Files
concrete/src/circuit_bootstrap.cu
2023-02-27 17:14:30 +01:00

318 lines
15 KiB
Plaintext

#include "circuit_bootstrap.cuh"
#include "circuit_bootstrap.h"
/*
* This scratch function allocates the necessary amount of data on the GPU for
* the circuit bootstrap on 32 bits inputs, into `cbs_buffer`. It also
* configures SM options on the GPU in case FULLSM mode is going to be used.
*/
void scratch_cuda_circuit_bootstrap_32(
void *v_stream, uint32_t gpu_index, int8_t **cbs_buffer,
uint32_t glwe_dimension, uint32_t lwe_dimension, uint32_t polynomial_size,
uint32_t level_count_cbs, uint32_t number_of_inputs,
uint32_t max_shared_memory, bool allocate_gpu_memory) {
switch (polynomial_size) {
case 256:
scratch_circuit_bootstrap<uint32_t, int32_t, Degree<256>>(
v_stream, gpu_index, cbs_buffer, glwe_dimension, lwe_dimension,
polynomial_size, level_count_cbs, number_of_inputs, max_shared_memory,
allocate_gpu_memory);
break;
case 512:
scratch_circuit_bootstrap<uint32_t, int32_t, Degree<512>>(
v_stream, gpu_index, cbs_buffer, glwe_dimension, lwe_dimension,
polynomial_size, level_count_cbs, number_of_inputs, max_shared_memory,
allocate_gpu_memory);
break;
case 1024:
scratch_circuit_bootstrap<uint32_t, int32_t, Degree<1024>>(
v_stream, gpu_index, cbs_buffer, glwe_dimension, lwe_dimension,
polynomial_size, level_count_cbs, number_of_inputs, max_shared_memory,
allocate_gpu_memory);
break;
case 2048:
scratch_circuit_bootstrap<uint32_t, int32_t, Degree<2048>>(
v_stream, gpu_index, cbs_buffer, glwe_dimension, lwe_dimension,
polynomial_size, level_count_cbs, number_of_inputs, max_shared_memory,
allocate_gpu_memory);
break;
case 4096:
scratch_circuit_bootstrap<uint32_t, int32_t, Degree<4096>>(
v_stream, gpu_index, cbs_buffer, glwe_dimension, lwe_dimension,
polynomial_size, level_count_cbs, number_of_inputs, max_shared_memory,
allocate_gpu_memory);
break;
case 8192:
scratch_circuit_bootstrap<uint32_t, int32_t, Degree<8192>>(
v_stream, gpu_index, cbs_buffer, glwe_dimension, lwe_dimension,
polynomial_size, level_count_cbs, number_of_inputs, max_shared_memory,
allocate_gpu_memory);
break;
default:
break;
}
}
/*
* This scratch function allocates the necessary amount of data on the GPU for
* the circuit bootstrap on 64 bits inputs, into `cbs_buffer`. It also
* configures SM options on the GPU in case FULLSM mode is going to be used.
*/
void scratch_cuda_circuit_bootstrap_64(
void *v_stream, uint32_t gpu_index, int8_t **cbs_buffer,
uint32_t glwe_dimension, uint32_t lwe_dimension, uint32_t polynomial_size,
uint32_t level_count_cbs, uint32_t number_of_inputs,
uint32_t max_shared_memory, bool allocate_gpu_memory) {
switch (polynomial_size) {
case 256:
scratch_circuit_bootstrap<uint64_t, int64_t, Degree<256>>(
v_stream, gpu_index, cbs_buffer, glwe_dimension, lwe_dimension,
polynomial_size, level_count_cbs, number_of_inputs, max_shared_memory,
allocate_gpu_memory);
break;
case 512:
scratch_circuit_bootstrap<uint64_t, int64_t, Degree<512>>(
v_stream, gpu_index, cbs_buffer, glwe_dimension, lwe_dimension,
polynomial_size, level_count_cbs, number_of_inputs, max_shared_memory,
allocate_gpu_memory);
break;
case 1024:
scratch_circuit_bootstrap<uint64_t, int64_t, Degree<1024>>(
v_stream, gpu_index, cbs_buffer, glwe_dimension, lwe_dimension,
polynomial_size, level_count_cbs, number_of_inputs, max_shared_memory,
allocate_gpu_memory);
break;
case 2048:
scratch_circuit_bootstrap<uint64_t, int64_t, Degree<2048>>(
v_stream, gpu_index, cbs_buffer, glwe_dimension, lwe_dimension,
polynomial_size, level_count_cbs, number_of_inputs, max_shared_memory,
allocate_gpu_memory);
break;
case 4096:
scratch_circuit_bootstrap<uint64_t, int64_t, Degree<4096>>(
v_stream, gpu_index, cbs_buffer, glwe_dimension, lwe_dimension,
polynomial_size, level_count_cbs, number_of_inputs, max_shared_memory,
allocate_gpu_memory);
break;
case 8192:
scratch_circuit_bootstrap<uint64_t, int64_t, Degree<8192>>(
v_stream, gpu_index, cbs_buffer, glwe_dimension, lwe_dimension,
polynomial_size, level_count_cbs, number_of_inputs, max_shared_memory,
allocate_gpu_memory);
break;
default:
break;
}
}
/*
* Perform circuit bootstrapping for the batch of 32 bit LWE ciphertexts.
* Head out to the equivalent operation on 64 bits for more details.
*/
void cuda_circuit_bootstrap_32(
void *v_stream, uint32_t gpu_index, void *ggsw_out, void *lwe_array_in,
void *fourier_bsk, void *fp_ksk_array, void *lut_vector_indexes,
int8_t *cbs_buffer, uint32_t delta_log, uint32_t polynomial_size,
uint32_t glwe_dimension, uint32_t lwe_dimension, uint32_t level_bsk,
uint32_t base_log_bsk, uint32_t level_pksk, uint32_t base_log_pksk,
uint32_t level_cbs, uint32_t base_log_cbs, uint32_t number_of_inputs,
uint32_t max_shared_memory) {
assert(("Error (GPU circuit bootstrap): polynomial_size should be one of "
"256, 512, 1024, 2048, 4096, 8192",
polynomial_size == 256 || polynomial_size == 512 ||
polynomial_size == 1024 || polynomial_size == 2048 ||
polynomial_size == 4096 || polynomial_size == 8192));
// The number of samples should be lower than the number of streaming
// multiprocessors divided by (4 * (k + 1) * l) (the factor 4 being related
// to the occupancy of 50%). The only supported value for k is 1, so
// k + 1 = 2 for now.
int number_of_sm = 0;
cudaDeviceGetAttribute(&number_of_sm, cudaDevAttrMultiProcessorCount, 0);
assert(("Error (GPU extract bits): the number of input LWEs must be lower or "
"equal to the "
"number of streaming multiprocessors on the device divided by 8 * "
"level_count_bsk",
number_of_inputs <= number_of_sm / 4. / 2. / level_bsk));
switch (polynomial_size) {
case 256:
host_circuit_bootstrap<uint32_t, Degree<256>>(
v_stream, gpu_index, (uint32_t *)ggsw_out, (uint32_t *)lwe_array_in,
(double2 *)fourier_bsk, (uint32_t *)fp_ksk_array,
(uint32_t *)lut_vector_indexes, cbs_buffer, delta_log, polynomial_size,
glwe_dimension, lwe_dimension, level_bsk, base_log_bsk, level_pksk,
base_log_pksk, level_cbs, base_log_cbs, number_of_inputs,
max_shared_memory);
break;
case 512:
host_circuit_bootstrap<uint32_t, Degree<512>>(
v_stream, gpu_index, (uint32_t *)ggsw_out, (uint32_t *)lwe_array_in,
(double2 *)fourier_bsk, (uint32_t *)fp_ksk_array,
(uint32_t *)lut_vector_indexes, cbs_buffer, delta_log, polynomial_size,
glwe_dimension, lwe_dimension, level_bsk, base_log_bsk, level_pksk,
base_log_pksk, level_cbs, base_log_cbs, number_of_inputs,
max_shared_memory);
break;
case 1024:
host_circuit_bootstrap<uint32_t, Degree<1024>>(
v_stream, gpu_index, (uint32_t *)ggsw_out, (uint32_t *)lwe_array_in,
(double2 *)fourier_bsk, (uint32_t *)fp_ksk_array,
(uint32_t *)lut_vector_indexes, cbs_buffer, delta_log, polynomial_size,
glwe_dimension, lwe_dimension, level_bsk, base_log_bsk, level_pksk,
base_log_pksk, level_cbs, base_log_cbs, number_of_inputs,
max_shared_memory);
break;
case 2048:
host_circuit_bootstrap<uint32_t, Degree<2048>>(
v_stream, gpu_index, (uint32_t *)ggsw_out, (uint32_t *)lwe_array_in,
(double2 *)fourier_bsk, (uint32_t *)fp_ksk_array,
(uint32_t *)lut_vector_indexes, cbs_buffer, delta_log, polynomial_size,
glwe_dimension, lwe_dimension, level_bsk, base_log_bsk, level_pksk,
base_log_pksk, level_cbs, base_log_cbs, number_of_inputs,
max_shared_memory);
break;
case 4096:
host_circuit_bootstrap<uint32_t, Degree<4096>>(
v_stream, gpu_index, (uint32_t *)ggsw_out, (uint32_t *)lwe_array_in,
(double2 *)fourier_bsk, (uint32_t *)fp_ksk_array,
(uint32_t *)lut_vector_indexes, cbs_buffer, delta_log, polynomial_size,
glwe_dimension, lwe_dimension, level_bsk, base_log_bsk, level_pksk,
base_log_pksk, level_cbs, base_log_cbs, number_of_inputs,
max_shared_memory);
break;
case 8192:
host_circuit_bootstrap<uint32_t, Degree<8192>>(
v_stream, gpu_index, (uint32_t *)ggsw_out, (uint32_t *)lwe_array_in,
(double2 *)fourier_bsk, (uint32_t *)fp_ksk_array,
(uint32_t *)lut_vector_indexes, cbs_buffer, delta_log, polynomial_size,
glwe_dimension, lwe_dimension, level_bsk, base_log_bsk, level_pksk,
base_log_pksk, level_cbs, base_log_cbs, number_of_inputs,
max_shared_memory);
break;
default:
break;
}
}
/*
* Perform circuit bootstrapping on a batch of 64 bit input LWE ciphertexts.
* - `v_stream` is a void pointer to the Cuda stream to be used in the kernel
* launch
* - `gpu_index` is the index of the GPU to be used in the kernel launch
* - 'ggsw_out' output batch of ggsw with size:
* 'number_of_inputs' * 'level_cbs' * ('glwe_dimension' + 1)^2 *
* polynomial_size * sizeof(u64)
* - 'lwe_array_in' input batch of lwe ciphertexts, with size:
* 'number_of_inputs' * '(lwe_dimension' + 1) * sizeof(u64)
* - 'fourier_bsk' bootstrapping key in fourier domain with size:
* 'lwe_dimension' * 'level_bsk' * ('glwe_dimension' + 1)^2 *
* 'polynomial_size' / 2 * sizeof(double2)
* - 'fp_ksk_array' batch of fp-keyswitch keys with size:
* ('polynomial_size' + 1) * 'level_pksk' * ('glwe_dimension' + 1)^2 *
* 'polynomial_size' * sizeof(u64)
* - 'cbs_buffer': buffer used during calculations, it is not an actual
* inputs of the function, just allocated memory for calculation
* process, like this, memory can be allocated once and can be used as much
* as needed for different calls of circuit_bootstrap function
*
* This function calls a wrapper to a device kernel that performs the
* circuit bootstrap. The kernel is templatized based on integer discretization
* and polynomial degree.
*/
void cuda_circuit_bootstrap_64(
void *v_stream, uint32_t gpu_index, void *ggsw_out, void *lwe_array_in,
void *fourier_bsk, void *fp_ksk_array, void *lut_vector_indexes,
int8_t *cbs_buffer, uint32_t delta_log, uint32_t polynomial_size,
uint32_t glwe_dimension, uint32_t lwe_dimension, uint32_t level_bsk,
uint32_t base_log_bsk, uint32_t level_pksk, uint32_t base_log_pksk,
uint32_t level_cbs, uint32_t base_log_cbs, uint32_t number_of_inputs,
uint32_t max_shared_memory) {
assert(("Error (GPU circuit bootstrap): polynomial_size should be one of "
"256, 512, 1024, 2048, 4096, 8192",
polynomial_size == 256 || polynomial_size == 512 ||
polynomial_size == 1024 || polynomial_size == 2048 ||
polynomial_size == 4096 || polynomial_size == 8192));
// The number of samples should be lower than the number of streaming
// multiprocessors divided by (4 * (k + 1) * l) (the factor 4 being related
// to the occupancy of 50%). The only supported value for k is 1, so
// k + 1 = 2 for now.
int number_of_sm = 0;
cudaDeviceGetAttribute(&number_of_sm, cudaDevAttrMultiProcessorCount, 0);
assert(("Error (GPU extract bits): the number of input LWEs must be lower or "
"equal to the "
"number of streaming multiprocessors on the device divided by 8 * "
"level_count_bsk",
number_of_inputs <= number_of_sm / 4. / 2. / level_bsk));
// The number of samples should be lower than the number of streaming
switch (polynomial_size) {
case 256:
host_circuit_bootstrap<uint64_t, Degree<256>>(
v_stream, gpu_index, (uint64_t *)ggsw_out, (uint64_t *)lwe_array_in,
(double2 *)fourier_bsk, (uint64_t *)fp_ksk_array,
(uint64_t *)lut_vector_indexes, cbs_buffer, delta_log, polynomial_size,
glwe_dimension, lwe_dimension, level_bsk, base_log_bsk, level_pksk,
base_log_pksk, level_cbs, base_log_cbs, number_of_inputs,
max_shared_memory);
break;
case 512:
host_circuit_bootstrap<uint64_t, Degree<512>>(
v_stream, gpu_index, (uint64_t *)ggsw_out, (uint64_t *)lwe_array_in,
(double2 *)fourier_bsk, (uint64_t *)fp_ksk_array,
(uint64_t *)lut_vector_indexes, cbs_buffer, delta_log, polynomial_size,
glwe_dimension, lwe_dimension, level_bsk, base_log_bsk, level_pksk,
base_log_pksk, level_cbs, base_log_cbs, number_of_inputs,
max_shared_memory);
break;
case 1024:
host_circuit_bootstrap<uint64_t, Degree<1024>>(
v_stream, gpu_index, (uint64_t *)ggsw_out, (uint64_t *)lwe_array_in,
(double2 *)fourier_bsk, (uint64_t *)fp_ksk_array,
(uint64_t *)lut_vector_indexes, cbs_buffer, delta_log, polynomial_size,
glwe_dimension, lwe_dimension, level_bsk, base_log_bsk, level_pksk,
base_log_pksk, level_cbs, base_log_cbs, number_of_inputs,
max_shared_memory);
break;
case 2048:
host_circuit_bootstrap<uint64_t, Degree<2048>>(
v_stream, gpu_index, (uint64_t *)ggsw_out, (uint64_t *)lwe_array_in,
(double2 *)fourier_bsk, (uint64_t *)fp_ksk_array,
(uint64_t *)lut_vector_indexes, cbs_buffer, delta_log, polynomial_size,
glwe_dimension, lwe_dimension, level_bsk, base_log_bsk, level_pksk,
base_log_pksk, level_cbs, base_log_cbs, number_of_inputs,
max_shared_memory);
break;
case 4096:
host_circuit_bootstrap<uint64_t, Degree<4096>>(
v_stream, gpu_index, (uint64_t *)ggsw_out, (uint64_t *)lwe_array_in,
(double2 *)fourier_bsk, (uint64_t *)fp_ksk_array,
(uint64_t *)lut_vector_indexes, cbs_buffer, delta_log, polynomial_size,
glwe_dimension, lwe_dimension, level_bsk, base_log_bsk, level_pksk,
base_log_pksk, level_cbs, base_log_cbs, number_of_inputs,
max_shared_memory);
break;
case 8192:
host_circuit_bootstrap<uint64_t, Degree<8192>>(
v_stream, gpu_index, (uint64_t *)ggsw_out, (uint64_t *)lwe_array_in,
(double2 *)fourier_bsk, (uint64_t *)fp_ksk_array,
(uint64_t *)lut_vector_indexes, cbs_buffer, delta_log, polynomial_size,
glwe_dimension, lwe_dimension, level_bsk, base_log_bsk, level_pksk,
base_log_pksk, level_cbs, base_log_cbs, number_of_inputs,
max_shared_memory);
break;
default:
break;
}
}
/*
* This cleanup function frees the data for the circuit bootstrap on GPU in
* cbs_buffer for 32 or 64 bits inputs.
*/
void cleanup_cuda_circuit_bootstrap(void *v_stream, uint32_t gpu_index,
int8_t **cbs_buffer) {
auto stream = static_cast<cudaStream_t *>(v_stream);
// Free memory
cuda_drop_async(*cbs_buffer, stream, gpu_index);
}