Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userCo-Optimizing Performance and Memory FootprintVia Integrated CPU/GPU Memory Management, anImplementation on Autonomous Driving Platform
220
0
0.0
(
0
)
Added by
Yang Hu
Publication date
2020
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
Cutting-edge embedded system applications, such as self-driving cars and unmanned drone software, are reliant on integrated CPU/GPU platforms for their DNNs-driven workload, such as perception and other highly parallel components. In this work, we set out to explore the hidden performance implication of GPU memory management methods of integrated CPU/GPU architecture. Through a series of experiments on micro-benchmarks and real-world workloads, we find that the performance under different memory management methods may vary according to application characteristics. Based on this observation, we develop a performance model that can predict system overhead for each memory management method based on application characteristics. Guided by the performance model, we further propose a runtime scheduler. By conducting per-task memory management policy switching and kernel overlapping, the scheduler can significantly relieve the system memory pressure and reduce the multitasking co-run response time. We have implemented and extensively evaluated our system prototype on the NVIDIA Jetson TX2, Drive PX2, and Xavier AGX platforms, using both Rodinia benchmark suite and two real-world case studies of drone software and autonomous driving software.
rate research
Read More
GPU (graphics processing unit) has been used for many data-intensive applications. Among them, deep learning systems are one of the most important consumer systems for GPU nowadays. As deep learning applications impose deeper and larger models in order to achieve higher accuracy, memory management becomes an important research topic for deep learning systems, given that GPU has limited memory size. Many approaches have been proposed towards this issue, e.g., model compression and memory swapping. However, they either degrade the model accuracy or require a lot of manual intervention. In this paper, we propose two orthogonal approaches to reduce the memory cost from the system perspective. Our approaches are transparent to the models, and thus do not affect the model accuracy. They are achieved by exploiting the iterative nature of the training algorithm of deep learning to derive the lifetime and read/write order of all variables. With the lifetime semantics, we are able to implement a memory pool with minimal fragments. However, the optimization problem is NP-complete. We propose a heuristic algorithm that reduces up to 13.3% of memory compared with Nvidias default memory pool with equal time complexity. With the read/write semantics, the variables that are not in use can be swapped out from GPU to CPU to reduce the memory footprint. We propose multiple swapping strategies to automatically decide which variable to swap and when to swap out (in), which reduces the memory cost by up to 34.2% without communication overhead.
The current mobile applications have rapidly growing memory footprints, posing a great challenge for memory system design. Insufficient DRAM main memory will incur frequent data swaps between memory and storage, a process that hurts performance, consumes energy and deteriorates the write endurance of typical flash storage devices. Alternately, a larger DRAM has higher leakage power and drains the battery faster. Further, DRAM scaling trends make further growth of DRAMin the mobile space prohibitive due to cost. Emerging non-volatile memory (NVM) has the potential to alleviate these issues due to its higher capacity per cost than DRAM and mini-mal static power. Recently, a wide spectrum of NVM technologies, including phase-change memories (PCM), memristor, and 3D XPoint have emerged. Despite the mentioned advantages, NVM has longer access latency compared to DRAMand NVM writes can incur higher latencies and wear costs. Therefore integration of these new memory technologies in the memory hierarchy requires a fundamental rearchitect-ing of traditional system designs. In this work, we propose a hardware-accelerated memory manager (HMMU) that addresses both types of memory in a flat space address space. We design a set of data placement and data migration policies within this memory manager, such that we may exploit the advantages of each memory technology. By augmenting the system with this HMMU, we reduce the overall memory latency while also reducing writes to the NVM. Experimental results show that our design achieves a 39% reduction in energy consumption with only a 12% performance degradation versus an all-DRAM baseline that is likely untenable in the future.
Plenty of research efforts have been devoted to FPGA-based acceleration, due to its low latency and high energy efficiency. However, using the original low-level hardware description languages like Verilog to program FPGAs requires generally good knowledge of hardware design details and hand-on experiences. Fortunately, the FPGA community intends to address this low programmability issues. For example, , with the intention that programming FPGAs is just as easy as programming GPUs. Even though Vitis is proven to increase programmability, we cannot directly obtain high performance without careful design regarding hardware pipeline and memory subsystem.In this paper, we focus on the memory subsystem, comprehensively and systematically benchmarking the effect of optimization methods on memory performance. Upon benchmarking, we quantitatively analyze the typical memory access patterns for a broad range of applications, including AI, HPC, and database. Further, we also provide the corresponding optimization direction for each memory access pattern so as to improve overall performance.
Much of the current focus in high-performance computing is on multi-threading, multi-computing, and graphics processing unit (GPU) computing. However, vectorization and non-parallel optimization techniques, which can often be employed additionally, are less frequently discussed. In this paper, we present an analysis of several optimizations done on both central processing unit (CPU) and GPU implementations of a particular computationally intensive Metropolis Monte Carlo algorithm. Explicit vectorization on the CPU and the equivalent, explicit memory coalescing, on the GPU are found to be critical to achieving good performance of this algorithm in both environments. The fully-optimized CPU version achieves a 9x to 12x speedup over the original CPU version, in addition to speedup from multi-threading. This is 2x faster than the fully-optimized GPU version.
The proliferation of fast, dense, byte-addressable nonvolatile memory suggests that data might be kept in pointer-rich in-memory format across program runs and even process and system crashes. For full generality, such data requires dynamic memory allocation, and while the allocator could in principle rolled into each data structure, it is desirable to make it a separate abstraction. Toward this end, we introduce recoverability, a correctness criterion for persistent allocators, together with a nonblocking allocator, Ralloc, that satisfies this criterion. Ralloc is based on the LRMalloc of Leite and Rocha, with three key innovations. First, we persist just enough information during normal operation to permit correct reconstruction of the heap after a full-system crash. Our reconstruction mechanism performs garbage collection (GC) to identify and remedy any failure-induced memory leaks. Second, we introduce the notion of filter functions, which identify the locations of pointers within persistent blocks to mitigate the limitations of conservative GC. Third, to allow persistent regions to be mapped at an arbitrary address, we employ position-independent (offset-based) pointers for both data and metadata. Experiments show Ralloc to be performance-competitive with both Makalu, the state-of-the-art lock-based persistent allocator, and such transient allocators as LRMalloc and JEMalloc. In particular, reliance on GC and offline metadata reconstruction allows Ralloc to pay almost nothing for persistence during normal operation.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
