1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
|
#include "stdafx.h"
#include "Reshaper.h"
#include "../D3D/MappedResource.h"
#include "../D3D/Binder.h"
#include "../D3D/shaders.h"
#include "reshapedMultiply.h"
namespace
{
using namespace DirectCompute;
struct Constants
{
// Size and strides of the source tensor
TensorShape arg0;
uint32_t zzPadding;
// Count of elements per panel
uint32_t panelSize;
// Layer strides of the output matrix
std::array<uint32_t, 2> layerStrides;
};
}
HRESULT DirectCompute::Reshaper::createConstants()
{
constexpr uint32_t cb = sizeof( Constants );
CD3D11_BUFFER_DESC desc{ cb, D3D11_BIND_CONSTANT_BUFFER, D3D11_USAGE_DYNAMIC, D3D11_CPU_ACCESS_WRITE };
return device()->CreateBuffer( &desc, nullptr, &constantBuffer );
}
HRESULT DirectCompute::Reshaper::makePanels( Tensor& tensor, eDataType dataType )
{
if( !constantBuffer )
CHECK( createConstants() );
constexpr uint32_t TILE_SIZE = ReshapedMultiply::TILE_SIZE;
// Reshaping into column major horizontal panels, height = TILE_SIZE, width = width of the source matrix
std::array<uint32_t, 4> ne = tensor.ne;
const uint32_t groupsX = ( ne[ 1 ] + TILE_SIZE - 1 ) / TILE_SIZE;
ne[ 1 ] = groupsX * TILE_SIZE;;
// Each panel has [ size.x, TILE_SIZE ] elements
const uint32_t panelSize = ne[ 0 ] * TILE_SIZE;
Tensor result;
result.create( dataType, ne );
{
MappedResource mapped;
CHECK( mapped.map( constantBuffer, false ) );
Constants& cb = *(Constants*)mapped.data();
store( cb.arg0.ne, tensor.sizeVec() );
store( cb.arg0.nb, tensor.stridesVec() );
cb.panelSize = panelSize;
cb.layerStrides[ 0 ] = result.nb[ 2 ];
cb.layerStrides[ 1 ] = result.nb[ 3 ];
}
csSetCB( constantBuffer );
{
Binder bind;
bind.bind( tensor, result );
bindShader( eComputeShader::matReshapePanels );
context()->Dispatch( groupsX, tensor.ne[ 2 ], tensor.ne[ 3 ] );
}
tensor.nb[ 0 ] = 0;
tensor.nb[ 1 ] = panelSize;
tensor.nb[ 2 ] = result.nb[ 2 ];
tensor.nb[ 3 ] = result.nb[ 3 ];
tensor.setGpuViews( result );
return S_OK;
}
DirectCompute::Reshaper::~Reshaper()
{
if( constantBuffer )
csSetCB( nullptr );
}
|
