diff options
author | Brecht Van Lommel <brecht@blender.org> | 2022-06-16 20:39:13 +0300 |
---|---|---|
committer | Brecht Van Lommel <brecht@blender.org> | 2022-06-17 15:08:14 +0300 |
commit | 2c1bffa286b9551c0533081cb16f497058974b03 (patch) | |
tree | d0b12d9a78bc798fcfa1754cbc7077f9b3bbf9f7 /intern/cycles/device | |
parent | 24246d98707096f16d5ab48f673f49354eac87a1 (diff) |
Cleanup: add verbose logging category names instead of numbers
And use them more consistently than before.
Diffstat (limited to 'intern/cycles/device')
-rw-r--r-- | intern/cycles/device/cpu/device_impl.cpp | 22 | ||||
-rw-r--r-- | intern/cycles/device/cuda/device.cpp | 28 | ||||
-rw-r--r-- | intern/cycles/device/cuda/device_impl.cpp | 41 | ||||
-rw-r--r-- | intern/cycles/device/cuda/queue.cpp | 6 | ||||
-rw-r--r-- | intern/cycles/device/device.cpp | 6 | ||||
-rw-r--r-- | intern/cycles/device/hip/device.cpp | 25 | ||||
-rw-r--r-- | intern/cycles/device/hip/device_impl.cpp | 36 | ||||
-rw-r--r-- | intern/cycles/device/hip/queue.cpp | 6 | ||||
-rw-r--r-- | intern/cycles/device/metal/device_impl.mm | 18 | ||||
-rw-r--r-- | intern/cycles/device/metal/queue.mm | 4 | ||||
-rw-r--r-- | intern/cycles/device/optix/device.cpp | 6 | ||||
-rw-r--r-- | intern/cycles/device/optix/device_impl.cpp | 6 | ||||
-rw-r--r-- | intern/cycles/device/queue.cpp | 21 |
13 files changed, 115 insertions, 110 deletions
diff --git a/intern/cycles/device/cpu/device_impl.cpp b/intern/cycles/device/cpu/device_impl.cpp index 612c391f7d5..0a4eb089037 100644 --- a/intern/cycles/device/cpu/device_impl.cpp +++ b/intern/cycles/device/cpu/device_impl.cpp @@ -55,8 +55,8 @@ CPUDevice::CPUDevice(const DeviceInfo &info_, Stats &stats_, Profiler &profiler_ { /* Pick any kernel, all of them are supposed to have same level of microarchitecture * optimization. */ - VLOG(1) << "Using " << get_cpu_kernels().integrator_init_from_camera.get_uarch_name() - << " CPU kernels."; + VLOG_INFO << "Using " << get_cpu_kernels().integrator_init_from_camera.get_uarch_name() + << " CPU kernels."; if (info.cpu_threads == 0) { info.cpu_threads = TaskScheduler::max_concurrency(); @@ -111,9 +111,9 @@ void CPUDevice::mem_alloc(device_memory &mem) } else { if (mem.name) { - VLOG(1) << "Buffer allocate: " << mem.name << ", " - << string_human_readable_number(mem.memory_size()) << " bytes. (" - << string_human_readable_size(mem.memory_size()) << ")"; + VLOG_WORK << "Buffer allocate: " << mem.name << ", " + << string_human_readable_number(mem.memory_size()) << " bytes. (" + << string_human_readable_size(mem.memory_size()) << ")"; } if (mem.type == MEM_DEVICE_ONLY || !mem.host_pointer) { @@ -205,9 +205,9 @@ void CPUDevice::const_copy_to(const char *name, void *host, size_t size) void CPUDevice::global_alloc(device_memory &mem) { - VLOG(1) << "Global memory allocate: " << mem.name << ", " - << string_human_readable_number(mem.memory_size()) << " bytes. (" - << string_human_readable_size(mem.memory_size()) << ")"; + VLOG_WORK << "Global memory allocate: " << mem.name << ", " + << string_human_readable_number(mem.memory_size()) << " bytes. (" + << string_human_readable_size(mem.memory_size()) << ")"; kernel_global_memory_copy(&kernel_globals, mem.name, mem.host_pointer, mem.data_size); @@ -227,9 +227,9 @@ void CPUDevice::global_free(device_memory &mem) void CPUDevice::tex_alloc(device_texture &mem) { - VLOG(1) << "Texture allocate: " << mem.name << ", " - << string_human_readable_number(mem.memory_size()) << " bytes. (" - << string_human_readable_size(mem.memory_size()) << ")"; + VLOG_WORK << "Texture allocate: " << mem.name << ", " + << string_human_readable_number(mem.memory_size()) << " bytes. (" + << string_human_readable_size(mem.memory_size()) << ")"; mem.device_pointer = (device_ptr)mem.host_pointer; mem.device_size = mem.memory_size(); diff --git a/intern/cycles/device/cuda/device.cpp b/intern/cycles/device/cuda/device.cpp index 400490336d6..5a213c45b71 100644 --- a/intern/cycles/device/cuda/device.cpp +++ b/intern/cycles/device/cuda/device.cpp @@ -29,24 +29,25 @@ bool device_cuda_init() initialized = true; int cuew_result = cuewInit(CUEW_INIT_CUDA); if (cuew_result == CUEW_SUCCESS) { - VLOG(1) << "CUEW initialization succeeded"; + VLOG_INFO << "CUEW initialization succeeded"; if (CUDADevice::have_precompiled_kernels()) { - VLOG(1) << "Found precompiled kernels"; + VLOG_INFO << "Found precompiled kernels"; result = true; } else if (cuewCompilerPath() != NULL) { - VLOG(1) << "Found CUDA compiler " << cuewCompilerPath(); + VLOG_INFO << "Found CUDA compiler " << cuewCompilerPath(); result = true; } else { - VLOG(1) << "Neither precompiled kernels nor CUDA compiler was found," - << " unable to use CUDA"; + VLOG_INFO << "Neither precompiled kernels nor CUDA compiler was found," + << " unable to use CUDA"; } } else { - VLOG(1) << "CUEW initialization failed: " - << ((cuew_result == CUEW_ERROR_ATEXIT_FAILED) ? "Error setting up atexit() handler" : - "Error opening the library"); + VLOG_WARNING << "CUEW initialization failed: " + << ((cuew_result == CUEW_ERROR_ATEXIT_FAILED) ? + "Error setting up atexit() handler" : + "Error opening the library"); } return result; @@ -121,7 +122,8 @@ void device_cuda_info(vector<DeviceInfo> &devices) int major; cuDeviceGetAttribute(&major, CU_DEVICE_ATTRIBUTE_COMPUTE_CAPABILITY_MAJOR, num); if (major < 3) { - VLOG(1) << "Ignoring device \"" << name << "\", this graphics card is no longer supported."; + VLOG_INFO << "Ignoring device \"" << name + << "\", this graphics card is no longer supported."; continue; } @@ -166,21 +168,21 @@ void device_cuda_info(vector<DeviceInfo> &devices) * Windows 10 even when it is, due to an issue in application profiles. * Detect case where we expect it to be available and override. */ if (preempt_attr == 0 && (major >= 6) && system_windows_version_at_least(10, 17134)) { - VLOG(1) << "Assuming device has compute preemption on Windows 10."; + VLOG_INFO << "Assuming device has compute preemption on Windows 10."; preempt_attr = 1; } if (timeout_attr && !preempt_attr) { - VLOG(1) << "Device is recognized as display."; + VLOG_INFO << "Device is recognized as display."; info.description += " (Display)"; info.display_device = true; display_devices.push_back(info); } else { - VLOG(1) << "Device has compute preemption or is not used for display."; + VLOG_INFO << "Device has compute preemption or is not used for display."; devices.push_back(info); } - VLOG(1) << "Added device \"" << name << "\" with id \"" << info.id << "\"."; + VLOG_INFO << "Added device \"" << name << "\" with id \"" << info.id << "\"."; } if (!display_devices.empty()) diff --git a/intern/cycles/device/cuda/device_impl.cpp b/intern/cycles/device/cuda/device_impl.cpp index cb7e909a2d5..e75224abe90 100644 --- a/intern/cycles/device/cuda/device_impl.cpp +++ b/intern/cycles/device/cuda/device_impl.cpp @@ -244,9 +244,9 @@ string CUDADevice::compile_kernel(const uint kernel_features, if (!use_adaptive_compilation()) { if (!force_ptx) { const string cubin = path_get(string_printf("lib/%s_sm_%d%d.cubin", name, major, minor)); - VLOG(1) << "Testing for pre-compiled kernel " << cubin << "."; + VLOG_INFO << "Testing for pre-compiled kernel " << cubin << "."; if (path_exists(cubin)) { - VLOG(1) << "Using precompiled kernel."; + VLOG_INFO << "Using precompiled kernel."; return cubin; } } @@ -256,9 +256,9 @@ string CUDADevice::compile_kernel(const uint kernel_features, while (ptx_major >= 3) { const string ptx = path_get( string_printf("lib/%s_compute_%d%d.ptx", name, ptx_major, ptx_minor)); - VLOG(1) << "Testing for pre-compiled kernel " << ptx << "."; + VLOG_INFO << "Testing for pre-compiled kernel " << ptx << "."; if (path_exists(ptx)) { - VLOG(1) << "Using precompiled kernel."; + VLOG_INFO << "Using precompiled kernel."; return ptx; } @@ -287,9 +287,9 @@ string CUDADevice::compile_kernel(const uint kernel_features, const string cubin_file = string_printf( "cycles_%s_%s_%d%d_%s.%s", name, kernel_arch, major, minor, kernel_md5.c_str(), kernel_ext); const string cubin = path_cache_get(path_join("kernels", cubin_file)); - VLOG(1) << "Testing for locally compiled kernel " << cubin << "."; + VLOG_INFO << "Testing for locally compiled kernel " << cubin << "."; if (path_exists(cubin)) { - VLOG(1) << "Using locally compiled kernel."; + VLOG_INFO << "Using locally compiled kernel."; return cubin; } @@ -323,7 +323,7 @@ string CUDADevice::compile_kernel(const uint kernel_features, } const int nvcc_cuda_version = cuewCompilerVersion(); - VLOG(1) << "Found nvcc " << nvcc << ", CUDA version " << nvcc_cuda_version << "."; + VLOG_INFO << "Found nvcc " << nvcc << ", CUDA version " << nvcc_cuda_version << "."; if (nvcc_cuda_version < 101) { printf( "Unsupported CUDA version %d.%d detected, " @@ -399,7 +399,8 @@ bool CUDADevice::load_kernels(const uint kernel_features) */ if (cuModule) { if (use_adaptive_compilation()) { - VLOG(1) << "Skipping CUDA kernel reload for adaptive compilation, not currently supported."; + VLOG_INFO + << "Skipping CUDA kernel reload for adaptive compilation, not currently supported."; } return true; } @@ -481,8 +482,8 @@ void CUDADevice::reserve_local_memory(const uint kernel_features) cuMemGetInfo(&free_after, &total); } - VLOG(1) << "Local memory reserved " << string_human_readable_number(free_before - free_after) - << " bytes. (" << string_human_readable_size(free_before - free_after) << ")"; + VLOG_INFO << "Local memory reserved " << string_human_readable_number(free_before - free_after) + << " bytes. (" << string_human_readable_size(free_before - free_after) << ")"; # if 0 /* For testing mapped host memory, fill up device memory. */ @@ -513,7 +514,7 @@ void CUDADevice::init_host_memory() } } else { - VLOG(1) << "Mapped host memory disabled, failed to get system RAM"; + VLOG_WARNING << "Mapped host memory disabled, failed to get system RAM"; map_host_limit = 0; } @@ -524,8 +525,8 @@ void CUDADevice::init_host_memory() device_working_headroom = 32 * 1024 * 1024LL; // 32MB device_texture_headroom = 128 * 1024 * 1024LL; // 128MB - VLOG(1) << "Mapped host memory limit set to " << string_human_readable_number(map_host_limit) - << " bytes. (" << string_human_readable_size(map_host_limit) << ")"; + VLOG_INFO << "Mapped host memory limit set to " << string_human_readable_number(map_host_limit) + << " bytes. (" << string_human_readable_size(map_host_limit) << ")"; } void CUDADevice::load_texture_info() @@ -593,7 +594,7 @@ void CUDADevice::move_textures_to_host(size_t size, bool for_texture) * multiple CUDA devices could be moving the memory. The * first one will do it, and the rest will adopt the pointer. */ if (max_mem) { - VLOG(1) << "Move memory from device to host: " << max_mem->name; + VLOG_WORK << "Move memory from device to host: " << max_mem->name; static thread_mutex move_mutex; thread_scoped_lock lock(move_mutex); @@ -701,9 +702,9 @@ CUDADevice::CUDAMem *CUDADevice::generic_alloc(device_memory &mem, size_t pitch_ } if (mem.name) { - VLOG(1) << "Buffer allocate: " << mem.name << ", " - << string_human_readable_number(mem.memory_size()) << " bytes. (" - << string_human_readable_size(mem.memory_size()) << ")" << status; + VLOG_WORK << "Buffer allocate: " << mem.name << ", " + << string_human_readable_number(mem.memory_size()) << " bytes. (" + << string_human_readable_size(mem.memory_size()) << ")" << status; } mem.device_pointer = (device_ptr)device_pointer; @@ -1008,9 +1009,9 @@ void CUDADevice::tex_alloc(device_texture &mem) desc.NumChannels = mem.data_elements; desc.Flags = 0; - VLOG(1) << "Array 3D allocate: " << mem.name << ", " - << string_human_readable_number(mem.memory_size()) << " bytes. (" - << string_human_readable_size(mem.memory_size()) << ")"; + VLOG_WORK << "Array 3D allocate: " << mem.name << ", " + << string_human_readable_number(mem.memory_size()) << " bytes. (" + << string_human_readable_size(mem.memory_size()) << ")"; cuda_assert(cuArray3DCreate(&array_3d, &desc)); diff --git a/intern/cycles/device/cuda/queue.cpp b/intern/cycles/device/cuda/queue.cpp index 38c71866ad0..5912e68a92b 100644 --- a/intern/cycles/device/cuda/queue.cpp +++ b/intern/cycles/device/cuda/queue.cpp @@ -39,12 +39,12 @@ int CUDADeviceQueue::num_concurrent_states(const size_t state_size) const num_states = max((int)(num_states * factor), 1024); } else { - VLOG(3) << "CYCLES_CONCURRENT_STATES_FACTOR evaluated to 0"; + VLOG_DEVICE_STATS << "CYCLES_CONCURRENT_STATES_FACTOR evaluated to 0"; } } - VLOG(3) << "GPU queue concurrent states: " << num_states << ", using up to " - << string_human_readable_size(num_states * state_size); + VLOG_DEVICE_STATS << "GPU queue concurrent states: " << num_states << ", using up to " + << string_human_readable_size(num_states * state_size); return num_states; } diff --git a/intern/cycles/device/device.cpp b/intern/cycles/device/device.cpp index ea5b3c6dc8c..82c7881da5f 100644 --- a/intern/cycles/device/device.cpp +++ b/intern/cycles/device/device.cpp @@ -325,8 +325,8 @@ DeviceInfo Device::get_multi_device(const vector<DeviceInfo> &subdevices, int orig_cpu_threads = (threads) ? threads : TaskScheduler::max_concurrency(); int cpu_threads = max(orig_cpu_threads - (subdevices.size() - 1), size_t(0)); - VLOG(1) << "CPU render threads reduced from " << orig_cpu_threads << " to " << cpu_threads - << ", to dedicate to GPU."; + VLOG_INFO << "CPU render threads reduced from " << orig_cpu_threads << " to " + << cpu_threads << ", to dedicate to GPU."; if (cpu_threads >= 1) { DeviceInfo cpu_device = device; @@ -338,7 +338,7 @@ DeviceInfo Device::get_multi_device(const vector<DeviceInfo> &subdevices, } } else { - VLOG(1) << "CPU render threads disabled for interactive render."; + VLOG_INFO << "CPU render threads disabled for interactive render."; continue; } } diff --git a/intern/cycles/device/hip/device.cpp b/intern/cycles/device/hip/device.cpp index d6a5ed9c419..3c9c73e7db0 100644 --- a/intern/cycles/device/hip/device.cpp +++ b/intern/cycles/device/hip/device.cpp @@ -29,30 +29,31 @@ bool device_hip_init() initialized = true; int hipew_result = hipewInit(HIPEW_INIT_HIP); if (hipew_result == HIPEW_SUCCESS) { - VLOG(1) << "HIPEW initialization succeeded"; + VLOG_INFO << "HIPEW initialization succeeded"; if (HIPDevice::have_precompiled_kernels()) { - VLOG(1) << "Found precompiled kernels"; + VLOG_INFO << "Found precompiled kernels"; result = true; } else if (hipewCompilerPath() != NULL) { - VLOG(1) << "Found HIPCC " << hipewCompilerPath(); + VLOG_INFO << "Found HIPCC " << hipewCompilerPath(); result = true; } else { - VLOG(1) << "Neither precompiled kernels nor HIPCC was found," - << " unable to use HIP"; + VLOG_INFO << "Neither precompiled kernels nor HIPCC was found," + << " unable to use HIP"; } } else { if (hipew_result == HIPEW_ERROR_ATEXIT_FAILED) { - VLOG(1) << "HIPEW initialization failed: Error setting up atexit() handler"; + VLOG_WARNING << "HIPEW initialization failed: Error setting up atexit() handler"; } else if (hipew_result == HIPEW_ERROR_OLD_DRIVER) { - VLOG(1) << "HIPEW initialization failed: Driver version too old, requires AMD Radeon Pro " - "21.Q4 driver or newer"; + VLOG_WARNING + << "HIPEW initialization failed: Driver version too old, requires AMD Radeon Pro " + "21.Q4 driver or newer"; } else { - VLOG(1) << "HIPEW initialization failed: Error opening HIP dynamic library"; + VLOG_WARNING << "HIPEW initialization failed: Error opening HIP dynamic library"; } } @@ -165,16 +166,16 @@ void device_hip_info(vector<DeviceInfo> &devices) hipDeviceGetAttribute(&timeout_attr, hipDeviceAttributeKernelExecTimeout, num); if (timeout_attr && !preempt_attr) { - VLOG(1) << "Device is recognized as display."; + VLOG_INFO << "Device is recognized as display."; info.description += " (Display)"; info.display_device = true; display_devices.push_back(info); } else { - VLOG(1) << "Device has compute preemption or is not used for display."; + VLOG_INFO << "Device has compute preemption or is not used for display."; devices.push_back(info); } - VLOG(1) << "Added device \"" << name << "\" with id \"" << info.id << "\"."; + VLOG_INFO << "Added device \"" << name << "\" with id \"" << info.id << "\"."; } if (!display_devices.empty()) diff --git a/intern/cycles/device/hip/device_impl.cpp b/intern/cycles/device/hip/device_impl.cpp index ea68c821166..652c1001f85 100644 --- a/intern/cycles/device/hip/device_impl.cpp +++ b/intern/cycles/device/hip/device_impl.cpp @@ -233,9 +233,9 @@ string HIPDevice::compile_kernel(const uint kernel_features, const char *name, c /* Attempt to use kernel provided with Blender. */ if (!use_adaptive_compilation()) { const string fatbin = path_get(string_printf("lib/%s_%s.fatbin", name, arch)); - VLOG(1) << "Testing for pre-compiled kernel " << fatbin << "."; + VLOG_INFO << "Testing for pre-compiled kernel " << fatbin << "."; if (path_exists(fatbin)) { - VLOG(1) << "Using precompiled kernel."; + VLOG_INFO << "Using precompiled kernel."; return fatbin; } } @@ -265,9 +265,9 @@ string HIPDevice::compile_kernel(const uint kernel_features, const char *name, c const string include_path = source_path; const string fatbin_file = string_printf("cycles_%s_%s_%s", name, arch, kernel_md5.c_str()); const string fatbin = path_cache_get(path_join("kernels", fatbin_file)); - VLOG(1) << "Testing for locally compiled kernel " << fatbin << "."; + VLOG_INFO << "Testing for locally compiled kernel " << fatbin << "."; if (path_exists(fatbin)) { - VLOG(1) << "Using locally compiled kernel."; + VLOG_INFO << "Using locally compiled kernel."; return fatbin; } @@ -301,7 +301,7 @@ string HIPDevice::compile_kernel(const uint kernel_features, const char *name, c } const int hipcc_hip_version = hipewCompilerVersion(); - VLOG(1) << "Found hipcc " << hipcc << ", HIP version " << hipcc_hip_version << "."; + VLOG_INFO << "Found hipcc " << hipcc << ", HIP version " << hipcc_hip_version << "."; if (hipcc_hip_version < 40) { printf( "Unsupported HIP version %d.%d detected, " @@ -361,7 +361,7 @@ bool HIPDevice::load_kernels(const uint kernel_features) */ if (hipModule) { if (use_adaptive_compilation()) { - VLOG(1) << "Skipping HIP kernel reload for adaptive compilation, not currently supported."; + VLOG_INFO << "Skipping HIP kernel reload for adaptive compilation, not currently supported."; } return true; } @@ -444,8 +444,8 @@ void HIPDevice::reserve_local_memory(const uint kernel_features) hipMemGetInfo(&free_after, &total); } - VLOG(1) << "Local memory reserved " << string_human_readable_number(free_before - free_after) - << " bytes. (" << string_human_readable_size(free_before - free_after) << ")"; + VLOG_INFO << "Local memory reserved " << string_human_readable_number(free_before - free_after) + << " bytes. (" << string_human_readable_size(free_before - free_after) << ")"; # if 0 /* For testing mapped host memory, fill up device memory. */ @@ -476,7 +476,7 @@ void HIPDevice::init_host_memory() } } else { - VLOG(1) << "Mapped host memory disabled, failed to get system RAM"; + VLOG_WARNING << "Mapped host memory disabled, failed to get system RAM"; map_host_limit = 0; } @@ -487,8 +487,8 @@ void HIPDevice::init_host_memory() device_working_headroom = 32 * 1024 * 1024LL; // 32MB device_texture_headroom = 128 * 1024 * 1024LL; // 128MB - VLOG(1) << "Mapped host memory limit set to " << string_human_readable_number(map_host_limit) - << " bytes. (" << string_human_readable_size(map_host_limit) << ")"; + VLOG_INFO << "Mapped host memory limit set to " << string_human_readable_number(map_host_limit) + << " bytes. (" << string_human_readable_size(map_host_limit) << ")"; } void HIPDevice::load_texture_info() @@ -556,7 +556,7 @@ void HIPDevice::move_textures_to_host(size_t size, bool for_texture) * multiple HIP devices could be moving the memory. The * first one will do it, and the rest will adopt the pointer. */ if (max_mem) { - VLOG(1) << "Move memory from device to host: " << max_mem->name; + VLOG_WORK << "Move memory from device to host: " << max_mem->name; static thread_mutex move_mutex; thread_scoped_lock lock(move_mutex); @@ -658,9 +658,9 @@ HIPDevice::HIPMem *HIPDevice::generic_alloc(device_memory &mem, size_t pitch_pad } if (mem.name) { - VLOG(1) << "Buffer allocate: " << mem.name << ", " - << string_human_readable_number(mem.memory_size()) << " bytes. (" - << string_human_readable_size(mem.memory_size()) << ")" << status; + VLOG_WORK << "Buffer allocate: " << mem.name << ", " + << string_human_readable_number(mem.memory_size()) << " bytes. (" + << string_human_readable_size(mem.memory_size()) << ")" << status; } mem.device_pointer = (device_ptr)device_pointer; @@ -966,9 +966,9 @@ void HIPDevice::tex_alloc(device_texture &mem) desc.NumChannels = mem.data_elements; desc.Flags = 0; - VLOG(1) << "Array 3D allocate: " << mem.name << ", " - << string_human_readable_number(mem.memory_size()) << " bytes. (" - << string_human_readable_size(mem.memory_size()) << ")"; + VLOG_WORK << "Array 3D allocate: " << mem.name << ", " + << string_human_readable_number(mem.memory_size()) << " bytes. (" + << string_human_readable_size(mem.memory_size()) << ")"; hip_assert(hipArray3DCreate((hArray *)&array_3d, &desc)); diff --git a/intern/cycles/device/hip/queue.cpp b/intern/cycles/device/hip/queue.cpp index 6c2c2c29624..8b3d963a32f 100644 --- a/intern/cycles/device/hip/queue.cpp +++ b/intern/cycles/device/hip/queue.cpp @@ -39,12 +39,12 @@ int HIPDeviceQueue::num_concurrent_states(const size_t state_size) const num_states = max((int)(num_states * factor), 1024); } else { - VLOG(3) << "CYCLES_CONCURRENT_STATES_FACTOR evaluated to 0"; + VLOG_DEVICE_STATS << "CYCLES_CONCURRENT_STATES_FACTOR evaluated to 0"; } } - VLOG(3) << "GPU queue concurrent states: " << num_states << ", using up to " - << string_human_readable_size(num_states * state_size); + VLOG_DEVICE_STATS << "GPU queue concurrent states: " << num_states << ", using up to " + << string_human_readable_size(num_states * state_size); return num_states; } diff --git a/intern/cycles/device/metal/device_impl.mm b/intern/cycles/device/metal/device_impl.mm index 086bf0af979..a0ac677beda 100644 --- a/intern/cycles/device/metal/device_impl.mm +++ b/intern/cycles/device/metal/device_impl.mm @@ -411,9 +411,9 @@ MetalDevice::MetalMem *MetalDevice::generic_alloc(device_memory &mem) } if (mem.name) { - VLOG(2) << "Buffer allocate: " << mem.name << ", " - << string_human_readable_number(mem.memory_size()) << " bytes. (" - << string_human_readable_size(mem.memory_size()) << ")"; + VLOG_WORK << "Buffer allocate: " << mem.name << ", " + << string_human_readable_number(mem.memory_size()) << " bytes. (" + << string_human_readable_size(mem.memory_size()) << ")"; } mem.device_size = metal_buffer.allocatedSize; @@ -800,9 +800,9 @@ void MetalDevice::tex_alloc(device_texture &mem) desc.textureType = MTLTextureType3D; desc.depth = mem.data_depth; - VLOG(2) << "Texture 3D allocate: " << mem.name << ", " - << string_human_readable_number(mem.memory_size()) << " bytes. (" - << string_human_readable_size(mem.memory_size()) << ")"; + VLOG_WORK << "Texture 3D allocate: " << mem.name << ", " + << string_human_readable_number(mem.memory_size()) << " bytes. (" + << string_human_readable_size(mem.memory_size()) << ")"; mtlTexture = [mtlDevice newTextureWithDescriptor:desc]; assert(mtlTexture); @@ -834,9 +834,9 @@ void MetalDevice::tex_alloc(device_texture &mem) desc.storageMode = storage_mode; desc.usage = MTLTextureUsageShaderRead; - VLOG(2) << "Texture 2D allocate: " << mem.name << ", " - << string_human_readable_number(mem.memory_size()) << " bytes. (" - << string_human_readable_size(mem.memory_size()) << ")"; + VLOG_WORK << "Texture 2D allocate: " << mem.name << ", " + << string_human_readable_number(mem.memory_size()) << " bytes. (" + << string_human_readable_size(mem.memory_size()) << ")"; mtlTexture = [mtlDevice newTextureWithDescriptor:desc]; assert(mtlTexture); diff --git a/intern/cycles/device/metal/queue.mm b/intern/cycles/device/metal/queue.mm index 0e260886abb..55db7c5afce 100644 --- a/intern/cycles/device/metal/queue.mm +++ b/intern/cycles/device/metal/queue.mm @@ -311,8 +311,8 @@ bool MetalDeviceQueue::enqueue(DeviceKernel kernel, return false; } - VLOG(3) << "Metal queue launch " << device_kernel_as_string(kernel) << ", work_size " - << work_size; + VLOG_DEVICE_STATS << "Metal queue launch " << device_kernel_as_string(kernel) << ", work_size " + << work_size; id<MTLComputeCommandEncoder> mtlComputeCommandEncoder = get_compute_encoder(kernel); diff --git a/intern/cycles/device/optix/device.cpp b/intern/cycles/device/optix/device.cpp index 70810bae10d..68ca21374fd 100644 --- a/intern/cycles/device/optix/device.cpp +++ b/intern/cycles/device/optix/device.cpp @@ -31,12 +31,12 @@ bool device_optix_init() const OptixResult result = optixInit(); if (result == OPTIX_ERROR_UNSUPPORTED_ABI_VERSION) { - VLOG(1) << "OptiX initialization failed because the installed NVIDIA driver is too old. " - "Please update to the latest driver first!"; + VLOG_WARNING << "OptiX initialization failed because the installed NVIDIA driver is too old. " + "Please update to the latest driver first!"; return false; } else if (result != OPTIX_SUCCESS) { - VLOG(1) << "OptiX initialization failed with error code " << (unsigned int)result; + VLOG_WARNING << "OptiX initialization failed with error code " << (unsigned int)result; return false; } diff --git a/intern/cycles/device/optix/device_impl.cpp b/intern/cycles/device/optix/device_impl.cpp index 9ab9bbb59c5..53697db5c04 100644 --- a/intern/cycles/device/optix/device_impl.cpp +++ b/intern/cycles/device/optix/device_impl.cpp @@ -278,7 +278,7 @@ OptiXDevice::OptiXDevice(const DeviceInfo &info, Stats &stats, Profiler &profile }; # endif if (DebugFlags().optix.use_debug) { - VLOG(1) << "Using OptiX debug mode."; + VLOG_INFO << "Using OptiX debug mode."; options.validationMode = OPTIX_DEVICE_CONTEXT_VALIDATION_MODE_ALL; } optix_assert(optixDeviceContextCreate(cuContext, &options, &context)); @@ -1392,11 +1392,11 @@ bool OptiXDevice::build_optix_bvh(BVHOptiX *bvh, /* The build flags have to match the ones used to query the built-in curve intersection program (see optixBuiltinISModuleGet above) */ build_input.type == OPTIX_BUILD_INPUT_TYPE_CURVES) { - VLOG(2) << "Using fast to trace OptiX BVH"; + VLOG_INFO << "Using fast to trace OptiX BVH"; options.buildFlags = OPTIX_BUILD_FLAG_PREFER_FAST_TRACE | OPTIX_BUILD_FLAG_ALLOW_COMPACTION; } else { - VLOG(2) << "Using fast to update OptiX BVH"; + VLOG_INFO << "Using fast to update OptiX BVH"; options.buildFlags = OPTIX_BUILD_FLAG_PREFER_FAST_BUILD | OPTIX_BUILD_FLAG_ALLOW_UPDATE; } diff --git a/intern/cycles/device/queue.cpp b/intern/cycles/device/queue.cpp index de65047ed6a..cc0cf0ccf84 100644 --- a/intern/cycles/device/queue.cpp +++ b/intern/cycles/device/queue.cpp @@ -19,7 +19,7 @@ DeviceQueue::DeviceQueue(Device *device) DeviceQueue::~DeviceQueue() { - if (VLOG_IS_ON(3)) { + if (VLOG_DEVICE_STATS_IS_ON) { /* Print kernel execution times sorted by time. */ vector<pair<DeviceKernelMask, double>> stats_sorted; for (const auto &stat : stats_kernel_time_) { @@ -32,17 +32,18 @@ DeviceQueue::~DeviceQueue() return a.second > b.second; }); - VLOG(3) << "GPU queue stats:"; + VLOG_DEVICE_STATS << "GPU queue stats:"; for (const auto &[mask, time] : stats_sorted) { - VLOG(3) << " " << std::setfill(' ') << std::setw(10) << std::fixed << std::setprecision(5) - << std::right << time << "s: " << device_kernel_mask_as_string(mask); + VLOG_DEVICE_STATS << " " << std::setfill(' ') << std::setw(10) << std::fixed + << std::setprecision(5) << std::right << time + << "s: " << device_kernel_mask_as_string(mask); } } } void DeviceQueue::debug_init_execution() { - if (VLOG_IS_ON(3)) { + if (VLOG_DEVICE_STATS_IS_ON) { last_sync_time_ = time_dt(); } @@ -51,9 +52,9 @@ void DeviceQueue::debug_init_execution() void DeviceQueue::debug_enqueue(DeviceKernel kernel, const int work_size) { - if (VLOG_IS_ON(3)) { - VLOG(4) << "GPU queue launch " << device_kernel_as_string(kernel) << ", work_size " - << work_size; + if (VLOG_DEVICE_STATS_IS_ON) { + VLOG_DEVICE_STATS << "GPU queue launch " << device_kernel_as_string(kernel) << ", work_size " + << work_size; } last_kernels_enqueued_ |= (uint64_t(1) << (uint64_t)kernel); @@ -61,10 +62,10 @@ void DeviceQueue::debug_enqueue(DeviceKernel kernel, const int work_size) void DeviceQueue::debug_synchronize() { - if (VLOG_IS_ON(3)) { + if (VLOG_DEVICE_STATS_IS_ON) { const double new_time = time_dt(); const double elapsed_time = new_time - last_sync_time_; - VLOG(4) << "GPU queue synchronize, elapsed " << std::setw(10) << elapsed_time << "s"; + VLOG_DEVICE_STATS << "GPU queue synchronize, elapsed " << std::setw(10) << elapsed_time << "s"; stats_kernel_time_[last_kernels_enqueued_] += elapsed_time; |