What's wrong with Monero?

[Tigel]

FPGA switching from Raven to Monero?

[astraleureka]

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:

<@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

<sech1> 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:
<sech1> 320 cores (10 chips x 32 cores?)
<sech1> each core processes nonces with 2^22 step - from 0 to ~1.34e9
<sech1> Single core speed is 400 h/s - calculated from the area of spikes (which come with 2^22 step) on winning nonce graph
<sech1> Overall ASIC device speed = 128 kh/s (400*320)
<sech1> Around 4000-4500 ASICs online now

<sech1> M5M400 they do mine on public pools: f2pool, viabtc, poolin - they all submit blocks with winning nonces strictly in ASIC nonce ranges
<sech1> all 3 are Chinese pools ironically
<sech1> so we can just count them in "unknown hashrate" part
<sech1> which leaves only 230 MH/s known
<sech1> 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.

[Marvell2]

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:

<@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

<sech1> 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:
<sech1> 320 cores (10 chips x 32 cores?)
<sech1> each core processes nonces with 2^22 step - from 0 to ~1.34e9
<sech1> Single core speed is 400 h/s - calculated from the area of spikes (which come with 2^22 step) on winning nonce graph
<sech1> Overall ASIC device speed = 128 kh/s (400*320)
<sech1> Around 4000-4500 ASICs online now

<sech1> M5M400 they do mine on public pools: f2pool, viabtc, poolin - they all submit blocks with winning nonces strictly in ASIC nonce ranges
<sech1> all 3 are Chinese pools ironically
<sech1> so we can just count them in "unknown hashrate" part
<sech1> which leaves only 230 MH/s known
<sech1> 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.

5000 a month expenses thats where the problem is .  They should not spend more than 1k max

[astraleureka]

It's not feasible, they need to have servers to handle nonce validation.