X11 algorithm: the Dash network, ASIC miners and specs

X11 is a proof of work algorithm in which data passes through eleven different cryptographic hash functions one after another. The output of the first becomes the input of the second, the output of the second the input of the third, and so on down to the eleventh. The scheme was devised by Evan Duffield in 2014 for a network that was then called Darkcoin and is known today as Dash.

None of the eleven functions were picked at random. They are all entrants in the NIST competition for a new SHA-3 standard: BLAKE, BMW, Grøstl, Skein, JH, Keccak, Luffa, CubeHash, SHAvite-3, SIMD and ECHO. Five of them reached the final, and Keccak won the competition and became SHA-3. The point of the chain was insurance: if a weakness is ever found in one function, the other ten still cover the result.

The asic.es catalog holds 6 X11 models from four manufacturers. The efficiency spread here is a little over twenty three times: the Antminer D9 needs 1.6 joules per gigahash, the iBeLink DM22G needs 36.82. The unit is its own, J/GH, because hashrate on X11 machines is measured in gigahashes, not in terahashes.

The job is the same as everywhere else: find a nonce that makes the block header hash land below the current target. Only the length of the path is different. In Bitcoin the header goes through SHA-256 twice, in X11 it goes through eleven different functions in a row.

The order is fixed: BLAKE, then BMW, Grøstl, Skein, JH, Keccak, Luffa, CubeHash, SHAvite-3, SIMD and ECHO at the end. Each one works in its 512 bit variant and takes as input the 64 bytes the previous one produced. The first 32 bytes of the final result are taken, and that is the block hash.

For silicon this scheme means the following: eleven completely different compute blocks have to sit on the chip, and all of them work in sequence. Development comes out more expensive and slower than for SHA-256, where there is one circuit and you can repeat it a thousand times. That is exactly why X11 ASICs arrived later than Bitcoin ones, and why there were never many manufacturers in this market.

In the graphics card era the chain had a side effect that people valued at the time: cards on X11 ran noticeably cooler than on Scrypt and drew less power. Those advantages did not last long. The first X11 ASIC, the iBeLink DM384M, was announced in February 2016, and that was the end of the graphics card era.

1. Block headerVersion, previous block hash, Merkle root, time, target, nonce
2. First four functionsBLAKE, BMW, Grøstl, Skein, each in its 512 bit variant
3. Next fourJH, Keccak, Luffa, CubeHash, the input is the output of the previous one
4. Last threeSHAvite-3, SIMD, ECHO, the result is truncated to 32 bytes
5. Comparison with the targetBelow the target, the block is found. Above it, a new nonce and start again

• 2014On 18 January the network launches under the name XCoin. Evan Duffield presents X11 as an algorithm meant to hold out longer on graphics cards and heat them less.
• 2014The network is renamed Darkcoin. Masternodes appear with a collateral of 1000 coins, along with the first privacy services.
• 2015In March comes the second and last rename, Darkcoin becomes Dash, from digital cash.
• 2016In February the iBeLink DM384M arrives, the first production X11 ASIC: 384 MH/s at 715 watts. The graphics card era on the algorithm ends.
• 2017Across this year and 2018 Bitmain ships the Antminer D3 and D5, Innosilicon answers with the A5 line. Network hashrate grows by orders of magnitude.
• 2021October 2021 brings the Antminer D7 at 1286 GH/s, and in 2022 the D9 arrives at 1770 GH/s. These are the last new machines for the algorithm; no new chips came after them.
• 2023In December comes hard fork version 20: the block reward shares change to 60 percent for masternodes, 20 for miners and 20 for the treasury. Before that it was 45, 45 and 10.
• 2024From July onwards Dash Platform rolls out, part of the masternode reward goes into the platform credit pool. In the summer of 2026 private balances on the Orchard technology from Zcash were added to the platform.

All 6 catalog models on X11, sorted by efficiency with the best on top. Hashrate in GH/s, efficiency in joules per gigahash. Click a name to open the profitability calculator for that model.

There is no custom firmware for X11. AsicBoost firmware is built for SHA-256 boards, on Scrypt two models are supported, and the Antminer D series is not on that list. Network installation through HashCore Toolkit belongs to the same supported lines and is not meant for Dash machines.

Third party solutions do not exist either. Not one of the known alternative firmware projects for Antminer claims support for the D7 and D9: their lists hold only SHA-256 and two Scrypt models. The reason is plain and economic. The whole X11 network today is on the order of two and a bit petahashes, the equivalent of a little over a thousand D9 machines. Nobody is going to develop separate firmware for a fleet that size.

Search results do turn up pages titled something like "firmware for Antminer D9". Check them and you find those shops carry no D9 in the catalog at all, so these are pages built for a search query, not for a product. Files of unknown origin do not belong on a machine worth several thousand euros: the risk of a brick and a lost warranty is real enough.

The conclusion is the same as on other algorithms without firmware: efficiency in J/GH is fixed at the moment of purchase and does not improve afterwards. It has to be worked out before you buy.

Income on X11 is set by Dash network difficulty, the DASH exchange rate and your efficiency in J/GH. But there is an important detail people often skip: the miner does not get the whole block reward.

Since the hard fork of December 2023 the reward is split like this: 60 percent to masternodes, 20 percent to miners, 20 percent to the network treasury, which funds community proposals. Before that the scheme was 45, 45 and 10. Plenty of calculators and reference sites still use the old shares and overstate miner income by more than double. Check which scheme your source is using.

Dash issuance is not built like Bitcoin either. Instead of a sharp halving once every four years, the reward drops by 7.14 percent every 210,240 blocks, roughly once every 383 days. The curve comes out smooth and predictable, with no one day step, but the miner also gets no breather in the form of a four year plateau.

Open the profitability calculator →

Efficiency of X11 machines is measured in joules per gigahash, J/GH, because hashrate here is measured in gigahashes. This is not a typo and not a different scale of the same thing: J/GH cannot be compared with the J/TH from the SHA-256 or Eaglesong pages under any conversion, because the work behind one hash differs between algorithms. The comparison only makes sense within X11.

Inside the algorithm the spread is wide. The Antminer D9 needs 1.6 J/GH, the D7 needs 2.45, the FusionSilicon X7 needs 5.42, while the Innosilicon A5 and iBeLink DM22G go past 23 and 36 respectively. Between the leader and the bottom of the list the gap is twenty three times. At the same output in DASH, the electricity bill of the owners of these machines will differ by those same twenty three times.

The arithmetic is the usual one: multiply efficiency by hashrate to get draw in watts, divide by 1000 to get kilowatts, multiply by your tariff and by 720 hours a month. Compare the result with the mining forecast in the calculator. And keep the reward shares in mind: the miner gets 20 percent, not a hundred.

Check which scheme the calculator uses. The reward shares changed in December 2023. A source that still calculates with the old 45 percent to the miner instead of the current 20 will overstate your income by more than double. That is the most expensive mistake on this algorithm.

Look at J/GH, not at gigahashes. Hashrate tells you about speed, efficiency about the power bill. Between the D9 and the A5 the efficiency gap is fourteen times, and no difference in purchase price covers that.

Accept that no new models are coming. The last X11 machine came out in 2022. There is nothing to plan a fleet refresh around two years from now, so calculate payback on the hardware you are buying today.

Plan a 3 kW supply. The D7 and D9 need a dedicated breaker, decent cable and exhaust ventilation. Noise is around 75 decibels, so these machines do not belong in a living space.

Do not count on firmware. There is nothing to improve on X11. The factory numbers are your numbers for the whole time you own the machine.

Calculate at current difficulty. Take today's difficulty, today's rate and your real tariff. Building a rising exchange rate into the plan is risky: difficulty grows far more reliably.

Counting income on the full block reward. The miner gets 20 percent, masternodes 60, the treasury 20. A calculation on the full subsidy gives a number that has nothing to do with your wallet.

Taking figures from outdated reference sites. Some large services still publish the 45, 45 and 10 scheme that was cancelled in December 2023, along with masternode counts inflated by roughly 50 percent. Check the publication date of your source.

Comparing J/GH with J/TH. These are different units for different work. Converting one into the other by multiplying by a thousand is pointless; you get a number with no physical meaning.

Buying an A5 or a DM22G for the low price. They are cheap because they spend fourteen and twenty three times more energy on the same work. The saving on the purchase burns off in the first months.

Counting on a second coin. Dash has no merge mining. Auto switching on multi algorithm pools is selling hashrate, not mining a second coin.

Ignoring the regulatory background. Dash is treated as a coin with privacy features, and European rules for such assets are tightening. This is not a ban on mining, but it can affect your plans for cashing out through EU exchanges.