Usage example

We present a typical benchmark workflow using octez-snoop. We’ll consider the case of the blake2b hashing function, which is used among other things to hash blocks, operations and contexts:

Tezos_crypto.Blake2B.hash_bytes : ?key:bytes -> bytes list -> Tezos_crypto.Blake2B.t

At the time of writing, this function is a thin wrapper which concatenates the list of bytes and passes it to the blake2b implementation provided by HACL*.

Step 1: Defining the benchmark

Benchmarks correspond to OCaml modules implementing the Benchmark.S signature. These must then be registered via the Registration.register function. Of course, for this registration to happen, the file containing the benchmark and the call to Registration.register should be linked with octez-snoop. See Architecture of octez-snoop for complementary details.

We’ll define the benchmark module chunk by chunk and describe each part. Benchmarks are referenced by name. The info field is a brief description of the benchmark. Finally, there’s also a system of tags that allows listing benchmarks by kind.

module Blake2b_bench : Benchmark.S = struct
  let name = "Blake2b_example"
  let info = "Illustrating tezos-benchmark by benchmarking blake2b"
  let tags = ["example"]

Typically, a benchmark will depend on a set of parameters corresponding e.g. to the parameters of the samplers used to generate input data to the function being benchmarked. This corresponds to the type config. A default_config is provided, which can be overridden by specifying a well-formatted JSON file. This is made possible by defining a config_encoding using the data-encoding library.

type config = {max_bytes : int}
let default_config = {max_bytes = 1 lsl 16}
let config_encoding =
  let open Data_encoding in
  conv
    (fun {max_bytes} -> max_bytes)
    (fun max_bytes -> {max_bytes})
    (obj1 (req "max_bytes" int31))

Benchmarking involves measuring the execution time of some piece of code and using the recorded execution time to fit a model. As explained in Architecture of octez-snoop, a model is in fact a function of two parameters: a workload and the vector of free parameters to be fitted. The workload corresponds to the information on the input of the function being benchmarked required to predict its execution time. Typically, it corresponds to some notion of “size” of the input. In order to be saved to disk, we must define a workload_encoding as well. The workload type is abstract from the outside of the module, however, for plotting purposes, it is necessary to exhibit a vector-like structure on these workloads. The workload_to_vector function maps workloads to sparse vectors. If one is not interested in plotting, this function can be made to always return Sparse_vec.String.zero.

type workload = {nbytes : int}
let workload_encoding =
  let open Data_encoding in
  conv
    (fun {nbytes} -> nbytes)
    (fun nbytes -> {nbytes})
    (obj1 (req "nbytes" int31))
let workload_to_vector {nbytes} =
  Sparse_vec.String.of_list [("nbytes", float_of_int nbytes)]

We expect the execution time of Blake2b.hash_bytes to be proportional to the number of bytes being hashed, with possibly a small constant-time overhead. Hence, we pick an affine model. The affine model is generic, of the form \(\text{affine}(n) = \theta_0 + \theta_1 \times n\) with \(\theta_i\) the free parameters. One must explain how to convert the workload to the argument n. This is the purpose of the conv parameter.

let models =
  [ ( "blake2b",
      Model.make
        ~conv:(fun {nbytes} -> (nbytes, ()))
        ~model:
          (Model.affine
             ~intercept:(Free_variable.of_string "blake2b_const")
             ~coeff:(Free_variable.of_string "blake2b_ns_p_byte")) ) ]

Finally, we can define the actual benchmark. The function to be defined is create_benchmarks, which expects to be given an rng_state, a bench_num and a config and returns a list of suspensions, each suspension yielding a benchmark when evaluated.

One might wonder why this particular signature was been chosen, instead of returning directly a list of benchmarks, or requiring simply a benchmark generator to be defined.

• The current signature allows for setup code to be shared by all benchmarks being generated (not the case here).

• Returning a list of suspensions allows to delay the sampling process and the memory allocation associated to benchmark generation until actually needed, hence preventing memory leaks.

The auxiliary function blake2b_benchmark is in charge of preparing a closure, corresponding to a call to Blake2b.hash_bytes applied to a random bytes, and the associated workload, containing the size of the random bytes. We want benchmarks to be easily replayable given a seed, hence the closure-generation function is parameterized with an explicit rng_state of type Random.State.t.

  let blake2b_benchmark rng_state config () =
    let nbytes =
      Base_samplers.sample_in_interval
        rng_state
        ~range:{min = 1; max = config.max_bytes}
    in
    let bytes = Base_samplers.uniform_bytes rng_state ~nbytes in
    let workload = {nbytes} in
    (* The closure here is the piece of code to be benchmarked. *)
    let closure () = ignore (Tezos_crypto.Blake2B.hash_bytes [bytes]) in
    Generator.Plain {workload; closure}
  let create_benchmarks ~rng_state ~bench_num config =
    List.repeat bench_num (blake2b_benchmark rng_state config)
end (* module Blake2b_bench *)

This concludes the definition of the benchmark. Let’s register it:

let () = Registration.register (module Blake2b_bench)

For illustrative purposes, we also make the blake2b available for code generation.

let () =
  Registration.register_for_codegen
    "blake2b_codegen"
    (Model.For_codegen (List.assoc "blake2b" Blake2b_bench.models))

Step 2: Checking the timer

Before we perform the benchmarks, we need to ensure that the system timer is sufficiently precise. This data is also useful to subtract the latency of the timer for benchmarks of very small duration (which is not required here). We invoke the tool on the built-in benchmark TIMER_LATENCY and specify (through --bench-num) that we want only one closure to benchmark (since all closures are identical for this benchmark) but to execute this closure 100000 times (through --nsamples).

octez-snoop benchmark TIMER_LATENCY and save to timer.workload --bench-num 1 --nsamples 100000

The tool returns the following on standard output:

Benchmarking with the following options:
{ options = { flush_cache=false;
              stabilize_gc=false;
              seed=self-init;
              bench #=1;
              nsamples/bench=100000;
              determinizer=percentile 50;
              cpu_affinity=none;
              minor_heap_size=262144 words;
              config directory=None };
   save_file = timer.workload;
   storage = Mem }
Using default configuration for benchmark TIMER_LATENCY
{}
benchmarking 1/1
stats over all benchmarks: { max_time = 25.000000 ; min_time = 25.000000 ; mean_time = 25.000000 ; sigma = 0.000000 }

This commands measures 100000 times the latency of the timer, that is the minimum time between two timing measurements. This yields an empirical distribution on timings. The tool takes the 50th percentile (i.e. the median) of the empirical distribution and returns the result: 25ns latency. This is reasonable. Since there’s only one benchmark (with many samples), the standard deviation is by definition zero. One could also run many benchmarks with fewer samples per benchmark:

octez-snoop benchmark TIMER_LATENCY and save to timer.workload --bench-num 1000 --nsamples 100

This yields on standard output:

Benchmarking with the following options:
{ options = { flush_cache=false;
              stabilize_gc=false;
              seed=self-init;
              bench #=1000;
              nsamples/bench=100;
              determinizer=percentile 50;
              cpu_affinity=none;
              minor_heap_size=262144 words;
              config directory=None };
   save_file = timer.workload;
   storage = Mem }
Using default configuration for benchmark TIMER_LATENCY
{}
benchmarking 1000/1000
stats over all benchmarks: { max_time = 40.000000 ; min_time = 23.000000 ; mean_time = 24.130000 ; sigma = 0.653529 }

This is consistent with the previous results.

A reliable timer should have a latency of the order of 20 to 30 nanoseconds, with a very small standard deviation. It can happen on some hardware or software configurations that the timer latency is of the order of microseconds or worse: this makes benchmarking short-lived computations impossible.

Step 3: Benchmarking

If the results obtained in the previous section are reasonable, we can proceed to the generation of raw timing data. We want to invoke the Blake2b_example benchmark and save the resulting data to ./blake2b.workload. We want 500 distinct random inputs, and for each input we will perform the timing measurement 3000 times. The --determinizer option specifies how the empirical timing distribution corresponding to the per-input 3000 samples will be converted to a fixed value: here we pick the 50th percentile, i.e. the median (which happens to also be the default, so this option could have been omitted). We also use an explicit random seed in case we want to reproduce the exact same benchmarks. If not specified, the PRNG will self-initialize using an unknown seed.

octez-snoop benchmark Blake2b_example and save to blake2b.workload --bench-num 500 --nsamples 3000 --determinizer percentile@50 --seed 12897

Here’s the output:

Benchmarking with the following options:
{ options = { flush_cache=false;
              stabilize_gc=false;
              seed=12897;
              bench #=500;
              nsamples/bench=3000;
              determinizer=percentile 50;
              cpu_affinity=none;
              minor_heap_size=262144 words;
              config directory=None };
   save_file = blake2b.workload;
   storage = Mem }
Using default configuration for benchmark Blake2b_example
{ "max_bytes": 65536 }
benchmarking 500/500
stats over all benchmarks: { max_time = 71957.000000 ; min_time = 284.000000 ; mean_time = 34750.532000 ; sigma = 20155.604394 }

Since the size of inputs varies a lot, the statistics over all benchmarks are less useful.

Step 3.5: (optional) Removing outliers

It is possible to remove outliers from the raw benchmark data. The command is the following:

octez-snoop remove outliers from data ./blake2b.workload above 3 sigmas and save to blake2b-cleaned.workload

In this particular example, the data seems clean though:

Measure.load: loaded ./blake2b.workload
Removing outliers.
Stats: { max_time = 71925.000000 ; min_time = 289.000000 ; mean_time = 34988.436000 ; sigma = 20766.341788 }
Validity interval: [-27310.589365, 97287.461365].
Removed 0 outliers out of 500 elements.

The best defense against outliers is to have clean data in the first place: use a stable environment for benchmarking.

Step 4: Fitting the model

We can now proceed to inferring the free parameters from the model using the data. At the time of writing, the tool offloads the regression problem to the scikit-learn (aka sklearn) and the statmodels Python libraries: install them before proceeding.

pip install scikit-learn statsmodels

Let’s execute the following command:

octez-snoop infer parameters for model blake2b on data blake2b.workload using lasso --lasso-positive --dump-csv blake2b.csv --save-solution blake2b.sol --plot

Initializing python... Done.
Measure.load: loaded blake2b.workload
Applying model to workload data 500/500
Initializing matrices 500/500
Importing blake2b.csv
Exporting to blake2b.csv
Saved solution to blake2b.sol

The command performed the following tasks:

• load the workload data from blake2b.workload;

• construct a linear regression problem using the chosen model: here, the Blake2b_example benchmark only provides the blake2b model;

• solve this problem using the specified lasso algorithm, with the constraint that the inferred coefficients must be positive;

• dump the result of inference to a csv file named blake2b.csv;

• save a structured solution (useful for code generation) to blake2b.sol;

• plot the result of inference.

Let’s first have a look at the contents of the CSV solution blake2b.csv.

Inference results

blake2b_const	blake2b_ns_p_byte
129.279086813	1.09627036127

The columns correspond to the inferred values for the free variables of the blake2b cost model. The units are respectively ns/bytes for blake2b_ns_p_byte and ns for blake2b_const.

The tool also produces a plot:

The leftmost figure plots the empirical data, i.e. the raw execution time (in nanoseconds) as a function of the input size (here, in bytes – other data structures might use different notions of sizes). The rightmost figure plots the empirical data along the predicted execution time. If the model is good and the parameters were correctly fitted, these should match. The central plot is useful when using complex nonlinearities to model the execution time of some piece of code: the tool will project back the raw data in the linear space spanned by the chosen nonlinearities and if the model is good, one should observe that the empirical data lies along a linear subspace. Here, the model is trivial so the central plot is less interesting.

Step 5: Generating code

As a final step, we demonstrate how to generate code corresponding to the model. This is typically used to generate gas consumption functions for Michelson instructions and not for raw functions like blake2b but the principle is similar.

octez-snoop generate code using solution blake2b.sol and model blake2b_codegen

By default, the tool produces integer code by casting floating point constant to integers. The tool produces the following code on stdout:

let model_blake2b_codegen size =
    (int_of_float 144.753899773) + (int_of_float 1.17988921492) * size

It is also possible to generate code implementing the cost function using fixed-point arithmetic. This requires specifying some codegen parameters in a JSON file. For instance, we can require to consider 5 bits of precision and use rounding to nearest to convert constants, failing if we make more than 10% relative error when casting. The inverse_scaling and resolution parameters respectively specify the fraction of digits considered to be not significant, and the resolution of the grid used when prettifying constants (in nanoseconds).

{ "precision": 5, "max_relative_error": 0.1, "cast_mode": "Round", "inverse_scaling": 3, "resolution": 5 }

Calling the tool:

octez-snoop generate code using solution blake2b.sol and model blake2b_codegen --fixed-point codegen_params.json

We get:

let model_blake2b_codegen size =
    let v0 = size in
    150 + ((v0 + (v0 lsr 3)) + (v0 lsr 5))