necessary and its not open source, so the best we might be able to do is include the file in the dload package and make a script for the install.