Comprehensive ASIC Miner Comparison Table

[alpinemo]

Hi Everyone.

I have attempted to put together a comprehensive comparison chart of all ASIC miners that are available on the market today (as of January, 2018). Some things to keep in mind:

   1) The chart is sorted by algorithm, then by miner efficiency
   2) I have calculated the efficiency a bit differently than all the vendors, choosing instead to divide the hashpower by the electrical power instead of the other way around. This way a larger value means a more efficient miner
   3) This chart includes all the devices I was personally able to identify. If there are other miners you think should be there but aren't please let me know
   4) I have this chart as an excel sheet but was not sure how to best imbed an excel sheet, so I am sharing it as an image for now. If you all find this useful I may transfer it to a google spreadsheet document.

Comments/suggestions are welcome.

I couldn't make the image show in the post, so here's the link.

https://ibb.co/giTCU6

[Beaucoupnice]

Lot of work done there sir. Thank you so very much. It's very helpful. 

I'm experiencing the same confusion whenever I look up similar (but smaller) tables online - in that there is a mixture of differing rates.

In your table, I can see GH/J & MH/J in the efficiency one column which makes just looking down the efficiency column confusing.

In the Hashrate column, there is GH/s & MH/s & TH/s - again a little confusing.

You have organised the column by vender. Personally, I would have organised it by the best efficiency, so viewers can immediately see the best at the top.

A column that calculates Hash per $ might be useful as well. To see what gets the most bang for the buck.

Maybe the Efficiency calculation can be taken a step further and build in the unit cost to give a single figure comparison?

Thanks again.

[alpinemo]

Thanks for your comments. Allow me to address what I can.

I'm experiencing the same confusion whenever I look up similar (but smaller) tables online - in that there is a mixture of differing rates.

Indeed there are different figures for the devices. Sometimes the rates are different when the device is operated under different conditions, etc. I have tried to go with the most common reported rate.

In your table, I can see GH/J & MH/J in the efficiency one column which makes just looking down the efficiency column confusing.

In the Hashrate column, there is GH/s & MH/s & TH/s - again a little confusing.

I guess my intention was not to compare (each miner) to (all the other miners), but to compare each miner (for a certain algorithm) to all miners that (work with that algorithm). In this case, the best way to read the chart is to actually go to the algorithm you are interested in miners for, then compare efficiency specs for the miners under that algorithm. That's why I also left the hashrate units as they are typically reported, because it is not useful to compare the hashrate of say the antminer L3+ to that of antminer S9, since the former is for litecoin (scrypt algorithm) while the latter is for Bitcoin (SHA-256 algorithm), etc....

You have organised the column by vender. Personally, I would have organised it by the best efficiency, so viewers can immediately see the best at the top.

I did address this in the previous point. Efficiencies were meant to be compared only within the same algorithm, not across all algorithms/devices.

A column that calculates Hash per $ might be useful as well. To see what gets the most bang for the buck.

Maybe the Efficiency calculation can be taken a step further and build in the unit cost to give a single figure comparison?

I think this is a good point and I've been trying to implement it, but the problem again is that it would only be relevant within each algorithm, not across algorithms. In order to compare profit per hash across all algorithms, it needs to turn into a full ROI calculator which would need live feeds of current prices/difficulties/etc.....

[QuintLeo]

As a specific recent example, many of the early S9 batches had slightly different configurations in the software that resulted in different hashrates and different power draws, due to variations in the chips themselves.
The more recent versions have different software that does some "auto-adapt" to try to maximise the hashrate, but results in even wider variation WITHIN the batch on hashrate and power draw.

 Some older miners (especially the Spondoolies models) were configurable, able to tailor power draw and resulting hashrate quite a bit.


 Hash/$ is going to be a moving target, as miner pricing shifts in response to market changes.

[locolive]

I was looking for a comparison table like this one, thank you.

Soon you can add DragonMint https://halongmining.com

[CryptoWatcher420]

Hi Everyone.

I have attempted to put together a comprehensive comparison chart of all ASIC miners that are available on the market today (as of January, 2018). Some things to keep in mind:

   1) The chart is sorted by algorithm, then by miner efficiency
   2) I have calculated the efficiency a bit differently than all the vendors, choosing instead to divide the hashpower by the electrical power instead of the other way around. This way a larger value means a more efficient miner
   3) This chart includes all the devices I was personally able to identify. If there are other miners you think should be there but aren't please let me know
   4) I have this chart as an excel sheet but was not sure how to best imbed an excel sheet, so I am sharing it as an image for now. If you all find this useful I may transfer it to a google spreadsheet document.

Comments/suggestions are welcome.

I couldn't make the image show in the post, so here's the link.

https://ibb.co/giTCU6

lol the price on the Baikal giant b isn't correct, from Baikal there 3,600 dollars usd before shipping

[Beaucoupnice]

Hi Everyone.

I have attempted to put together a comprehensive comparison chart of all ASIC miners that are available on the market today (as of January, 2018). Some things to keep in mind:

   1) The chart is sorted by algorithm, then by miner efficiency
   2) I have calculated the efficiency a bit differently than all the vendors, choosing instead to divide the hashpower by the electrical power instead of the other way around. This way a larger value means a more efficient miner
   3) This chart includes all the devices I was personally able to identify. If there are other miners you think should be there but aren't please let me know
   4) I have this chart as an excel sheet but was not sure how to best imbed an excel sheet, so I am sharing it as an image for now. If you all find this useful I may transfer it to a google spreadsheet document.

Comments/suggestions are welcome.

I couldn't make the image show in the post, so here's the link.

https://ibb.co/giTCU6

lol the price on the Baikal giant b isn't correct, from Baikal there 3,600 dollars usd before shipping

True but minimum order is 6 making initial investment over $20K. I know it just adds confusion but I guess the price in the table reflects the cost of an individual trying to buying it via the only means possible of buying a single unit - being on ebay. And $15K is the price they are there.

[Beaucoupnice]

Hey alpinemo  - any chance of a copy of the spreadsheet, for personal use?

[NameTaken]

DragonMint coming out with a 16Th/s SHA256 ASIC.

http://cryptomining-blog.com/9295-new-16-ths-dragonmint-16t-bitcoin-asic-miner-coming-in-2018/

[QuintLeo]

DragonMint has ANNOUNCED a miner, there is considerable doubt at this point if it actually exists.

Let's wait for WORKING HARDWARE to show up and be widely reported as REAL before we add stuff to a list like this?

[maqu]

Hi,

Thanks for sharing this! I got some new models to my list found here: https://hodl.eu/mining/asic-mining/profitable-asic-miners/

[QuintLeo]

The SP20 Jackson RARELY managed to achieve 1.7 THash - more commonly they would top out around 1.5ish.

They needed their full MAXIMUM of 1152 watts to have a PRAYER of getting there, and a VERY cool room.

They were almost identical on specs vs the Antminer S5 when tuned to match hashrate - 1.15 Thash at right around 600 watts.

They could be tuned quite a bit further down than that, for higher efficiency, but I don't have my "tuning chart" any more so I don't remember the exact numbers.

I THINK the bottom stable figure on mine was around 850 Ghash at a 350 or so watts, but that could be off a bit on both figures.


The original Innosilicon A2 was rated for 88 Mhash/sec (Scrypt)  in it's "Turbo" setting - mine pulls 980 watts at that hashrate.
The later version was rated for 110 Mhash/sec in it's "Max Turbo" setting - I have 2, one at 1240 watts the other at 1260, both with Seasonic X-1250 supplies I added to replace the original power supplies (one of which was DOA when I bought the units).
That later version running at 100 Mhash/sec (Low Turbo setting, I have 2 of THOSE) pull 1055 or 1085 watts - this was the PRACTICAL limit for them on the original 1100 watt Gold power supply (which proved quite robust if you didn't overclock the unit past the 100 Mhash setting).

The Gridseed "80 blade" was good for about 2.7 Scrypt Mhash/sec per "side" for 5.4 total, and would draw about 45 watts per side at that setting.
They were a LOT more reliable if you didn't push them that hard though, 2.5 Mhash at about 40 watts per side ran the voltage conversion circuitry a LOT cooler.