Guess I should have waited for about an hour before posting the last article, but that’s how it always is – only after you give up hope do you finally find your breakthrough :-).
Anyway; quick update: Good news – I got code correctness for the upcoming v4r change. It’s still “somewhat less performant than I’d like it to be” (just by an order of magnitude or so, so …. meh…), but from here on out it’s no longer the extreme pain of trying to find bugs in seemingly random code sequences – now it’s only an optimization problem. Really happy – done for today, time for a beer.
Happy mining, guys.
The lukMiner blog 
Good job. Do we still have any competitive advantage over regular CPU/GPU with this algo?
LikeLike
We’ll see – I haven’t seen any performance numbers for GPUs, yet, that’ll be interesting. GPUs should be slightly better at the random math part of v4r (in particular it’s easier to compile per block because the driver has a built-in compiler, anyway), but most of the memory pattern is closer to what the original xmr was, so we should be reasonably good. We’ll see.
LikeLike
F Yes! Thanks !
LikeLike
Great news Luk!
LikeLike
You’re a trooper, man.
How bout that donation address?
LikeLike
Please put one up, Luk. You deserve it.
LikeLike
LOL. I think I did once put one up, but then decided to take it down. After all, I do get a miner share from people that run that code, so it kind of “felt” wrong to also ask for donations.
IMHO donations should be for those that freely donate their skills and time for a greater good (e.g., somebody should _really_ donate to that killallasics guy for running that test net!) … and though you could argue that I’m probably spending more time than what it pays (in particular at current prices :-/) it still felt wrong.
That said – if somebody absoltely _does_ want to donate: 45Pw3bWFJXQiR1hN97huM6BbNPpnYdPytWnTUbkEm6KS9MExU7Gtr3nBsVoZA746qhCEwqVsFuPdwfXVtZwXxHQ6LDyfBaL
🙂
LikeLike