lowram: Eliminate y vector in REDUCE_RAM mode in sign by mkannwischer · Pull Request #1031 · pq-code-package/mldsa-native

mkannwischer · 2026-04-08T09:40:14Z

Add mld_yvec following the lazy polyvec pattern: in REDUCE_RAM mode it
stores only the seed and nonce and regenerates y[i] on demand. The
fused matrix-vector helper mld_polyvec_matrix_pointwise_montgomery_yvec
samples y column-by-column in REDUCE_RAM mode.

CBMC proofs updated for the eager path; lazy variant has no proof yet.

Depends on lowram: Stream matrix A element-by-element to reduce memory #1019
Hoisted out from PoC: Reduce large struct allocations to <= 13/17/21 KiB for ML-DSA-44/65/87 #1005

oqs-bot · 2026-04-08T09:58:00Z

CBMC Results (ML-DSA-65)

⚠️ Attention Required

Proof	Status	Current	Previous	Change
`polymat_permute_bitrev_to_custom`	⚠️	495s	37s	+1238%

Full Results (185 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	✅	2834s	2329s	+21.7%
`polymat_permute_bitrev_to_custom`	⚠️	495s	37s	+1238%
`polyvecl_pointwise_acc_montgomery_c`	✅	341s	280s	+22%
`sign_verify_internal`	✅	275s	286s	-4%
`poly_pointwise_montgomery_c`	✅	170s	156s	+9%
`rej_uniform_native`	✅	155s	148s	+5%
`mld_invntt_layer`	✅	99s	94s	+5%
`polyvec_matrix_expand_eager`	✅	99s	-	new
`mld_ct_memcmp`	✅	82s	79s	+4%
`mld_attempt_signature_generation`	✅	72s	107s	-33%
`mld_ntt_layer`	✅	59s	55s	+7%
`polyvec_matrix_expand_eager_serial`	✅	52s	-	new
`sign_signature_internal`	✅	32s	25s	+28%
`rej_uniform`	✅	23s	22s	+5%
`poly_chknorm_c`	✅	21s	19s	+11%
`polyveck_decompose`	✅	20s	16s	+25%
`fqmul`	✅	19s	18s	+6%
`mld_compute_t0_t1_tr_from_sk_components`	✅	18s	28s	-36%
`poly_uniform_4x`	✅	17s	15s	+13%
`poly_uniform_eta_4x`	✅	17s	16s	+6%
`polyvecl_chknorm`	✅	17s	5s	+240%
`rej_uniform_c`	✅	16s	15s	+7%
`keccakf1600x4_permute_native`	✅	14s	13s	+8%
`polyt0_unpack`	✅	14s	15s	-7%
`mld_check_pct`	✅	13s	13s	+0%
`mld_ntt_butterfly_block`	✅	13s	13s	+0%
`polyvec_matrix_pointwise_montgomery_eager`	✅	13s	-	new
`polyveck_power2round`	✅	13s	12s	+8%
`keccak_absorb_once_x4`	✅	12s	10s	+20%
`polyveck_invntt_tomont`	✅	12s	9s	+33%
`poly_add`	✅	11s	10s	+10%
`polyveck_add`	✅	11s	10s	+10%
`keccak_absorb`	✅	10s	5s	+100%
`polyveck_ntt`	✅	10s	7s	+43%
`keccakf1600_permute`	✅	9s	10s	-10%
`poly_invntt_tomont_c`	✅	9s	7s	+29%
`polyveck_caddq`	✅	9s	11s	-18%
`polyveck_pointwise_poly_montgomery`	✅	9s	6s	+50%
`polyveck_use_hint`	✅	9s	22s	-59%
`polyvecl_ntt`	✅	9s	5s	+80%
`sign`	✅	9s	7s	+29%
`keccakf1600_permute_native`	✅	8s	8s	+0%
`poly_decompose_c`	✅	8s	7s	+14%
`polyveck_reduce`	✅	8s	5s	+60%
`sign_keypair_internal`	✅	8s	6s	+33%
`sign_pk_from_sk`	✅	8s	9s	-11%
`keccak_squeezeblocks_x4`	✅	7s	6s	+17%
`mld_polyvecl_permute_bitrev_to_custom_native`	✅	7s	8s	-12%
`poly_caddq_c`	✅	7s	5s	+40%
`polyveck_chknorm`	✅	7s	8s	-12%
`mld_compute_pack_z`	✅	6s	6s	+0%
`mld_sample_s1_s2`	✅	6s	7s	-14%
`poly_uniform_gamma1_4x`	✅	6s	5s	+20%
`polyeta_unpack`	✅	6s	6s	+0%
`polyt0_pack`	✅	6s	3s	+100%
`polyveck_shiftl`	✅	6s	7s	-14%
`polyveck_sub`	✅	6s	8s	-25%
`polyveck_unpack_eta`	✅	6s	2s	+200%
`rej_eta_native`	✅	6s	5s	+20%
`shake128x4_squeezeblocks`	✅	6s	3s	+100%
`sign_open`	✅	6s	4s	+50%
`unpack_sk`	✅	6s	9s	-33%
`keccakf1600_extract_bytes (big endian)`	✅	5s	4s	+25%
`mld_h`	✅	5s	5s	+0%
`ntt_native_aarch64`	✅	5s	3s	+67%
`poly_caddq_native`	✅	5s	2s	+150%
`poly_decompose_native`	✅	5s	4s	+25%
`poly_ntt`	✅	5s	4s	+25%
`poly_ntt_native`	✅	5s	3s	+67%
`poly_power2round`	✅	5s	7s	-29%
`poly_uniform`	✅	5s	5s	+0%
`poly_uniform_eta`	✅	5s	3s	+67%
`poly_use_hint_c`	✅	5s	4s	+25%
`polyveck_pack_eta`	✅	5s	3s	+67%
`polyvecl_pointwise_acc_montgomery_native`	✅	5s	3s	+67%
`polyvecl_uniform_gamma1_serial`	✅	5s	2s	+150%
`unpack_hints`	✅	5s	5s	+0%
`keccak_squeeze`	✅	4s	6s	-33%
`keccakf1600_xor_bytes`	✅	4s	2s	+100%
`mld_ct_cmask_nonzero_u32`	✅	4s	3s	+33%
`mld_prepare_domain_separation_prefix`	✅	4s	6s	-33%
`mld_sample_s1_s2_serial`	✅	4s	3s	+33%
`pack_pk`	✅	4s	3s	+33%
`pack_sig_h_poly`	✅	4s	4s	+0%
`pointwise_native_x86_64`	✅	4s	1s	+300%
`poly_caddq_native_aarch64`	✅	4s	5s	-20%
`poly_chknorm_native`	✅	4s	3s	+33%
`poly_chknorm_native_aarch64`	✅	4s	4s	+0%
`poly_ntt_c`	✅	4s	2s	+100%
`poly_pointwise_montgomery_native`	✅	4s	2s	+100%
`poly_use_hint_native`	✅	4s	4s	+0%
`polyvec_matrix_pointwise_montgomery_yvec_eager`	✅	4s	-	new
`polyvecl_pack_eta`	✅	4s	3s	+33%
`polyvecl_pointwise_acc_montgomery`	✅	4s	4s	+0%
`polyvecl_unpack_z`	✅	4s	2s	+100%
`polyw1_pack`	✅	4s	3s	+33%
`rej_eta`	✅	4s	4s	+0%
`shake128_init`	✅	4s	2s	+100%
`shake256`	✅	4s	2s	+100%
`shake256_squeeze`	✅	4s	5s	-20%
`sign_keypair`	✅	4s	2s	+100%
`sign_signature`	✅	4s	4s	+0%
`sign_signature_extmu`	✅	4s	4s	+0%
`sign_signature_pre_hash_internal`	✅	4s	7s	-43%
`decompose`	✅	3s	4s	-25%
`intt_native_x86_64`	✅	3s	2s	+50%
`keccak_f1600_x4_native_aarch64_v84a`	✅	3s	3s	+0%
`keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid`	✅	3s	1s	+200%
`keccak_init`	✅	3s	2s	+50%
`keccakf1600_xor_bytes (big endian)`	✅	3s	3s	+0%
`keccakf1600x4_xor_bytes`	✅	3s	3s	+0%
`mld_ct_cmask_neg_i32`	✅	3s	4s	-25%
`mld_value_barrier_u32`	✅	3s	1s	+200%
`mld_value_barrier_u8`	✅	3s	4s	-25%
`montgomery_reduce`	✅	3s	5s	-40%
`ntt_native_x86_64`	✅	3s	5s	-40%
`pack_sig_z`	✅	3s	2s	+50%
`pack_sk`	✅	3s	4s	-25%
`poly_challenge`	✅	3s	6s	-50%
`poly_decompose`	✅	3s	3s	+0%
`poly_make_hint`	✅	3s	3s	+0%
`poly_pointwise_montgomery`	✅	3s	6s	-50%
`poly_reduce`	✅	3s	2s	+50%
`polyt1_pack`	✅	3s	3s	+0%
`polyt1_unpack`	✅	3s	5s	-40%
`polyveck_pack_t0`	✅	3s	4s	-25%
`polyveck_unpack_t0`	✅	3s	2s	+50%
`polyvecl_permute_bitrev_to_custom`	✅	3s	4s	-25%
`polyvecl_uniform_gamma1`	✅	3s	3s	+0%
`polyvecl_unpack_eta`	✅	3s	3s	+0%
`polyz_unpack`	✅	3s	4s	-25%
`polyz_unpack_c`	✅	3s	2s	+50%
`polyz_unpack_native`	✅	3s	5s	-40%
`power2round`	✅	3s	2s	+50%
`rej_eta_c`	✅	3s	3s	+0%
`shake128_absorb`	✅	3s	4s	-25%
`shake128_finalize`	✅	3s	1s	+200%
`shake128_release`	✅	3s	3s	+0%
`shake128x4_absorb_once`	✅	3s	3s	+0%
`shake256_finalize`	✅	3s	2s	+50%
`shake256_release`	✅	3s	3s	+0%
`shake256x4_absorb_once`	✅	3s	1s	+200%
`shake256x4_squeezeblocks`	✅	3s	3s	+0%
`sign_signature_pre_hash_shake256`	✅	3s	5s	-40%
`sign_verify_extmu`	✅	3s	2s	+50%
`sign_verify_pre_hash_internal`	✅	3s	6s	-50%
`sys_check_capability`	✅	3s	3s	+0%
`unpack_pk`	✅	3s	4s	-25%
`use_hint`	✅	3s	3s	+0%
`fqscale`	✅	2s	4s	-50%
`keccak_f1600_x1_native_aarch64_v84a`	✅	2s	3s	-33%
`keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid`	✅	2s	4s	-50%
`keccak_finalize`	✅	2s	1s	+100%
`keccakf1600x4_extract_bytes`	✅	2s	1s	+100%
`keccakf1600x4_permute`	✅	2s	5s	-60%
`make_hint`	✅	2s	2s	+0%
`mld_ct_abs_i32`	✅	2s	2s	+0%
`mld_ct_cmask_nonzero_u8`	✅	2s	2s	+0%
`mld_ct_get_optblocker_u8`	✅	2s	2s	+0%
`mld_ct_sel_int32`	✅	2s	4s	-50%
`mld_keccakf1600_extract_bytes`	✅	2s	1s	+100%
`pack_sig_c`	✅	2s	2s	+0%
`pointwise_native_aarch64`	✅	2s	3s	-33%
`poly_caddq`	✅	2s	2s	+0%
`poly_chknorm`	✅	2s	2s	+0%
`poly_invntt_tomont`	✅	2s	6s	-67%
`poly_invntt_tomont_native`	✅	2s	3s	-33%
`poly_shiftl`	✅	2s	2s	+0%
`poly_sub`	✅	2s	3s	-33%
`poly_uniform_gamma1`	✅	2s	5s	-60%
`poly_use_hint`	✅	2s	1s	+100%
`polyeta_pack`	✅	2s	4s	-50%
`polyveck_pack_w1`	✅	2s	3s	-33%
`polyz_pack`	✅	2s	2s	+0%
`reduce32`	✅	2s	3s	-33%
`shake128_squeeze`	✅	2s	2s	+0%
`shake256_absorb`	✅	2s	3s	-33%
`sign_verify`	✅	2s	2s	+0%
`sign_verify_pre_hash_shake256`	✅	2s	5s	-60%
`unpack_sig`	✅	2s	2s	+0%
`caddq`	✅	1s	2s	-50%
`keccak_f1600_x1_native_aarch64`	✅	1s	2s	-50%
`mld_ct_get_optblocker_i64`	✅	1s	2s	-50%
`mld_ct_get_optblocker_u32`	✅	1s	2s	-50%
`mld_value_barrier_i64`	✅	1s	4s	-75%
`shake256_init`	✅	1s	2s	-50%

oqs-bot · 2026-04-08T09:59:09Z

CBMC Results (ML-DSA-87)

Full Results (185 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	✅	2300s	2411s	-4.6%
`polyvecl_pointwise_acc_montgomery_c`	✅	229s	218s	+5%
`polyvec_matrix_expand_eager`	✅	191s	-	new
`poly_pointwise_montgomery_c`	✅	169s	186s	-9%
`rej_uniform_native`	✅	157s	155s	+1%
`sign_verify_internal`	✅	149s	147s	+1%
`mld_attempt_signature_generation`	✅	100s	96s	+4%
`mld_invntt_layer`	✅	99s	104s	-5%
`mld_ct_memcmp`	✅	80s	89s	-10%
`polyvec_matrix_expand_eager_serial`	✅	65s	-	new
`sign_signature_internal`	✅	61s	41s	+49%
`mld_ntt_layer`	✅	59s	59s	+0%
`mld_compute_t0_t1_tr_from_sk_components`	✅	29s	25s	+16%
`rej_uniform`	✅	23s	24s	-4%
`poly_chknorm_c`	✅	21s	23s	-9%
`polymat_permute_bitrev_to_custom`	✅	21s	27s	-22%
`fqmul`	✅	19s	23s	-17%
`poly_uniform_eta_4x`	✅	17s	16s	+6%
`polyeta_unpack`	✅	17s	19s	-11%
`rej_uniform_c`	✅	16s	19s	-16%
`mld_polyvecl_permute_bitrev_to_custom_native`	✅	15s	14s	+7%
`poly_uniform_4x`	✅	15s	17s	-12%
`keccak_absorb_once_x4`	✅	14s	12s	+17%
`mld_ntt_butterfly_block`	✅	14s	16s	-12%
`keccakf1600x4_permute_native`	✅	13s	14s	-7%
`poly_add`	✅	13s	11s	+18%
`polyt0_unpack`	✅	12s	20s	-40%
`polyveck_add`	✅	11s	11s	+0%
`polyveck_shiftl`	✅	11s	8s	+38%
`polyveck_decompose`	✅	10s	11s	-9%
`polyveck_power2round`	✅	10s	20s	-50%
`polyvecl_ntt`	✅	10s	9s	+11%
`sign_keypair_internal`	✅	10s	6s	+67%
`mld_check_pct`	✅	9s	17s	-47%
`mld_sample_s1_s2_serial`	✅	9s	7s	+29%
`polyveck_ntt`	✅	9s	8s	+12%
`sign_pk_from_sk`	✅	9s	7s	+29%
`unpack_sk`	✅	9s	8s	+12%
`keccak_absorb`	✅	8s	7s	+14%
`keccakf1600_permute_native`	✅	8s	9s	-11%
`poly_invntt_tomont_c`	✅	8s	7s	+14%
`polyvec_matrix_pointwise_montgomery_eager`	✅	8s	-	new
`polyveck_invntt_tomont`	✅	8s	10s	-20%
`polyveck_pointwise_poly_montgomery`	✅	8s	6s	+33%
`polyveck_sub`	✅	8s	7s	+14%
`polyvecl_permute_bitrev_to_custom`	✅	8s	4s	+100%
`polyz_unpack_c`	✅	8s	12s	-33%
`keccakf1600_permute`	✅	7s	6s	+17%
`poly_caddq_c`	✅	7s	7s	+0%
`polyveck_reduce`	✅	7s	8s	-12%
`polyveck_use_hint`	✅	7s	12s	-42%
`sign`	✅	7s	7s	+0%
`sign_verify_pre_hash_internal`	✅	7s	2s	+250%
`keccak_squeezeblocks_x4`	✅	6s	9s	-33%
`mld_prepare_domain_separation_prefix`	✅	6s	4s	+50%
`poly_uniform_eta`	✅	6s	7s	-14%
`polyveck_caddq`	✅	6s	8s	-25%
`sign_open`	✅	6s	5s	+20%
`sign_signature_pre_hash_shake256`	✅	6s	4s	+50%
`unpack_hints`	✅	6s	6s	+0%
`keccak_f1600_x1_native_aarch64_v84a`	✅	5s	2s	+150%
`mld_h`	✅	5s	3s	+67%
`mld_sample_s1_s2`	✅	5s	5s	+0%
`pack_sk`	✅	5s	6s	-17%
`pointwise_native_aarch64`	✅	5s	4s	+25%
`poly_decompose_c`	✅	5s	7s	-29%
`poly_ntt_c`	✅	5s	6s	-17%
`polyt0_pack`	✅	5s	5s	+0%
`polyvec_matrix_pointwise_montgomery_yvec_eager`	✅	5s	-	new
`polyveck_pack_w1`	✅	5s	4s	+25%
`polyveck_unpack_t0`	✅	5s	5s	+0%
`polyvecl_pointwise_acc_montgomery`	✅	5s	4s	+25%
`polyvecl_uniform_gamma1`	✅	5s	2s	+150%
`polyvecl_uniform_gamma1_serial`	✅	5s	2s	+150%
`polyvecl_unpack_z`	✅	5s	5s	+0%
`power2round`	✅	5s	2s	+150%
`rej_eta_native`	✅	5s	4s	+25%
`sign_keypair`	✅	5s	3s	+67%
`sign_signature_extmu`	✅	5s	6s	-17%
`sign_signature_pre_hash_internal`	✅	5s	4s	+25%
`sign_verify_pre_hash_shake256`	✅	5s	7s	-29%
`unpack_sig`	✅	5s	3s	+67%
`keccak_f1600_x1_native_aarch64`	✅	4s	1s	+300%
`keccakf1600x4_extract_bytes`	✅	4s	2s	+100%
`make_hint`	✅	4s	4s	+0%
`mld_compute_pack_z`	✅	4s	10s	-60%
`mld_ct_cmask_nonzero_u32`	✅	4s	3s	+33%
`mld_ct_cmask_nonzero_u8`	✅	4s	2s	+100%
`ntt_native_aarch64`	✅	4s	4s	+0%
`pack_sig_c`	✅	4s	2s	+100%
`pack_sig_z`	✅	4s	4s	+0%
`poly_caddq`	✅	4s	4s	+0%
`poly_caddq_native_aarch64`	✅	4s	5s	-20%
`poly_challenge`	✅	4s	5s	-20%
`poly_chknorm`	✅	4s	1s	+300%
`poly_chknorm_native`	✅	4s	4s	+0%
`poly_pointwise_montgomery`	✅	4s	1s	+300%
`poly_sub`	✅	4s	3s	+33%
`poly_uniform`	✅	4s	5s	-20%
`poly_use_hint`	✅	4s	3s	+33%
`poly_use_hint_native`	✅	4s	3s	+33%
`polyvecl_chknorm`	✅	4s	8s	-50%
`polyvecl_unpack_eta`	✅	4s	5s	-20%
`polyz_unpack`	✅	4s	4s	+0%
`polyz_unpack_native`	✅	4s	2s	+100%
`rej_eta`	✅	4s	3s	+33%
`shake128_absorb`	✅	4s	3s	+33%
`shake128_finalize`	✅	4s	2s	+100%
`shake256`	✅	4s	3s	+33%
`shake256_absorb`	✅	4s	4s	+0%
`shake256x4_squeezeblocks`	✅	4s	2s	+100%
`sign_signature`	✅	4s	5s	-20%
`sign_verify`	✅	4s	6s	-33%
`sys_check_capability`	✅	4s	3s	+33%
`unpack_pk`	✅	4s	5s	-20%
`caddq`	✅	3s	2s	+50%
`fqscale`	✅	3s	3s	+0%
`intt_native_x86_64`	✅	3s	1s	+200%
`keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid`	✅	3s	1s	+200%
`keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid`	✅	3s	1s	+200%
`keccakf1600_xor_bytes`	✅	3s	2s	+50%
`keccakf1600x4_permute`	✅	3s	2s	+50%
`mld_ct_cmask_neg_i32`	✅	3s	2s	+50%
`mld_ct_get_optblocker_i64`	✅	3s	4s	-25%
`mld_ct_sel_int32`	✅	3s	2s	+50%
`mld_value_barrier_i64`	✅	3s	2s	+50%
`mld_value_barrier_u8`	✅	3s	3s	+0%
`montgomery_reduce`	✅	3s	3s	+0%
`pack_sig_h_poly`	✅	3s	4s	-25%
`pointwise_native_x86_64`	✅	3s	2s	+50%
`poly_caddq_native`	✅	3s	2s	+50%
`poly_chknorm_native_aarch64`	✅	3s	2s	+50%
`poly_decompose_native`	✅	3s	7s	-57%
`poly_make_hint`	✅	3s	6s	-50%
`poly_ntt`	✅	3s	4s	-25%
`poly_power2round`	✅	3s	6s	-50%
`poly_reduce`	✅	3s	4s	-25%
`poly_shiftl`	✅	3s	5s	-40%
`poly_uniform_gamma1_4x`	✅	3s	2s	+50%
`polyeta_pack`	✅	3s	5s	-40%
`polyt1_pack`	✅	3s	3s	+0%
`polyveck_chknorm`	✅	3s	7s	-57%
`polyveck_pack_eta`	✅	3s	4s	-25%
`polyveck_pack_t0`	✅	3s	4s	-25%
`polyveck_unpack_eta`	✅	3s	6s	-50%
`polyvecl_pointwise_acc_montgomery_native`	✅	3s	3s	+0%
`polyw1_pack`	✅	3s	3s	+0%
`reduce32`	✅	3s	2s	+50%
`rej_eta_c`	✅	3s	3s	+0%
`shake128x4_absorb_once`	✅	3s	2s	+50%
`shake256_finalize`	✅	3s	6s	-50%
`shake256_init`	✅	3s	1s	+200%
`shake256_release`	✅	3s	3s	+0%
`shake256_squeeze`	✅	3s	4s	-25%
`sign_verify_extmu`	✅	3s	4s	-25%
`keccak_finalize`	✅	2s	2s	+0%
`keccak_init`	✅	2s	3s	-33%
`keccak_squeeze`	✅	2s	3s	-33%
`keccakf1600_extract_bytes (big endian)`	✅	2s	2s	+0%
`keccakf1600_xor_bytes (big endian)`	✅	2s	3s	-33%
`keccakf1600x4_xor_bytes`	✅	2s	3s	-33%
`mld_ct_get_optblocker_u32`	✅	2s	3s	-33%
`mld_ct_get_optblocker_u8`	✅	2s	2s	+0%
`mld_keccakf1600_extract_bytes`	✅	2s	2s	+0%
`mld_value_barrier_u32`	✅	2s	1s	+100%
`ntt_native_x86_64`	✅	2s	3s	-33%
`pack_pk`	✅	2s	4s	-50%
`poly_decompose`	✅	2s	4s	-50%
`poly_invntt_tomont`	✅	2s	4s	-50%
`poly_invntt_tomont_native`	✅	2s	3s	-33%
`poly_ntt_native`	✅	2s	3s	-33%
`poly_pointwise_montgomery_native`	✅	2s	4s	-50%
`poly_use_hint_c`	✅	2s	2s	+0%
`polyt1_unpack`	✅	2s	3s	-33%
`polyvecl_pack_eta`	✅	2s	3s	-33%
`polyz_pack`	✅	2s	2s	+0%
`shake128_init`	✅	2s	4s	-50%
`shake128_release`	✅	2s	3s	-33%
`shake128_squeeze`	✅	2s	2s	+0%
`shake256x4_absorb_once`	✅	2s	2s	+0%
`use_hint`	✅	2s	2s	+0%
`decompose`	✅	1s	2s	-50%
`keccak_f1600_x4_native_aarch64_v84a`	✅	1s	1s	+0%
`mld_ct_abs_i32`	✅	1s	3s	-67%
`poly_uniform_gamma1`	✅	1s	2s	-50%
`shake128x4_squeezeblocks`	✅	1s	2s	-50%

oqs-bot · 2026-04-08T10:01:07Z

CBMC Results (ML-DSA-44)

Full Results (185 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	✅	1994s	2039s	-2.2%
`polyvecl_pointwise_acc_montgomery_c`	✅	322s	336s	-4%
`sign_verify_internal`	✅	228s	213s	+7%
`poly_pointwise_montgomery_c`	✅	152s	148s	+3%
`rej_uniform_native`	✅	140s	139s	+1%
`mld_invntt_layer`	✅	87s	85s	+2%
`mld_ct_memcmp`	✅	74s	72s	+3%
`mld_attempt_signature_generation`	✅	61s	103s	-41%
`mld_ntt_layer`	✅	54s	52s	+4%
`rej_uniform`	✅	21s	20s	+5%
`sign_signature_internal`	✅	20s	19s	+5%
`fqmul`	✅	19s	20s	-5%
`mld_compute_t0_t1_tr_from_sk_components`	✅	18s	11s	+64%
`poly_chknorm_c`	✅	18s	20s	-10%
`polymat_permute_bitrev_to_custom`	✅	18s	26s	-31%
`polyeta_unpack`	✅	17s	16s	+6%
`polyvec_matrix_expand_eager`	✅	17s	-	new
`poly_uniform_4x`	✅	16s	16s	+0%
`polyt0_unpack`	✅	16s	14s	+14%
`poly_uniform_eta_4x`	✅	15s	17s	-12%
`keccakf1600x4_permute_native`	✅	13s	13s	+0%
`mld_ntt_butterfly_block`	✅	13s	11s	+18%
`poly_add`	✅	13s	10s	+30%
`rej_uniform_c`	✅	13s	15s	-13%
`keccak_absorb_once_x4`	✅	11s	9s	+22%
`polyvec_matrix_pointwise_montgomery_eager`	✅	11s	-	new
`keccakf1600_permute`	✅	10s	8s	+25%
`polyvec_matrix_expand_eager_serial`	✅	10s	-	new
`polyz_unpack_c`	✅	10s	12s	-17%
`keccak_absorb`	✅	9s	5s	+80%
`polyveck_ntt`	✅	9s	6s	+50%
`polyvecl_chknorm`	✅	9s	5s	+80%
`sign`	✅	9s	8s	+12%
`mld_check_pct`	✅	8s	12s	-33%
`sign_pk_from_sk`	✅	8s	8s	+0%
`unpack_sk`	✅	8s	7s	+14%
`keccak_squeezeblocks_x4`	✅	7s	6s	+17%
`mld_compute_pack_z`	✅	7s	7s	+0%
`mld_polyvecl_permute_bitrev_to_custom_native`	✅	7s	8s	-12%
`polyveck_chknorm`	✅	7s	3s	+133%
`polyveck_decompose`	✅	7s	6s	+17%
`polyveck_shiftl`	✅	7s	4s	+75%
`polyveck_sub`	✅	7s	4s	+75%
`rej_eta_native`	✅	7s	6s	+17%
`sign_signature`	✅	7s	5s	+40%
`keccakf1600_permute_native`	✅	6s	9s	-33%
`poly_invntt_tomont_c`	✅	6s	6s	+0%
`poly_make_hint`	✅	6s	5s	+20%
`polyveck_power2round`	✅	6s	6s	+0%
`polyveck_reduce`	✅	6s	4s	+50%
`sign_keypair_internal`	✅	6s	4s	+50%
`sign_open`	✅	6s	3s	+100%
`mld_ct_get_optblocker_i64`	✅	5s	2s	+150%
`mld_h`	✅	5s	4s	+25%
`ntt_native_x86_64`	✅	5s	4s	+25%
`poly_challenge`	✅	5s	7s	-29%
`poly_chknorm`	✅	5s	2s	+150%
`poly_decompose`	✅	5s	6s	-17%
`poly_decompose_native`	✅	5s	3s	+67%
`poly_uniform_gamma1_4x`	✅	5s	4s	+25%
`polyveck_add`	✅	5s	8s	-38%
`polyveck_use_hint`	✅	5s	7s	-29%
`polyvecl_uniform_gamma1`	✅	5s	3s	+67%
`polyvecl_unpack_eta`	✅	5s	3s	+67%
`sign_verify_pre_hash_internal`	✅	5s	6s	-17%
`use_hint`	✅	5s	2s	+150%
`caddq`	✅	4s	2s	+100%
`keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid`	✅	4s	1s	+300%
`keccak_finalize`	✅	4s	4s	+0%
`keccakf1600_extract_bytes (big endian)`	✅	4s	5s	-20%
`keccakf1600_xor_bytes`	✅	4s	2s	+100%
`mld_ct_cmask_nonzero_u8`	✅	4s	3s	+33%
`mld_ct_get_optblocker_u32`	✅	4s	2s	+100%
`mld_sample_s1_s2_serial`	✅	4s	4s	+0%
`pack_sig_c`	✅	4s	3s	+33%
`pack_sig_z`	✅	4s	5s	-20%
`pointwise_native_aarch64`	✅	4s	2s	+100%
`pointwise_native_x86_64`	✅	4s	2s	+100%
`poly_caddq_c`	✅	4s	5s	-20%
`poly_chknorm_native`	✅	4s	2s	+100%
`poly_pointwise_montgomery`	✅	4s	3s	+33%
`poly_sub`	✅	4s	3s	+33%
`poly_uniform_eta`	✅	4s	5s	-20%
`poly_uniform_gamma1`	✅	4s	3s	+33%
`poly_use_hint_c`	✅	4s	4s	+0%
`poly_use_hint_native`	✅	4s	6s	-33%
`polyt1_unpack`	✅	4s	5s	-20%
`polyveck_invntt_tomont`	✅	4s	3s	+33%
`polyveck_pointwise_poly_montgomery`	✅	4s	3s	+33%
`polyvecl_ntt`	✅	4s	6s	-33%
`rej_eta_c`	✅	4s	5s	-20%
`shake128x4_squeezeblocks`	✅	4s	2s	+100%
`shake256`	✅	4s	2s	+100%
`shake256_init`	✅	4s	2s	+100%
`shake256x4_absorb_once`	✅	4s	4s	+0%
`sign_keypair`	✅	4s	6s	-33%
`sign_signature_extmu`	✅	4s	6s	-33%
`sign_signature_pre_hash_shake256`	✅	4s	4s	+0%
`sign_verify`	✅	4s	6s	-33%
`sys_check_capability`	✅	4s	4s	+0%
`decompose`	✅	3s	2s	+50%
`intt_native_x86_64`	✅	3s	3s	+0%
`keccak_f1600_x1_native_aarch64_v84a`	✅	3s	3s	+0%
`keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid`	✅	3s	4s	-25%
`keccakf1600x4_permute`	✅	3s	2s	+50%
`mld_ct_cmask_nonzero_u32`	✅	3s	2s	+50%
`mld_keccakf1600_extract_bytes`	✅	3s	3s	+0%
`mld_prepare_domain_separation_prefix`	✅	3s	6s	-50%
`mld_sample_s1_s2`	✅	3s	6s	-50%
`mld_value_barrier_i64`	✅	3s	2s	+50%
`pack_sig_h_poly`	✅	3s	5s	-40%
`pack_sk`	✅	3s	3s	+0%
`poly_caddq`	✅	3s	3s	+0%
`poly_caddq_native`	✅	3s	3s	+0%
`poly_chknorm_native_aarch64`	✅	3s	4s	-25%
`poly_ntt_native`	✅	3s	1s	+200%
`poly_power2round`	✅	3s	2s	+50%
`poly_reduce`	✅	3s	3s	+0%
`poly_shiftl`	✅	3s	3s	+0%
`poly_uniform`	✅	3s	4s	-25%
`polyeta_pack`	✅	3s	2s	+50%
`polyt0_pack`	✅	3s	7s	-57%
`polyt1_pack`	✅	3s	4s	-25%
`polyvec_matrix_pointwise_montgomery_yvec_eager`	✅	3s	-	new
`polyveck_caddq`	✅	3s	3s	+0%
`polyveck_pack_eta`	✅	3s	4s	-25%
`polyveck_pack_t0`	✅	3s	3s	+0%
`polyveck_pack_w1`	✅	3s	3s	+0%
`polyveck_unpack_t0`	✅	3s	3s	+0%
`polyvecl_pack_eta`	✅	3s	4s	-25%
`polyvecl_pointwise_acc_montgomery`	✅	3s	3s	+0%
`polyvecl_uniform_gamma1_serial`	✅	3s	3s	+0%
`polyw1_pack`	✅	3s	2s	+50%
`polyz_pack`	✅	3s	2s	+50%
`polyz_unpack_native`	✅	3s	3s	+0%
`power2round`	✅	3s	5s	-40%
`shake256_finalize`	✅	3s	3s	+0%
`shake256_squeeze`	✅	3s	3s	+0%
`shake256x4_squeezeblocks`	✅	3s	1s	+200%
`sign_verify_extmu`	✅	3s	4s	-25%
`sign_verify_pre_hash_shake256`	✅	3s	3s	+0%
`unpack_hints`	✅	3s	6s	-50%
`unpack_pk`	✅	3s	2s	+50%
`unpack_sig`	✅	3s	3s	+0%
`fqscale`	✅	2s	3s	-33%
`keccak_f1600_x4_native_aarch64_v84a`	✅	2s	2s	+0%
`keccak_init`	✅	2s	2s	+0%
`keccak_squeeze`	✅	2s	6s	-67%
`keccakf1600_xor_bytes (big endian)`	✅	2s	2s	+0%
`keccakf1600x4_extract_bytes`	✅	2s	3s	-33%
`keccakf1600x4_xor_bytes`	✅	2s	3s	-33%
`make_hint`	✅	2s	1s	+100%
`mld_ct_cmask_neg_i32`	✅	2s	1s	+100%
`mld_ct_get_optblocker_u8`	✅	2s	1s	+100%
`mld_ct_sel_int32`	✅	2s	2s	+0%
`mld_value_barrier_u32`	✅	2s	1s	+100%
`montgomery_reduce`	✅	2s	2s	+0%
`ntt_native_aarch64`	✅	2s	5s	-60%
`poly_caddq_native_aarch64`	✅	2s	4s	-50%
`poly_decompose_c`	✅	2s	3s	-33%
`poly_invntt_tomont`	✅	2s	4s	-50%
`poly_invntt_tomont_native`	✅	2s	2s	+0%
`poly_ntt`	✅	2s	2s	+0%
`poly_ntt_c`	✅	2s	2s	+0%
`poly_pointwise_montgomery_native`	✅	2s	3s	-33%
`poly_use_hint`	✅	2s	3s	-33%
`polyveck_unpack_eta`	✅	2s	4s	-50%
`polyvecl_permute_bitrev_to_custom`	✅	2s	5s	-60%
`polyvecl_pointwise_acc_montgomery_native`	✅	2s	4s	-50%
`polyvecl_unpack_z`	✅	2s	3s	-33%
`reduce32`	✅	2s	2s	+0%
`rej_eta`	✅	2s	4s	-50%
`shake128_init`	✅	2s	5s	-60%
`shake128_release`	✅	2s	5s	-60%
`shake128_squeeze`	✅	2s	3s	-33%
`sign_signature_pre_hash_internal`	✅	2s	2s	+0%
`keccak_f1600_x1_native_aarch64`	✅	1s	2s	-50%
`mld_ct_abs_i32`	✅	1s	3s	-67%
`mld_value_barrier_u8`	✅	1s	2s	-50%
`pack_pk`	✅	1s	4s	-75%
`polyz_unpack`	✅	1s	2s	-50%
`shake128_absorb`	✅	1s	3s	-67%
`shake128_finalize`	✅	1s	3s	-67%
`shake128x4_absorb_once`	✅	1s	4s	-75%
`shake256_absorb`	✅	1s	2s	-50%
`shake256_release`	✅	1s	3s	-67%

Replace the row-level matrix buffer (mld_polyvecl) with a single-poly buffer in REDUCE_RAM mode. In the lazy path, matrix elements A[k][l] are sampled on demand one at a time, and the matrix-vector product accumulates element-by-element instead of row-by-row. Restructure polymat into eager/lazy variants following the same pattern as s1hat/s2hat/t0hat: - mld_polymat_eager: stores full K x L matrix - mld_polymat_lazy: stores rho + single poly_buffer + tmp - mld_polyvec_matrix_expand_eager/_lazy: separate implementations - mld_polyvec_matrix_pointwise_montgomery_eager/_lazy: separate implementations with CBMC contracts only on the eager variants Move all polymat-related code from polyvec.h/polyvec.c into polyvec_lazy.h/polyvec_lazy.c. Signed-off-by: Matthias J. Kannwischer <matthias@zerorisc.com>

Add mld_yvec following the lazy polyvec pattern: in REDUCE_RAM mode it stores only the seed and nonce and regenerates y[i] on demand. The fused matrix-vector helper mld_polyvec_matrix_pointwise_montgomery_yvec samples y column-by-column in REDUCE_RAM mode. CBMC proofs updated for the eager path; lazy variant has no proof yet. Signed-off-by: Matthias J. Kannwischer <matthias@zerorisc.com>

Signed-off-by: Matthias J. Kannwischer <matthias@zerorisc.com>

mkannwischer added the low-ram label Apr 8, 2026

mkannwischer force-pushed the lowram-sign-eliminate-y branch 2 times, most recently from de53526 to 4b9fcd1 Compare April 9, 2026 01:43

mkannwischer added 3 commits April 9, 2026 10:28

CBMC: Restore polyvecl_pointwise_acc_montgomery_c proof flags

cef9d32

Signed-off-by: Matthias J. Kannwischer <matthias@zerorisc.com>

mkannwischer force-pushed the lowram-sign-eliminate-y branch from 4b9fcd1 to cef9d32 Compare April 9, 2026 02:29

mkannwischer marked this pull request as ready for review April 9, 2026 03:05

mkannwischer requested a review from a team as a code owner April 9, 2026 03:05

mkannwischer assigned hanno-becker Apr 9, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

lowram: Eliminate y vector in REDUCE_RAM mode in sign#1031

lowram: Eliminate y vector in REDUCE_RAM mode in sign#1031
mkannwischer wants to merge 3 commits intomainfrom
lowram-sign-eliminate-y

mkannwischer commented Apr 8, 2026

Uh oh!

oqs-bot commented Apr 8, 2026 •

edited

Loading

Uh oh!

oqs-bot commented Apr 8, 2026 •

edited

Loading

Uh oh!

oqs-bot commented Apr 8, 2026 •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

mkannwischer commented Apr 8, 2026

Uh oh!

oqs-bot commented Apr 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CBMC Results (ML-DSA-65)

Uh oh!

oqs-bot commented Apr 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CBMC Results (ML-DSA-87)

Uh oh!

oqs-bot commented Apr 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CBMC Results (ML-DSA-44)

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

oqs-bot commented Apr 8, 2026 •

edited

Loading

oqs-bot commented Apr 8, 2026 •

edited

Loading

oqs-bot commented Apr 8, 2026 •

edited

Loading