An official website of the United States government
Recent trends in high-performance computing show an increase in the adoption of performance portable frameworks such as Kokkos and interpreted languages such as Python. PyKokkos follows these trends and enables programmers to write performance-portable kernels in Python which greatly increases productivity. One issue that programmers still face is how to organize parallel code, as splitting code into separate kernels simplifies testing and debugging but may result in suboptimal performance. To enable programmers to organize kernels in any way they prefer while ensuring good performance, we present PyFuser, a program analysis framework for automatic fusion of performance portable PyKokkos kernels. PyFuser dynamically traces kernel calls and lazily fuses them once the result is requested by the application. PyFuser generates fused kernels that execute faster due to better reuse of data, improved compiler optimizations, and reduced kernel launch overhead, while not requiring any changes to existing PyKokkos code. We also introduce automated code transformations that further optimize the fused kernels generated by PyFuser. Our experiments show that on average PyFuser achieves speedups compared to unfused kernels of 3.8x on NVIDIA and AMD GPUs, as well as Intel and AMD CPUs.
more »
« less
Runtime Composition of Iterations for Fusing Loop-carried Sparse Dependence
Dependence between iterations in sparse computations causes inefficient use of memory and computation resources. This paper proposes sparse fusion, a technique that generates efficient parallel code for the combination of two sparse matrix kernels, where at least one of the kernels has loop-carried dependencies. Existing implementations optimize individual sparse kernels separately. However, this approach leads to synchronization overheads and load imbalance due to the irregular dependence patterns of sparse kernels, as well as inefficient cache usage due to their irregular memory access patterns. Sparse fusion uses a novel inspection strategy and code transformation to generate parallel fused code optimized for data locality and load balance. Sparse fusion outperforms the best of unfused implementations using ParSy and MKL by an average of 4.2× and is faster than the best of fused implementations using existing scheduling algorithms, such as LBC, DAGP, and wavefront by an average of 4× for various kernel combinations.
more »
« less
- PAR ID:
- 10482910
- Publisher / Repository:
- ACM
- Date Published:
- ISBN:
- 9798400701092
- Page Range / eLocation ID:
- 1 to 15
- Format(s):
- Medium: X
- Location:
- Denver CO USA
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Neighborhood attention reduces the cost of self attention by restricting each token’s attention span to its nearest neighbors. This restriction, parameterized by a window size and dilation factor, draws a spectrum of possible attention patterns between linear projection and self attention. Neighborhood attention, and more generally sliding window attention patterns, have long been bounded by infrastructure, particularly in higher-rank spaces (2-D and 3-D), calling for the development of custom kernels, which have been limited in either functionality, or performance, if not both. In this work, we aim to massively improve upon existing infrastructure by providing two new methods for implementing neighborhood attention. We first show that neighborhood attention can be represented as a batched GEMM problem, similar to standard attention, and implement it for 1-D and 2-D neighborhood attention. These kernels on average provide 895% and 272% improvement in full precision runtime compared to existing naive CUDA kernels for 1-D and 2-D neighborhood attention respectively. We find that aside from being heavily bound by memory bandwidth, certain inherent inefficiencies exist in all unfused implementations of neighborhood attention, which in most cases undo their theoretical efficiency gain. Motivated by the progress made into fused dot-product attention kernels, we developed fused neighborhood attention; an adaptation of fused dot-product attention kernels that allow fine-grained control over attention across different spatial axes. Known for reducing the quadratic time complexity of self attention to a linear complexity, neighborhood attention can now enjoy a reduced and constant memory footprint, and record-breaking half precision runtime. We observe that our fused implementation successfully circumvents some of the unavoidable inefficiencies in unfused implementations. While our unfused GEMM-based kernels only improve half precision performance compared to naive kernels by an average of 548% and 193% in 1-D and 2-D problems respectively, our fused kernels improve naive kernels by an average of 1759% and 958% in 1-D and 2-D problems respectively. These improvements translate into up to 104% improvement in inference and 39% improvement in training existing models based on neighborhood attention, and additionally extend its applicability to image and video perception, as well as other modalities.more » « less
-
null (Ed.)Applications targeting digital signal processors (DSPs) benefit from fast implementations of small linear algebra kernels. While existing auto-vectorizing compilers are effective at extracting performance from large kernels, they struggle to invent the complex data movements necessary to optimize small kernels. To get the best performance, DSP engineers must hand-write and tune specialized small kernels for a wide spectrum of applications and architectures. We present Diospyros, a search-based compiler that automatically finds efficient vectorizations and data layouts for small linear algebra kernels. Diospyros combines symbolic evaluation and equality saturation to vectorize computations with irregular structure. We show that a collection of Diospyros-compiled kernels outperform implementations from existing DSP libraries by 3.1× on average, that Diospyros can generate kernels that are competitive with expert-tuned code, and that optimizing these small kernels offers end-to-end speedup for a DSP application.more » « less
-
We show how to build a compiler for a sparse array language that supports shape operators such as reshaping or concatenating arrays, in addition to compute operators. Existing sparse array programming systems implement generic shape operators for only some sparse data structures, reduce shape operators on other data structures to those, and do not support fusion. Our system compiles sparse array expressions to code that efficiently iterates over reshaped views of irregular sparse data structures, without needing to materialize temporary storage for intermediates. Our evaluation shows that our approach generates sparse array code competitive with popular sparse array libraries: our generated shape operators achieve geometric mean speed-ups of 1.66×–15.3× when compared to hand-written kernels in scipy.sparse and 1.67×–651× when compared to generic implementations in pydata/sparse. For operators that require data structure conversions in these libraries, our generated code achieves geometric mean speed-ups of 7.29×–13.0× when compared to scipy.sparse and 21.3×–511× when compared to pydata/sparse. Finally, our evaluation demonstrates that fusing shape and compute operators improves the performance of several expressions by geometric mean speed-ups of 1.22×–2.23×.more » « less
-
Cathie Olschanowsky (Ed.)Hydrologists must process many gigabytes of data for hydrologic simulations, which takes time and resources degrading performance. The performance issues are caused mainly by domain scientists’ preference for using Python, which trades performance for productivity. In my thesis, I demonstrate that using the static compilation technique to compile Python to generate C code along with several optimizations reduces time and resources for hydrologic data processing. I developed a Domain Specific Library (DSL) which is a subset of Python and compiles to Sparse Polyhedral Framework - Intermediate Representation (SPF-IR), which allows opportunities for optimizations like read reduction fusion which are not available in Python. We fused the file I/O to perform computation on small chunks of data (stream computation) in order to reduce the memory footprint. The C code we generated from SPF-IR shows an average speed-up of 2.58x over the existing hand-optimized implementations and can totally eliminate the tempo- rary storage required. DSL users can still enjoy the ease of use of Python but get performance better than the C code.more » « less
