Superposition AMM gas analysis (randomly generated)

Trades were randomly generated using a sizeable dataset.

Methodology for comparison (IGNORE ME - NOT USED IN PRACTICE)

1. The random dataset was generated with some Q code. We had some issues with abi.q, so we used an intermediate step of converting it to a CSV to stdout, to be handled by a Go program. This is the creation of the liquidity amounts for stresstesting.

/ you would need to invoke this script with q -q -S 123, then \\l superposition_data.q

calcdelta:... / stubbed (IGNORE ME)

length:10000000

gentick:{[] {(rand 1774544) - 887272}@'length?1}
amount0MaxAmt:10000*1e6 / per trade!
amount1MaxAmt:100000*1e18

main:{
	c:update l:?[l>u;u;l], u:?[l>u;l;u] from([] amount0:length?amount0MaxAmt; amount1:length?amount1MaxAmt; l:gentick[]; u:gentick[]);
	-1","0:update delta:{calcdelta[x`amount0;x`amount1]}@'c from c;
	exit 0}

main[]

1. A local Stylus instance needs to be run locally.
2. This is a Go program that reads from the CSV file using mmap, sending it out to specific goroutines that execute the calldata sequentially. It first allocates the “buckets” of liquidity that will need to be minted using a specific token, first transferring the amounts to the addresses derived from the private keys. It then begins to distribute the tasks to the workers given. Gist here
3. This is a program that generates public/private keys that can be used to create the LP positions with. Gist here . It could be composed into function arguments using $(tail -n +2 addresses-used.csv | cut -f 1)
4. You would then need to call the function on the AMM createPool .
5. This program should be run against the local Stylus instance to populate the ticks. This will populate the liquidity to facilitate the swaps we’ll make. This will stress the local environment, and will take time.

Uniswap 3 gas analysis (from 71876737 to 169909567)

Specific approach taken for comparison

1. Functions exactInput(tuple params) (0xb85818), exactInputSingle(tuple params) (0x04e45a ), exactInputSingle(tuple params) (0x414bf3), exactInput(tuple params) (0xc04b8d), exactOutput(tuple params) (0x09b813), exactOutputSingle(tuple params) (0x5023b4) that interacted with the SwapRouter02 were selected for in the final dataset.
2. The pools were found using Arbiscan’s transaction history on the factory contract, then taking those transaction hashes, and going through the topics/data that was created. Using that as a frame of reference was the approach that was taken. Then some scripting was used to get the address of the pool was used (ie, jq -r '.[] | [.tx.hash, .receipt.logs[0].topics[0], .receipt.logs[0].data] | @csv' | grep 0x783cca1c0412dd0d695e784568c96da2e9c22ff989357a2e8b1d9b2b4e6b7118 | sed 's/"//g' | cut -f3 -d, | cut -c91-)
3. Each pool was scanned using a filter log query the same way as before. These were then filtered for the Swap event.

Camelot V3 gas analysis (from 103164316 to 178277479)