Skip to content

refactor riscv-v implementation. #1069

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Cyan4973 merged 1 commit into from
Sep 9, 2025
+28 −42
Merged

refactor riscv-v implementation. #1069

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
70 changes: 28 additions & 42 deletions xxhash.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5894,68 +5894,54 @@ XXH3_scrambleAcc_rvv(void* XXH_RESTRICT acc, const void* XXH_RESTRICT secret)
// Try to set vector lenght to 512 bits.
// If this length is unavailable, then maximum available will be used
size_t vl = RVV_OP(vsetvl_e64m2)(8);
uint64_t* const xacc = (uint64_t*) acc;
const uint64_t* const xsecret = (const uint64_t*) secret;

uint64_t prime[16] = {XXH_PRIME32_1, XXH_PRIME32_1, XXH_PRIME32_1, XXH_PRIME32_1, XXH_PRIME32_1, XXH_PRIME32_1, XXH_PRIME32_1, XXH_PRIME32_1,\
XXH_PRIME32_1, XXH_PRIME32_1, XXH_PRIME32_1, XXH_PRIME32_1, XXH_PRIME32_1, XXH_PRIME32_1, XXH_PRIME32_1, XXH_PRIME32_1};
vuint64m2_t vprime = RVV_OP(vle64_v_u64m2)(prime, vl);
uint64_t* xacc = (uint64_t*) acc;
const uint64_t* xsecret = (const uint64_t*) secret;

// vuint64m2_t is sizeless.
// But we can assume that vl can be only 4(vlen=128) or 8(vlen=256,512)
for(size_t i = 0; i < XXH_STRIPE_LEN/(8 * vl); i++){
size_t i = 0;
for (; i < XXH_STRIPE_LEN / 8; i += vl) {
/* xacc[i] ^= (xacc[i] >> 47) */
vuint64m2_t acc_vec = RVV_OP(vle64_v_u64m2)(xacc + vl * i, vl);
vuint64m2_t acc_vec = RVV_OP(vle64_v_u64m2)(xacc + i, vl);
vuint64m2_t shifted = RVV_OP(vsrl_vx_u64m2)(acc_vec, 47, vl);
vuint64m2_t data_vec = RVV_OP(vxor_vv_u64m2)(acc_vec, shifted, vl);
vuint64m2_t key_vec = RVV_OP(vreinterpret_v_u8m2_u64m2)(RVV_OP(vle8_v_u8m2)((const uint8_t*)(xsecret + i), vl * 8));
acc_vec = RVV_OP(vxor_vv_u64m2)(acc_vec, shifted, vl);
/* xacc[i] ^= xsecret[i]; */
vuint64m2_t key_vec = RVV_OP(vreinterpret_v_u8m2_u64m2)(RVV_OP(vle8_v_u8m2)((const uint8_t*)(xsecret + vl * i), vl * 8));
vuint64m2_t data_key = RVV_OP(vxor_vv_u64m2)(data_vec, key_vec, vl);
acc_vec = RVV_OP(vxor_vv_u64m2)(acc_vec, key_vec, vl);

/* xacc[i] *= XXH_PRIME32_1; */
vuint64m2_t prod_even = RVV_OP(vmul_vv_u64m2)(RVV_OP(vand_vx_u64m2)(data_key, 0xffffffff, vl), vprime, vl);
vuint64m2_t prod_odd = RVV_OP(vmul_vv_u64m2)(RVV_OP(vsrl_vx_u64m2)(data_key, 32, vl), vprime, vl);
vuint64m2_t prod = RVV_OP(vadd_vv_u64m2)(prod_even, RVV_OP(vsll_vx_u64m2)(prod_odd, 32, vl), vl);
RVV_OP(vse64_v_u64m2)(xacc + vl * i, prod, vl);
acc_vec = RVV_OP(vmul_vx_u64m2)(acc_vec, (xxh_u64)XXH_PRIME32_1, vl);
RVV_OP(vse64_v_u64m2)(xacc + i, acc_vec, vl);
}
}
}

XXH_FORCE_INLINE void
XXH3_initCustomSecret_rvv(void* XXH_RESTRICT customSecret, xxh_u64 seed64)
{
XXH_STATIC_ASSERT((XXH_SECRET_DEFAULT_SIZE & 63) == 0);
XXH_STATIC_ASSERT((XXH_SECRET_DEFAULT_SIZE & 15) == 0);
XXH_STATIC_ASSERT(XXH_SEC_ALIGN == 64);
XXH_ASSERT(((size_t)customSecret & 63) == 0);
{
uint64_t* const xcustomSecret = (uint64_t*)customSecret;
const uint8_t* kSecretPtr = XXH3_kSecret;
size_t XXH3_kSecret_64b_len = XXH_SECRET_DEFAULT_SIZE / 16;
size_t i = 0;
for(;i < XXH3_kSecret_64b_len;) {
size_t vl = RVV_OP(vsetvl_e64m2)(XXH3_kSecret_64b_len - i);
{
vuint64m2_t vlo = RVV_OP(vlse64_v_u64m2)((const uint64_t*)(const void*)kSecretPtr, 16, vl);
vuint64m2_t vhi = RVV_OP(vlse64_v_u64m2)((const uint64_t*)(const void*)(kSecretPtr + 8), 16, vl);

(void)(&XXH_writeLE64);
{
// Calculate the number of 64-bit elements in the `XXH3_kSecret` secret
size_t XXH3_kSecret_64b_len = XXH_SECRET_DEFAULT_SIZE / 8;
// Create an array of repeated seed values, alternating between seed64 and -seed64.
uint64_t seed_pos[16] = {seed64, (uint64_t)(-(int64_t)seed64), \
seed64, (uint64_t)(-(int64_t)seed64), \
seed64, (uint64_t)(-(int64_t)seed64), \
seed64, (uint64_t)(-(int64_t)seed64), \
seed64, (uint64_t)(-(int64_t)seed64), \
seed64, (uint64_t)(-(int64_t)seed64), \
seed64, (uint64_t)(-(int64_t)seed64), \
seed64, (uint64_t)(-(int64_t)seed64)};
// Cast the default secret to a signed 64-bit pointer for vectorized access
const int64_t* const xXXH3_kSecret = (const int64_t*)((const void*)XXH3_kSecret);
size_t vl = 0;
for (size_t i=0; i < XXH3_kSecret_64b_len; i += vl) {

vl = RVV_OP(vsetvl_e64m2)(XXH3_kSecret_64b_len - i);
{
vint64m2_t seed = RVV_OP(vle64_v_i64m2)((int64_t*)seed_pos, vl);
vint64m2_t src = RVV_OP(vle64_v_i64m2)((const int64_t*)&xXXH3_kSecret[i], vl);
vint64m2_t res = RVV_OP(vadd_vv_i64m2)(src, seed, vl);
RVV_OP(vse64_v_i64m2)((int64_t*)&xcustomSecret[i], res, vl);
}
// vlo += seed64, vhi -= seed64
vuint64m2_t lo = RVV_OP(vadd_vx_u64m2)(vlo, seed64, vl);
vuint64m2_t hi = RVV_OP(vsub_vx_u64m2)(vhi, seed64, vl);

RVV_OP(vsse64_v_u64m2)((uint64_t*)customSecret, 16, lo, vl);
RVV_OP(vsse64_v_u64m2)((uint64_t*)(void*)((uint8_t*)customSecret + 8), 16, hi, vl);
}
i += vl;
kSecretPtr += vl * 16;
customSecret = (uint8_t*)customSecret + vl * 16;
}
}
}
Expand Down
Loading