Does that look vaguely accurate for a calculation?

Someone in another project is running some cross-algo numbers and came up with that for Monero with a little help

seems reasonable

While others prattled on about how ASIC-resistance is futile (but failed to produce any clear reason/evidence)

Yeah it seemed within scope

It's using a 3900X as a baseline

I wouldn't say without clear reason

there are arguments that were made on both sides

I'm not saying there isn't truth to the statement

Merely that they couldn't/wouldn't produce any clear backing for their statements

I think simply saying that if/when randomx fails then it will be tweaked or swapped for sha3 inherently shows that this is kind of a more centralized and temporary approach

I think that's pretty clear, and using facts that both sides probably agree about

IMO switching to SHA3 would be the most decentralized and permanent approach if RandomX fails

IDK how that would be centralized

More attempts at ASIC-resistance would be centralizing, IMO, as the previous were

That's why I'm also in the camp of "if RandomX fails move to SHA3/similar ASIC-friendly algo"

I just think that with the way computer hardware and manufacturing have been advancing for decades, it is going to need to be tweaked or swapped

so I think the simpler solution would have been to just go directly to sha3

Yeah I guess that falls back to "when has RandomX failed" issue

but...

.shrug

Because I guess small tweaks could be fine, but would easily fall back into scheduled changes...

right

Unless it was just an optimization/improvement

But thats a fine line lol

having a handful or less of people fiddling with proof-of-work in never ending game of cat-and-mouse seems less than fully decentralized

Yup

Absolutely agree

i dunno. If the situation arises where a heretofore unknown entity can create what is effectively a general purpose CPU that performs more efficiently than the existing competition, AND the mining of monero has a greater ROI than selling these hyper efficient CPUs to the world, then .....

i just think the playing field at that point is something that no one has imagined yet

I don't think that necessarily is the case gingeropolous

maybe it's just in a year or three when current gen fabs are not cutting edge anymore someone trims some fat and adds a bunch of cache and they can get 50% or 250% more efficiency

then core team deliberates and makes announcement on whether or not changes will be forthcoming via their scribe rehrar and it just gives the appearance of centralization

so how much is it to license amd or intel stuff to make your own fat trimmed thingy?

good q

does arm still suck butthole?

for randomx?

yes

i think its just a matter of cache

I don't think you can really license x86 stuff if you want

i mean, in one sense, randomx may have done to cpu manufacturing what PoW in general has done to energy

I think intel and amd like their duopoly

i.e., in energy, its now possible to store excess energy in the form of a cryptocurrency

because now there's always a demand for energy - bitcoin mining

same with cpu manufacturing. at some point, it may be realized that you can always churn out CPUs and call em Monero Miners if cpu demand in this or that sector is low

and then when its high, you just sell less "monero miners"

if monero mining becomes that profitable

so now there's a way to store excess CPU manufacturing ability in the form of cryptocurrency

if the tubes are pointed in the right directions

if monero marketcap increases 100x or so

if monero mining becomes that profitable <<>> but this again comes down to energy cost just like any other POW

but for now we're effectively protected by AMD and intel

except my cpu can do other things besides mining

it doesn't become a door stop

which is a helluva .... utilization of resources that have been re-purposed for goals not originally designed for

i mean, i guess in one sense, bitcoin asics was a similar repurposing.... of an unprecedented era in fabrication

i just don't think they had the knobs turned quite right. hopefully we do

thats what i don't get with the whole thing. like, we saw what happens with the ASIC path. bitcoin. and anyone thats come in the wake of that has been even worse for wear.

honestly I don't think it's looking that bad for bitcoin

there's five or six companies making asics, several of which are now publicly traded

there's operations opening around the world at pretty large scale

.shrug

yeah i still don't get that. still not gettin one at bestbuy anytime soon

I guess I wouldn't buy a computer/monero miner there either

yeah, u woulnd't. but your neighbor 2 doors down that read about the bitcorns and then sees a bitcorn miner at his local costco might buy one

geet ya bitcorn err!

le sigh

dayum

250 MH/s

That chart is nice, cost $33 per XMR ,and XMR is worth $60-ish, so 2:1 profit

those 5-6 ASIC mfrs in bitcoin land are all going to consolidate, they're not all viable on their own

there are other x86 companies. VIA is one. I think there's another licensee in China, can't recall the name

and ARM is actually quite good at the computational part of RandomX. it just lacks in memory perf and cache, as ginger already noted

I believe if you had an ARM-based chip with Ryzen-level memory controller and cache subsystem, it would slightly beat Ryzen efficiency

as for SHA3, ASIC-friendliness - I think the track record is pretty clear, the cat'n'mouse game continues except more of it occurs in secret, which is overall detrimental to the ecosysstem

all you ASIC-friendly advocates are basically shitheads who haven't learned from history

the CPU game is no less competitive but more of it occurs in public. Intel and AMD present at conferences like ISSCC and brag about their latest designs

ASIC designers generally keep trade secrets forever

Intel bragging: "Our innovative new speculative execution system..."

all this pro-ASIC stuff is antithetical to the open source nature of projects like Monero

it makes a mockery of the principle of permissionless access to the technology

...

I dunno, I buy the idea that if it ever commoditized, it'd be a good thing.

But that's long term.

then that should be one of the conditions before pivoting to ASIC-friendliness. see how the Bitcoin ecosystem shakes out

it will inevitably centralize back down to only 1 or 2 mfrs

the only question will be how much time they spend self-mining before they release limited quantities of their miner to the public

Maybe that 250 MH/s miner is AMD "testing" Zen 3?

I guess it's not impossible

RavenCoin's fix: RavenProject/Ravencoin-critical-fix 3bba0a9#diff-ca81084f62961a188f5c1e86a5ff1d7cR311

Has nothing to do with PoW though, just transaction checking

raven reports burning 2.8M. out of 300M counterfeited coins. I guess it's a start

291 MH/s

its not ZEN3

its a space station!

rome can go 45kh

single one,

but thats still like 6500 nodes

that would be around 50mln USD in hardware to propel such hashrate

samples for Milan are out videocardz.com/newz/amd-zen3-based-epyc-milan-processor-spotted