اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدEnabling High-Capacity, Latency-Tolerant, and Highly-Concurrent GPU Register Files via Software/Hardware Cooperation
120
0
0.0
(
0
)
نشر من قبل
Mohammad Sadrosadati
تاريخ النشر
2020
مجال البحث
الهندسة المعلوماتية
والبحث باللغة
English
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Graphics Processing Units (GPUs) employ large register files to accommodate all active threads and accelerate context switching. Unfortunately, register files are a scalability bottleneck for future GPUs due to long access latency, high power consumption, and large silicon area provisioning. Prior work proposes hierarchical register file to reduce the register file power consumption by caching registers in a smaller register file cache. Unfortunately, this approach does not improve register access latency due to the low hit rate in the register file cache. In this paper, we propose the Latency-Tolerant Register File (LTRF) architecture to achieve low latency in a two-level hierarchical structure while keeping power consumption low. We observe that compile-time interval analysis enables us to divide GPU program execution into intervals with an accurate estimate of a warps aggregate register working-set within each interval. The key idea of LTRF is to prefetch the estimated register working-set from the main register file to the register file cache under software control, at the beginning of each interval, and overlap the prefetch latency with the execution of other warps. We observe that register bank conflicts while prefetching the registers could greatly reduce the effectiveness of LTRF. Therefore, we devise a compile-time register renumbering technique to reduce the likelihood of register bank conflicts. Our experimental results show that LTRF enables high-capacity yet long-latency main GPU register files, paving the way for various optimizations. As an example optimization, we implement the main register file with emerging high-density high-latency memory technologies, enabling 8X larger capacity and improving overall GPU performance by 34%.
قيم البحث
اقرأ أيضاً
Putting the DRAM on the same package with a processor enables several times higher memory bandwidth than conventional off-package DRAM. Yet, the latency of in-package DRAM is not appreciably lower than that of off-package DRAM. A promising use of in-
package DRAM is as a large cache. Unfortunately, most previous DRAM cache designs mainly optimize for hit latency and do not consider off-chip bandwidth efficiency as a first-class design constraint. Hence, as we show in this paper, these designs are suboptimal for use with in-package DRAM. We propose a new DRAM cache design, Banshee, that optimizes for both in- and off-package DRAM bandwidth efficiency without degrading access latency. The key ideas are to eliminate the in-package DRAM bandwidth overheads due to costly tag accesses through virtual memory mechanism and to incorporate a bandwidth-aware frequency-based replacement policy that is biased to reduce unnecessary traffic to off-package DRAM. Our extensive evaluation shows that Banshee provides significant performance improvement and traffic reduction over state-of-the-art latency-optimized DRAM cache designs.
DRAM is the prevalent main memory technology, but its long access latency can limit the performance of many workloads. Although prior works provide DRAM designs that reduce DRAM access latency, their reduced storage capacities hinder the performance
of workloads that need large memory capacity. Because the capacity-latency trade-off is fixed at design time, previous works cannot achieve maximum performance under very different and dynamic workload demands. This paper proposes Capacity-Latency-Reconfigurable DRAM (CLR-DRAM), a new DRAM architecture that enables dynamic capacity-latency trade-off at low cost. CLR-DRAM allows dynamic reconfiguration of any DRAM row to switch between two operating modes: 1) max-capacity mode, where every DRAM cell operates individually to achieve approximately the same storage density as a density-optimized commodity DRAM chip and 2) high-performance mode, where two adjacent DRAM cells in a DRAM row and their sense amplifiers are coupled to operate as a single low-latency logical cell driven by a single logical sense amplifier. We implement CLR-DRAM by adding isolation transistors in each DRAM subarray. Our evaluations show that CLR-DRAM can improve system performance and DRAM energy consumption by 18.6% and 29.7% on average with four-core multiprogrammed workloads. We believe that CLR-DRAM opens new research directions for a system to adapt to the diverse and dynamically changing memory capacity and access latency demands of workloads.
An exponential growth in data volume, combined with increasing demand for real-time analysis (i.e., using the most recent data), has resulted in the emergence of database systems that concurrently support transactions and data analytics. These hybrid
transactional and analytical processing (HTAP) database systems can support real-time data analysis without the high costs of synchronizing across separate single-purpose databases. Unfortunately, for many applications that perform a high rate of data updates, state-of-the-art HTAP systems incur significant drops in transactional (up to 74.6%) and/or analytical (up to 49.8%) throughput compared to performing only transactions or only analytics in isolation, due to (1) data movement between the CPU and memory, (2) data update propagation, and (3) consistency costs. We propose Polynesia, a hardware-software co-designed system for in-memory HTAP databases. Polynesia (1) divides the HTAP system into transactional and analytical processing islands, (2) implements custom algorithms and hardware to reduce the costs of update propagation and consistency, and (3) exploits processing-in-memory for the analytical islands to alleviate data movement. Our evaluation shows that Polynesia outperforms three state-of-the-art HTAP systems, with average transactional/analytical throughput improvements of 1.70X/3.74X, and reduces energy consumption by 48% over the prior lowest-energy system.
Hardware flaws are permanent and potent: hardware cannot be patched once fabricated, and any flaws may undermine any software executing on top. Consequently, verification time dominates implementation time. The gold standard in hardware Design Verifi
cation (DV) is concentrated at two extremes: random dynamic verification and formal verification. Both struggle to root out the subtle flaws in complex hardware that often manifest as security vulnerabilities. The root problem with random verification is its undirected nature, making it inefficient, while formal verification is constrained by the state-space explosion problem, making it infeasible against complex designs. What is needed is a solution that is directed, yet under-constrained. Instead of making incremental improvements to existing DV approaches, we leverage the observation that existing software fuzzers already provide such a solution, and adapt them for hardware DV. Specifically, we translate RTL hardware to a software model and fuzz that model. The central challenge we address is how best to mitigate the differences between the hardware execution model and software execution model. This includes: 1) how to represent test cases, 2) what is the hardware equivalent of a crash, 3) what is an appropriate coverage metric, and 4) how to create a general-purpose fuzzing harness for hardware. To evaluate our approach, we fuzz four IP blocks from Googles OpenTitan SoC. Our experiments reveal a two orders-of-magnitude reduction in run time to achieve Finite State Machine (FSM) coverage over traditional dynamic verification schemes. Moreover, with our design-agnostic harness, we achieve over 88% HDL line coverage in three out of four of our designs -- even without any initial seeds.
Despite the superb performance of State-Of-The-Art (SOTA) DNNs, the increasing computational cost makes them very challenging to meet real-time latency and accuracy requirements. Although DNN runtime latency is dictated by model property (e.g., archi
tecture, operations), hardware property (e.g., utilization, throughput), and more importantly, the effective mapping between these two, many existing approaches focus only on optimizing model property such as FLOPS reduction and overlook the mismatch between DNN model and hardware properties. In this work, we show that the mismatch between the varied DNN computation workloads and GPU capacity can cause the idle GPU tail effect, leading to GPU under-utilization and low throughput. As a result, the FLOPs reduction cannot bring effective latency reduction, which causes sub-optimal accuracy versus latency trade-offs. Motivated by this, we propose a GPU runtime-aware DNN optimization methodology to eliminate such GPU tail effect adaptively on GPU platforms. Our methodology can be applied on top of existing SOTA DNN optimization approaches to achieve better latency and accuracy trade-offs. Experiments show 11%-27% latency reduction and 2.5%-4.0% accuracy improvement over several SOTA DNN pruning and NAS methods, respectively
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
