An official website of the United States government
Neural architecture search (NAS) is a promising technique to design efficient and high-performance deep neural networks (DNNs). As the performance requirements of ML applications grow continuously, the hardware accelerators start playing a central role in DNN design. This trend makes NAS even more complicated and time-consuming for most real applications. This paper proposes FLASH, a very fast NAS methodology that co-optimizes the DNN accuracy and performance on a real hardware platform. As the main theoretical contribution, we first propose the NN-Degree, an analytical metric to quantify the topological characteristics of DNNs with skip connections (e.g., DenseNets, ResNets, Wide-ResNets, and MobileNets). The newly proposed NN-Degree allows us to do training-free NAS within one second and build an accuracy predictor by training as few as 25 samples out of a vast search space with more than 63 billion configurations. Second, by performing inference on the target hardware, we fine-tune and validate our analytical models to estimate the latency, area, and energy consumption of various DNN architectures while executing standard ML datasets. Third, we construct a hierarchical algorithm based on simplicial homology global optimization (SHGO) to optimize the model-architecture co-design process, while considering the area, latency, and energy consumption of the target hardware. We demonstrate that, compared to the state-of-the-art NAS approaches, our proposed hierarchical SHGO-based algorithm enables more than four orders of magnitude speedup (specifically, the execution time of the proposed algorithm is about 0.1 seconds). Finally, our experimental evaluations show that FLASH is easily transferable to different hardware architectures, thus enabling us to do NAS on a Raspberry Pi-3B processor in less than 3 seconds.
more »
« less
Unifying Spatial Accelerator Compilation with Idiomatic and Modular Transformations
Spatial accelerators provide high performance, energy efficiency, and flexibility. Recent design frameworks enable these architectures to be quickly designed and customized to a domain. However, constructing a compiler for this immense design space is challenging, first because accelerators express programs with high-level idioms that are difficult to recognize. Second, it is unpredictable whether certain transformations are beneficial or will lead to infeasible hardware mappings. Our work develops a general spatial-accelerator compiler with two key ideas. First, we propose an approach to recognize and represent useful dataflow idioms, along with a novel idiomatic memory representation. Second, we propose the principle of modular compilation, which combines hardware-aware transformation selection and an iterative approach to handle uncertainty. Our compiler achieves 2.3× speedup, and 98.7× area-normalized speedup over high-end server CPU.
more »
« less
- PAR ID:
- 10354847
- Date Published:
- Journal Name:
- IEEE Micro
- ISSN:
- 0272-1732
- Page Range / eLocation ID:
- 1 to 12
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
The architecture of a coarse-grained reconfigurable array (CGRA) processing element (PE) has a significant effect on the performance and energy-efficiency of an application running on the CGRA. This paper presents APEX, an automated approach for generating specialized PE architectures for an application or an application domain. APEX first analyzes application domain benchmarks using frequent subgraph mining to extract commonly occurring computational subgraphs. APEX then generates specialized PEs by merging subgraphs using a datapath graph merging algorithm. The merged datapath graphs are translated into a PE specification from which we automatically generate the PE hardware description in Verilog along with a compiler that maps applications to the PE. The PE hardware and compiler are inserted into a flexible CGRA generation and compilation toolchain that allows for agile evaluation of CGRAs. We evaluate APEX for two domains, machine learning and image processing. For image processing applications, our automatically generated CGRAs with specialized PEs achieve from 5% to 30% less area and from 22% to 46% less energy compared to a general-purpose CGRA. For machine learning applications, our automatically generated CGRAs consume 16% to 59% less energy and 22% to 39% less area than a general-purpose CGRA. This work paves the way for creation of application domain-driven design-space exploration frameworks that automatically generate efficient programmable accelerators, with a much lower design effort for both hardware and compiler generation.more » « less
-
MAESTRO: A Data-Centric Approach to Understand Reuse, Performance, and Hardware Cost of DNN MappingsThe efficiency of an accelerator depends on three factors—mapping, deep neural network (DNN) layers, and hardware—constructing extremely complicated design space of DNN accelerators. To demystify such complicated design space and guide the DNN accelerator design for better efficiency, we propose an analytical cost model, MAESTRO. MAESTRO receives DNN model description and hardware resources information as a list, and mapping described in a data-centric representation we propose as inputs. The data centric representation consists of three directives that enable concise description of mappings in a compiler-friendly form. MAESTRO analyzes various forms of data reuse in an accelerator based on inputs quickly and generates more than 20 statistics including total latency, energy, throughput, etc., as outputs. MAESTRO’s fast analysis enables various optimization tools for DNN accelerators such as hardware design exploration tool we present as an example.more » « less
-
As customized accelerator design has become increasingly popular to keep up with the demand for high performance computing, it poses challenges for modern simulator design to adapt to such a large variety of accelerators. Existing simulators tend to two extremes: low-level and general approaches, such as RTL simulation, that can model any hardware but require substantial effort and long execution times; and higher-level application-specific models that can be much faster and easier to use but require one-off engineering effort.This work proposes a compiler-driven simulation workflow that can model configurable hardware accelerator. The key idea is to separate structure representation from simulation by developing an intermediate language that can flexibly represent a wide variety of hardware constructs. We design the Event Queue (EQueue) dialect of MLIR, a dialect that can model arbitrary hardware accelerators with explicit data movement and distributed event-based control; we also implement a generic simulation engine to model EQueue programs with hybrid MLIR dialects representing different abstraction levels. We demonstrate two case studies of EQueue-implemented accelerators: the systolic array of convolution and SIMD processors in a modern FPGA. In the former we show EQueue simulation is as accurate as a state-of-the-art simulator, while offering higher extensibility and lower iteration cost via compiler passes. In the latter we demonstrate our simulation flow can guide designer efficiently improve their design using visualizable simulation outputs.more » « less
-
In recent years, the MLIR framework has had explosive growth due to the need for extensible deep learning compilers for hardware accelerators. Such examples include Triton, CIRCT, and ONNX-MLIR. MLIR compilers introduce significant complexities in localizing bugs or inefficiencies because of their layered optimization and transformation process with compilation passes. While existing delta debugging techniques can be used to identify a minimum subset of IR code that reproduces a given bug symptom, their naive application to MLIR is time-consuming because real-world MLIR compilers usually involve a large number of compilation passes. Compiler developers must identify a minimized set of relevant compilation passes to reduce the footprint of MLIR compiler code to be inspected for a bug fix. We propose DuoReduce, a dual-dimensional reduction approach for MLIR bug localization. DuoReduce leverages three key ideas in tandem to design an efficient MLIR delta debugger. First, DuoReduce reduces compiler passes that are irrelevant to the bug by identifying ordering dependencies among the different compilation passes. Second, DuoReduce uses MLIR-semantics-aware transformations to expedite IR code reduction. Finally, DuoReduce leverages cross-dependence between the IR code dimension and the compilation pass dimension by accounting for which IR code segments are related to which compilation passes to reduce unused passes. Experiments with three large-scale MLIR compiler projects find that DuoReduce outperforms syntax-aware reducers such as Perses and Vulcan in terms of IR code reduction by 31.6% and 21.5% respectively. If one uses these reducers by enumerating all possible compilation passes (on average 18 passes), it could take up to 145 hours. By identifying ordering dependencies among compilation passes, DuoReduce reduces this time to 9.5 minutes. By identifying which compilation passes are unused for compiling reduced IR code, DuoReduce reduces the number of passes by 14.6%. This translates to not needing to examine 281 lines of MLIR compiler code on average to fix the bugs. DuoReduce has the potential to significantly reduce debugging effort in MLIR compilers, which serves as the foundation for the current landscape of machine learning and hardware accelerators.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1109/MM.2022.3189976
