How to optimize Plutus smart contracts?

Question

Context

I'm creating scripts with hardcoded parameters in my smartcontract and I would like to reduce footprint of them to reduce transaction fees. Right now i'm doing so using parameterized smartcontracts. I can't provide source but I read or heard somewhere that using parameterized smartcontracts use up a lot of ExMemory. When building plutus scripts i'm using evaluateScriptCounting which after compilation shows me ExCPU and ExMemory of my script. So far I have found the only way to hardcode parameters without using parameterized smartcontracts with BuiltinByteStrings is doing something like this:

{-# INLINABLE flattenBuiltinByteString #-}
flattenBuiltinByteString :: [BuiltinByteString] -> BuiltinByteString
flattenBuiltinByteString [] = emptyByteString 
flattenBuiltinByteString (x:xs) = appendByteString x (flattenBuiltinByteString xs)

{-# INLINABLE getPkh #-}
getPkh :: PubKeyHash
getPkh = PubKeyHash { getPubKeyHash = flattenBuiltinByteString [ consByteString x emptyByteString |x <- [128,164,244,91,86,184,141,17,57,218,35,188,76,60,117,236,109,50,148,60,8,127,37,11,134,25,60,167]]}

I was expecting that this way I would get lower ExCPU and ExMemory units. But I did some tests and to my surprise my smartcontract without parameterized parameters was bigger!

Tests

1. ExCPU -> 8366313, ExMemory -> 28200. Plain always succeeds.
2. ExCPU -> 8366313, ExMemory -> 28200. BuiltinByteString, no checks, no datum.
3. ExCPU -> 9080865, ExMemory -> 30600. Parameterized no checks, no datum.
4. ExCPU -> 9438141, ExMemory -> 31800. Parameterized no checks, added typed datum with 4 fields.
5. ExCPU -> 10152693, ExMemory -> 34200. Parameterized 4 checks against typed datum.
6. ExCPU -> 11135202, ExMemory -> 37500. BuiltinByteString 4 checks against typed datum.

Questions

1. My flattenBuiltinByteString is doing the flattening foldr style which is ineffective and I suppose is why i'm having such big jump in ExMemory units. I was looking for foldl' in Plutus library but couldn't find anything. Does Plutus have something simillar to seq so that I could make my own foldl' implementation?
2. Are there other more effective ways to flatten BuiltinByteString list on-chain?
3. What over tricks I could do besides using BuiltinData types instead of my own custom typed data types to optimize my smartcontracts?
4. Should I even be bothering with optimizations like this, whats the upper ExCPU and ExMemory limit?

Answer 1

I am building a dApp myself and I am running into similar issues related to hitting the current mainnet's maxTxExUnits upper boundaries of exUnitsMem and exUnitsSteps.

In regards to question 4:

You can find the current upper limits by querying your cardano node for protocol parameters like so:

cardano-cli query protocol-parameters $MAGIC --out-file protocol-parameters.json

The limits can be found under: maxTxExecutionUnits

I resolved some of my issues related to exceeding the ExMemory upper limit by:

• simplifying/ optimizing the contract logic
• removing traceIfFalse statements or at least decreasing long log message string down to error codes
• using Bang Pattern in let-bindings
• avoid exhaustive case patterns and instead rely on Plutus' throwing an error for a missing pattern match

There are many more resources you can read below for further possible code improvements to optimize your script.

Sources:

• MinSwap Article
• Plutonomicon

Answer 2

Mlabs created excelent tool to quickly check and analyse size of smartcontracts. Github repo is here