Quote from: qwizzie on February 09, 2019, 02:08:27 PM

Quote from: astraleureka on February 07, 2019, 10:47:07 AM

I heard a rumor a few weeks back (Jan 19th) regarding ASICs on the network - supposedly there was a batch of 5000 miners manufactured for a private Chinese farm that went online just recently. It seems to be pretty feasible after comparing against some analytics done on the nonce ranges of historic blocks:

https://github.com/noncesense-research-lab/nonce_distribution/blob/master/nonce_distribution.ipynb

Conveniently enough, this analysis concludes the miners came online around Jan 9th. Some chit-chat from IRC matches up as well:

Quote

<@solar> i got some confirmation about asics too
<@solar> there's at least 1 group with cnv2 asics, 65nm
<@solar> supposedly a batch of 5000

So after a thorough analysis of the winning nonces for blocks 1760000-1765000 I can say this about ASICs that are on the network now:
320 cores (10 chips x 32 cores?)
each core processes nonces with 2^22 step - from 0 to ~1.34e9
Single core speed is 400 h/s - calculated from the area of spikes (which come with 2^22 step) on winning nonce graph
Overall ASIC device speed = 128 kh/s (400*320)
Around 4000-4500 ASICs online now

M5M400 they do mine on public pools: f2pool, viabtc, poolin - they all submit blocks with winning nonces strictly in ASIC nonce ranges
all 3 are Chinese pools ironically
so we can just count them in "unknown hashrate" part
which leaves only 230 MH/s known
and 641 MH/s = almost exactly 5000 * 128 kh/s solar

At the time of this post, network hashrate is 850 MH/s (640 MH/s being 75% unknown), putting us squarely in the danger zone for network-level attacks. This is also a major drain on public pools - with only 230MH/s of "legitimate" miners popular pools are operating at a significant loss, having only 60-70 MH/s on average (e.g. SupportXMR at 0.6% is just below $1300/mo, while their server expenses are approximately $5000/mo). With the next hard fork still a few months out this is a completely unsustainable scenario and will eventually lead to the closure and consolidation of the public pools, not to mention a mass exodus of "legitimate" miners.

Another Hard Fork Coming for Monero [XMR]? Analysis says ASIC Dominates Monero Hashrate
https://coingape.com/asic-dominates-monero-hashrate/

Quote

the current network hashrate likely consists of 85.2% ASICs (5400 ASIC machines) and some die-hard GPU miners and botnets.

This is why hard forks purely for banning ASIC miners are dangerous in my opinion, they lower your network strength (making it vulnerable to 51% attacks) while ASIC manufacturers
will find ways to circumvent it. Best is to just allow ASIC miners and have them spread their hashrate evenly across pools (difficult to arrange, but not impossible).

Monero tries to be fair for all so ASICS (which are not easily obtainable for normal miners) are unfair advantage. So it is not like ASICS are bad but more like they are bad because so few can get them.

This situation is not new. It was the same year ago. As soon as it is known that algo will be changed, those ASICS will be sold. I just hope people are not getting fooled into buying them.