Add support for latest ZTEX firmware.#15
Add support for latest ZTEX firmware.#15dpifke wants to merge 3 commits intoTheSeven:testingfrom dpifke:newer-ztex-firmware
Conversation
This adds support for a newer version of the ZTEX firmware, including the version supported on the 1.15y quad miner.
In a future commit, I will be adding support for mining using all of the FPGAs on a ZTEX 1.15y. For now, we simply detect the presence of multiple FPGAs and log a message indicating that mining will only take place on the currently-selected chip.
With the previous value of 100ms, I was seeing the occasional communication timeout talking to my ZTEX boards. Looking at the ZTEX SDK, it uses a value of 1000ms. This commit changes MPBM to use the same value.
ghost
assigned
|
I'm not sure what to do about this. |
|
It was written to match the behavior of the current BTCMiner.java code. Without these changes I couldn't mine at all using my MPBM and my 1.15y. Since applying the changes, I've been able to do so successfully, albeit using only the first FPGA. I believe the intent of the extra nonce stuff is to support bitstreams which have more than one hasher working in parallel. Rather than pipeline everything so you get 1 hash per clock, it's possible to design code that performs X hashes per Y clocks. I'm not aware of any publicly-released bitstreams that do this, but I've played with this technique and it makes the place-and-route a lot easier. Agree that the multiple FPGA stuff should be straightforward; my intent was to push one piece at a time for ease of testing and merging. |
2 participants
After applying these changes, I can succesfully mine on both my 1.15x
and my 1.15y boards. The latter currently only uses a single FPGA;
I'm currently working on a (much more invasive) patch to use all four
chips, to be applied on top of these (fairly safe) changes.