#define CPU_JIT_AVX512_COMMON_1x1_CONVOLUTION_HPP
#include "c_types_map.hpp"
+#include "memory_tracking.hpp"
+#include "mkldnn_thread.hpp"
+#include "utils.hpp"
+
#include "cpu_convolution_pd.hpp"
#include "cpu_engine.hpp"
#include "cpu_reducer.hpp"
+
#include "jit_avx512_common_1x1_conv_kernel.hpp"
#include "jit_uni_1x1_conv_utils.hpp"
#include "jit_transpose_src_utils.hpp"
-#include "mkldnn_thread.hpp"
-#include "utils.hpp"
namespace mkldnn {
namespace impl {
namespace cpu {
-template <bool with_relu, impl::data_type_t src_type,
+template <impl::data_type_t src_type,
impl::data_type_t wei_type = src_type,
impl::data_type_t dst_type = src_type>
-struct _jit_avx512_common_1x1_convolution_fwd_t : public cpu_primitive_t {
+struct jit_avx512_common_1x1_convolution_fwd_t : public cpu_primitive_t {
// TODO: (Roma) Code duplication duplication! Remove with templates
// (maybe...)!
- struct pd_t: public _cpu_convolution_fwd_pd_t<with_relu> {
- pd_t(engine_t *engine,
- const typename pd_t::base_desc_t *adesc,
+ struct pd_t: public cpu_convolution_fwd_pd_t {
+ pd_t(engine_t *engine, const convolution_desc_t *adesc,
const primitive_attr_t *attr,
const typename pd_t::base_class *hint_fwd_pd)
- : _cpu_convolution_fwd_pd_t<with_relu>(engine, adesc, attr,
- hint_fwd_pd)
+ : cpu_convolution_fwd_pd_t(engine, adesc, attr, hint_fwd_pd)
, jcp_(), rtus_() {}
DECLARE_COMMON_PD_T(
JIT_IMPL_NAME_HELPER("jit_1x1:", avx512_common, ""),
- _jit_avx512_common_1x1_convolution_fwd_t);
+ jit_avx512_common_1x1_convolution_fwd_t);
virtual status_t init() override {
using namespace prop_kind;
assert(this->engine()->kind() == engine_kind::cpu);
bool ok = true
&& this->set_default_params() == status::success
- && utils::one_of(this->cdesc_().prop_kind, forward_training,
+ && utils::one_of(this->desc()->prop_kind, forward_training,
forward_inference)
- && this->cdesc_().alg_kind == alg_kind::convolution_direct
+ && utils::one_of(this->desc()->alg_kind,
+ alg_kind::convolution_auto,
+ alg_kind::convolution_direct)
&& !this->has_zero_dim_memory()
- && this->cdesc_().src_desc.data_type == src_type
- && this->cdesc_().weights_desc.data_type == wei_type
- && this->cdesc_().dst_desc.data_type == dst_type
+ && this->desc()->src_desc.data_type == src_type
+ && this->desc()->weights_desc.data_type == wei_type
+ && this->desc()->dst_desc.data_type == dst_type
&& IMPLICATION(this->with_bias(),
- dst_type == this->cdesc_().bias_desc.data_type)
- && IMPLICATION(with_relu && dst_type == data_type::s32
- && everyone_is(data_type::s16, src_type, wei_type),
- this->negative_slope() == 0.);
+ dst_type == this->desc()->bias_desc.data_type);
if (!ok) return status::unimplemented;
- const convolution_desc_t *conv_d = &this->cdesc_();
+ const convolution_desc_t *conv_d = this->desc();
const memory_desc_t *src_d = this->src_pd_.desc();
rtus_prepare(this, conv_d, src_d, this->dst_pd_.desc());
- return jit_avx512_common_1x1_conv_kernel::init_conf(jcp_,
- *conv_d, *src_d, *this->weights_pd_.desc(),
+
+ status_t status = jit_avx512_common_1x1_conv_kernel::init_conf(
+ jcp_, *conv_d, *src_d, *this->weights_pd_.desc(),
*this->dst_pd_.desc(), *this->attr(),
- with_relu, this->negative_slope(),
mkldnn_get_max_threads(), rtus_.reduce_src_);
+ if (status != status::success) return status;
+
+ auto scratchpad = scratchpad_registry().registrar();
+ jit_avx512_common_1x1_conv_kernel::init_scratchpad(scratchpad,
+ jcp_);
+
+ rtus_prepare_space_info(this, scratchpad);
+
+ return status::success;
}
jit_1x1_conv_conf_t jcp_;
- struct reduce_to_unit_stride_t {
- convolution_desc_t conv_d_;
- bool reduce_src_;
- } rtus_;
+ reduce_to_unit_stride_t rtus_;
- protected:
+ protected:
virtual status_t set_default_params() override {
using namespace memory_format;
if (this->src_pd_.desc()->format == any)
}
if (this->bias_pd_.desc()->format == any)
CHECK(this->bias_pd_.set_format(x));
+ if (this->desc()->alg_kind == alg_kind::convolution_auto)
+ CHECK(this->set_alg_kind(alg_kind::convolution_direct));
return status::success;
}
};
template <cpu_isa_t isa, typename conv_t>
friend void init_rtus_driver(conv_t *self);
- _jit_avx512_common_1x1_convolution_fwd_t(const pd_t *pd,
- const input_vector &inputs,
- const output_vector &outputs)
- : cpu_primitive_t(&conf_, inputs, outputs), conf_(*pd)
- , kernel_(nullptr), rtus_driver_(nullptr), ws_per_thread_(0)
- , scratch_(nullptr), padded_bias_(nullptr)
- {
- kernel_ = new jit_avx512_common_1x1_conv_kernel(conf_.jcp_,
- *conf_.attr());
+ jit_avx512_common_1x1_convolution_fwd_t(const pd_t *apd,
+ const input_vector &inputs, const output_vector &outputs)
+ : cpu_primitive_t(apd, inputs, outputs)
+ , kernel_(nullptr), rtus_driver_(nullptr)
+ {
+ kernel_ =
+ new jit_avx512_common_1x1_conv_kernel(pd()->jcp_, *pd()->attr());
init_rtus_driver<avx512_common>(this);
-
- if (conf_.want_padded_bias()) {
- const auto &j = conf_.jcp_;
- assert(j.ngroups == 1);
- padded_bias_ = (dst_data_t *)malloc(sizeof(dst_data_t) * j.oc, 64);
- for (int oc = j.oc_without_padding; oc < j.oc; ++oc)
- padded_bias_[oc] = 0;
- }
}
- ~_jit_avx512_common_1x1_convolution_fwd_t() {
+ ~jit_avx512_common_1x1_convolution_fwd_t() {
delete kernel_;
delete rtus_driver_;
- free(scratch_);
- free(padded_bias_);
}
typedef typename prec_traits<src_type>::type src_data_t;
typedef typename prec_traits<wei_type>::type wei_data_t;
typedef typename prec_traits<dst_type>::type dst_data_t;
- virtual void execute(event_t *e) {
+ virtual void execute(event_t *e) const {
execute_forward();
e->set_state(event_t::ready);
}
private:
- void execute_forward();
+ void execute_forward() const;
void execute_forward_thr(const int ithr, const int nthr,
const src_data_t *src, const wei_data_t *weights,
- const dst_data_t *bias, dst_data_t *dst);
- pd_t conf_;
+ const dst_data_t *bias, dst_data_t *dst,
+ const memory_tracking::grantor_t &scratchpad) const;
+ const pd_t *pd() const { return (const pd_t *)primitive_t::pd(); }
+
jit_avx512_common_1x1_conv_kernel *kernel_;
- /* reduction to unit stride */
rtus_driver_t<avx512_common> *rtus_driver_;
- size_t ws_per_thread_;
- src_data_t *scratch_;
- dst_data_t *padded_bias_;
};
using jit_avx512_common_1x1_convolution_fwd_f32_t
- = _jit_avx512_common_1x1_convolution_fwd_t<false, data_type::f32>;
-using jit_avx512_common_1x1_convolution_relu_f32_t
- = _jit_avx512_common_1x1_convolution_fwd_t<true, data_type::f32>;
+ = jit_avx512_common_1x1_convolution_fwd_t<data_type::f32>;
using jit_avx512_common_1x1_convolution_fwd_s16s16s32_t
- = _jit_avx512_common_1x1_convolution_fwd_t<false, data_type::s16,
- data_type::s16, data_type::s32>;
-using jit_avx512_common_1x1_convolution_relu_s16s16s32_t
- = _jit_avx512_common_1x1_convolution_fwd_t<true, data_type::s16,
+ = jit_avx512_common_1x1_convolution_fwd_t<data_type::s16,
data_type::s16, data_type::s32>;
template <impl::data_type_t diff_dst_type,
impl::data_type_t wei_type = diff_dst_type,
impl::data_type_t diff_src_type = diff_dst_type>
-struct _jit_avx512_common_1x1_convolution_bwd_data_t : public cpu_primitive_t {
+struct jit_avx512_common_1x1_convolution_bwd_data_t : public cpu_primitive_t {
struct pd_t : public cpu_convolution_bwd_data_pd_t {
pd_t(engine_t *engine,
const convolution_desc_t *adesc,
DECLARE_COMMON_PD_T(
JIT_IMPL_NAME_HELPER("jit_1x1:", avx512_common, ""),
- _jit_avx512_common_1x1_convolution_bwd_data_t);
+ jit_avx512_common_1x1_convolution_bwd_data_t);
virtual status_t init() override {
using namespace prop_kind;
bool ok = true
&& this->set_default_params() == status::success
&& this->desc()->prop_kind == backward_data
- && this->desc()->alg_kind == alg_kind::convolution_direct
+ && utils::one_of(this->desc()->alg_kind, alg_kind::convolution_auto,
+ alg_kind::convolution_direct)
&& !this->has_zero_dim_memory()
&& this->desc()->diff_dst_desc.data_type == diff_dst_type
&& this->desc()->weights_desc.data_type == wei_type
const convolution_desc_t *conv_d = this->desc();
const memory_desc_t *diff_src_d = this->diff_src_pd_.desc();
rtus_prepare(this, conv_d, diff_src_d, this->diff_dst_pd_.desc());
- return jit_avx512_common_1x1_conv_kernel::init_conf(jcp_,
- *conv_d, *diff_src_d, *this->weights_pd_.desc(),
- *this->diff_dst_pd_.desc(), *this->attr(),
- mkldnn_get_max_threads(), rtus_.reduce_src_);
+
+ status_t status = jit_avx512_common_1x1_conv_kernel::init_conf(
+ jcp_, *conv_d, *diff_src_d, *this->weights_pd_.desc(),
+ *this->diff_dst_pd_.desc(), *this->attr(),
+ mkldnn_get_max_threads(), rtus_.reduce_src_);
+ if (status != status::success) return status;
+
+ auto scratchpad = scratchpad_registry().registrar();
+ jit_avx512_common_1x1_conv_kernel::init_scratchpad(scratchpad,
+ jcp_);
+
+ rtus_prepare_space_info(this, scratchpad);
+
+ return status::success;
}
// TODO (Roma): structs conf header cleanup
jit_1x1_conv_conf_t jcp_;
- struct reduce_to_unit_stride_t {
- convolution_desc_t conv_d_;
- bool reduce_src_;
- } rtus_;
+ reduce_to_unit_stride_t rtus_;
protected:
virtual status_t set_default_params() override {
? pick(this->ndims() - 3, gOIw8o16i2o, gOIhw8o16i2o)
: pick(this->ndims() - 3, OIw8o16i2o, OIhw8o16i2o)));
}
+ if (this->desc()->alg_kind == alg_kind::convolution_auto)
+ CHECK(this->set_alg_kind(alg_kind::convolution_direct));
return status::success;
}
template <cpu_isa_t isa, typename conv_t>
friend void init_rtus_driver(conv_t *self);
- _jit_avx512_common_1x1_convolution_bwd_data_t(const pd_t *pd,
- const input_vector &inputs,
- const output_vector &outputs)
- : cpu_primitive_t(&conf_, inputs, outputs), conf_(*pd)
- , kernel_(nullptr), rtus_driver_(nullptr), ws_per_thread_(0)
- , scratch_(nullptr)
+
+ jit_avx512_common_1x1_convolution_bwd_data_t(const pd_t *apd,
+ const input_vector &inputs, const output_vector &outputs)
+ : cpu_primitive_t(apd, inputs, outputs)
+ , kernel_(nullptr), rtus_driver_(nullptr)
{
- kernel_ = new jit_avx512_common_1x1_conv_kernel(conf_.jcp_,
- *conf_.attr());
+ kernel_ = new jit_avx512_common_1x1_conv_kernel(pd()->jcp_,
+ *pd()->attr());
init_rtus_driver<avx512_common>(this);
}
- ~_jit_avx512_common_1x1_convolution_bwd_data_t()
- {
+
+ ~jit_avx512_common_1x1_convolution_bwd_data_t() {
delete kernel_;
delete rtus_driver_;
- free(scratch_);
}
typedef typename prec_traits<diff_dst_type>::type diff_dst_data_t;
typedef typename prec_traits<wei_type>::type wei_data_t;
typedef typename prec_traits<diff_src_type>::type diff_src_data_t;
- virtual void execute(event_t *e) {
- switch (conf_.desc()->prop_kind) {
+ virtual void execute(event_t *e) const {
+ switch (pd()->desc()->prop_kind) {
case prop_kind::backward_data:
execute_backward_data();
break;
}
private:
- void execute_backward_data();
- pd_t conf_;
+ void execute_backward_data() const;
+ const pd_t *pd() const { return (const pd_t *)primitive_t::pd(); }
+
jit_avx512_common_1x1_conv_kernel *kernel_;
- /* reduction to unit stride */
rtus_driver_t<avx512_common> *rtus_driver_;
- size_t ws_per_thread_;
- diff_src_data_t *scratch_;
};
using jit_avx512_common_1x1_convolution_bwd_data_f32_t
- = _jit_avx512_common_1x1_convolution_bwd_data_t<data_type::f32>;
+ = jit_avx512_common_1x1_convolution_bwd_data_t<data_type::f32>;
using jit_avx512_common_1x1_convolution_bwd_data_s16s16s32_t
- = _jit_avx512_common_1x1_convolution_bwd_data_t<data_type::s16,
+ = jit_avx512_common_1x1_convolution_bwd_data_t<data_type::s16,
data_type::s16, data_type::s32>;
struct jit_avx512_common_1x1_convolution_bwd_weights_t : public cpu_primitive_t
bool ok = true
&& this->set_default_params() == status::success
&& this->desc()->prop_kind == backward_weights
- && this->desc()->alg_kind == alg_kind::convolution_direct
+ && utils::one_of(this->desc()->alg_kind,
+ alg_kind::convolution_auto,
+ alg_kind::convolution_direct)
&& !this->has_zero_dim_memory()
&& utils::everyone_is(data_type::f32,
this->desc()->src_desc.data_type,
const convolution_desc_t *conv_d = this->desc();
const memory_desc_t *src_d = this->src_pd_.desc();
rtus_prepare(this, conv_d, src_d, this->diff_dst_pd_.desc());
- return jit_avx512_common_1x1_conv_kernel::init_conf(jcp_,
- *conv_d, *src_d, *this->diff_weights_pd_.desc(),
- *this->diff_dst_pd_.desc(), *this->attr(),
- mkldnn_get_max_threads(), rtus_.reduce_src_);
+
+ status_t status = jit_avx512_common_1x1_conv_kernel::init_conf(
+ jcp_, *conv_d, *src_d, *this->diff_weights_pd_.desc(),
+ *this->diff_dst_pd_.desc(), *this->attr(),
+ mkldnn_get_max_threads(), rtus_.reduce_src_);
+ if (status != status::success) return status;
+
+ init_balancers();
+
+ auto scratchpad = scratchpad_registry().registrar();
+ jit_avx512_common_1x1_conv_kernel::init_scratchpad(scratchpad,
+ jcp_);
+
+ auto reducer_bia_scratchpad = memory_tracking::registrar_t(
+ scratchpad, memory_tracking::names::prefix_reducer_bia);
+ reducer_bia_conf_.init_scratchpad(reducer_bia_scratchpad);
+
+ rtus_prepare_space_info(this, scratchpad);
+
+ return status::success;
}
// TODO (Roma): structs conf header cleanup
jit_1x1_conv_conf_t jcp_;
-
- struct reduce_to_unit_stride_t {
- convolution_desc_t conv_d_;
- bool reduce_src_;
- } rtus_;
+ cpu_reducer_t<data_type::f32>::conf_t reducer_bia_conf_;
+ reduce_to_unit_stride_t rtus_;
protected:
virtual status_t set_default_params() override {
: pick(this->ndims() - 3, OIw16i16o, OIhw16i16o)));
if (this->diff_bias_pd_.desc()->format == any)
CHECK(this->diff_bias_pd_.set_format(x));
+ if (this->desc()->alg_kind == alg_kind::convolution_auto)
+ CHECK(this->set_alg_kind(alg_kind::convolution_direct));
return status::success;
}
+
+ private:
+ void init_balancers() {
+ const size_t max_buffer_size = jcp_.nthr * 3 * 5 * 5 * 16 * 16;
+ if (with_bias()) {
+ reducer_bia_conf_.init(reduce_balancer_t(jcp_.nthr,
+ jcp_.oc_block, jcp_.ngroups * jcp_.nb_load,
+ jcp_.mb, max_buffer_size));
+ }
+ }
};
template <cpu_isa_t isa, typename conv_t>
friend void init_rtus_driver(conv_t *self);
- jit_avx512_common_1x1_convolution_bwd_weights_t(const pd_t *pd,
- const input_vector &inputs,
- const output_vector &outputs);
+
+ jit_avx512_common_1x1_convolution_bwd_weights_t(const pd_t *apd,
+ const input_vector &inputs, const output_vector &outputs);
+
~jit_avx512_common_1x1_convolution_bwd_weights_t() {
delete kernel_;
delete acc_ker_;
delete reducer_bias_;
delete rtus_driver_;
delete trans_kernel_;
- free(bctx_);
- free(ws_reduction_);
- free(scratch_);
- free(tr_src_);
- free(padded_bias_);
}
typedef typename prec_traits<data_type::f32>::type data_t;
- virtual void execute(event_t *e) {
- switch (conf_.desc()->prop_kind) {
+ virtual void execute(event_t *e) const {
+ switch (pd()->desc()->prop_kind) {
case prop_kind::backward_weights:
execute_backward_weights();
break;
}
private:
- void execute_backward_weights();
+ void execute_backward_weights() const;
+ const pd_t *pd() const { return (const pd_t *)primitive_t::pd(); }
- pd_t conf_;
jit_avx512_common_1x1_conv_kernel *kernel_;
cpu_accumulator_1d_t<data_type::f32> *acc_ker_;
cpu_reducer_t<data_type::f32> *reducer_bias_;
jit_transpose4x16_src *trans_kernel_;
-
- /* reduction to unit stride */
rtus_driver_t<avx512_common> *rtus_driver_;
- size_t ws_per_thread_;
- data_t *scratch_;
- data_t *padded_bias_;
-
- simple_barrier::ctx_t *bctx_;
- data_t *tr_src_;
- data_t *ws_reduction_;
};
}
projects / platform / upstream / dldt.git / blobdiff