An official website of the United States government
With the increase in GPU register file (RF) size, its power consumption has also increased. Since RF exists at the highest level in cache hierarchy, designing it with memories with high write latency/energy (e.g., spin transfer torque RAM) can lead to large energy loss. In this paper, we present an spin orbit torque RAM (SOT-RAM) based RF design which provides higher energy efficiency than SRAM and STT-RAM RFs while maintaining performance same as that of SRAM RF. To further improve energy efficiency of SOT-RAM based RF, we propose avoiding redundant bit-writes to RF. Compared to SRAM RF, SOT-RAM RF saves 18.6% energy and by using our technique for avoiding redundant writes, the energy saving can be increased to 44.3%, without harming performance.
more »
« less
Design and Analysis of Soft-Error Resilience Mechanisms for GPU Register File
Modern graphics processing units (GPUs) are using increasingly larger register file (RF) which occupies a large fraction of GPU core area and is very frequently accessed. This makes RF vulnerable to soft-errors (SE). In this paper, we present two techniques for improving SE resilience of GPU RF. First, we propose compressing the RF values for reducing the number of vulnerable bits. We leverage value similarity and the presence of narrow-width values to perform compression at warp or thread-level, respectively. Second, we propose selective hardening to design a portion of register entry with SE immune circuits. By collectively using these techniques, higher resilience can be provided with lower overhead. Without hardening, our warp and thread-level compression techniques bring 47.0% and 40.8% reduction in SE vulnerability, respectively.
more »
« less
- Award ID(s):
- 1657336
- PAR ID:
- 10048266
- Date Published:
- Journal Name:
- International Conference on VLSI Design and International Conference on Embedded Systems (VLSID)
- Page Range / eLocation ID:
- 409 to 414
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Today’s large-scale scientific applications running on high-performance computing (HPC) systems generate vast data volumes. Thus, data compression is becoming a critical technique to mitigate the storage burden and data-movement cost. However, existing lossy compressors for scientific data cannot achieve a high compression ratio and throughput simultaneously, hindering their adoption in many applications requiring fast compression, such as in-memory compression. To this end, in this work, we develop a fast and high-ratio error-bounded lossy compressor on GPUs for scientific data (called FZ-GPU). Specifically, we first design a new compression pipeline that consists of fully parallelized quantization, bitshuffle, and our newly designed fast encoding. Then, we propose a series of deep architectural optimizations for each kernel in the pipeline to take full advantage of CUDA architectures. We propose a warp-level optimization to avoid data conflicts for bit-wise operations in bitshuffle, maximize shared memory utilization, and eliminate unnecessary data movements by fusing different compression kernels. Finally, we evaluate FZ-GPU on two NVIDIA GPUs (i.e., A100 and RTX A4000) using six representative scientific datasets from SDRBench. Results on the A100 GPU show that FZ-GPU achieves an average speedup of 4.2× over cuSZ and an average speedup of 37.0× over a multi-threaded CPU implementation of our algorithm under the same error bound. FZ-GPU also achieves an average speedup of 2.3× and an average compression ratio improvement of 2.0× over cuZFP under the same data distortion.more » « less
-
Domain-specific languages that execute image processing pipelines on GPUs, such as Halide and Forma, operate by 1) dividing the image into overlapped tiles, and 2) fusing loops to improve memory locality. However, current approaches have limitations: 1) they require intra thread block synchronization, which has a nontrivial cost, 2) they must choose between small tiles that require more overlapped computations or large tiles that increase shared memory access (and lowers occupancy), and 3) their autoscheduling algorithms use simplified GPU models that can result in inefficient global memory accesses. We present a new approach for executing image processing pipelines on GPUs that addresses these limitations as follows. 1) We fuse loops to form overlapped tiles that fit in a single warp, which allows us to use lightweight warp synchronization. 2) We introduce hybrid tiling, which stores overlapped regions in a combination of thread-local registers and shared memory. Thus hybrid tiling either increases occupancy by decreasing shared memory usage or decreases overlapping computations using larger tiles. 3) We present an automatic loop fusion algorithm that considers several factors that affect the performance of GPU kernels. We implement these techniques in PolyMage-GPU, which is a new GPU backend for PolyMage. Our approach produces code that is faster than Halide's manual schedules: 1.65x faster on an NVIDIA GTX 1080Ti and 1.33x faster on an NVIDIA Tesla V100.more » « less
-
We design and implement a fully concurrent dynamic hash table for GPUs with comparable performance to the state of the art static hash tables. We propose a warp-cooperative work sharing strategy that reduces branch divergence and provides an efficient alternative to the traditional way of per-thread (or per-warp) work assignment and processing. By using this strategy, we build a dynamic non-blocking concurrent linked list, the slab list, that supports asynchronous, concurrent updates (insertions and deletions) as well as search queries. We use the slab list to implement a dynamic hash table with chaining (the slab hash). On an NVIDIA Tesla K40c GPU, the slab hash performs updates with up to 512 M updates/s and processes search queries with up to 937 M queries/s. We also design a warp-synchronous dynamic memory allocator, SlabAlloc, that suits the high performance needs of the slab hash. SlabAlloc dynamically allocates memory at a rate of 600 M allocations/s, which is up to 37x faster than alternative methods in similar scenarios.more » « less
-
We design and implement a fully concurrent dynamic hash table for GPUs with comparable performance to the state of the art static hash tables. We propose a warp-cooperative work sharing strategy that reduces branch divergence and provides an efficient alternative to the tradi- tional way of per-thread (or per-warp) work assignment and processing. By using this strategy, we build a dynamic non- blocking concurrent linked list, the slab list, that supports asynchronous, concurrent updates (insertions and deletions) as well as search queries. We use the slab list to implement a dynamic hash table with chaining (the slab hash). On an NVIDIA Tesla K40c GPU, the slab hash performs updates with up to 512 M updates/s and processes search queries with up to 937 M queries/s. We also design a warp-synchronous dynamic memory allocator, SlabAlloc, that suits the high performance needs of the slab hash. SlabAlloc dynamically allocates memory at a rate of 600 M allocations/s, which is up to 37x faster than alternative methods in similar scenarios.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.1109/VLSID.2017.14
