Comments (1)
There is no such instruction, so there will be a round-trip through registers and no vectorization, but have you tried UniversalCopy<cute::float_e4m3_t, cute::half_t>
?
For vectorization + conversion, you would write the copy-to-rmem -> transform -> copy-to-smem yourself:
Tensor gA = ... // FP8
Tensor sA = ... // FP16
Tensor tAgA = ... // Partitioned
Tensor tAsA = ... // Partitioned
Tensor tArA_8 = make_fragment_like(tAgA);
Tensor tArA_16 = make_fragment_like(tAsA);
cute::copy_aligned(tAgA, tArA_8);
cute::transform(tArA_8, tArA_16, FP8->FP16);
cute::copy_aligned(tArA_16, tAsA);
from cutlass.
Related Issues (20)
- [QST] could you please help me understand how right_inverse work? HOT 2
- [QST] `retile_D` in mainloop HOT 2
- [QST] `TiledCopy` HOT 2
- CuTe documentation -- 02_layout_algebra.md has a wrong example in Complement section HOT 3
- [QST] How to pass a cute::Tensor as parameter to a device function? HOT 4
- [QST] `SmemLayoutAtom` Layouts HOT 3
- [QST] MMA_Traits shape mismatch HOT 9
- [QST] `Cutlass 3.0` `GemmUniversal` Interface HOT 6
- [BUG] Error Internal with large batch size in gemm::device::GemmBatched HOT 3
- [QST] how can i do w4a8 (int4 * int8) using cutlass? HOT 4
- [QST] Checking Tensor Shape returns 0 HOT 4
- [BUG] Build failed with nvcc "-G" on H100 HOT 3
- [cute gemm] HOT 2
- [QST] Constraint of Tensor Layout in FP8 GEMM Kernel HOT 4
- [QST] Build hangs on AWS p3.2xlarge Ubuntu HOT 7
- [QST] Epilogue Swizzle HOT 3
- [QST] Confusing stride definition in example_57
- [QST] Understanding `retile_*` HOT 4
- [BUG] Failure in TensorRef.rank2_column_major_interleaved test with Intel LLVM compiler HOT 3
- [QST] `uint4b / int4` Mixed Type GEMM using `Cutlass 3.x` for `Ampere`
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.