Even FPGA would be expensive for CryptoNight because most standard FPGA have close to zero memory. Your second point will, therefore, not only decrease the relative advantage of ASICs or FPGA over GPUs and CPUs but greatly increase their cost.
Even a basic FPGA mining board would be an expensive custom job because of this memory requirement.
I would ...
In terms of power efficiency, the reigning champ was the Nvidia 750Ti. That GPU will get you 250H/s and use about 30-40W power. For efficiency, that puts it around 6H/s/W.
With the optimized ARM64 code I've been working on, a small ARM64 box (like a high end TV box) can get you 22H/s using less than 5W power. Much slower mining, but much higher efficiency, ...
There are no ASICS yet, and GPUs are about 2-4x faster than CPUs. Monero's PoW algorithm has a large memory requirement that would make it difficult to use an ASIC and less advantageous to use a GPU over a CPU than in Bitcoin.
From the Monero Missives:
We're also quite happy to announce the addition of a new feature to
Monero: Smart Mining. This is a ...
Right now Wolf's OpenCL XMR Miner for AMD GPUs (open source) and Claymore CryptoNote GPU Miner (proprietary) are probably the most efficient.
It is believed that Monero is still quite far away from ASIC mining largely because of the memory requirement and that CryptoNight takes advantage of the AES-NI instruction set on modern PCs. The relative advantage ...
People visiting this question in the future will find the answers provided are out of date. This website lists user-submitted hardware benchmarks that include Monero mining hashrate, power consumption, and configuration settings. The creator told me they were adding features to make it easier to search and sort the data.
EDIT: the original bytehost5 URL ...
Currently network hashrates are near historically high levels. How much further hash rate moves up in the future depends heavily on mining profitability based on current XMR prices and coinbase reward.
Similarly the following factors will have a large impact on the attractiveness of future FPGA and ASIC development for Monero.
The emission ...
In versions 1 to 6 of the protocol, the CryptoNight algorithm was very roughly:
state = keccak(block_data)
scratchpad = fill_scratchpad(state)
loop 524,288 times
address = compute_address(scratchpad, state)
address = compute_address(scratchpad, state)
text = reduce(scratchpad, ...
Currently the rx400 series is the most efficient, powerful, and cost-effective mining equipment. I have gotten an rx470 to 700 h/s using about 80-90 watts. It required many hours of bios modification, but this is all doable with readily available software.
Monero could be mined by ASIC. The Bitmain Antminer X3 is U$1255, ships Aug. 21-31 2018, and weighs 7 kg, it does 220000 H/s using 465 W +7% (2.27 J/kH +7%) - but they clearly mention:
There are financial risks associated with mining cryptocurrencies. These risks can be related to changes in exchange rate of the cryptocurrency or to changes in the algorithm ...
Centralization is a continuum. Actions and circumstances move a particular system along that continuum.
There are several points which are topical here:
Monero development is made by a smallish number of people (more centralized)
Anybody with skills is free to join Monero development (less centralized)
A single party being able to dictate what gets mined ...
This depends on whether you refer to "performance per watt" or "performance per purchase price".
Raw hash rates for different CPUs and GPUs can be found, for example, at http://monerobenchmarks.info/index.php
In terms of performance per watt (which is easiest to measure/calculate):
a Xeon Phi 7210 gets 2770H/s @ 225W, that's 12.3 H/J
a Dual AMD Epyc gets ...
There may have been others, but at least Baikalminer[archive] and Bitmain[archive] launched Monero ASIC miners in 2018.
It's worth noting Bitmain allowed people to buy the ASICs, but didn't actually ship them until after the Monero hardfork - meaning only Bitmain themselves were able to mine XMR on their own ASICs, and knowingly scammed their customers ...
What are the main differences between ProgPOW and RandomX?
The design goal of ProgPoW is to have the algorithm’s requirements match what is available on commodity GPUs, while RandomX is inefficient (as stated on https://github.com/tevador/RandomX) on GPUs.
can some parts of it also benefit Monero's development in a longer-term ASIC resistant mining ...
This Google docs spreadsheet has a nice list of GPU/CPU hash rate reports to help you make up your mind!
Here are top-10 CPUs:
CPU ~H/s Threads TDP, W Miner app / version Settings / command line
AMD Opteron 32x6276 15000.00 400+ 115 wolf -a cryptonight -t25 (x16)
Intel Xeon E7-8880v3 1805.00 27 150 wolf's latest cpuminer
AMD Opteron 6272 ...
The most efficient hardware for mining Monero, at this time, are AMD GPU cards. High hashrate power and low power consumption is the perfect combination.
Truly, the more cost-efficient hardware for Monero mining is the AMD Radeon Vega card. Both 56 and 64 VEGA cards are notably profitable in mining Monero.
Radeon VEGA 56 can mine with stock BIOS around ...
The idea behind this draft is more like an extension to PoW changes like currently to prevent ASICs overtaking a serious amount of hash rate.
However, it's proposing to time lock all mining rewards during a full period of time where one algorithm is active, between two PoW hard forks and release them only after another challenge period which might be 1-3 ...
To evaluate whether one alternative is more secure than the other we have to answer the question: Which alternative makes an attack more costly?
In the Cryptonight-only case the cost of an attack is that of gaining a 51% advantage over all mining nodes on the network.
In the Cryptonight+ASIC case you'd need to get some combination of regular miners and ...
The main changes are:
reading a whole 64 byte chache line at a time: many recent processors will read 64 bytes at a time when fetching from RAM, so access to the whole set is virtually free after it's cached. Since ASICs don't have to behave like this, this will increase the memory bandwidth an ASIC needs by 4, since Cryptonight acts on 16 bytes at a time. ...
The difficulty is a numerical representation of how long on average it
will take to create a valid block that satisfies the Proof-of-Work
algorithm being used, it does not represent minutes or time per se.
but instead is an integer number that makes the PoW calculation take
longer to perform, this is adjusted by the nodes and miners so that on
Cryptonight really only exploited one particular feature of modern CPUs, fast cache access. When it was designed, it was thought ASICs would be at a significant disadvantage because of their general lack (at the time) of access to fast memory. It was also not imagined the ways in which ASIC manufacturers could overcome the memory constraints.
The fork was moved to today (April 6, 2018), so it is hard to tell. Early indications based on block-times are about 80% of the hash-rate is missing. Some of that is non-ASIC miners failing to update on time. We should know more in a week or so.
Based on above comments ... No. CryptoNight is the exact opposite of what "hardware-agnostic" means. In particular, CryptoNight is designed to run best on CPUs, and run less efficiently on GPUs and other accelerated compute devices.
An algorithm that is truly hardware-agnostic would run efficiently on any compute device.
Do we know who exactly created these asics during the big hash run up
during 2019-Jan to 2019-Mar?
It's hard to say, but the names are the usual known.
And why would they create these asics, when they know for a fact that
the monero community would easily fork the algo to brick all of their
asics. It just doesn't make sense why they would spend so ...
No hard evidence I'm aware of. The changes this time were made to thwart any ASICs that may have been under development since the last fork. By changing the algorithm every 6 months, this severely impacts the time to develop and ship any new ASICs to market, making it an unprofitable venture. Until we come up with an alternative algorithm (such as the ...
I wouldn't say it's "almost finished". There is a lot of discussion on this particular PR here. On reading, it looks like there is a lot to iron out still on this particular proposal if it were to ever get merged.
There are alternative PoW changes being worked on also. Any merged change would ensure CPU/GPU mining would have negligible performance changes ...
As of writing this answer (2018-04-07 03:00 UTC), ~18.5 hours after network upgrade, there are 65 blocks mined (latest is 1546065). It takes on average 17 minutes to mine a block, which is 8.5 times the expected 2 minutes.
The difficulty is calculated based on past 720 blocks. However not all blocks are used.
DiffCut: the number of highest and lowest ...
The network hash rate was around 1123 MH/s at the time of the fork. The difficulty hasn't changed yet, so blocks are coming in much slower(lower hash rate at same "asic difficulty"), and while it will slowly decrease with each new block, it will take 720 blocks before the true adjustment takes place(a few more days at minimum). As the difficulty is a ...
Say 50% is cryptonight and 50% is an asic friendly pow
That scenario would bring mining centralization very quickly because once an ASIC friendly PoW is introduced only miners using ASICs can effectively compete. This is true whether you are only using 1 PoW algorithm or several.
A better solution is to stick with a tested and working algorithms (...