An official website of the United States government
Graph structures are a natural representation of important and pervasive data. While graph applications have significant parallelism, their characteristic pointer indirect loads to neighbor data hinder scalability to large datasets on multicore systems. A scalable and efficient system must tolerate latency while leveraging data parallelism across millions of vertices. Modern Out-of-Order (OoO) cores inherently tolerate a fraction of long latencies, but become clogged when running severely memory-bound applications. Combined with large power/area footprints, this limits their parallel scaling potential and, consequently, the gains that existing software frameworks can achieve. Conversely, accelerator and memory hierarchy designs provide performant hardware specializations, but cannot support diverse application demands. To address these shortcomings, we present GraphAttack, a hardware-software data supply approach that accelerates graph applications on in-order multicore architectures. GraphAttack proposes compiler passes to (1) identify idiomatic long-latency loads and (2) slice programs along these loads into data Producer/ Consumer threads to map onto pairs of parallel cores. Each pair shares a communication queue; the Producer asynchronously issues long-latency loads, whose results are buffered in the queue and used by the Consumer. This scheme drastically increases memory-level parallelism (MLP) to mitigate latency bottlenecks. In equal-area comparisons, GraphAttack outperforms OoO cores, do-all parallelism, prefetching, and prior decoupling approaches, achieving a 2.87× speedup and 8.61× gain in energy efficiency across a range of graph applications. These improvements scale; GraphAttack achieves a 3× speedup over 64 parallel cores. Lastly, it has pragmatic design principles; it enhances in-order architectures that are gaining increasing open-source support.
more »
« less
Trireme: Exploration of Hierarchical Multi-Level Parallelism for Hardware Acceleration
The design of heterogeneous systems that include domain specific accelerators is a challenging and time-consuming process. While taking into account area constraints, designers must decide which parts of an application to accelerate in hardware and which to leave in software. Moreover, applications in domains such as Extended Reality (XR) offer opportunities for various forms of parallel execution, including loop level, task level and pipeline parallelism. To assist the design process and expose every possible level of parallelism, we present Trireme , a fully automated tool-chain that explores multiple levels of parallelism and produces domain specific accelerator designs and configurations that maximize performance, given an area budget. FPGA SoCs were used as target platforms and Catapult HLS [7] was used to synthesize RTL using a commercial 12nm FinFET technology. Experiments on demanding benchmarks from the XR domain revealed a speedup of up to 20 ×, as well as a speedup of up to 37 × for smaller applications, compared to software-only implementations.
more »
« less
- Award ID(s):
- 1718160
- PAR ID:
- 10392900
- Date Published:
- Journal Name:
- ACM Transactions on Embedded Computing Systems
- ISSN:
- 1539-9087
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Finite-state automata serve as compute kernels for many application domains such as pattern matching and data analytics. Existing approaches on GPUs exploit three levels of parallelism in automata processing tasks: 1)~input stream level, 2)~automaton-level and 3)~state-level. Among these, only state-level parallelism is intrinsic to automata while the other two levels of parallelism depend on the number of automata and input streams to be processed. As GPU resources increase, a parallelism-limited automata processing task can underutilize GPU compute resources. To this end, we propose AsyncAP, a low-overhead approach that optimizes for both scalability and throughput. Our insight is that most automata processing tasks have an additional source of parallelism originating from the input symbols which has not been leveraged before. Making the matching process associated with the automata tasks asynchronous, i.e., parallel GPU threads start processing an input stream from different input locations instead of processing it serially, improves throughput significantly and scales with input length. When the task does not have enough parallelism to utilize all the GPU cores, detailed evaluation across 12 evaluated applications shows that AsyncAP achieves up to 58× speedup on average over the state-of-the-art GPU automata processing engine. When the tasks have enough parallelism to utilize GPU cores, AsyncAP still achieves 2.4× speedup.more » « less
-
Shuffle is an indispensable process in distributed online analytical processing systems to enable task-level parallelism exploitation via multiple nodes. As a data-intensive data reorganization process, shuffle implemented on general-purpose CPUs not only incurs data traffic back and forth between the computing and storage resources, but also pollutes the cache hierarchy with almost zero data reuse. As a result, shuffle can easily become the bottleneck of distributed analysis pipelines.Our PSACS approach attacks these bottlenecks with the rising computational storage paradigm. Shuffle is offloaded to the storage-side PSACS accelerator to avoid polluting computing node memory hierarchy and enjoy the latency, bandwidth and energy benefits of near-data computing. Further, the microarchitecture of PSACS exploits data-, subtask-, and task-level parallelism for high performance and a customized scratchpad for fast on-chip random access.PSACS achieves 4.6x—5.7x shuffle throughput at kernel-level and up to 1.3x overall shuffle throughput with only a twentieth of CPU utilization comparing to software baselines. These mount up to 23% end-to-end OLAP query speedup on average.more » « less
-
Polynomial multiplication over the quotient ring is a critical operation in Ring Learning with Errors (Ring-LWE) based cryptosystems that are used for post-quantum cryptography and homomorphic encryption. This operation can be efficiently implemented using number-theoretic transform (NTT)-based architectures. Among these, pipelined parallel NTTbased polynomial multipliers are attractive for cloud computing as these are well suited for high throughput and low latency applications. For a given polynomial length, a pipelined parallel NTT-based multiplier can be designed with varying degrees of parallelism, resulting in different tradeoffs. Higher parallelism reduces latency but increases area and power consumption,and vice versa. In this paper, we develop a predictive model based on synthesized results for pipelined parallel NTT-based polynomial multipliers and analyze design tradeoffs in terms of area, power, energy, area-time product, and area-energy product across parallelism levels up to 128. We predict that, for very long polynomials, area and power differences between designs with varying levels of parallelism become negligible. In contrast, areatime product and energy per polynomial multiplication decrease with increased parallelism. Our findings suggest that, given area and power constraints, the highest feasible level of parallelism optimizes latency, area-time product, and energy per polynomial multiplication.more » « less
-
Abstract Decoder-only Transformer models such as Generative Pre-trained Transformers (GPT) have demonstrated exceptional performance in text generation by autoregressively predicting the next token. However, the efficiency of running GPT on current hardware systems is bounded by low compute-to-memory-ratio and high memory access. In this work, we propose a Process-in-memory (PIM) GPT accelerator, PIM-GPT, which achieves end-to-end acceleration of GPT inference with high performance and high energy efficiency. PIM-GPT leverages DRAM-based PIM designs for executing multiply-accumulate (MAC) operations directly in the DRAM chips, eliminating the need to move matrix data off-chip. Non-linear functions and data communication are supported by an application specific integrated chip (ASIC). At the software level, mapping schemes are designed to maximize data locality and computation parallelism. Overall, PIM-GPT achieves 41 − 137 × , 631 − 1074 × speedup and 123 − 383 × , 320 − 602 × energy efficiency over GPU and CPU baseline on 8 GPT models with up to 1.4 billion parameters.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1145/3580394
