chore(bench): make bits to prove customizable in zk benchmarks

[soonum]

Some application like blockchain, may wants to prove less bits than CRS size allows to.

---

This change is 

[nsarlin-zama]

Just so I understand, the goal is to be able to benchmark with proofs that do not completely fill the max size supported by the crs ?

Reviewable status: 0 of 2 files reviewed, 1 unresolved discussion (waiting on @IceTDrinker)

---

tfhe/web_wasm_parallel_tests/worker.js line 703 at r1 (raw file):

      [64, [64]],
      [640, [640]],
      [2048, [2048, 64 * 3]],

I guess this comes from a real life usecase ?

[IceTDrinker]

@soonum currently seeing a conflict on that PR, if you can rebase ?

[IceTDrinker]

@nsarlin-zama yes

Reviewable status: 0 of 2 files reviewed, 3 unresolved discussions (waiting on @nsarlin-zama and @soonum)

[soonum]

Reviewable status: 0 of 2 files reviewed, 3 unresolved discussions (waiting on @agnesLeroy, @IceTDrinker, and @nsarlin-zama)

---

tfhe/web_wasm_parallel_tests/worker.js line 703 at r1 (raw file):

Previously, nsarlin-zama (Nicolas Sarlin) wrote…

I guess this comes from a real life usecase ?

Yes, from blockchain use case.

[IceTDrinker]

A few questions

@IceTDrinker reviewed 2 of 2 files at r2, all commit messages.
Reviewable status: all files reviewed, 4 unresolved discussions (waiting on @agnesLeroy, @IceTDrinker, and @nsarlin-zama)

---

tfhe-benchmark/benches/integer/zk_pke.rs line 52 at r2 (raw file):

    let usable_cpu_threads = (rayon::current_num_threads() as u64 / max_threads).max(1) + 1;

    #[cfg(feature = "gpu")]

prefer if cfg!(feature = "gpu")

---

tfhe/web_wasm_parallel_tests/worker.js line 700 at r2 (raw file):

      {"crs_bit_size": 640, "bits_to_encrypt": [640]},
      {"crs_bit_size": 2048, "bits_to_encrypt": [2048, 64 * 4]}  // 64 * 4 is a production use-case
      //{"crs_bit_size": 4096, "bits_to_encrypt": [4096]}, // This case is too big and make webdriver timeout

that's odd, might be a crash ? does it work with integer ? Or the CRS gen is timing out in the browser, did you try with the integer bench ?

how is the timeout managed currently ? It's an old system, so could it be that the test

[soonum]

Reviewable status: all files reviewed, 3 unresolved discussions (waiting on @agnesLeroy, @IceTDrinker, and @nsarlin-zama)

---

tfhe/web_wasm_parallel_tests/worker.js line 700 at r2 (raw file):

Previously, IceTDrinker (Arthur Meyre) wrote…

that's odd, might be a crash ? does it work with integer ? Or the CRS gen is timing out in the browser, did you try with the integer bench ?

how is the timeout managed currently ? It's an old system, so could it be that the test

The timeout for this test currently is 3600 seconds.
I guess the CRS generation time plus the benchmarks exceed that duration.
I have to measure the 4096 case only to see how long it takes.

[nsarlin-zama]

small comment

Reviewable status: 0 of 2 files reviewed, 5 unresolved discussions (waiting on @agnesLeroy and @IceTDrinker)

---

tfhe-benchmark/benches/integer/zk_pke.rs line 110 at r4 (raw file):

                // Packing, so we take the message and carry modulus to compute our block count
                let num_block = 64usize.div_ceil(
                    (param_pke.message_modulus.0 * param_pke.carry_modulus.0).ilog2() as usize,

You could reuse msg_bits computed above here

---

tfhe-benchmark/benches/integer/zk_pke.rs line 245 at r4 (raw file):

                // Packing, so we take the message and carry modulus to compute our block count
                let num_block = 64usize.div_ceil(
                    (param_pke.message_modulus.0 * param_pke.carry_modulus.0).ilog2() as usize,

same could use msg_bits

---

tfhe-benchmark/benches/integer/zk_pke.rs line 505 at r4 (raw file):

                    // Packing, so we take the message and carry modulus to compute our block count
                    let num_block = 64usize.div_ceil(
                        (param_pke.message_modulus.0 * param_pke.carry_modulus.0).ilog2() as usize,

same

[IceTDrinker]

build is red currently

[IceTDrinker]

@IceTDrinker reviewed 2 of 2 files at r5, all commit messages.
Reviewable status: all files reviewed, 5 unresolved discussions (waiting on @agnesLeroy, @IceTDrinker, and @soonum)

---

tfhe/web_wasm_parallel_tests/worker.js line 700 at r2 (raw file):

Previously, soonum (David Testé) wrote…

The timeout for this test currently is 3600 seconds.
I guess the CRS generation time plus the benchmarks exceed that duration.
I have to measure the 4096 case only to see how long it takes.

you reverted the comment, so this cases passes now in 1h ?

[IceTDrinker]

Thanks !

[IceTDrinker]

Reviewable status: 1 of 2 files reviewed, 5 unresolved discussions (waiting on @agnesLeroy and @soonum)