| 20 Dec 2024 |
 hacker1024 | https://github.com/NixOS/nixpkgs/blob/316d2c82570270e90536e5a8a8947d5ac6159b5f/nixos/modules/virtualisation/qemu-vm.nix#L513 | 22:12:49 |
| hacker1024 | https://nixos.org/manual/nixos/stable/#sec-running-nixos-tests-interactively | 22:13:36 |
 SomeoneSerge (back on matrix) | This isn't all that pretty but it works https://github.com/sinanmohd/evanix/blob/master/nixos/tests/dsl.nix | 22:20:30 |
| 21 Dec 2024 |
|  @stablejoy:matrix.org left the room. | 05:08:26 |
|  @dmiskovic:matrix.org left the room. | 05:13:53 |
| @stablejoy:matrix.org joined the room. | 06:43:21 |
 connor (he/him) | Anyone have any suggestions for benchmarking https://github.com/NixOS/nix/pull/12097? I mean, I expect it to be no slower than it is currently, but how do I show that…
Would running a nix eval in QEMU with -icount and reporting the wall time from within the VM be a good way to do that? | 06:52:04 |
 matthewcroughan | SomeoneSerge (utc+3):
rocmlir-rock> [2830/2949] Building CXX object external/llvm-project/llvm/tools/mlir/lib/Dialect/Tosa/Transforms/CMakeFiles/obj.MLIRTosaTransforms.dir/TosaMakeBroadcastable.cpp.o
rocmlir-rock> [2831/2949] Building CXX object mlir/lib/Dialect/Rock/Transforms/CMakeFiles/obj.MLIRRockTransforms.dir/ViewToTransform.cpp.o
rocmlir-rock> FAILED: mlir/lib/Dialect/Rock/Transforms/CMakeFiles/obj.MLIRRockTransforms.dir/ViewToTransform.cpp.o
rocmlir-rock> /nix/store/imlsgdankmk7vxdrm49sx23bkmh37r1q-rocm-llvm-clang-wrapper-5.7.1/bin/clang++ -DGTEST_HAS_RTTI=0 -D_DEBUG -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS -I/build/source/build/mlir/lib/Dialect/Rock/Transforms -I/build/source/mlir/lib/Dialect/Rock/Transforms -I/build/source/external/llvm-project/llvm/include -I/build/source/build/external/llvm-project/llvm/include -I/build/source/external/llvm-project/mlir/include -I/build/source/build/external/llvm-project/llvm/tools/mlir/include -I/build/source/mlir/include -I/build/source/build/mlir/include -fPIC -fno-semantic-interposition -fvisibility-inlines-hidden -Werror=date-time -Werror=unguarded-availability-new -Wall -Wextra -Wno-unused-parameter -Wwrite-strings -Wcast-qual -Wmissing-field-initializers -pedantic -Wno-long-long -Wc++98-compat-extra-semi -Wimplicit-fallthrough -Wcovered-switch-default -Wno-noexcept-type -Wnon-virtual-dtor -Wdelete-non-virtual-dtor -Wsuggest-override -Wstring-conversion -Wmisleading-indentation -Wctad-maybe-unsupported -fdiagnostics-color -ffunction-sections -fdata-sections -Werror=global-constructors -O3 -DNDEBUG -std=gnu++17 -fno-exceptions -fno-rtti -UNDEBUG -MD -MT mlir/lib/Dialect/Rock/Transforms/CMakeFiles/obj.MLIRRockTransforms.dir/ViewToTransform.cpp.o -MF mlir/lib/Dialect/Rock/Transforms/CMakeFiles/obj.MLIRRockTransforms.dir/ViewToTransform.cpp.o.d -o mlir/lib/Dialect/Rock/Transforms/CMakeFiles/obj.MLIRRockTransforms.dir/ViewToTransform.cpp.o -c /build/source/mlir/lib/Dialect/Rock/Transforms/ViewToTransform.cpp
rocmlir-rock> In file included from /build/source/mlir/lib/Dialect/Rock/Transforms/ViewToTransform.cpp:14:
rocmlir-rock> /build/source/mlir/include/mlir/Conversion/TosaToRock/TosaToRock.h:21:10: fatal error: 'mlir/Conversion/RocMLIRPasses.h.inc' file not found
rocmlir-rock> #include "mlir/Conversion/RocMLIRPasses.h.inc"
rocmlir-rock> ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
rocmlir-rock> 1 error generated.
rocmlir-rock> [2832/2949] Building CXX object mlir/lib/Dialect/Rock/Transforms/CMakeFiles/obj.MLIRRockTransforms.dir/CleanMath.cpp.o
| 18:54:49 |
| matthewcroughan | trying to use rocm | 18:54:55 |
| matthewcroughan | rocmlir-rock-5.7.1 | 18:55:09 |
| matthewcroughan | https://github.com/NixOS/nixpkgs/issues/197885#issuecomment-2416994704 | 19:00:58 |
| matthewcroughan | ah does torch-bin only work for cuda? | 19:17:31 |
| matthewcroughan | https://github.com/ROCm/rocMLIR/releases/tag/rocm-6.3.1 | 19:40:29 |
| matthewcroughan | looks like it got an update to 6.3.1 yesterday | 19:40:43 |
| matthewcroughan | 6.3.1 compiles fine just by overriding to use this new source | 20:42:05 |
| matthewcroughan | but then I need magma in rocmPackages 6, which doesn't seem to build either | 20:42:28 |
| matthewcroughan | rocm in nixpkgs is close to working, but not quite there | 20:42:37 |
| 22 Dec 2024 |
| SomeoneSerge (back on matrix) | 👍️ cuda and rocm in pytorch are mutually exclusive, so I guess there must be a separate wheel | 00:19:20 |
| matthewcroughan | https://hydra.nixos.org/job/nixpkgs/trunk/zluda.x86_64-linux | 10:51:11 |
| matthewcroughan | when is the last time this even built | 10:51:18 |
| matthewcroughan | it seems like literally never? | 10:51:30 |
| matthewcroughan | how was this merged if it never built? | 10:51:34 |
| SomeoneSerge (back on matrix) | Back when rocmPackages weren't broken | 11:28:26 |
| SomeoneSerge (back on matrix) | * Back when rocmPackages weren't broken exceeding the hydra quotas | 11:28:39 |
| matthewcroughan | SomeoneSerge (utc+3): I would love to try out zluda, but do not want to wait over 7 hours for the build, or whatever it takes | 12:21:49 |
| matthewcroughan | got a cache? | 12:21:52 |
| SomeoneSerge (back on matrix) | No, sorry. I haven't got any amd hardware at home unfortunately | 12:37:10 |
| matthewcroughan | Yeah I think there's a lack of investment in the AMD community. | 12:37:30 |
| matthewcroughan | In a fairer world, we would be given the hardware to support this, but we don't have such a thing. And given that nobody is willing to give, then they won't get. | 12:37:49 |
| matthewcroughan | The state of things, I think reflects a lack of support from the people who need to give it. | 12:38:27 |
