Cr8-xyz/blender
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Cycles: Use special debug panel to fine-tune debug flags

Browse Source 
This panel is only visible when debug_value is set to 256 and has no
affect in other cases. However, if debug value is not set to this
value, environment variables will be used to control which features
are enabled, so there's no visible changes to anyone in fact.

There are some changes needed to prevent devices re-enumeration on
every Cycles session create.

Reviewers: juicyfruit, lukasstockner97, dingto, brecht

Reviewed By: lukasstockner97, dingto

Differential Revision: https://developer.blender.org/D1720
This commit is contained in:
Sergey Sharybin
2016-01-12 16:00:48 +05:00
parent c6c223ade6
commit ac7aefd7c2
12 changed files with 561 additions and 89 deletions
Download Patch File Download Diff File Expand all files Collapse all files

62
intern/cycles/device/device_cpu.cpp
View File

@@ -78,6 +78,40 @@ public:
system_cpu_support_sse41();
system_cpu_support_avx();
system_cpu_support_avx2();
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX2
if(system_cpu_support_avx2()) {
VLOG(1) << "Will be using AVX2 kernels.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX
if(system_cpu_support_avx()) {
VLOG(1) << "Will be using AVX kernels.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE41
if(system_cpu_support_sse41()) {
VLOG(1) << "Will be using SSE4.1 kernels.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE3
if(system_cpu_support_sse3()) {
VLOG(1) << "Will be using SSE3kernels.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE2
if(system_cpu_support_sse2()) {
VLOG(1) << "Will be using SSE2 kernels.";
}
else
#endif
{
VLOG(1) << "Will be using regular kernels.";
}
}
~CPUDevice()
@@ -181,8 +215,6 @@ public:
void thread_path_trace(DeviceTask& task)
{
static bool cpu_type_logged = false;
if(task_pool.canceled()) {
if(task.need_finish_queue == false)
return;
@@ -201,41 +233,35 @@ public:
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX2
if(system_cpu_support_avx2()) {
path_trace_kernel = kernel_cpu_avx2_path_trace;
VLOG_ONCE(1, cpu_type_logged) << "Path tracing using AVX2 kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX
if(system_cpu_support_avx()) {
path_trace_kernel = kernel_cpu_avx_path_trace;
VLOG_ONCE(1, cpu_type_logged) << "Path tracing using AVX kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE41
if(system_cpu_support_sse41()) {
path_trace_kernel = kernel_cpu_sse41_path_trace;
VLOG_ONCE(1, cpu_type_logged) << "Path tracing using SSE4.1 kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE3
if(system_cpu_support_sse3()) {
path_trace_kernel = kernel_cpu_sse3_path_trace;
VLOG_ONCE(1, cpu_type_logged) << "Path tracing using SSE3 kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE2
if(system_cpu_support_sse2()) {
path_trace_kernel = kernel_cpu_sse2_path_trace;
VLOG_ONCE(1, cpu_type_logged) << "Path tracing using SSE2 kernel.";
}
else
#endif
{
path_trace_kernel = kernel_cpu_path_trace;
VLOG_ONCE(1, cpu_type_logged) << "Path tracing using regular kernel.";
}
while(task.acquire_tile(this, tile)) {
@@ -277,7 +303,6 @@ public:
void thread_film_convert(DeviceTask& task)
{
static bool cpu_type_logged = false;
float sample_scale = 1.0f/(task.sample + 1);
if(task.rgba_half) {
@@ -285,41 +310,35 @@ public:
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX2
if(system_cpu_support_avx2()) {
convert_to_half_float_kernel = kernel_cpu_avx2_convert_to_half_float;
VLOG_ONCE(1, cpu_type_logged) << "Converting to half float using AVX2 kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX
if(system_cpu_support_avx()) {
convert_to_half_float_kernel = kernel_cpu_avx_convert_to_half_float;
VLOG_ONCE(1, cpu_type_logged) << "Converting to half float using AVX kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE41
if(system_cpu_support_sse41()) {
convert_to_half_float_kernel = kernel_cpu_sse41_convert_to_half_float;
VLOG_ONCE(1, cpu_type_logged) << "Converting to half float using SSE4.1 kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE3
if(system_cpu_support_sse3()) {
convert_to_half_float_kernel = kernel_cpu_sse3_convert_to_half_float;
VLOG_ONCE(1, cpu_type_logged) << "Converting to half float using SSE3 kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE2
if(system_cpu_support_sse2()) {
convert_to_half_float_kernel = kernel_cpu_sse2_convert_to_half_float;
VLOG_ONCE(1, cpu_type_logged) << "Converting to half float using SSE2 kernel.";
}
else
#endif
{
convert_to_half_float_kernel = kernel_cpu_convert_to_half_float;
VLOG_ONCE(1, cpu_type_logged) << "Converting to half float using regular kernel.";
}
for(int y = task.y; y < task.y + task.h; y++)
@@ -332,41 +351,35 @@ public:
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX2
if(system_cpu_support_avx2()) {
convert_to_byte_kernel = kernel_cpu_avx2_convert_to_byte;
VLOG_ONCE(1, cpu_type_logged) << "Converting to byte using AVX2 kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX
if(system_cpu_support_avx()) {
convert_to_byte_kernel = kernel_cpu_avx_convert_to_byte;
VLOG_ONCE(1, cpu_type_logged) << "Converting to byte using AVX kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE41
if(system_cpu_support_sse41()) {
convert_to_byte_kernel = kernel_cpu_sse41_convert_to_byte;
VLOG_ONCE(1, cpu_type_logged) << "Converting to byte using SSE4.1 kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE3
if(system_cpu_support_sse3()) {
convert_to_byte_kernel = kernel_cpu_sse3_convert_to_byte;
VLOG_ONCE(1, cpu_type_logged) << "Converting to byte using SSE3 kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE2
if(system_cpu_support_sse2()) {
convert_to_byte_kernel = kernel_cpu_sse2_convert_to_byte;
VLOG_ONCE(1, cpu_type_logged) << "Converting to byte using SSE2 kernel.";
}
else
#endif
{
convert_to_byte_kernel = kernel_cpu_convert_to_byte;
VLOG_ONCE(1, cpu_type_logged) << "Converting to byte using regular kernel.";
}
for(int y = task.y; y < task.y + task.h; y++)
@@ -380,7 +393,6 @@ public:
void thread_shader(DeviceTask& task)
{
KernelGlobals kg = kernel_globals;
static bool cpu_type_logged = false;
#ifdef WITH_OSL
OSLShader::thread_init(&kg, &kernel_globals, &osl_globals);
@@ -390,41 +402,35 @@ public:
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX2
if(system_cpu_support_avx2()) {
shader_kernel = kernel_cpu_avx2_shader;
VLOG_ONCE(1, cpu_type_logged) << "Shading using AVX2 kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX
if(system_cpu_support_avx()) {
shader_kernel = kernel_cpu_avx_shader;
VLOG_ONCE(1, cpu_type_logged) << "Shading using AVX kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE41
if(system_cpu_support_sse41()) {
shader_kernel = kernel_cpu_sse41_shader;
VLOG_ONCE(1, cpu_type_logged) << "Shading using SSE4.1 kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE3
if(system_cpu_support_sse3()) {
shader_kernel = kernel_cpu_sse3_shader;
VLOG_ONCE(1, cpu_type_logged) << "Shading using SSE3 kernel.";
}
else
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE2
if(system_cpu_support_sse2()) {
shader_kernel = kernel_cpu_sse2_shader;
VLOG_ONCE(1, cpu_type_logged) << "Shading using SSE2 kernel.";
}
else
#endif
{
shader_kernel = kernel_cpu_shader;
VLOG_ONCE(1, cpu_type_logged) << "Shading using regular kernel.";
}
for(int sample = 0; sample < task.num_samples; sample++) {