def forward(self, pos): if self.initial_density_map is None: self.compute_initial_density_map(pos) # plot(0, self.initial_density_map.clone().div(self.bin_size_x*self.bin_size_y).cpu().numpy(), self.padding, 'summary/initial_potential_map') logger.info("fixed density map: average %g, max %g, bin area %g" % (self.initial_density_map.mean(), self.initial_density_map.max(), self.bin_size_x * self.bin_size_y)) # expk M = self.num_bins_x N = self.num_bins_y self.exact_expkM = precompute_expk(M, dtype=pos.dtype, device=pos.device) self.exact_expkN = precompute_expk(N, dtype=pos.dtype, device=pos.device) # init dct2, idct2, idct_idxst, idxst_idct with expkM and expkN self.dct2 = dct.DCT2(self.exact_expkM, self.exact_expkN) if not self.fast_mode: self.idct2 = dct.IDCT2(self.exact_expkM, self.exact_expkN) self.idct_idxst = dct.IDCT_IDXST(self.exact_expkM, self.exact_expkN) self.idxst_idct = dct.IDXST_IDCT(self.exact_expkM, self.exact_expkN) # wu and wv wu = torch.arange(M, dtype=pos.dtype, device=pos.device).mul( 2 * np.pi / M).view([M, 1]) # scale wv because the aspect ratio of a bin may not be 1 wv = torch.arange(N, dtype=pos.dtype, device=pos.device).mul(2 * np.pi / N).view( [1, N]).mul_(self.bin_size_x / self.bin_size_y) wu2_plus_wv2 = wu.pow(2) + wv.pow(2) wu2_plus_wv2[0, 0] = 1.0 # avoid zero-division, it will be zeroed out self.inv_wu2_plus_wv2 = 1.0 / wu2_plus_wv2 self.inv_wu2_plus_wv2[0, 0] = 0.0 self.wu_by_wu2_plus_wv2_half = wu.mul(self.inv_wu2_plus_wv2).mul_( 1. / 2) self.wv_by_wu2_plus_wv2_half = wv.mul(self.inv_wu2_plus_wv2).mul_( 1. / 2) return ElectricPotentialFunction.apply( pos, self.node_size_x_clamped, self.node_size_y_clamped, self.offset_x, self.offset_y, self.ratio, self.bin_center_x, self.bin_center_y, self.initial_density_map, self.buf, self.target_density, self.xl, self.yl, self.xh, self.yh, self.bin_size_x, self.bin_size_y, self.num_movable_nodes, self.num_filler_nodes, self.padding, self.padding_mask, self.num_bins_x, self.num_bins_y, self.num_movable_impacted_bins_x, self.num_movable_impacted_bins_y, self.num_filler_impacted_bins_x, self.num_filler_impacted_bins_y, self.deterministic_flag, self.sorted_node_map, self.exact_expkM, self.exact_expkN, self.inv_wu2_plus_wv2, self.wu_by_wu2_plus_wv2_half, self.wv_by_wu2_plus_wv2_half, self.dct2, self.idct2, self.idct_idxst, self.idxst_idct, self.fast_mode, self.num_threads)
def forward(self, pos, mode="density"): assert mode in {"density", "overflow" }, "Only support density mode or overflow mode" if (self.region_id is not None): ### reconstruct pos, only extract cells in this electric field pos = pos[self.pos_mask] if self.initial_density_map is None: num_nodes = pos.size(0) // 2 if (self.fence_regions is not None): if (self.placedb.num_terminals > 0): ### merge fence region density and macro density together as initial density map ### pay attention to the number of nodes, must use data from self ### here pos is reconstructed pos ! self.initial_density_map = self.compute_fence_region_map( self.fence_regions, pos[self.num_movable_nodes:self.num_movable_nodes + self.num_terminals], pos[num_nodes + self.num_movable_nodes:num_nodes + self.num_movable_nodes + self.num_terminals], self.node_size_x[self.num_movable_nodes:self. num_movable_nodes + self.num_terminals], self.node_size_y[self.num_movable_nodes:self. num_movable_nodes + self.num_terminals]) else: self.initial_density_map = self.compute_fence_region_map( self.fence_regions) else: self.compute_initial_density_map(pos) ## sync the initial density map with # self.compute_initial_density_map(pos) # plot(0, self.initial_density_map.clone().div(self.bin_size_x*self.bin_size_y).cpu().numpy(), self.padding, 'summary/initial_potential_map') logger.info("fixed density map: average %g, max %g, bin area %g" % (self.initial_density_map.mean(), self.initial_density_map.max(), self.bin_size_x * self.bin_size_y)) # expk M = self.num_bins_x N = self.num_bins_y self.exact_expkM = precompute_expk(M, dtype=pos.dtype, device=pos.device) self.exact_expkN = precompute_expk(N, dtype=pos.dtype, device=pos.device) # init dct2, idct2, idct_idxst, idxst_idct with expkM and expkN self.dct2 = dct.DCT2(self.exact_expkM, self.exact_expkN) if not self.fast_mode: self.idct2 = dct.IDCT2(self.exact_expkM, self.exact_expkN) self.idct_idxst = dct.IDCT_IDXST(self.exact_expkM, self.exact_expkN) self.idxst_idct = dct.IDXST_IDCT(self.exact_expkM, self.exact_expkN) # wu and wv wu = torch.arange(M, dtype=pos.dtype, device=pos.device).mul( 2 * np.pi / M).view([M, 1]) # scale wv because the aspect ratio of a bin may not be 1 wv = torch.arange(N, dtype=pos.dtype, device=pos.device).mul(2 * np.pi / N).view( [1, N]).mul_(self.bin_size_x / self.bin_size_y) wu2_plus_wv2 = wu.pow(2) + wv.pow(2) wu2_plus_wv2[0, 0] = 1.0 # avoid zero-division, it will be zeroed out self.inv_wu2_plus_wv2 = 1.0 / wu2_plus_wv2 self.inv_wu2_plus_wv2[0, 0] = 0.0 self.wu_by_wu2_plus_wv2_half = wu.mul(self.inv_wu2_plus_wv2).mul_( 1. / 2) self.wv_by_wu2_plus_wv2_half = wv.mul(self.inv_wu2_plus_wv2).mul_( 1. / 2) if (mode == "density"): return ElectricPotentialFunction.apply( pos, self.node_size_x_clamped, self.node_size_y_clamped, self.offset_x, self.offset_y, self.ratio, self.bin_center_x, self.bin_center_y, self.initial_density_map, self.target_density, self.xl, self.yl, self.xh, self.yh, self.bin_size_x, self.bin_size_y, self.num_movable_nodes, self.num_filler_nodes, self.padding, self.padding_mask, self.num_bins_x, self.num_bins_y, self.num_movable_impacted_bins_x, self.num_movable_impacted_bins_y, self.num_filler_impacted_bins_x, self.num_filler_impacted_bins_y, self.deterministic_flag, self.sorted_node_map, self.exact_expkM, self.exact_expkN, self.inv_wu2_plus_wv2, self.wu_by_wu2_plus_wv2_half, self.wv_by_wu2_plus_wv2_half, self.dct2, self.idct2, self.idct_idxst, self.idxst_idct, self.fast_mode) elif (mode == "overflow"): ### num_filler_nodes is set 0 density_map = ElectricDensityMapFunction.forward( pos, self.node_size_x_clamped, self.node_size_y_clamped, self.offset_x, self.offset_y, self.ratio, self.bin_center_x, self.bin_center_y, self.initial_density_map, self.target_density, self.xl, self.yl, self.xh, self.yh, self.bin_size_x, self.bin_size_y, self.num_movable_nodes, 0, self.padding, self.padding_mask, self.num_bins_x, self.num_bins_y, self.num_movable_impacted_bins_x, self.num_movable_impacted_bins_y, self.num_filler_impacted_bins_x, self.num_filler_impacted_bins_y, self.deterministic_flag, self.sorted_node_map) bin_area = self.bin_size_x * self.bin_size_y density_cost = (density_map - self.target_density * bin_area).clamp_( min=0.0).sum() return density_cost, density_map.max() / bin_area
def compare_different_methods(cuda_flag, M=1024, N=1024, dtype=torch.float64): density_map = torch.empty(M, N, dtype=dtype).uniform_(0, 10.0) if cuda_flag: density_map = density_map.cuda() expkM = discrete_spectral_transform.get_expk(M, dtype, density_map.device) expkN = discrete_spectral_transform.get_expk(N, dtype, density_map.device) exact_expkM = discrete_spectral_transform.get_exact_expk(M, dtype, density_map.device) exact_expkN = discrete_spectral_transform.get_exact_expk(N, dtype, density_map.device) print("M = {}, N = {}".format(M, N)) wu = torch.arange(M, dtype=density_map.dtype, device=density_map.device).mul(2 * np.pi / M).view([M, 1]) wv = torch.arange(N, dtype=density_map.dtype, device=density_map.device).mul(2 * np.pi / N).view([1, N]) wu2_plus_wv2 = wu.pow(2) + wv.pow(2) wu2_plus_wv2[0, 0] = 1.0 # avoid zero-division, it will be zeroed out inv_wu2_plus_wv2_2X = 2.0 / wu2_plus_wv2 inv_wu2_plus_wv2_2X[0, 0] = 0.0 wu_by_wu2_plus_wv2_2X = wu.mul(inv_wu2_plus_wv2_2X) wv_by_wu2_plus_wv2_2X = wv.mul(inv_wu2_plus_wv2_2X) # the first approach is used as the ground truth auv_golden = dct.dct2(density_map, expk0=expkM, expk1=expkN) auv = auv_golden.clone() auv[0, :].mul_(0.5) auv[:, 0].mul_(0.5) auv_by_wu2_plus_wv2_wu = auv.mul(wu_by_wu2_plus_wv2_2X) auv_by_wu2_plus_wv2_wv = auv.mul(wv_by_wu2_plus_wv2_2X) field_map_x_golden = dct.idsct2(auv_by_wu2_plus_wv2_wu, expkM, expkN) field_map_y_golden = dct.idcst2(auv_by_wu2_plus_wv2_wv, expkM, expkN) # compute potential phi # auv / (wu**2 + wv**2) auv_by_wu2_plus_wv2 = auv.mul(inv_wu2_plus_wv2_2X).mul_(2) #potential_map = discrete_spectral_transform.idcct2(auv_by_wu2_plus_wv2, expkM, expkN) potential_map_golden = dct.idcct2(auv_by_wu2_plus_wv2, expkM, expkN) # compute energy energy_golden = potential_map_golden.mul(density_map).sum() if density_map.is_cuda: torch.cuda.synchronize() # the second approach uses the idxst_idct and idct_idxst dct2 = dct2_fft2.DCT2(exact_expkM, exact_expkN) idct2 = dct2_fft2.IDCT2(exact_expkM, exact_expkN) idct_idxst = dct2_fft2.IDCT_IDXST(exact_expkM, exact_expkN) idxst_idct = dct2_fft2.IDXST_IDCT(exact_expkM, exact_expkN) inv_wu2_plus_wv2 = 1.0 / wu2_plus_wv2 inv_wu2_plus_wv2[0, 0] = 0.0 wu_by_wu2_plus_wv2_half = wu.mul(inv_wu2_plus_wv2).mul_(0.5) wv_by_wu2_plus_wv2_half = wv.mul(inv_wu2_plus_wv2).mul_(0.5) buv = dct2.forward(density_map) buv_by_wu2_plus_wv2_wu = buv.mul(wu_by_wu2_plus_wv2_half) buv_by_wu2_plus_wv2_wv = buv.mul(wv_by_wu2_plus_wv2_half) field_map_x = idxst_idct.forward(buv_by_wu2_plus_wv2_wu) field_map_y = idct_idxst.forward(buv_by_wu2_plus_wv2_wv) buv_by_wu2_plus_wv2 = buv.mul(inv_wu2_plus_wv2) potential_map = idct2.forward(buv_by_wu2_plus_wv2) energy = potential_map.mul(density_map).sum() if density_map.is_cuda: torch.cuda.synchronize() # compare results np.testing.assert_allclose(buv.data.cpu().numpy(), auv_golden.data.cpu().numpy(), rtol=1e-6, atol=1e-5) np.testing.assert_allclose(field_map_x.data.cpu().numpy(), field_map_x_golden.data.cpu().numpy(), rtol=1e-6, atol=1e-5) np.testing.assert_allclose(field_map_y.data.cpu().numpy(), field_map_y_golden.data.cpu().numpy(), rtol=1e-6, atol=1e-5) np.testing.assert_allclose(potential_map.data.cpu().numpy(), potential_map_golden.data.cpu().numpy(), rtol=1e-6, atol=1e-5) np.testing.assert_allclose(energy.data.cpu().numpy(), energy_golden.data.cpu().numpy(), rtol=1e-6, atol=1e-5) # the third approach uses the dct.idxst_idct and dct.idxst_idct dct2 = dct.DCT2(expkM, expkN) idct2 = dct.IDCT2(expkM, expkN) idct_idxst = dct.IDCT_IDXST(expkM, expkN) idxst_idct = dct.IDXST_IDCT(expkM, expkN) cuv = dct2.forward(density_map) cuv_by_wu2_plus_wv2_wu = cuv.mul(wu_by_wu2_plus_wv2_half) cuv_by_wu2_plus_wv2_wv = cuv.mul(wv_by_wu2_plus_wv2_half) field_map_x = idxst_idct.forward(cuv_by_wu2_plus_wv2_wu) field_map_y = idct_idxst.forward(cuv_by_wu2_plus_wv2_wv) cuv_by_wu2_plus_wv2 = cuv.mul(inv_wu2_plus_wv2) potential_map = idct2.forward(cuv_by_wu2_plus_wv2) energy = potential_map.mul(density_map).sum() if density_map.is_cuda: torch.cuda.synchronize() # compare results np.testing.assert_allclose(cuv.data.cpu().numpy(), auv_golden.data.cpu().numpy(), rtol=1e-6, atol=1e-5) np.testing.assert_allclose(field_map_x.data.cpu().numpy(), field_map_x_golden.data.cpu().numpy(), rtol=1e-6, atol=1e-5) np.testing.assert_allclose(field_map_y.data.cpu().numpy(), field_map_y_golden.data.cpu().numpy(), rtol=1e-6, atol=1e-5) np.testing.assert_allclose(potential_map.data.cpu().numpy(), potential_map_golden.data.cpu().numpy(), rtol=1e-6, atol=1e-5) np.testing.assert_allclose(energy.data.cpu().numpy(), energy_golden.data.cpu().numpy(), rtol=1e-6, atol=1e-5)
def forward(self, pos): if self.initial_density_map is None: if self.num_terminals == 0: num_fixed_impacted_bins_x = 0 num_fixed_impacted_bins_y = 0 else: num_fixed_impacted_bins_x = int( ((self.node_size_x[ self.num_movable_nodes:self.num_movable_nodes + self.num_terminals].max() + self.bin_size_x) / self.bin_size_x).ceil().clamp(max=self.num_bins_x)) num_fixed_impacted_bins_y = int( ((self.node_size_y[ self.num_movable_nodes:self.num_movable_nodes + self.num_terminals].max() + self.bin_size_y) / self.bin_size_y).ceil().clamp(max=self.num_bins_y)) if pos.is_cuda: self.initial_density_map = electric_potential_cuda.fixed_density_map( pos.view(pos.numel()), self.node_size_x, self.node_size_y, self.bin_center_x, self.bin_center_y, self.xl, self.yl, self.xh, self.yh, self.bin_size_x, self.bin_size_y, self.num_movable_nodes, self.num_terminals, self.num_bins_x, self.num_bins_y, num_fixed_impacted_bins_x, num_fixed_impacted_bins_y) else: self.initial_density_map = electric_potential_cpp.fixed_density_map( pos.view(pos.numel()), self.node_size_x, self.node_size_y, self.bin_center_x, self.bin_center_y, self.xl, self.yl, self.xh, self.yh, self.bin_size_x, self.bin_size_y, self.num_movable_nodes, self.num_terminals, self.num_bins_x, self.num_bins_y, num_fixed_impacted_bins_x, num_fixed_impacted_bins_y, self.num_threads) # plot(0, self.initial_density_map.clone().div(self.bin_size_x*self.bin_size_y).cpu().numpy(), self.padding, 'summary/initial_potential_map') # scale density of fixed macros self.initial_density_map.mul_(self.target_density) # expk M = self.num_bins_x N = self.num_bins_y self.exact_expkM = precompute_expk(M, dtype=pos.dtype, device=pos.device) self.exact_expkN = precompute_expk(N, dtype=pos.dtype, device=pos.device) # init dct2, idct2, idct_idxst, idxst_idct with expkM and expkN self.dct2 = dct.DCT2(self.exact_expkM, self.exact_expkN) if not self.fast_mode: self.idct2 = dct.IDCT2(self.exact_expkM, self.exact_expkN) self.idct_idxst = dct.IDCT_IDXST(self.exact_expkM, self.exact_expkN) self.idxst_idct = dct.IDXST_IDCT(self.exact_expkM, self.exact_expkN) # wu and wv wu = torch.arange(M, dtype=pos.dtype, device=pos.device).mul( 2 * np.pi / M).view([M, 1]) # scale wv because the aspect ratio of a bin may not be 1 wv = torch.arange(N, dtype=pos.dtype, device=pos.device).mul(2 * np.pi / N).view( [1, N]).mul_(self.bin_size_x / self.bin_size_y) wu2_plus_wv2 = wu.pow(2) + wv.pow(2) wu2_plus_wv2[0, 0] = 1.0 # avoid zero-division, it will be zeroed out self.inv_wu2_plus_wv2 = 1.0 / wu2_plus_wv2 self.inv_wu2_plus_wv2[0, 0] = 0.0 self.wu_by_wu2_plus_wv2_half = wu.mul(self.inv_wu2_plus_wv2).mul_( 1. / 2) self.wv_by_wu2_plus_wv2_half = wv.mul(self.inv_wu2_plus_wv2).mul_( 1. / 2) return ElectricPotentialFunction.apply( pos, self.node_size_x_clamped, self.node_size_y_clamped, self.offset_x, self.offset_y, self.ratio, self.bin_center_x, self.bin_center_y, self.initial_density_map, self.target_density, self.xl, self.yl, self.xh, self.yh, self.bin_size_x, self.bin_size_y, self.num_movable_nodes, self.num_filler_nodes, self.padding, self.padding_mask, self.num_bins_x, self.num_bins_y, self.num_movable_impacted_bins_x, self.num_movable_impacted_bins_y, self.num_filler_impacted_bins_x, self.num_filler_impacted_bins_y, self.sorted_node_map, self.exact_expkM, self.exact_expkN, self.inv_wu2_plus_wv2, self.wu_by_wu2_plus_wv2_half, self.wv_by_wu2_plus_wv2_half, self.dct2, self.idct2, self.idct_idxst, self.idxst_idct, self.fast_mode, self.num_threads)
def eval_dct2d(x, expk0, expk1, expkM, expkN, runs): x_numpy = x.data.cpu().numpy() torch.cuda.synchronize() tt = time.time() y = fftpack.dct(fftpack.dct(x_numpy.T, norm=None).T/x.size(1), norm=None)/x.size(0) torch.cuda.synchronize() print("CPU scipy.fftpack.dct2d takes %.7f ms" % ((time.time()-tt)*1000)) # 9s for 200 iterations 1024x1024 on GTX 1080 torch.cuda.synchronize() tt = time.time() # with torch.autograd.profiler.profile(use_cuda=True) as prof: for i in range(runs): y_2N = discrete_spectral_transform.dct2_2N(x, expk0=expk0, expk1=expk1) torch.cuda.synchronize() # print(prof) print("PyTorch: dct2d_2N takes %.7f ms" % ((time.time()-tt)/runs*1000)) # 11s for 200 iterations 1024x1024 on GTX 1080 perm0 = discrete_spectral_transform.get_perm(x.size(-2), dtype=torch.int64, device=x.device) perm1 = discrete_spectral_transform.get_perm(x.size(-1), dtype=torch.int64, device=x.device) torch.cuda.synchronize() tt = time.time() # with torch.autograd.profiler.profile(use_cuda=True) as prof: for i in range(runs): y_N = discrete_spectral_transform.dct2_N(x, perm0=perm0, expk0=expk0, perm1=perm1, expk1=expk1) torch.cuda.synchronize() # print(prof) print("PyTorch: dct2d_N takes %.7f ms" % ((time.time()-tt)/runs*1000)) dct2func = dct.DCT2(expk0, expk1, algorithm='2N') torch.cuda.synchronize() tt = time.time() # with torch.autograd.profiler.profile(use_cuda=True) as prof: for i in range(runs): y_2N = dct2func.forward(x) torch.cuda.synchronize() # print(prof) print("DCT2d_2N Function takes %.7f ms" % ((time.time()-tt)/runs*1000)) dct2func = dct.DCT2(expk0, expk1, algorithm='N') y_N = dct2func.forward(x) torch.cuda.synchronize() # with torch.autograd.profiler.profile(use_cuda=True) as prof: tt = time.time() for i in range(runs): y_N = dct2func.forward(x) torch.cuda.synchronize() # print(prof) print("DCT2d_N Function takes %.7f ms" % ((time.time()-tt)/runs*1000)) # The implementation below only supports float64 by now dct2func = dct_lee.DCT2(expk0, expk1) torch.cuda.synchronize() tt = time.time() # with torch.autograd.profiler.profile(use_cuda=True) as prof: for i in range(runs): y_N = dct2func.forward(x) torch.cuda.synchronize() # print(prof) print("DCT2d_Lee Function takes %.7f ms" % ((time.time()-tt)/runs*1000)) dct2func = dct2_fft2.DCT2(expkM, expkN) y = dct2func.forward(x) torch.cuda.synchronize() tt = time.time() for i in range(runs): y_test = dct2func.forward(x) torch.cuda.synchronize() print("DCT2_FFT2 Function takes %.7f ms" % ((time.time()-tt)/runs*1000)) print("")
def test_dct2Random(self): torch.manual_seed(10) M = 4 N = 8 x = torch.empty(M, N, dtype=dtype).uniform_(0, 10.0) expkM = discrete_spectral_transform.get_exact_expk(M, dtype=x.dtype, device=x.device) expkN = discrete_spectral_transform.get_exact_expk(N, dtype=x.dtype, device=x.device) golden_value = discrete_spectral_transform.dct2_N(x).data.numpy() print("2D DCT golden_value") print(golden_value) # test cpu using N-FFT # pdb.set_trace() custom = dct.DCT2(algorithm='N') dct_value = custom.forward(x) print("2D dct_value") print(dct_value.data.numpy()) np.testing.assert_allclose(dct_value.data.numpy(), golden_value, rtol=1e-6, atol=1e-5) # test cpu using 2N-FFT # pdb.set_trace() custom = dct.DCT2(algorithm='2N') dct_value = custom.forward(x) print("2D dct_value") print(dct_value.data.numpy()) np.testing.assert_allclose(dct_value.data.numpy(), golden_value, rtol=1e-6, atol=1e-5) # test cpu using dct_lee # pdb.set_trace() custom = dct_lee.DCT2() dct_value = custom.forward(x) print("2D dct_value") print(dct_value.data.numpy()) np.testing.assert_allclose(dct_value.data.numpy(), golden_value, rtol=1e-6, atol=1e-5) # test cpu using fft2 custom = dct2_fft2.DCT2(expkM, expkN) dct_value = custom.forward(x) print("2D dct_value") print(dct_value.data.numpy()) np.testing.assert_allclose(dct_value.data.numpy(), golden_value, rtol=1e-6, atol=1e-5) if torch.cuda.device_count(): # test gpu custom = dct.DCT2(algorithm='N') dct_value = custom.forward(x.cuda()).cpu() print("2D dct_value cuda") print(dct_value.data.numpy()) np.testing.assert_allclose(dct_value.data.numpy(), golden_value, rtol=1e-6, atol=1e-5) # test gpu custom = dct.DCT2(algorithm='2N') dct_value = custom.forward(x.cuda()).cpu() print("2D dct_value cuda") print(dct_value.data.numpy()) np.testing.assert_allclose(dct_value.data.numpy(), golden_value, rtol=1e-6, atol=1e-5) # test gpu custom = dct_lee.DCT2() dct_value = custom.forward(x.cuda()).cpu() print("2D dct_value cuda") print(dct_value.data.numpy()) np.testing.assert_allclose(dct_value.data.numpy(), golden_value, rtol=1e-6, atol=1e-5) # test gpu using fft2 custom = dct2_fft2.DCT2(expkM.cuda(), expkN.cuda()) dct_value = custom.forward(x.cuda()).cpu() print("2D dct_value cuda") print(dct_value.data.numpy()) np.testing.assert_allclose(dct_value.data.numpy(), golden_value, rtol=1e-6, atol=1e-5)