I am trying to explore various optimization available in the XLA and MLIR. Is there a way, I can use certain flags like -o in clang and other compilers, or say LLVM optimizer manager? In the XLA documentation, I could find a few flags but that doesn’t provide means to play with the optimizations individually.
I highly appreciate any leads provided.
I am trying to explore various optimization available in the XLA and MLIR. Is there a way, I can use certain flags like -o in clang and other compilers, or say LLVM optimizer manager? In the XLA documentation, I could find a few flags but that doesn’t provide means to play with the optimizations individually.
I highly appreciate any leads provided.
--
You received this message because you are subscribed to the Google Groups "XLA development" group.
To unsubscribe from this group and stop receiving emails from it, send an email to xla-dev+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/xla-dev/320d508a-2efa-4c38-9e65-b71d44f3a7d9%40googlegroups.com.
Hi Gaurav,Are you concerned with the compile time? For those there is `env XLA_FLAGS=--xla_backend_optimization_level=0/1/2/3` for LLVM optimizations, as those are usually the costliest.There is also `--xla_gpu_disable_gpuasm_optimizations` for ptxas.In general, flags are passed using the environment variable XLA_FLAGS, and you can see all flags at xla.proto: https://cs.opensource.google/tensorflow/tensorflow/+/master:tensorflow/compiler/xla/xla.proto?q=file:xla.protoGeorge
On Thu, Apr 30, 2020 at 12:44 PM Gaurav Verma <gaura...@stonybrook.edu> wrote:
--I am trying to explore various optimization available in the XLA and MLIR. Is there a way, I can use certain flags like -o in clang and other compilers, or say LLVM optimizer manager? In the XLA documentation, I could find a few flags but that doesn’t provide means to play with the optimizations individually.
I highly appreciate any leads provided.
You received this message because you are subscribed to the Google Groups "XLA development" group.
To unsubscribe from this group and stop receiving emails from it, send an email to xla...@googlegroups.com.
| |
int32 opt_level = hlo_module_config.debug_options().xla_backend_optimization_level();
if (opt_level < 2) {
LOG(ERROR) << std::string(80, '*');
LOG(ERROR) << "The XLA GPU backend doesn't support unoptimized code "
"generation but ";
LOG(ERROR) << "--xla_backend_optimization_level is set to " << opt_level
<< "!";
LOG(ERROR) << "(Supported configuration is "
"--xla_backend_optimization_level >= 2.)";
LOG(ERROR) << std::string(80, '*');
}I understand that 0 and 1 will have a hit on the performance. But was just curious to know. I wanted to visualize how XLA will perform the fusion under those conditions.
To unsubscribe from this group and stop receiving emails from it, send an email to xla-dev+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/xla-dev/21346fbe-a788-45ef-919c-f633464caca0%40googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/xla-dev/21346fbe-a788-45ef-919c-f633464caca0%40googlegroups.com.