Design and implementation of dual processor block with shared external cache memory

The availability of low cost, high performance microprocessors has led to various designs of shared memory multiprocessor systems. As a result, commercial products which are based on shared memory have been proliferated. Such a multiprocessor system is heavily influenced by the structure of memory system and it is not difficult to find that most configurations include local cache memories. The more processors a system carries, the larger local cache memory is needed to maintain the traffic to and from the shared memory at reasonable level. The implementation of local cache memories, however, is not a simple task because of environmental limitations. In particular, the general lack of board space availability presents a formidable problem. A cache memory system usually needs space mostly to support its complex control logic circuits for the cache itself and network interfaces like snooping logic circuits for shared bus. Although packaging can be made denser to reduce system size, there are still multiple processors per board. It requires a more area-efficient cache memory architecture. This paper presents a design of shared cache for dual processor board of bus-based symmetric multiprocessors. The design and implementation issues are described first and then the evaluation and measurement results are discussed. The shared cache proposed in this paper has been determined to be quite area-efficient without the significant loss of throughput and scalability. It has been implemented as a plug-in unit for TICOM, a prevalent commercial multiprocessor system.     

Formal automatic verification of cache coherence in multiprocessors with relaxed memory models

State-based, formal methods have been successfully applied to the automatic verification of cache coherence in sequentially consistent systems. However, coherence in shared memory multiprocessors under a relaxed memory model is much more complex to verify automatically. With relaxed memory models, incoming invalidations and outgoing updates can be delayed in each cache while processors are allowed to race ahead. This buffering of memory accesses considerably increases the amount of state in each cache and the complexity of protocol interactions. Moreover, because caches can hold inconsistent copies of the same data for long periods of time, coherence cannot be verified by simply checking that cached copies are identical at all times. This paper makes two major contributions. First, we demonstrate how to model and verify cache coherence under a relaxed memory model in the context of state-based verification methods. Frameworks for modeling the hardware and for generating correct memory access sequences driving the hardware model are developed. We also show correctness properties which must be verified on the hardware model. Second, we demonstrate a successful application of a state-based verification tool called SSM for the verification of the delayed protocol, an aggressive protocol for relaxed memory models. SSM is based on an abstraction technique preserving the properties to verify. We show that with classical, explicit approaches the verification of cache coherence is realistically unfeasible because of the state space explosion problem, whereas SSM is able to verify protocols both at both behavioral and message-passing levels.     

High Performance Software Coherence for Current and Future Architectures

Shared memory provides an attractive and intuitive programming model for large-scale parallel computing, but requires a coherence mechanism to allow caching for performance while ensuring that processors do not use stale data in their computation. Implementation options range from distributed shared memory emulations on networks of workstations to tightly coupled fully cache-coherent distributed shared memory multiprocessors. Previous work indicates that performance varies dramatically from one end of this spectrum to the other. Hardware cache coherence is fast, but also costly and time-consuming to design and implement, while DSM systems provide acceptable performance on only a limit class of applications. We claim that an intermediate hardware option-memory-mapped network interfaces that support a global physical address space, without cache coherence-can provide most of the performance benefits of fully cache-coherent hardware, at a fraction of the cost. To support this claim we present a software coherence protocol that runs on this class of machines, and use simulation to conduct a performance study. We look at both programming and architectural issues in the context of software and hardware coherence protocols. Our results suggest that software coherence on NCC-NUMA machines in a more cost-effective approach to large-scale shared-memory multiprocessing than either pure distributed shared memory or hardware cache coherence.     

A Lock-Based Cache Coherence Protocol for Scope Consistency

Directory protocols are widely adopted to maintain cache coherence of distributed shared memory multiprocessors.Although scalable to a certain extent,directory protocols are complex enough to prevent it from being used in very large scale multiprocessors with tens of thousands of nodes.his paper proposes a lock-based cache coherence protocol for scope consistency.In does not rely on directory information to maintain cache coherence.Instead,cache coherence is maintained through requiring the releasing processor of a lock to stroe all write-notices generated in the associated critical section to the lock and the acquiring processor invalidates or updates its locally cached data copies according to the write notices of the lock.To evaluate the performance of the lock-based cache coherence protocol,a software SDM system named JIAJIA is built on network of workstations.Besides the lock-based cache coherence protocol,JIAJIA also characterizes itself with its shared memory organization scheme which combines the physical memories of multiple workstations to form a large shared space.Performance measurements with SPLASH2 program suite and NAS benchmarks indicate that,compared to recent SVM systems such as CVM,higher speedup is achieved by JIAJIA.Besides,JIAJIA can solve large scale problems that cannot be solved by other SVM systems due to memory size limitation.     

ARP:同时多线程处理器中共享Cache自适应运行时划分机制

同时多线程是一种延迟容忍的体系结构,采用共享的二级Cache,在每个周期内可以执行多个线程的多条指令,这就会增加对存储层次的压力,文中主要研究了SMT处理器中多个并发执行的线程之间共享Cache的划分问题,尤其是Cache共享中的公平性问题以及它和吞吐量之间的关系,传统的LRU策略会根据线程的需要隐式地划分共享Cache,给具有较高需求的线程分配较多的Cache空间,对Cache的管理具有不公平性,从而会引起线程饿死、优先级反转等问题,实现了一种自适应、运行时划分机制(ARP)来管理共享Cache.ARP采用公平性作为划分的度量,并且使用动态划分算法来优化公平性,该算法具有易于实现,所需剖析较少的特点,硬件上使用经典的监控器来收集每个线程的栈距离信息,其存储开销不到0.25%.实验结果显示,与基于LRU的Cache划分相比,ARP可以将一个2路SMT处理器的公平性提高2.26倍,而将吞吐量平均提高14.75%.     

Lightweight dynamic partitioning for last-level cache of multicore processor on real system

With rapid development of multi/many-core processors, contention in shared cache becomes more and more serious that restricts performance improvement of parallel programs. Recent researches have employed page coloring mechanism to realize cache partitioning on real system and to reduce contentions in shared cache. However, page coloring-based cache partitioning has some side effects, one is page coloring restricts memory space that an application can allocate, from which may lead to memory pressure, another is changing cache partition dynamically needs massive page copying which will incur large overhead. To make page coloring-based cache partition more practical, this paper proposes a malloc allocator-based dynamic cache partitioning mechanism with page coloring. Memory allocated by our malloc allocator can be dynamically partitioned among different applications according to partitioning policy. Only coloring the dynamically allocated pages can remit memory pressure and reduce page copying overhead led by re-coloring compared to all-page coloring. To further alleviate the overhead, we introduce minimum distance page copying strategy and lazy flush strategy. We conduct experiments on real system to evaluate these strategies and results show that they work well for reducing cache misses and re-coloring overhead.     

Precise control of page cache for containers

Container-based virtualization is becoming increasingly popular in cloud computing due to its efficiency and flexibility. Resource isolation is a fundamental property of containers. Existing works have indicated weak resource isolation could cause significant performance degradation for containerized applications and enhanced resource isolation. However, current studies have almost not discussed the isolation problems of page cache which is a key resource for containers. Containers leverage memory cgroup to control page cache usage. Unfortunately, existing policy introduces two major problems in a container-based environment. First, containers can utilize more memory than limited by their cgroup, effectively breaking memory isolation. Second, the OS kernel has to evict page cache to make space for newly-arrived memory requests, slowing down containerized applications. This paper performs an empirical study of these problems and demonstrates the performance impacts on containerized applications. Then we propose pCache (precise control of page cache) to address the problems by dividing page cache into private and shared and controlling both kinds of page cache separately and precisely. To do so, pCache leverages two new technologies: fair account (f-account) and evict on demand (EoD). F-account splits the shared page cache charging based on per-container share to prevent containers from using memory for free, enhancing memory isolation. And EoD reduces unnecessary page cache evictions to avoid the performance impacts. The evaluation results demonstrate that our system can effectively enhance memory isolation for containers and achieve substantial performance improvement over the original page cache management policy.     

存储模型仿真器的设计与实现

存储一致性问题和高速缓存一致性问题是共享存储并行计算机中两个最关键的问题,通过仿真器对它们进行了量化研究,设计并实现了一个存储模型仿真器MMS．基于MMS仿真了不同并行机结构模型下多种存储一致性模型的行为;针对不同类型的计算问题比较了不同的存储一致性模型,并对实验结果进行了分析;实现了几个不同的高速缓存一致性协议,并比较了它们的性能．     

Locally parallel cache design based on KL1 memory access characteristics

The parallel inference machine (PIM) is now being developed at ICOT. It consists of a dozen or more clusters, each of which is a tightly coupled multiprocessor (comprising about eight processing elements) with shared global memory and a common bus. Kernel language 1 (KL1), a parallel logic programming language based on Guarded Horn Clauses (GHC), is executed on each PIM cluster. This paper describes the memory access characteristics in KL1 parallel execution and a locally parallel cache mechanism with hardware lock. The most important issue of locally parallel cache design is how to reduce common bus traffic. A write-back cache protocol having five cache states specially optimized for KL1 execution on each PIM cluster is described. We introduced new software controlled memory access commands, named DW, ER, and RP. A hardware lock mechanism is attached to the cache on each processor. This lock mechanism enables efficient word-by-word locking, reducing common bus traffic by using the cache states.     

Reusability-aware cache memory sharing for chip multiprocessors with private L2 caches

In this paper, we propose a novel on-chip L2 cache organization for chip multiprocessors (CMPs) with private L2 caches. The proposed approach, called reusability-aware cache sharing (RACS), combines the advantages of both a private L2 cache and a shared L2 cache. Since a private L2 cache organization has a short access latency, the RACS scheme employs a private L2 cache organization. However, when a cache block in a private L2 cache is selected for eviction, RACS first evaluates its reusability. If the block is likely to be reused in the near future, it may be saved to a peer L2 cache which has space available. In this way, the RACS scheme effectively simulates the larger capacity of a shared L2 cache. Simulation results show that RACS reduced the number of off-chip memory accesses by 24% compared to a pure private L2 cache organization on average for the SPLASH 2 multi-threaded benchmarks, and by 16% for multi-programmed benchmarks.     

Virtual-Cache: A cache-line borrowing technique for efficient GPU cache architectures

GPUs provide megabytes of registers and shared memories to maintain the contexts for thousands of threads and enable fast data sharing amongst threads of a thread block, respectively. Besides, GPUs employ L1 cache to provide the high bandwidth service for memory requests. However, the average L1 cache capacity per thread is very limited, resulting in cache thrashing which in turn impairs the performance. Meanwhile, many registers and shared memories are unassigned to any warps or thread blocks. Moreover, registers and shared memories that are assigned can be idle when warps or thread blocks are finished. Exploiting the above insights, we propose Virtual-Cache to cost-effectively increase the effective size of L1 cache by utilizing the unassigned and released registers and shared memories as cache-lines in this paper. Specifically, we leverage the unassigned registers and shared memories to serve cache requests directly. Regarding the registers assigned to a warp, they can work as cache-lines after the warp completes the execution and before they are accessed again by a new launched warp. Regarding the shared memories of a thread block, they are enabled to serve cache requests when the thread block is finished till they are referenced by shared memory instructions of the relaunched thread block. The register file, shared memory and L1 cache are physically independent but logically unified as a large virtual cache with redesigned cache-line management. We develop the control and data path for the register file, making the register file accessible for cache requests by borrowing an operand collector to serve the cache requests. We also expand the control and data path for the shared memory to serve the cache requests. Our evaluation results show that Virtual-Cache makes the performance improved by 28% over the previously proposed cache management technique for cache-sensitive applications.     

A superlinear speedup region for matrix multiplication

The realization of modern processors is based on a multicore architecture with increasing number of cores per processor. Multicore processors are often designed such that some level of the cache hierarchy is shared among cores. Usually, last level cache is shared among several or all cores (e.g., L3 cache) and each core possesses private low level caches (e.g., L1 and L2 caches). Superlinear speedup is possible for matrix multiplication algorithm executed in a shared memory multiprocessor due to the existence of a superlinear region. It is a region where cache requirements for matrix storage of the sequential execution incur more cache misses than in parallel execution. This paper shows theoretically and experimentally that there is a region, where the superlinear speedup can be achieved. We provide a theoretical proof of existence of a superlinear speedup and determine boundaries of the region where it can be achieved. The experiments confirm our theoretical results. Therefore, these results will have impact on future software development and exploitation of parallel hardware on the basis of a shared memory multiprocessor architecture. Copyright © 2013 John Wiley & Sons, Ltd.     

An argument for simple COMA

We present design details and some initial performance results of a novel scalable shared memory multiprocessor architecture. This architecture features the automatic data migration and replication capabilities of cache-only memory architecture (COMA) machines, without the accompanying hardware complexity. A software layer manages cache space allocation at a page-granularity — similarly to distributed virtual shared memory (DVSM) systems —leaving simpler hardware to maintain shared memory coherence at a cache line granularity.

By reducing the hardware complexity, the machine cost and development time are reduced. We call the resulting hybrid hardware and software multiprocessor architecture Simple COMA. Preliminary results indicate that the performance of Simple COMA is comparable to that of more complex contemporary all-hardware designs.     

Timing analysis of concurrent programs running on shared cache multi-cores

Memory accesses form an important source of timing unpredictability. Timing analysis of real-time embedded software thus requires bounding the time for memory accesses. Multiprocessing, a popular approach for performance enhancement, opens up the opportunity for concurrent execution. However due to contention for any shared memory by different processing cores, memory access behavior becomes more unpredictable, and hence harder to analyze. In this paper, we develop a timing analysis method for concurrent software running on multi-cores with a shared instruction cache. Communication across tasks is by message passing. Our method progressively improves the lifetime estimates of tasks that execute concurrently on multiple cores, in order to estimate potential conflicts in the shared cache. Possible conflicts arising from overlapping task lifetimes are accounted for in the hit-miss classification of accesses to the shared cache, to provide safe execution time bounds. We show that our method produces lower worst-case response time (WCRT) estimates than existing shared-cache analysis on a real-world embedded application. Furthermore, we also exploit instruction cache locking to improve WCRT. By locking some beneficial memory blocks into L1 cache, the WCET of the tasks and L2 cache conflicts are reduced, resulting in better WCRT. Experiments demonstrate that significant WCRT reduction is achieved through cache locking.     

Cache-Based Synchronization in Shared Memory Multiprocessors

In shared memory multiprocessors, efficient synchronization is imperative to assure good performance. There are two aspects to the “cost” of a synchronization operation: the first is the waiting time at synchronization points, and the second is the intrinsic overhead in performing the operation. The overhead has two components. The first component deals with contention resolution for synchronization operation among competing processors. The second component deals with the shared data accesses that the processor has to perform once it enters a synchronization region. We present a mechanism to reduce the overhead of performing synchronization operations in a cache-based shared memory multiprocessor. The mechanism is based on the intuitive notion that parallel programs invariably use synchronization operations to govern the access to shared data. Traditional multiprocessor cache protocols treat synchronization accesses the same way as normal read/write memory accesses, leading to inefficiencies in performing synchronization operations which ultimately limit the scalability of such systems. The key idea in our approach is to combine synchronization with the coherence maintenance for the cached data. Each cache line maintains states for synchronization as well as for cache coherence, and the cache protocol ensures the correctness of the synchronization operations and the coherence of the data at these synchronization points. To assess the performance gain due to the proposed mechanism, simulation studies are performed using a workload model that represents a dynamic scheduling paradigm which forms the core of several parallel programs. Results from simulation studies show that the proposed cache-based synchronization performs significantly better than traditional cache coherence approaches.     

Impact of level-2 cache sharing on the performance and power requirements of homogeneous multicore embedded systems

In order to satisfy the needs for increasing computer processing power, there are significant changes in the design process of modern computing systems. Major chip-vendors are deploying multicore or manycore processors to their product lines. Multicore architectures offer a tremendous amount of processing speed. At the same time, they bring challenges for embedded systems which suffer from limited resources. Various cache memory hierarchies have been proposed to satisfy the requirements for different embedded systems. Normally, a level-1 cache (CL1) memory is dedicated to each core. However, the level-2 cache (CL2) can be shared (like Intel Xeon and IBM Cell) or distributed (like AMD Athlon). In this paper, we investigate the impact of the CL2 organization type (shared Vs distributed) on the performance and power consumption of homogeneous multicore embedded systems. We use VisualSim and Heptane tools to model and simulate the target architectures running FFT, MI, and DFT applications. Experimental results show that by replacing a single-core system with an 8-core system, reductions in mean delay per core of 64% for distributed CL2 and 53% for shared CL2 are possible with little additional power (15% for distributed CL2 and 18% for shared CL2) for FFT. Results also reveal that the distributed CL2 hierarchy outperforms the shared CL2 hierarchy for all three applications considered and for other applications with similar code characteristics.     

Hybrid address spaces: A methodology for implementing scalable high-level programming models on non-coherent many-core architectures

This paper introduces hybrid address spaces as a fundamental design methodology for implementing scalable runtime systems on many-core architectures without hardware support for cache coherence. We use hybrid address spaces for an implementation of MapReduce, a programming model for large-scale data processing, and the implementation of a remote memory access (RMA) model. Both implementations are available on the Intel SCC and are portable to similar architectures. We present the design and implementation of HyMR, a MapReduce runtime system whereby different stages and the synchronization operations between them alternate between a distributed memory address space and a shared memory address space, to improve performance and scalability. We compare HyMR to a reference implementation and we find that HyMR improves performance by a factor of 1.71× over a set of representative MapReduce benchmarks. We also compare HyMR with Phoenix++, a state-of-art implementation for systems with hardware-managed cache coherence in terms of scalability and sustained to peak data processing bandwidth, where HyMR demonstrates improvements of a factor of 3.1× and 3.2× respectively. We further evaluate our hybrid remote memory access (HyRMA) programming model and assess its performance to be superior of that of message passing.     

Multiprocessor Cache Coherence Based on Virtual Memory Support

Virtual-memory-based cache coherence is a mechanism that relies only on hardware that already exists on the microprocessors of a shared-memory multiprocessor system, yet dynamically detects and resolves potential cache inconsistencies using virtual-memory techniques. The key feature of the approach is that the virtual-memory translation hardware on each processor is used to detect shared accesses that could lead to memory incoherencies, and VM page fault handlers execute the appropriate actions to maintain cache coherence. VM-based cache coherence basically trades off design simplicity for increased software overheads. The work presented in this paper evaluates this trade-off. We show that VM-based cache coherence performs well for scientific applications that require significant aggregate memory bandwidth.     

Filtering directory lookups in CMPs

Coherence protocols consume an important fraction of power to determine which coherence action to perform. Specifically, on CMPs with shared cache and directory-based coherence protocol implemented as a duplicate of local caches tags, we have observed that a big fraction of directory lookups cause a miss, because the block looked up is not allocated in any local cache. To reduce the number of directory lookups and therefore the power consumption, we propose to add a filter before the directory access.We introduce two filter implementations. In the first one, filtering information is explicitly kept in the shared cache for every block. In the second one, filtering information is decoupled from the shared cache organization, so the filter size does not depend on the shared cache size.We evaluate our filters in a CMP with 8 in-order processors with 4 threads each and a memory hierarchy with write-through local caches and a shared cache. We show that, for SPLASH2 benchmarks, the proposed filters reduce the number of directory lookups performed by 60% while power consumption is reduced by ∼28%. For Specweb2005, the number of directory lookups performed is reduced by 68% (44%), while directory power consumption is reduced by 19% (9%) using the first (second) filter implementation.     

Latency hiding on COMA multiprocessors

Cache-only memory access (COMA) multiprocessors support scalable coherent shared memory with a uniform memory access programming model. The local portion of shared memory associated with a processor is organized as a cache. This cache-based organization of memory results in long remote memory access latencies. Latency-hiding mechanisms can reduce effective remote memory access latency by making data present in a processor's local memory by the time the data are needed. In this paper we study the effectiveness of latency-hiding mechanisms on the KSR2 multiprocessor in improving the performance of three programs. The communication patterns of each program are analyzed and the mechanisms for latency hiding are applied. Results from a 52-processor system indicate that these mechanisms hide a significant portion of the latency of remote memory accesses. The results also quantify benefits in overall application performance.An earlier version of this paper was presented at the 1995 International Conference on Parallel Processing Techniques and Applications.