Comments (4)
Hi Tom!
After having looked at it, this is actually not strictly a bug. The backward pass modifies neg_logprobs in-place, so the auto-tuner will keep modifying the buffer over and over again as it tries different configurations to run. If you set num_warps=[2] or num_warps=[4], it will also work. I would suggest you disable auto-tuning for now. We should leave this issue open until a solution is implemented
Probably the solution will look something like:
triton.kernel( ... , autotune_clone = 'neg_logprobs'), then the auto-tuner will just clone automatically neg_logprobs before running.
I hope at some point pointers modified in-place can be determined automatically from static analysis.
from triton.
This is an old issue but still not addressed actually. I remember having a patch ready at some point so I'll dig in to see if I can find it somewhere. Now that the JIT API is back to taking pure torch.tensor, it shouldn't be too hard to clone them before the auto-tuner runs.
from triton.
Awesome!
from triton.
Should have been fixed in #205 . Now you can provide reset_to_zero(list_of_arg_names)
to the auto-tuner, and it will reset all arguments of provided name to zero before each benchmark run. Not ideal, but should fix this issue
from triton.
Related Issues (20)
- How to implement matmul with B in column major? HOT 2
- how to solve "cannot find -lNVGPUIR: No such file or directory"
- Use TRITON_CACHE in setup.py so that .triton must not have to be in home directory of user HOT 1
- Can't install triton HOT 1
- Flash Attention 3 --> Triton HOT 1
- Mistakes in `class DistributedEncoding`'s illustration
- Latest nightly triton causes my custom fused attention kernel to output incorrect results. HOT 3
- Unexpected segmentation fault with tl.sum in a simple loop HOT 3
- Support reduction operations on global memory with `red` ptx instruction
- Cannot use Triton interpreter with matrix multiply example HOT 4
- error: fp8e4nv data type is not supported on CUDA arch < 89
- Can I explicitly specify the "tl.load" to load data into shared memory? HOT 1
- breaking change to constexpr in triton 3.0.0 HOT 2
- AttributeError: 'InterpretedFunction' object has no attribute 'cache_key' HOT 2
- SWP: use of address of iterator in dist1Cluster HOT 1
- Introduce `tl.assume` or use `assert` expression in non-debug builds to guide optimization?
- run into dead loop when tuning the tma persistent kernel HOT 6
- [BUG] error load fp32 value from 2D tensor HOT 2
- [BUG] device_print - Triton nightly, 3.0 incorrect values (zero) when using pointer arithmetics(constexpr etc.) other than with triton.language.arange HOT 3
- Incompatible type error with "torch.onnx.export()"
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 triton.