hyc: i never doubted it was illicit. I just couldn't prove it. So I infiltrated, gathered info and relayed it to azure :)

0|proxy | 2020-06-09 17:04 +00:00: The proxy currently has 38385 miners connected at 201405703 h/s with an average diff of 78322

M5M400 you got into the proxy? How?

nice pick of the framework though

sech1: I asked for a root login. what else?

How convenient :D

well he couldn't figure out how to install xnp himself so I offered

then gathered info and wiped it. he later built another one himself it seems

because it came back few hours later

what a dolt

glad you took care of the problem ;)

there is a meeting about merge mining PoW in #tari

and?

the point of merge mining is any chain can merge mine with a parent chain without requiring any action on the parent chain's part

it's an interesting discussion

could a RandomX-type algorithm be created for GPUs? is that possible?

no

Yes. Though probably specific to one particular GPU architecture.

anything that a highly parallel GPU can do well, an FPGA or ASIC can do better

would that be randomx like then=

?*

Fair point. I interpreted "RandomX-type" as "generate random instruction streams and hash the end state".

But maybe that wasn't the intent :)

Might be harder since you need to generate a state at the end, I'm not sure how much control you get over that.

hmmm. well you may be right - you could probably tailor a randomx-like approach to a particular GPU family

but AFAICS, "GPGPU" are not really very general purpose

and generality is why RandomX works

I'm not saying an ASIC could not get ahead, it might well be able to easier than for CPU. Just answering the question as is.

yeah. there's another annoyance in that GPU instruction sets aren't publicly documented, and change dramatically from generation to generation

so the JIT approach that makes RandomX fast on CPUs is much harder to develop for GPUs

yeah I suppose "GPU" is too broad a target

talks a bit on why CPU was chosen

Oh yeah I understand why CPUs are a better option, I was just wondering whether something like RandomX could be created for other handware

RISC-V could be a thing eventually

RISC-V is just another CPU family, it will already work

it's already supported by gcc; writing the JIT for it wouldn't be hard

Isn't ProgPow "generate random instruction streams and hash the end state"?

not very large or varied instruction streams

leaves vast percentage of GPU silicon unused