32GiB synthetic PoRep lifecycle test fails

Description

Currently the 32GiB synthetic PoRep lifecycle test is failing.

The last line of the log are:

2024-04-10T15:56:54.722 INFO storage_proofs_porep::stacked::vanilla::proof > generating synthetic vanilla proofs in a single partition
test test_seal_lifecycle_32gib_porep_id_v1_2_top_8_8_0_api_v1_2 ... FAILED

failures:

---- test_seal_lifecycle_32gib_porep_id_v1_2_top_8_8_0_api_v1_2 stdout ----
thread 'test_seal_lifecycle_32gib_porep_id_v1_2_top_8_8_0_api_v1_2' panicked at 'assertion failed: `(left == right)`
  left: `10`,
 right: `1`', /home/vmx/rust-fil-proofs/storage-proofs-porep/src/stacked/vanilla/proof.rs:144:21
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

The same failure can also be reproduces with benchy (which might be quicker if you already have a preserved cache):

Create a cache:

RUST_LOG=info FIL_PROOFS_USE_GPU_COLUMN_BUILDER=1 FIL_PROOFS_USE_GPU_TREE_BUILDER=1 FIL_PROOFS_USE_MULTICORE_SDR=1 cargo run --release --features cuda --bin benchy -- porep --size 32GiB --api-features=synthetic-porep --cache /storage/d2/vmx/cache-32g --preserve-cache

Generate the syntheric PoRep:

RUST_LOG=info cargo run --release --features cuda --bin benchy -- porep --size 32GiB --api-features=synthetic-porep --cache /storage/d2/vmx/cache-32g --skip-precommit-phase1 --skip-precommit-phase2 --skip-commit-phase1 --skip-commit-phase2

Acceptance criteria

The test passes again.

Risks + pitfalls

Where to begin

Look why the partition count is different when used through the FFI. I possible fix could be to change the function that returns the PoRep configuration for the synthetic PoRep case.

It looks like the assertion at

rust-fil-proofs/storage-proofs-porep/src/stacked/vanilla/proof.rs

Line 144 in 8f5bd86

assert_eq!(partition_count, 1);

is just wrong. It was introduced in #1720.

I'll dig deeper on why I've added that. If I just remove it and run it with benchy, then the verification fails. I don't know if that's related or if it's just a problem with benchy.

It looks like the assertion at

rust-fil-proofs/storage-proofs-porep/src/stacked/vanilla/proof.rs

Line 144 in 8f5bd86

assert_eq!(partition_count, 1);

is just wrong. It was introduced in #1720.

I'll dig deeper on why I've added that. If I just remove it and run it with benchy, then the verification fails. I don't know if that's related or if it's just a problem with benchy.

Correct, it's wrong and doesn't make sense to assert there.