Skip to content

Lowram: Share buffers with non-overlapping lifetimes in keygen#1011

Open
mkannwischer wants to merge 1 commit intomainfrom
keygen-buffer-sharing
Open

Lowram: Share buffers with non-overlapping lifetimes in keygen#1011
mkannwischer wants to merge 1 commit intomainfrom
keygen-buffer-sharing

Conversation

@mkannwischer
Copy link
Copy Markdown
Contributor

Reuse t0 as the accumulator in mld_compute_t0_t1_tr_from_sk_components, and have the caller provide s1 already in NTT form, removing two allocations (s1hat and t) from the helper.

In mld_sign_keypair_internal, share the s1 and t1 buffers via a union since s1hat is consumed before t1 is produced. Pack s1 into the secret key before the in-place NTT so the original coefficients are preserved.

Split mld_pack_sk into mld_pack_sk_s1 and mld_pack_sk_rho_key_tr_s2_t0 to support packing s1 independently before the NTT.

@mkannwischer mkannwischer force-pushed the keygen-buffer-sharing branch from 0b96d02 to 0db7b80 Compare April 2, 2026 04:00
@oqs-bot
Copy link
Copy Markdown
Contributor

oqs-bot commented Apr 2, 2026

CBMC Results (ML-DSA-44)

Full Results (180 proofs)
Proof Status Current Previous Change
**TOTAL** 1951s 2072s -5.8%
mld_attempt_signature_generation 231s 253s -9%
polyvecl_pointwise_acc_montgomery_c 211s 216s -2%
poly_pointwise_montgomery_c 154s 155s -1%
sign_verify_internal 146s 184s -21%
rej_uniform_native 144s 145s -1%
mld_invntt_layer 81s 88s -8%
mld_ct_memcmp 72s 78s -8%
mld_ntt_layer 37s 56s -34%
polyvec_matrix_expand 22s 19s +16%
rej_uniform 22s 21s +5%
fqmul 19s 21s -10%
poly_chknorm_c 18s 20s -10%
polyeta_unpack 18s 16s +12%
sign_signature_internal 18s 18s +0%
poly_uniform_eta_4x 16s 18s -11%
rej_uniform_c 16s 16s +0%
mld_compute_t0_t1_tr_from_sk_components 15s 15s +0%
poly_uniform_4x 15s 17s -12%
polymat_permute_bitrev_to_custom 14s 17s -18%
keccakf1600x4_permute_native 13s 14s -7%
polyt0_unpack 13s 14s -7%
mld_ntt_butterfly_block 12s 12s +0%
polyvec_matrix_expand_serial 12s 16s -25%
polyz_unpack_c 12s 12s +0%
mld_check_pct 11s 6s +83%
polyvec_matrix_pointwise_montgomery 11s 12s -8%
polyveck_decompose 11s 4s +175%
keccak_absorb_once_x4 10s 12s -17%
poly_add 9s 12s -25%
sign 9s 11s -18%
sign_keypair_internal 9s 6s +50%
keccakf1600_permute 8s 8s +0%
keccakf1600_permute_native 8s 8s +0%
keccak_squeezeblocks_x4 7s 7s +0%
mld_compute_pack_z 7s 4s +75%
mld_polyvecl_permute_bitrev_to_custom_native 7s 10s -30%
polyveck_chknorm 7s 4s +75%
sign_open 7s 4s +75%
sign_pk_from_sk 7s 5s +40%
sign_verify 7s 6s +17%
keccak_absorb 6s 7s -14%
mld_sample_s1_s2 6s 7s -14%
poly_caddq_c 6s 4s +50%
poly_invntt_tomont_c 6s 6s +0%
poly_power2round 6s 4s +50%
polyveck_add 6s 9s -33%
polyveck_caddq 6s 6s +0%
polyveck_power2round 6s 13s -54%
polyveck_reduce 6s 5s +20%
polyveck_shiftl 6s 6s +0%
polyvecl_chknorm 6s 3s +100%
unpack_hints 6s 6s +0%
intt_native_x86_64 5s 3s +67%
poly_challenge 5s 5s +0%
poly_pointwise_montgomery 5s 5s +0%
poly_shiftl 5s 3s +67%
poly_sub 5s 3s +67%
poly_uniform 5s 4s +25%
poly_uniform_eta 5s 4s +25%
polyveck_ntt 5s 5s +0%
polyveck_pack_w1 5s 3s +67%
polyveck_sub 5s 6s -17%
polyveck_unpack_eta 5s 6s -17%
polyveck_use_hint 5s 7s -29%
polyvecl_ntt 5s 5s +0%
polyvecl_uniform_gamma1_serial 5s 3s +67%
polyw1_pack 5s 4s +25%
polyz_pack 5s 4s +25%
rej_eta_native 5s 5s +0%
shake128_absorb 5s 1s +400%
shake256_finalize 5s 3s +67%
sign_signature 5s 5s +0%
sign_signature_pre_hash_internal 5s 7s -29%
unpack_sk 5s 8s -38%
keccakf1600_extract_bytes (big endian) 4s 3s +33%
keccakf1600_xor_bytes 4s 1s +300%
mld_ct_abs_i32 4s 2s +100%
mld_ct_cmask_nonzero_u32 4s 3s +33%
mld_ct_cmask_nonzero_u8 4s 3s +33%
mld_h 4s 2s +100%
mld_sample_s1_s2_serial 4s 5s -20%
montgomery_reduce 4s 3s +33%
pack_pk 4s 3s +33%
poly_caddq 4s 5s -20%
poly_chknorm_native 4s 5s -20%
poly_decompose_native 4s 4s +0%
poly_invntt_tomont_native 4s 4s +0%
poly_uniform_gamma1_4x 4s 5s -20%
poly_use_hint_c 4s 3s +33%
poly_use_hint_native 4s 2s +100%
polyt0_pack 4s 5s -20%
polyveck_make_hint 4s 3s +33%
polyveck_pointwise_poly_montgomery 4s 5s -20%
polyveck_pointwise_poly_montgomery_s2 4s 4s +0%
polyveck_pointwise_poly_montgomery_t0 4s 4s +0%
polyvecl_pack_eta 4s 4s +0%
rej_eta 4s 3s +33%
sign_signature_extmu 4s 4s +0%
sign_signature_pre_hash_shake256 4s 7s -43%
sign_verify_pre_hash_shake256 4s 5s -20%
caddq 3s 5s -40%
fqscale 3s 2s +50%
mld_ct_get_optblocker_i64 3s 5s -40%
mld_ct_get_optblocker_u32 3s 2s +50%
mld_ct_sel_int32 3s 4s -25%
mld_prepare_domain_separation_prefix 3s 2s +50%
mld_value_barrier_u32 3s 3s +0%
mld_value_barrier_u8 3s 3s +0%
ntt_native_aarch64 3s 4s -25%
ntt_native_x86_64 3s 3s +0%
pack_sig_c_h 3s 5s -40%
pack_sk_s1 3s - new
poly_caddq_native 3s 6s -50%
poly_caddq_native_aarch64 3s 3s +0%
poly_chknorm 3s 3s +0%
poly_chknorm_native_aarch64 3s 4s -25%
poly_decompose 3s 5s -40%
poly_make_hint 3s 2s +50%
poly_ntt 3s 5s -40%
poly_pointwise_montgomery_native 3s 2s +50%
poly_reduce 3s 3s +0%
poly_use_hint 3s 2s +50%
polyeta_pack 3s 3s +0%
polyt1_unpack 3s 3s +0%
polyveck_invntt_tomont 3s 4s -25%
polyveck_pack_t0 3s 4s -25%
polyvecl_pointwise_acc_montgomery_native 3s 4s -25%
polyvecl_uniform_gamma1 3s 2s +50%
polyvecl_unpack_eta 3s 2s +50%
polyz_unpack_native 3s 3s +0%
power2round 3s 3s +0%
reduce32 3s 3s +0%
shake128_finalize 3s 3s +0%
shake256 3s 2s +50%
shake256_squeeze 3s 3s +0%
shake256x4_absorb_once 3s 2s +50%
shake256x4_squeezeblocks 3s 4s -25%
sign_keypair 3s 4s -25%
sign_verify_extmu 3s 6s -50%
sys_check_capability 3s 3s +0%
unpack_pk 3s 3s +0%
use_hint 3s 4s -25%
decompose 2s 2s +0%
keccak_finalize 2s 2s +0%
keccak_init 2s 1s +100%
keccak_squeeze 2s 3s -33%
keccakf1600_xor_bytes (big endian) 2s 3s -33%
keccakf1600x4_permute 2s 3s -33%
keccakf1600x4_xor_bytes 2s 2s +0%
make_hint 2s 2s +0%
mld_ct_cmask_neg_i32 2s 2s +0%
mld_ct_get_optblocker_u8 2s 2s +0%
mld_value_barrier_i64 2s 2s +0%
pack_sig_z 2s 2s +0%
pack_sk_rho_key_tr_s2_t0 2s - new
poly_decompose_c 2s 2s +0%
poly_ntt_c 2s 4s -50%
poly_ntt_native 2s 4s -50%
poly_uniform_gamma1 2s 2s +0%
polyt1_pack 2s 3s -33%
polyveck_pack_eta 2s 5s -60%
polyveck_unpack_t0 2s 4s -50%
polyvecl_permute_bitrev_to_custom 2s 4s -50%
polyvecl_pointwise_acc_montgomery 2s 2s +0%
polyvecl_unpack_z 2s 3s -33%
polyz_unpack 2s 3s -33%
rej_eta_c 2s 3s -33%
shake128_init 2s 2s +0%
shake128_release 2s 4s -50%
shake128_squeeze 2s 2s +0%
shake128x4_absorb_once 2s 2s +0%
shake128x4_squeezeblocks 2s 5s -60%
shake256_absorb 2s 2s +0%
shake256_init 2s 1s +100%
shake256_release 2s 2s +0%
sign_verify_pre_hash_internal 2s 5s -60%
unpack_sig 2s 2s +0%
keccakf1600x4_extract_bytes 1s 2s -50%
mld_keccakf1600_extract_bytes 1s 2s -50%
poly_invntt_tomont 1s 2s -50%

@oqs-bot
Copy link
Copy Markdown
Contributor

oqs-bot commented Apr 2, 2026

CBMC Results (ML-DSA-87)

Full Results (180 proofs)
Proof Status Current Previous Change
**TOTAL** 2613s 2668s -2.1%
polyvecl_pointwise_acc_montgomery_c 289s 311s -7%
mld_attempt_signature_generation 270s 239s +13%
sign_verify_internal 252s 216s +17%
polyvec_matrix_expand 197s 195s +1%
poly_pointwise_montgomery_c 165s 176s -6%
rej_uniform_native 149s 158s -6%
polyvec_matrix_expand_serial 83s 85s -2%
mld_ct_memcmp 82s 83s -1%
mld_invntt_layer 72s 104s -31%
mld_ntt_layer 55s 56s -2%
polymat_permute_bitrev_to_custom 46s 47s -2%
sign_signature_internal 37s 42s -12%
mld_compute_t0_t1_tr_from_sk_components 22s 25s -12%
fqmul 21s 21s +0%
rej_uniform 20s 24s -17%
poly_chknorm_c 18s 24s -25%
polyeta_unpack 17s 20s -15%
mld_ntt_butterfly_block 16s 13s +23%
rej_uniform_c 16s 16s +0%
poly_uniform_eta_4x 15s 17s -12%
keccakf1600x4_permute_native 14s 15s -7%
poly_uniform_4x 14s 16s -12%
polyt0_unpack 14s 15s -7%
polyveck_decompose 14s 60s -77%
polyveck_power2round 14s 8s +75%
polyz_unpack_c 13s 8s +62%
mld_polyvecl_permute_bitrev_to_custom_native 12s 13s -8%
polyveck_add 12s 9s +33%
sign_pk_from_sk 11s 9s +22%
mld_check_pct 10s 9s +11%
poly_add 10s 12s -17%
poly_invntt_tomont_c 10s 9s +11%
polyvec_matrix_pointwise_montgomery 10s 11s -9%
unpack_sk 10s 6s +67%
keccak_absorb_once_x4 9s 9s +0%
keccakf1600_permute 9s 7s +29%
keccakf1600_permute_native 9s 7s +29%
mld_compute_pack_z 9s 6s +50%
poly_decompose_c 9s 8s +12%
polyveck_reduce 9s 10s -10%
polyveck_use_hint 9s 9s +0%
polyvecl_ntt 9s 8s +12%
keccak_absorb 8s 7s +14%
mld_sample_s1_s2_serial 8s 6s +33%
poly_power2round 8s 8s +0%
polyveck_caddq 8s 11s -27%
polyveck_ntt 8s 11s -27%
polyveck_sub 8s 6s +33%
sign_keypair_internal 8s 5s +60%
poly_ntt_c 7s 4s +75%
polyveck_invntt_tomont 7s 8s -12%
polyveck_pointwise_poly_montgomery 7s 5s +40%
mld_sample_s1_s2 6s 5s +20%
pack_sig_c_h 6s 4s +50%
poly_decompose 6s 2s +200%
poly_decompose_native 6s 5s +20%
poly_ntt_native 6s 4s +50%
poly_uniform_eta 6s 4s +50%
polyveck_chknorm 6s 4s +50%
polyveck_pointwise_poly_montgomery_s2 6s 5s +20%
polyveck_shiftl 6s 7s -14%
polyveck_unpack_eta 6s 4s +50%
sign 6s 7s -14%
sign_signature_extmu 6s 6s +0%
sign_verify_pre_hash_internal 6s 2s +200%
fqscale 5s 7s -29%
intt_native_x86_64 5s 3s +67%
keccak_squeeze 5s 2s +150%
keccak_squeezeblocks_x4 5s 7s -29%
mld_ct_get_optblocker_u8 5s 2s +150%
mld_prepare_domain_separation_prefix 5s 4s +25%
poly_chknorm_native 5s 3s +67%
poly_shiftl 5s 5s +0%
polyveck_make_hint 5s 5s +0%
polyveck_pack_eta 5s 4s +25%
polyveck_pack_w1 5s 4s +25%
polyveck_pointwise_poly_montgomery_t0 5s 9s -44%
polyvecl_uniform_gamma1 5s 4s +25%
polyvecl_uniform_gamma1_serial 5s 4s +25%
sign_open 5s 2s +150%
sign_signature 5s 2s +150%
sign_signature_pre_hash_internal 5s 4s +25%
unpack_hints 5s 6s -17%
keccakf1600x4_extract_bytes 4s 5s -20%
keccakf1600x4_xor_bytes 4s 3s +33%
mld_ct_cmask_nonzero_u8 4s 2s +100%
ntt_native_x86_64 4s 3s +33%
pack_pk 4s 3s +33%
pack_sig_z 4s 4s +0%
poly_caddq_c 4s 6s -33%
poly_caddq_native 4s 3s +33%
poly_challenge 4s 5s -20%
poly_make_hint 4s 3s +33%
poly_pointwise_montgomery_native 4s 4s +0%
poly_reduce 4s 3s +33%
poly_uniform 4s 4s +0%
poly_uniform_gamma1 4s 5s -20%
poly_use_hint_c 4s 2s +100%
polyt1_unpack 4s 3s +33%
polyveck_pack_t0 4s 3s +33%
polyveck_unpack_t0 4s 7s -43%
polyvecl_permute_bitrev_to_custom 4s 3s +33%
polyvecl_pointwise_acc_montgomery 4s 4s +0%
polyvecl_pointwise_acc_montgomery_native 4s 5s -20%
polyvecl_unpack_eta 4s 6s -33%
polyvecl_unpack_z 4s 3s +33%
power2round 4s 4s +0%
shake256_init 4s 1s +300%
shake256_release 4s 4s +0%
shake256x4_squeezeblocks 4s 4s +0%
sign_signature_pre_hash_shake256 4s 4s +0%
sign_verify 4s 5s -20%
sign_verify_pre_hash_shake256 4s 6s -33%
caddq 3s 4s -25%
decompose 3s 3s +0%
keccak_finalize 3s 3s +0%
keccak_init 3s 2s +50%
keccakf1600_xor_bytes 3s 4s -25%
keccakf1600x4_permute 3s 4s -25%
make_hint 3s 4s -25%
mld_ct_abs_i32 3s 2s +50%
mld_ct_cmask_neg_i32 3s 1s +200%
mld_ct_cmask_nonzero_u32 3s 3s +0%
mld_ct_get_optblocker_u32 3s 2s +50%
mld_ct_sel_int32 3s 4s -25%
mld_value_barrier_i64 3s 2s +50%
poly_caddq 3s 4s -25%
poly_ntt 3s 2s +50%
poly_pointwise_montgomery 3s 6s -50%
poly_use_hint 3s 5s -40%
poly_use_hint_native 3s 4s -25%
polyeta_pack 3s 4s -25%
polyt0_pack 3s 3s +0%
polyvecl_chknorm 3s 4s -25%
polyvecl_pack_eta 3s 4s -25%
polyw1_pack 3s 2s +50%
polyz_unpack_native 3s 5s -40%
rej_eta_c 3s 4s -25%
rej_eta_native 3s 4s -25%
shake128_finalize 3s 2s +50%
shake256_finalize 3s 2s +50%
shake256x4_absorb_once 3s 2s +50%
sign_keypair 3s 7s -57%
sign_verify_extmu 3s 2s +50%
unpack_sig 3s 4s -25%
use_hint 3s 2s +50%
keccakf1600_extract_bytes (big endian) 2s 2s +0%
keccakf1600_xor_bytes (big endian) 2s 3s -33%
mld_ct_get_optblocker_i64 2s 4s -50%
mld_h 2s 5s -60%
mld_value_barrier_u32 2s 2s +0%
mld_value_barrier_u8 2s 2s +0%
montgomery_reduce 2s 2s +0%
ntt_native_aarch64 2s 4s -50%
pack_sk_rho_key_tr_s2_t0 2s - new
pack_sk_s1 2s - new
poly_caddq_native_aarch64 2s 2s +0%
poly_chknorm 2s 2s +0%
poly_chknorm_native_aarch64 2s 6s -67%
poly_invntt_tomont 2s 4s -50%
poly_invntt_tomont_native 2s 2s +0%
poly_sub 2s 3s -33%
poly_uniform_gamma1_4x 2s 4s -50%
polyt1_pack 2s 4s -50%
polyz_pack 2s 3s -33%
polyz_unpack 2s 2s +0%
reduce32 2s 4s -50%
rej_eta 2s 4s -50%
shake128_absorb 2s 3s -33%
shake128_init 2s 1s +100%
shake128_release 2s 3s -33%
shake128_squeeze 2s 4s -50%
shake128x4_absorb_once 2s 3s -33%
shake128x4_squeezeblocks 2s 2s +0%
shake256 2s 2s +0%
shake256_absorb 2s 3s -33%
shake256_squeeze 2s 2s +0%
sys_check_capability 2s 3s -33%
unpack_pk 2s 4s -50%
mld_keccakf1600_extract_bytes 1s 2s -50%

@oqs-bot
Copy link
Copy Markdown
Contributor

oqs-bot commented Apr 2, 2026

CBMC Results (ML-DSA-65)

Full Results (180 proofs)
Proof Status Current Previous Change
**TOTAL** 2396s 2509s -4.5%
sign_verify_internal 290s 346s -16%
mld_attempt_signature_generation 258s 285s -9%
polyvecl_pointwise_acc_montgomery_c 203s 191s +6%
poly_pointwise_montgomery_c 168s 166s +1%
rej_uniform_native 156s 149s +5%
polyvec_matrix_expand 113s 126s -10%
mld_ct_memcmp 80s 80s +0%
polyvec_matrix_expand_serial 73s 72s +1%
mld_invntt_layer 72s 98s -27%
mld_ntt_layer 59s 57s +4%
sign_signature_internal 24s 27s -11%
poly_chknorm_c 23s 24s -4%
rej_uniform 22s 20s +10%
fqmul 21s 23s -9%
poly_uniform_eta_4x 19s 17s +12%
polyt0_unpack 18s 13s +38%
polymat_permute_bitrev_to_custom 17s 31s -45%
rej_uniform_c 17s 17s +0%
mld_compute_t0_t1_tr_from_sk_components 15s 27s -44%
mld_ntt_butterfly_block 15s 12s +25%
poly_uniform_4x 15s 16s -6%
polyveck_decompose 15s 13s +15%
polyveck_power2round 15s 12s +25%
keccakf1600x4_permute_native 14s 13s +8%
polyveck_ntt 12s 10s +20%
polyvecl_chknorm 12s 12s +0%
poly_add 11s 11s +0%
polyvec_matrix_pointwise_montgomery 11s 10s +10%
polyveck_sub 11s 13s -15%
keccak_absorb_once_x4 10s 11s -9%
sign 10s 6s +67%
poly_invntt_tomont_c 9s 7s +29%
polyveck_add 9s 9s +0%
polyveck_invntt_tomont 9s 8s +12%
keccakf1600_permute 8s 9s -11%
keccakf1600_permute_native 8s 8s +0%
mld_polyvecl_permute_bitrev_to_custom_native 8s 8s +0%
poly_power2round 8s 6s +33%
polyveck_shiftl 8s 6s +33%
polyvecl_unpack_eta 8s 3s +167%
keccak_squeezeblocks_x4 7s 7s +0%
mld_check_pct 7s 8s -12%
poly_challenge 7s 5s +40%
poly_decompose_c 7s 8s -12%
polyveck_caddq 7s 8s -12%
polyveck_pointwise_poly_montgomery_s2 7s 7s +0%
polyveck_pointwise_poly_montgomery_t0 7s 8s -12%
sign_pk_from_sk 7s 8s -12%
sign_signature 7s 7s +0%
unpack_hints 7s 4s +75%
unpack_sk 7s 8s -12%
keccak_absorb 6s 11s -45%
mld_compute_pack_z 6s 8s -25%
poly_caddq 6s 5s +20%
poly_caddq_c 6s 4s +50%
poly_reduce 6s 3s +100%
poly_uniform_gamma1_4x 6s 5s +20%
polyeta_unpack 6s 5s +20%
polyveck_pointwise_poly_montgomery 6s 8s -25%
polyveck_use_hint 6s 11s -45%
polyvecl_ntt 6s 8s -25%
sign_keypair 6s 7s -14%
sign_keypair_internal 6s 7s -14%
sign_verify 6s 4s +50%
mld_prepare_domain_separation_prefix 5s 3s +67%
mld_sample_s1_s2_serial 5s 7s -29%
pack_pk 5s 2s +150%
poly_caddq_native 5s 4s +25%
poly_ntt_c 5s 6s -17%
poly_pointwise_montgomery_native 5s 3s +67%
poly_use_hint_c 5s 5s +0%
poly_use_hint_native 5s 5s +0%
polyt0_pack 5s 5s +0%
polyveck_make_hint 5s 4s +25%
polyveck_reduce 5s 11s -55%
shake256x4_absorb_once 5s 2s +150%
sign_open 5s 2s +150%
sign_signature_pre_hash_internal 5s 3s +67%
sign_verify_pre_hash_internal 5s 4s +25%
sys_check_capability 5s 3s +67%
caddq 4s 5s -20%
decompose 4s 2s +100%
mld_sample_s1_s2 4s 5s -20%
pack_sk_rho_key_tr_s2_t0 4s - new
poly_invntt_tomont_native 4s 3s +33%
poly_make_hint 4s 1s +300%
poly_pointwise_montgomery 4s 4s +0%
poly_shiftl 4s 2s +100%
poly_uniform_gamma1 4s 3s +33%
poly_use_hint 4s 2s +100%
polyveck_chknorm 4s 3s +33%
polyveck_pack_eta 4s 2s +100%
polyveck_pack_w1 4s 5s -20%
polyveck_unpack_t0 4s 3s +33%
polyvecl_pack_eta 4s 2s +100%
polyvecl_pointwise_acc_montgomery_native 4s 6s -33%
polyvecl_uniform_gamma1_serial 4s 2s +100%
polyw1_pack 4s 3s +33%
polyz_unpack_native 4s 2s +100%
rej_eta_c 4s 4s +0%
rej_eta_native 4s 5s -20%
shake128_finalize 4s 2s +100%
shake256 4s 3s +33%
shake256x4_squeezeblocks 4s 4s +0%
sign_verify_extmu 4s 3s +33%
sign_verify_pre_hash_shake256 4s 5s -20%
intt_native_x86_64 3s 4s -25%
keccak_finalize 3s 2s +50%
keccakf1600_extract_bytes (big endian) 3s 3s +0%
keccakf1600_xor_bytes 3s 3s +0%
keccakf1600x4_permute 3s 1s +200%
mld_ct_abs_i32 3s 2s +50%
mld_ct_get_optblocker_i64 3s 4s -25%
mld_ct_get_optblocker_u8 3s 2s +50%
mld_h 3s 5s -40%
mld_keccakf1600_extract_bytes 3s 5s -40%
mld_value_barrier_u8 3s 1s +200%
ntt_native_aarch64 3s 2s +50%
ntt_native_x86_64 3s 5s -40%
pack_sig_c_h 3s 4s -25%
pack_sig_z 3s 2s +50%
poly_caddq_native_aarch64 3s 4s -25%
poly_chknorm 3s 3s +0%
poly_chknorm_native 3s 3s +0%
poly_decompose_native 3s 5s -40%
poly_ntt_native 3s 3s +0%
poly_uniform_eta 3s 4s -25%
polyt1_pack 3s 3s +0%
polyt1_unpack 3s 3s +0%
polyveck_unpack_eta 3s 5s -40%
polyvecl_pointwise_acc_montgomery 3s 3s +0%
polyvecl_unpack_z 3s 5s -40%
polyz_unpack 3s 2s +50%
polyz_unpack_c 3s 3s +0%
power2round 3s 3s +0%
rej_eta 3s 6s -50%
shake128_absorb 3s 3s +0%
shake128_release 3s 2s +50%
shake128x4_absorb_once 3s 2s +50%
shake128x4_squeezeblocks 3s 2s +50%
shake256_init 3s 4s -25%
sign_signature_extmu 3s 3s +0%
unpack_pk 3s 5s -40%
unpack_sig 3s 4s -25%
fqscale 2s 2s +0%
keccak_init 2s 2s +0%
keccak_squeeze 2s 5s -60%
keccakf1600x4_extract_bytes 2s 4s -50%
keccakf1600x4_xor_bytes 2s 3s -33%
make_hint 2s 4s -50%
mld_ct_cmask_nonzero_u32 2s 4s -50%
mld_ct_cmask_nonzero_u8 2s 2s +0%
mld_ct_get_optblocker_u32 2s 3s -33%
mld_ct_sel_int32 2s 4s -50%
mld_value_barrier_i64 2s 1s +100%
mld_value_barrier_u32 2s 2s +0%
montgomery_reduce 2s 2s +0%
pack_sk_s1 2s - new
poly_chknorm_native_aarch64 2s 4s -50%
poly_decompose 2s 3s -33%
poly_invntt_tomont 2s 2s +0%
poly_ntt 2s 2s +0%
poly_uniform 2s 3s -33%
polyeta_pack 2s 2s +0%
polyveck_pack_t0 2s 4s -50%
polyvecl_permute_bitrev_to_custom 2s 3s -33%
polyvecl_uniform_gamma1 2s 3s -33%
polyz_pack 2s 3s -33%
reduce32 2s 3s -33%
shake128_init 2s 2s +0%
shake128_squeeze 2s 1s +100%
shake256_absorb 2s 3s -33%
shake256_finalize 2s 1s +100%
shake256_release 2s 1s +100%
shake256_squeeze 2s 3s -33%
sign_signature_pre_hash_shake256 2s 3s -33%
use_hint 2s 4s -50%
keccakf1600_xor_bytes (big endian) 1s 2s -50%
mld_ct_cmask_neg_i32 1s 2s -50%
poly_sub 1s 2s -50%

@mkannwischer mkannwischer marked this pull request as ready for review April 2, 2026 04:31
@mkannwischer mkannwischer requested a review from a team as a code owner April 2, 2026 04:31
@mkannwischer mkannwischer force-pushed the keygen-buffer-sharing branch from 0db7b80 to 7a7dd3c Compare April 3, 2026 13:02
Reuse t0 as the accumulator in mld_compute_t0_t1_tr_from_sk_components,
and have the caller provide s1 already in NTT form, removing two
allocations (s1hat and t) from the helper.

In mld_sign_keypair_internal, share the s1 and t1 buffers via a union
since s1hat is consumed before t1 is produced. Pack s1 into the secret
key before the in-place NTT so the original coefficients are preserved.

Split mld_pack_sk into mld_pack_sk_s1 and mld_pack_sk_rho_key_tr_s2_t0
to support packing s1 independently before the NTT.

Signed-off-by: Matthias J. Kannwischer <matthias@zerorisc.com>
@mkannwischer mkannwischer force-pushed the keygen-buffer-sharing branch from 7a7dd3c to 93856fc Compare April 3, 2026 13:49
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants