The 21st ACM SIGPLAN/SIGBED Conference on Languages, Compilers, and Tools for Embedded Systems最新文献

{"title":"Compiler 2.0: Using Machine Learning to Modernize Compiler Technology","authors":"Saman P. Amarasinghe","doi":"10.1145/3372799.3397167","DOIUrl":"https://doi.org/10.1145/3372799.3397167","url":null,"abstract":"Modern compilers are still built using technology that existed decades ago. These include basic algorithms and techniques for lexing, parsing, data-flow analysis, data dependence analysis, vectorization, register allocation, instruction selection, and instruction scheduling. It is high time that we modernize our compiler toolchain. In this talk, I will show the path to the modernization of one important compiler technique -- vectorization. Vectorization was first introduced in the era of Cray vector processors during the 1980's. In modernizing vectorization, I will first show how to use new techniques that better target modern hardware. While vector supercomputers need large vectors, which are only available by parallelizing loops, modern SIMD instructions efficiently work on short vectors. Thus, in 2000, we introduced Superword Level Parallelism (SLP) based vectorization. SLP finds short vector instructions within basic blocks, and by loop unrolling we can convert vector parallelism to SLP. Next, I will show how we can take advantage of the power of modern computers for compilation, by using more accurate but expensive techniques to improve SLP vectorization. Due to the hardware resource constraints of the era, like many other compiler optimizations, SLP implementation was a greedy algorithm. In 2018, we introduced goSLP, which uses integer linear programming to find an optimal instruction packing strategy and achieves 7.58% geomean performance improvement over the LLVM's SLP implementation on SPEC2017fp C/C++ programs. Finally, I will show how to truly modernize a compiler by automatically learning the necessary components of the compiler with Ithemal and Vemal. The optimality of goSLP is under LLVM's simple per instruction additive cost model that fits within the Integer programming framework. However, the actual cost of execution in a modern out-of-order, pipelined, superscalar processor is much more complex. Manually building such cost models as well as manually developing compiler optimizations is costly, tedious, error-prone and is hard to keep up with the architectural changes. Ithemal is the first learnt cost model for predicting the throughput of x86 basic blocks. It not only significantly outperforms (more than halves the error) state-of-the-art analytical hand-written tools like llvm-mca, but also is learnt from data requiring minimal human effort. Vemal is a learnt policy for end-to-end vectorization as opposed to tuning heuristics, which outperforms LLVM's SLP vectorizer. These data-driven techniques can help achieve state-of-the-art results while also reducing the development and maintenance burden of the compiler developer.","PeriodicalId":232907,"journal":{"name":"The 21st ACM SIGPLAN/SIGBED Conference on Languages, Compilers, and Tools for Embedded Systems","volume":"113 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117187050","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"FPGA-based Near Data Processing Platform Selection Using Fast Performance Modeling (WiP Paper)","authors":"Nazanin Farahpour, Zhenman Fang, Glenn D. Reinman","doi":"10.1145/3372799.3394373","DOIUrl":"https://doi.org/10.1145/3372799.3394373","url":null,"abstract":"With the trend of adopting FPGAs in data centers, various FPGA acceleration platforms have been developed in recent years. Each server could incorporate one or many of these FPGAs at different compute hierarchy levels to match its workload intensity. FPGAs could either be used as IO-attached accelerators or be closely integrated with CPU as on-chip co-processors. For a more data-centric approach, an FPGA could be moved closer to the data medium (RAM or disk) and serve as a near-memory or near-storage accelerator. In this work, we present a quantitative model and in-depth analysis of application characteristics to determine when an application is more suitable for each acceleration hierarchy. We analyze 18 benchmarks from six domains and create a preliminary guideline for both application and hardware developers.","PeriodicalId":232907,"journal":{"name":"The 21st ACM SIGPLAN/SIGBED Conference on Languages, Compilers, and Tools for Embedded Systems","volume":"122 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116859556","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Towards Real-time CNN Inference from a Video Stream on a Mobile GPU (WiP Paper)","authors":"Chanyoung Oh, G. Park, Sumin Kim, Dohee Kim, Youngmin Yi","doi":"10.1145/3372799.3394366","DOIUrl":"https://doi.org/10.1145/3372799.3394366","url":null,"abstract":"While there are several frameworks for CNN inference on mobile GPUs, they do not achieve real-time processing for the most of the CNNs that aim at reasonable accuracy since they all employ kernel-by-kernel execution model and do not effectively support INT8 quantization yet. In this paper, we reveal that mobile GPUs suffer from large kernel launch overhead unlike server GPUs, and then propose an on-device deep learning inference framework that can achieve real-time inference of CNNs on mobile GPUs by removing kernel launch overhead and by effectively exploiting INT8 quantization. We have evaluated the proposed framework with a state-of-the-art CNN based face detector (RetinaFace), and observed up to 2.01X of speedup compared to ARM Compute Library (ACL) on a commodity smartphone.","PeriodicalId":232907,"journal":{"name":"The 21st ACM SIGPLAN/SIGBED Conference on Languages, Compilers, and Tools for Embedded Systems","volume":"7 25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129597679","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Synthesis-Aided Compiler for DSP Architectures (WiP Paper)","authors":"Alexa VanHattum, Rachit Nigam, Vincent T. Lee, James Bornholt, Adrian Sampson","doi":"10.1145/3372799.3394358","DOIUrl":"https://doi.org/10.1145/3372799.3394358","url":null,"abstract":"Digital signal processors (DSPs) offer cutting-edge energy efficiency for embedded multimedia computations, but writing high-performance DSP code requires expert tuning. Programmers need to work at a low level of abstraction, manually tailoring vendor-specific instructions to enable vector and VLIW parallelism. Diospyros is a synthesizing compiler that searches for optimal data layouts to enable efficient vectorized code on DSPs. Preliminary results show that for small fixed-size matrix multiply and 2D convolution, Diospyros achieves a 6.4--7.6x speedup compared to vendor-provided optimized kernels, and a 6.5--31.3x speedup over loop-based kernels optimized with the vendor's included compiler.","PeriodicalId":232907,"journal":{"name":"The 21st ACM SIGPLAN/SIGBED Conference on Languages, Compilers, and Tools for Embedded Systems","volume":"93 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124672436","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"ApproxRefresh","authors":"Jinhua Cui, Junwei Liu, Jianhang Huang, Haoran Zhu, L. Yang","doi":"10.1145/3372799.3394362","DOIUrl":"https://doi.org/10.1145/3372799.3394362","url":null,"abstract":"With the increased density and the technology scaling, flash memory is more vulnerable to noise effects, overwhelmingly lowering the data retention time. The periodic data refresh technique is commonly used to retain the long-term data integrity. However, the frequent refresh requests also introduce the increased access conflict and severe write amplification, leading to sub-optimal performance and lifetime improvement. In this paper, we propose ApproxRefresh, which enables the uncorrectable data reuse with the assistance of approximate-computing applications in order to reduce the refresh costs. Specifically, we design a lightweight remapping-free refresh technique, called RFR, to periodically correct, compute an enhanced ECC and only remap new ECC parity bits, which dramatically reduces the refresh costs. Then, with approximate-read and precise-read hotness awareness, ApproxRefresh selectively adopts RFR or the traditional remapping-based refresh technique to reduce the refresh costs and the read disturbance. In addition, a corresponding ApproxRefresh-aware garbage collection algorithm is proposed to complete the design. Evaluations show that ApproxRefresh reduces the refresh latency by 35.09% over the state-of-the-art.","PeriodicalId":232907,"journal":{"name":"The 21st ACM SIGPLAN/SIGBED Conference on Languages, Compilers, and Tools for Embedded Systems","volume":"87 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121719099","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Beyond Base-2 Logarithmic Number Systems (WiP Paper)","authors":"Syed Asad Alam, David Gregg","doi":"10.1145/3372799.3394368","DOIUrl":"https://doi.org/10.1145/3372799.3394368","url":null,"abstract":"Logarithmic number systems (LNS) reduce hardware complexity for multiplication and division in embedded systems, at the cost of more complicated addition and subtraction. Existing LNS typically use base-2, meaning that representable numbers are some (often fractional) power of two. We argue that other bases should be considered. The base of the LNS determines the distribution of values and may reduce representation errors when converting inputs to LNS in domain-specific embedded hardware accelerators. Further, LNS addition and subtraction are normally implemented with lookup tables whose properties may be a function of the base. We show that other bases can lower both representation and addition and subtraction error. We consider the case of 8-bit LNS, when converting from 8-bit floating point (FP). We find that base-1.984 significantly reduces the conversion error. Where we can scale values, base-1.851 reduces the error to just 0.011 units of least precision (ULP). A suitable base can also reduce average arithmetic errors. For example, base-1.802 LNS has an average error of 0.242 ULP and 0.212 ULP as compared to 0.243 ULP and 0.226 ULP for addition and subtraction, respectively, for base-2.","PeriodicalId":232907,"journal":{"name":"The 21st ACM SIGPLAN/SIGBED Conference on Languages, Compilers, and Tools for Embedded Systems","volume":"60 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126987356","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"PAQSIM","authors":"Xiang Gong, Chunling Hu, C. Lim","doi":"10.1145/3372799.3394359","DOIUrl":"https://doi.org/10.1145/3372799.3394359","url":null,"abstract":"As the popularity of GPU in embedded systems keeps increasing, there is a growing demand for performance models for rapid estimation and tuning. One major challenge of developing a GPU performance model is the balance between accuracy and speed. The analytical model and the architectural model, two prevailing performance models, both have their weaknesses. The analytical model is fast to execute and simple to implement but usually suffers from low simulation accuracy. On the other hand, the cycle-level architectural model can offer high accuracy, but often at the expense of the execution time. In this work, we present a hybrid performance model for core-level performance studies. Our model takes advantage of the speed of the analytical model and the accuracy of the cycle-level architectural model. We model the resource contention as in traditional architectural models but reduce the pipeline stages when no contention is expected. The graphics workloads have shown uniform characteristics, which allows us to replace some detailed simulation with analytical models for latency estimation in key events such as memory accesses, texture fetches, and synchronizations. Such design greatly reduces the simulation time while maintains decent simulation accuracy. We evaluate our performance model against commercial mobile GPUs. The experiments using graphics workloads from popular games show great simulation speed and high accuracy in predicting the GPU performance. For simulations using the aggressive mode, the simulator can achieve an average 4.1x slowdown, with an average error rate at 6% and the peak error rate at 27.9%.","PeriodicalId":232907,"journal":{"name":"The 21st ACM SIGPLAN/SIGBED Conference on Languages, Compilers, and Tools for Embedded Systems","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121420011","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Synergistic Approach to Predictable Compilation and Scheduling on Commodity Multi-Cores","authors":"Björn Forsberg, Maxim Mattheeuws, Andreas Kurth, A. Marongiu, L. Benini","doi":"10.1145/3372799.3394369","DOIUrl":"https://doi.org/10.1145/3372799.3394369","url":null,"abstract":"Commodity multi-cores are still uncommon in real-time systems, as resource sharing complicates traditional timing analysis. The Predictable Execution Model (PREM) tackles this issue in software, through scheduling and code refactoring. State-of-the-art PREM compilers analyze tasks one at a time, maximizing task-level performance metrics, and are oblivious to system-level scheduling effects (e.g. memory serialization when tasks are co-scheduled). We propose a solution that allows PREM code generation and system scheduling to interact, based on a genetic algorithm aimed at maximizing overall system performance. Experiments on commodity hardware show that the performance increase can be as high as 31% compared to standard PREM code generation, without negatively impacting the predictability guarantees.","PeriodicalId":232907,"journal":{"name":"The 21st ACM SIGPLAN/SIGBED Conference on Languages, Compilers, and Tools for Embedded Systems","volume":"28 5","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114129808","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Exploiting the Trust Between Boundaries: Discovering Memory Corruptions in Printers via Driver-Assisted Testing","authors":"Xiaoyu He, Erick Bauman, Feng Li, Lei Yu, Linyu Li, Bingchang Liu, Aihua Piao, Kevin W. Hamlen, Wei Huo, Wei Zou","doi":"10.1145/3372799.3394363","DOIUrl":"https://doi.org/10.1145/3372799.3394363","url":null,"abstract":"TrustScope is a new, a practical approach to identifying vulnerabilities in printer firmware without actually touching the firmware. By exploiting the trust between the firmware and the device drivers, TrustScope analyzes driver software to identify the driver endpoints that output the page description language (PDL) code to be sent to the printer, extracts key constraints for this output, generates new inputs violating these constraints, and fuzzes the printer firmware with malicious PDL code composed with these inputs yet conforming to the grammar of the PDL accepted by the printer. To accommodate the black-box nature of printers, printer behavior is observed strictly externally, allowing TrustScope to detect more vulnerabilities than only those that produce crashes. A variety of key optimizations, such as fuzzing without consuming paper and ink, and offline test case generation, make printer vulnerability detection feasible and practical. An implementation of TrustScope tested with 8 different printers reveals at least one test case causing anomalous behavior in every printer tested. For most printers it finds multiple vulnerabilities, 6 of which have been assigned CVE numbers, including buffer overflow and information disclosure.","PeriodicalId":232907,"journal":{"name":"The 21st ACM SIGPLAN/SIGBED Conference on Languages, Compilers, and Tools for Embedded Systems","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123971696","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Performance Optimization on big.LITTLE Architectures: A Memory-latency Aware Approach","authors":"W. Wolff, Barry Porter","doi":"10.1145/3372799.3394370","DOIUrl":"https://doi.org/10.1145/3372799.3394370","url":null,"abstract":"The energy demands of modern mobile devices have driven a trend towards heterogeneous multi-core systems which include various types of core tuned for performance or energy efficiency, offering a rich optimization space for software. On such systems, data coherency between cores is automatically ensured by an interconnect between processors. On some chip designs the performance of this interconnect, and by extension of the entire CPU cluster, is highly dependent on the software's memory access characteristics and on the set of frequencies of each CPU core. Existing frequency scaling mechanisms in operating systems use a simple load-based heuristic to tune CPU frequencies, and so fail to achieve a holistically good configuration across such diverse clusters. We propose a new adaptive governor to solve this problem, which uses a simple trained hardware model of cache interconnect characteristics, along with real-time hardware monitors, to continually adjust core frequencies to maximize system performance. We evaluate our governor on the Exynos5422 SoC, as used in the Samsung Galaxy S5, across a range of standard benchmarks. This shows that our approach achieves a speedup of up to 40%, and a 70% energy saving, including a 30% speedup in common mobile applications such as video decoding and web browsing.","PeriodicalId":232907,"journal":{"name":"The 21st ACM SIGPLAN/SIGBED Conference on Languages, Compilers, and Tools for Embedded Systems","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128974932","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}