Update Instant-NGP & Density Grid with CUDA supports. Much faster and better PSNR. (#193)

This commit is contained in:
Ruilong Li(李瑞龙) authored and GitHub committed 2022-07-24 22:22:44 -07:00
1 parent 31b52ae186
commit 12cd321e1a
39 files changed
+1721 -650

No files matched your search

+53
View File
@@ -0,0 +1,53 @@
#pragma once
#ifdef __CUDACC__
#define CUDA_HOSTDEV __host__ __device__
#else
#define CUDA_HOSTDEV
#endif
#include <torch/extension.h>
inline constexpr CUDA_HOSTDEV float __SQRT3() { return 1.73205080757f; }
template <typename scalar_t>
inline CUDA_HOSTDEV void __swap(scalar_t &a, scalar_t &b)
{
scalar_t c = a;
a = b;
b = c;
}
inline CUDA_HOSTDEV float __clamp(float f, float a, float b) { return fmaxf(a, fminf(f, b)); }
inline CUDA_HOSTDEV int __clamp(int f, int a, int b) { return std::max(a, std::min(f, b)); }
inline CUDA_HOSTDEV float __sign(float x) { return copysignf(1.0, x); }
inline CUDA_HOSTDEV uint32_t __expand_bits(uint32_t v)
{
v = (v * 0x00010001u) & 0xFF0000FFu;
v = (v * 0x00000101u) & 0x0F00F00Fu;
v = (v * 0x00000011u) & 0xC30C30C3u;
v = (v * 0x00000005u) & 0x49249249u;
return v;
}
inline CUDA_HOSTDEV uint32_t __morton3D(uint32_t x, uint32_t y, uint32_t z)
{
uint32_t xx = __expand_bits(x);
uint32_t yy = __expand_bits(y);
uint32_t zz = __expand_bits(z);
return xx | (yy << 1) | (zz << 2);
}
inline CUDA_HOSTDEV uint32_t __morton3D_invert(uint32_t x)
{
x = x & 0x49249249;
x = (x | (x >> 2)) & 0xc30c30c3;
x = (x | (x >> 4)) & 0x0f00f00f;
x = (x | (x >> 8)) & 0xff0000ff;
x = (x | (x >> 16)) & 0x0000ffff;
return x;
}
+18
View File
@@ -0,0 +1,18 @@
#pragma once
#include "helpers.h"
#include <c10/cuda/CUDAGuard.h>
#define CHECK_CUDA(x) TORCH_CHECK(x.is_cuda(), #x " must be a CUDA tensor")
#define CHECK_CONTIGUOUS(x) \
TORCH_CHECK(x.is_contiguous(), #x " must be contiguous")
#define CHECK_INPUT(x) \
CHECK_CUDA(x); \
CHECK_CONTIGUOUS(x)
#define CUDA_GET_THREAD_ID(tid, Q) \
const int tid = blockIdx.x * blockDim.x + threadIdx.x; \
if (tid >= Q) \
return
#define CUDA_N_BLOCKS_NEEDED(Q, CUDA_N_THREADS) ((Q - 1) / CUDA_N_THREADS + 1)
#define DEVICE_GUARD(_ten) \
const at::cuda::OptionalCUDAGuard device_guard(device_of(_ten));