DNN accelerators are often developed and evaluated in isolation without considering the cross-stack, system-level effects in real-world environments. This makes it difficult to appreciate the impact of Systemon-Chip (SoC) resource contention, OS overheads, and programming-stack inefficiencies on overall performance/energy-efficiency. To address this challenge, we present Gemmini, an open-source, full-stack DNN accelerator generator. Gemmini generates a wide design-space of efficient ASIC accelerators from a flexible architectural template, together with flexible programming stacks and full SoCs with shared resources that capture system-level effects. Gemmini-generated accelerators have also been fabricated, delivering up to three orders-of-magnitude speedups over high-performance CPUs on various DNN benchmarks.

One of the key challenges arising when compilers vectorize loops for today’s SIMD-compatible architectures is to decide if vectorization or interleaving is beneficial. Then, the compiler has to determine the number of instructions to pack together and the interleaving level (stride). Compilers are designed today to use fixed-cost models that are based on heuristics to make vectorization decisions on loops. However, these models are unable to capture the data dependency, the computation graph, or the organization of instructions. Alternatively, software engineers often hand-write the vectorization factors of every loop. This, however, places a huge burden on them, since it requires prior experience and significantly increases the development time.

Domain specialization under energy constraints in deeply-scaled CMOS has been driving the need for agile development of Systems on a Chip (SoCs). While digital subsystems have design flows that are conducive to rapid iterations from specification to layout, analog and mixed-signal modules face the challenge of a long human-in-the-middle iteration loop that requires expert intuition to verify that post-layout circuit parameters meet the original design specification. Existing automated solutions that optimize circuit parameters for a given target design specification have limitations of being schematic-only, inaccurate, sample-inefficient or not generalizable. This work presents AutoCkt, a machine learning optimization framework trained using deep reinforcement learning that not only finds post-layout circuit parameters for a given target specification, but also gains knowledge about the entire design space through a sparse subsampling technique. Our results show that for multiple circuit topologies, AutoCkt is able to converge and meet all target specifications on at least 96.3% of tested design goals in schematic simulation, on average 40× faster than a traditional genetic algorithm. Using the Berkeley Analog Generator, AutoCkt is able to design 40 LVS passed operational amplifiers in 68 hours, 9.6× faster than the state-of-the-art when considering layout parasitics.

Network function virtualization (NFV) technology attracts tremendous interests from telecommunication industry and data center operators, as it allows service providers to assign resource for Virtual Network Functions (VNFs) on demand, achieving better flexibility, programmability, and scalability. To improve server utilization, one popular practice is to deploy best effort (BE) workloads along with high priority (HP) VNFs when high priority VNF's resource usage is detected to be low. The key challenge of this deployment scheme is to dynamically balance the Service level objective (SLO) and the total cost of ownership (TCO) to optimize the data center efficiency under inherently fluctuating workloads. With the recent advancement in deep reinforcement learning, we conjecture that it has the potential to solve this challenge by adaptively adjusting resource allocation to reach the improved performance and higher server utilization. In this paper, we present a closed-loop automation system RLDRM11RLDRM: Reinforcement Learning Dynamic Resource Management to dynamically adjust Last Level Cache allocation between HP VNFs and BE workloads using deep reinforcement learning. The results demonstrate improved server utilization while maintaining required SLO for the HP VNFs.

In-memory processing can dramatically improve the latency and energy consumption of computing systems by minimizing the data transfer between the memory and the processor. Efficient execution of processing operations within the memory is therefore, a highly motivated objective in modern computer architecture. This article presents a novel automatic framework for efficient implementation of arbitrary combinational logic functions within a memristive memory. Using tools from logic design, graph theory and compiler register allocation technology, we developed synthesis and in-memory mapping of logic execution in a single row (SIMPLER), a tool that optimizes the execution of in-memory logic operations in terms of throughput and area. Given a logical function, SIMPLER automatically generates a sequence of atomic memristor-aided logic (MAGIC) NOR operations and efficiently locates them within a single sizelimited memory row, reusing cells to save area when needed. This approach fully exploits the parallelism offered by the MAGIC NOR gates. It allows multiple instances of the logic function to be performed concurrently, each compressed into a single row of the memory. This virtue makes SIMPLER an attractive candidate for designing in-memory single instruction, multiple data (SIMD) operations. Compared to the previous work (that optimizes latency rather than throughput for a single function), SIMPLER achieves an average throughput improvement of 435×. When the previous tools are parallelized similarly to SIMPLER, SIMPLER achieves higher throughput of at least 5×, with 23× improvement in area and 20× improvement in area efficiency. These improvements more than fully compensate for the increase (up to 17% on average) in latency.

The performance of the code generated by a compiler depends on the order in which the optimization passes are applied. In high-level synthesis, the quality of the generated circuit relates directly to the code generated by the front-end compiler. Choosing a good order-often referred to as the phase-ordering problem-is an NP-hard problem. In this paper, we evaluate a new technique to address the phase-ordering problem: deep reinforcement learning. We implement a framework in the context of the LLVM compiler to optimize the ordering for HLS programs and compare the performance of deep reinforcement learning to state-of-the-art algorithms that address the phase-ordering problem. Overall, our framework runs one to two orders of magnitude faster than these algorithms, and achieves a 16% improvement in circuit performance over the -O3 compiler flag.

Processing-in-Memory (PiM) machines promise to overcome the von Neumann bottleneck in order to further scale performance and energy efficiency of computing systems by reducing the extent of data transfer and offering ample parallelism. In this paper, we take the memristive Memory Processing Unit (mMPU) as a case study of a PiM machine and scrutinize it in practical scenarios. Specifically, we explore the limitations of parallelism and data transfer elimination. We argue that lack of operand locality and arrangement might make data transfer inevitable in the mMPU. We then devise techniques to move data within the mMPU, without transferring it off-chip, and quantify their costs. Additionally, we present electrical parameters that might limit the parallelism offered by the mMPU and evaluate their impact. Using benchmarks from the LGsynth91 suite, their vector extensions, and a few synthetic data-parallel workloads, we show that the internal data transfer results in an increase of up to 1.5× in the execution time, while the parallelism can be limited in some cases to 256 gates, resulting in an increase in execution time by 1.1× to 2×.

The growing disparity between processor and memory performance poses significant limits on system performance and energy efficiency. To address this widely investigated problem, modern computing systems attempt to minimize data transfer by means of a memory hierarchy. Yet the benefit from such a solution for data-intensive applications is limited. Emerging non-volatile resistive memory technologies (memristors) offer the ability to both store and process data within the memristive memory cells, with almost no data transfer. In this paper, we propose algorithms for performing fixed point multiplication within the memristive memory using Memristor Aided Logic (MAGIC) gates and execute them in a cycle-accurate simulator to verify and evaluate them. Previously proposed implementations were not feasible for execution within memory because the required number of memory cells for the computation was too large to fit the size-limited memristive memory arrays. The algorithms proposed in this paper not only improve the latency as compared to previously proposed algorithms by 1.8× on average, but their significantly better area efficiency now makes it possible to perform numerous fixed point multiplications simultaneously within memristive memory arrays.