Comments (5)
Thanks, that's actually a pretty serious bug and affects all functions that have an overload for Tensor&&
and a generic overload (typically for IntTuple const&
). These appear to include
flatten
coalesce
filter_zeros
take
Previously, this has been solved by renaming functions -- see group_modes
vs group
and filter
versus filter_tuple
. Rather than that, we should probably just do the rref three-step dance for those functions
template <class Engine, class Layout>
CUTE_HOST_DEVICE constexpr auto
flatten(Tensor<Engine,Layout> const& tensor) {
return make_tensor(tensor.data(), flatten(tensor.layout()));
}
template <class Engine, class Layout>
CUTE_HOST_DEVICE constexpr auto
flatten(Tensor<Engine,Layout>& tensor) {
return make_tensor(tensor.data(), flatten(tensor.layout()));
}
template <class Engine, class Layout>
CUTE_HOST_DEVICE constexpr auto
flatten(Tensor<Engine,Layout>&& tensor) {
return make_tensor(tensor.data(), flatten(tensor.layout()));
}
I'll make sure this gets fixed as soon as possible.
from cutlass.
@thakkarV @ccecka Pls take a look.
from cutlass.
There's no pointer or tensor type.
Tensor a = make_tensor<float>(make_shape(_4{}, _1{})); // Owning tensor
Tensor b = make_tensor(my_float_ptr, make_shape(_4{}, _1{})); // Non-owning tensor
Your version should not compile, but could have a better error message.
from cutlass.
@ccecka I updated the issue code sample. The problem is coalesce(tensor)
will dispatch to coalesce(Shape s)
when tensor
is defined out side of function bad
a.k.a, the error message
"auto cute::coalesce(const Shape &) [with Shape=cute::Tensor<cute::ArrayEngine<...
from cutlass.
fixed with #1623
from cutlass.
Related Issues (20)
- [QST] Split-k in hopper gather scatter gemm
- [QST] kInternalError while increasing warp count in older SIMT GEMM kernels.
- [BUG] TMA Cooperative GeMM with Stream-K scheduler hangs for specific gemm shapes HOT 1
- [QST] Get a slice of the Tensor while keeping the dimension HOT 3
- [BUG] my code compiles well in `git reset --hard v3.2.1`, but errors `git reset --hard v3.4.1` HOT 1
- [BUG]
- [QST] Integrating Cutlass EVT to MLIR-Based compiler problems
- [QST] How does cutlass profiler test gemm performance?
- [QST] How does cutlass profiler test gemm performance?
- [QST] Internal error in cutlass gemm HOT 1
- [BUG] `#include "cutlass/gemm/device/gemm_universal_adapter.h"` is causing the named symbol to not be found HOT 6
- [QST]Can I create a null tensor? HOT 1
- [QST]How to understand composition?
- [QST] How do I import the `cutlass::gemm::kernel::GemmUniversal` HOT 2
- [BUG] cutlass-python does not work on H100 for CUDA 11.8 HOT 1
- [BUG] FAILED to compile example 47_ampere_gemm_universal_streamk
- [QST]Tensor Shape Mismatch in CUTLASS: Does Layout Information Attach to Pointers?
- cutlass hipblas
- [QST] conda or pip package for cutlass missing HOT 3
- [QST] Can synchronized TensorCore MMA operations overlap with CUDA Core operations in a single thread?
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from cutlass.