DRAM Row Research Articles

Flexible auto-refresh

DRAM cells require periodic refreshing to preserve data. In JEDEC DDRx devices, a refresh operation is performed via an auto-refresh command, which refreshes multiple rows in multiple banks simultaneously. The internal implementation of auto-refresh is completely opaque outside the DRAM --- all the memory controller can do is to instruct the DRAM to refresh itself --- the DRAM handles all else, in particular determining which rows in which banks are to be refreshed. This is in conflict with a large body of research on reducing the refresh overhead, in which the memory controller needs fine-grained control over which regions of the memory are refreshed. For example, prior works exploit the fact that a subset of DRAM rows can be refreshed at a slower rate than other rows due to access rate or retention period variations. However, such row-granularity approaches cannot use the standard auto-refresh command, which refreshes an entire batch of rows at once and does not permit skipping of rows. Consequently, prior schemes are forced to use explicit sequences of activate (ACT) and precharge (PRE) operations to mimic row-level refreshing. The drawback is that, compared to using JEDEC's auto-refresh mechanism, using explicit ACT and PRE commands is inefficient, both in terms of performance and power. In this paper, we show that even when skipping a high percentage of refresh operations, existing row-granurality refresh techniques are mostly ineffective due to the inherent efficiency disparity between ACT/PRE and the JEDEC auto-refresh mechanism. We propose a modification to the DRAM that extends its existing control-register access protocol to include the DRAM's internal refresh counter. We also introduce a new "dummy refresh" command that skips refresh operations and simply increments the internal counter. We show that these modifications allow a memory controller to reduce as many refreshes as in prior work, while achieving significant energy and performance advantages by using auto-refresh most of the time.

The dirty-block index

On-chip caches maintain multiple pieces of metadata about each cached block---e.g., dirty bit, coherence information, ECC. Traditionally, such metadata for each block is stored in the corresponding tag entry in the tag store. While this approach is simple to implement and scalable, it necessitates a full tag store lookup for any metadata query---resulting in high latency and energy consumption. We find that this approach is inefficient and inhibits several cache optimizations. In this work, we propose a new way of organizing the dirty bit information that enables simpler and more efficient implementations of several optimizations. In our proposed approach, we remove the dirty bits from the tag store and organize it differently in a separate structure, which we call the Dirty-Block Index (DBI). The organization of DBI is simple: it consists of multiple entries, each corresponding to some row in DRAM. A bit vector in each entry tracks whether or not each block in the corresponding DRAM row is dirty We demonstrate the benfits of DBI by using it to simultaneously and efficiently implement three optimizations proposed by prior work: 1) Aggressive DRAM-aware writeback, 2) Bypassing cache lookups, and 3) Heterogeneous ECC for clean/dirty blocks. DBI, with all three optimizations enabled, improves performance by 31% compared to the baseline (by 6% compared to the best previous mechanism) while reducing overall cache area cost by 8% compared to prior approaches.

Flipping bits in memory without accessing them

Memory isolation is a key property of a reliable and secure computing system--an access to one memory address should not have unintended side effects on data stored in other addresses. However, as DRAM process technology scales down to smaller dimensions, it becomes more difficult to prevent DRAM cells from electrically interacting with each other. In this paper, we expose the vulnerability of commodity DRAM chips to disturbance errors. By reading from the same address in DRAM, we show that it is possible to corrupt data in nearby addresses. More specifically, activating the same row in DRAM corrupts data in nearby rows. We demonstrate this phenomenon on Intel and AMD systems using a malicious program that generates many DRAM accesses. We induce errors in most DRAM modules (110 out of 129) from three major DRAM manufacturers. From this we conclude that many deployed systems are likely to be at risk. We identify the root cause of disturbance errors as the repeated toggling of a DRAM row's wordline, which stresses inter-cell coupling effects that accelerate charge leakage from nearby rows. We provide an extensive characterization study of disturbance errors and their behavior using an FPGA-based testing platform. Among our key findings, we show that (i) it takes as few as 139K accesses to induce an error and (ii) up to one in every 1.7K cells is susceptible to errors. After examining various potential ways of addressing the problem, we propose a low-overhead solution to prevent the errors

Reducing DRAM row activations with eager read/write clustering

This article describes and evaluates a new approach to optimizing DRAM performance and energy consumption that is based on eagerly writing dirty cache lines to DRAM. Under this approach, many dirty cache lines are written to DRAM before they are evicted. In particular, dirty cache lines that have not been recently accessed are eagerly written to DRAM when the corresponding row has been activated by an ordinary, noneager access, such as a read. This approach enables clustering of reads and writes that target the same row, resulting in a significant reduction in row activations. Specifically, for a variety of applications, it reduces the number of DRAM row activations by an average of 42% and a maximum of 82%. Moreover, the results from a full-system simulator show compelling performance improvements and energy consumption reductions. Out of 23 applications, 6 have overall performance improvements between 10% and 20%, and 3 have improvements in excess of 20%. Furthermore, 12 consume between 10% and 20% less DRAM energy, and 7 have energy consumption reductions in excess of 20%.

Reducing DRAM row activations with eager read/write clustering

This article describes and evaluates a new approach to optimizing DRAM performance and energy consumption that is based on eagerly writing dirty cache lines to DRAM. Under this approach, many dirty cache lines are written to DRAM before they are evicted. In particular, dirty cache lines that have not been recently accessed are eagerly written to DRAM when the corresponding row has been activated by an ordinary, noneager access, such as a read. This approach enables clustering of reads and writes that target the same row, resulting in a significant reduction in row activations. Specifically, for a variety of applications, it reduces the number of DRAM row activations by an average of 42% and a maximum of 82%. Moreover, the results from a full-system simulator show compelling performance improvements and energy consumption reductions. Out of 23 applications, 6 have overall performance improvements between 10% and 20%, and 3 have improvements in excess of 20%. Furthermore, 12 consume between 10% and 20% less DRAM energy, and 7 have energy consumption reductions in excess of 20%.

OWL

Emerging GPGPU architectures, along with programming models like CUDA and OpenCL, offer a cost-effective platform for many applications by providing high thread level parallelism at lower energy budgets. Unfortunately, for many general-purpose applications, available hardware resources of a GPGPU are not efficiently utilized, leading to lost opportunity in improving performance. A major cause of this is the inefficiency of current warp scheduling policies in tolerating long memory latencies. In this paper, we identify that the scheduling decisions made by such policies are agnostic to thread-block, or cooperative thread array (CTA), behavior, and as a result inefficient. We present a coordinated CTA-aware scheduling policy that utilizes four schemes to minimize the impact of long memory latencies. The first two schemes, CTA-aware two-level warp scheduling and locality aware warp scheduling, enhance per-core performance by effectively reducing cache contention and improving latency hiding capability. The third scheme, bank-level parallelism aware warp scheduling, improves overall GPGPU performance by enhancing DRAM bank-level parallelism. The fourth scheme employs opportunistic memory-side prefetching to further enhance performance by taking advantage of open DRAM rows. Evaluations on a 28-core GPGPU platform with highly memory-intensive applications indicate that our proposed mechanism can provide 33% average performance improvement compared to the commonly-employed round-robin warp scheduling policy.

OWL

Emerging GPGPU architectures, along with programming models like CUDA and OpenCL, offer a cost-effective platform for many applications by providing high thread level parallelism at lower energy budgets. Unfortunately, for many general-purpose applications, available hardware resources of a GPGPU are not efficiently utilized, leading to lost opportunity in improving performance. A major cause of this is the inefficiency of current warp scheduling policies in tolerating long memory latencies. In this paper, we identify that the scheduling decisions made by such policies are agnostic to thread-block, or cooperative thread array (CTA), behavior, and as a result inefficient. We present a coordinated CTA-aware scheduling policy that utilizes four schemes to minimize the impact of long memory latencies. The first two schemes, CTA-aware two-level warp scheduling and locality aware warp scheduling, enhance per-core performance by effectively reducing cache contention and improving latency hiding capability. The third scheme, bank-level parallelism aware warp scheduling, improves overall GPGPU performance by enhancing DRAM bank-level parallelism. The fourth scheme employs opportunistic memory-side prefetching to further enhance performance by taking advantage of open DRAM rows. Evaluations on a 28-core GPGPU platform with highly memory-intensive applications indicate that our proposed mechanism can provide 33% average performance improvement compared to the commonly-employed round-robin warp scheduling policy.

LOT-ECC

Memory system reliability is a serious and growing concern in modern servers. Existing chipkill-level memory protection mechanisms suffer from several drawbacks. They activate a large number of chips on every memory access -- this increases energy consumption, and reduces performance due to the reduction in rank-level parallelism. Additionally, they increase access granularity, resulting in wasted bandwidth in the absence of sufficient access locality. They also restrict systems to use narrow-I/O x4 devices, which are known to be less energy-efficient than the wider x8 DRAM devices. In this paper, we present LOT-ECC, a localized and multi-tiered protection scheme that attempts to solve these problems. We separate error detection and error correction functionality, and employ simple checksum and parity codes effectively to provide strong fault-tolerance, while simultaneously simplifying implementation. Data and codes are localized to the same DRAM row to improve access efficiency. We use system firmware to store correction codes in DRAM data memory and modify the memory controller to handle data mapping. We thus build an effective fault-tolerance mechanism that provides strong reliability guarantees, activates as few chips as possible (reducing power consumption by up to 44.8% and reducing latency by up to 46.9%), and reduces circuit complexity, all while working with commodity DRAMs and operating systems. Finally, we propose the novel concept of a heterogeneous DIMM that enables the extension of LOT-ECC to x16 and wider DRAM parts.

Enabling Efficient and Scalable Hybrid Memories Using Fine-Granularity DRAM Cache Management

Hybrid main memories composed of DRAM as a cache to scalable non-volatile memories such as phase-change memory (PCM) can provide much larger storage capacity than traditional main memories. A key challenge for enabling high-performance and scalable hybrid memories, though, is efficiently managing the metadata (e.g., tags) for data cached in DRAM at a fine granularity. Based on the observation that storing metadata off-chip in the same row as their data exploits DRAM row buffer locality, this paper reduces the overhead of fine-granularity DRAM caches by only caching the metadata for recently accessed rows on-chip using a small buffer. Leveraging the flexibility and efficiency of such a fine-granularity DRAM cache, we also develop an adaptive policy to choose the best granularity when migrating data into DRAM. On a hybrid memory with a 512MB DRAM cache, our proposal using an 8KB on-chip buffer can achieve within 6% of the performance of, and 18% better energy efficiency than, a conventional 8MB SRAM metadata store, even when the energy overhead due to large SRAM metadata storage is not considered.

Micro-pages

Power consumption and DRAM latencies are serious concerns in modern chip-multiprocessor (CMP or multi-core) based compute systems. The management of the DRAM row buffer can significantly impact both power consumption and latency. Modern DRAM systems read data from cell arrays and populate a row buffer as large as 8 KB on a memory request. But only a small fraction of these bits are ever returned back to the CPU. This ends up wasting energy and time to read (and subsequently write back) bits which are used rarely. Traditionally, an open-page policy has been used for uni-processor systems and it has worked well because of spatial and temporal locality in the access stream. In future multi-core processors, the possibly independent access streams of each core are interleaved, thus destroying the available locality and significantly under-utilizing the contents of the row buffer. In this work, we attempt to improve row-buffer utilization for future multi-core systems. The schemes presented here are motivated by our observations that a large number of accesses within heavily accessed OS pages are to small, contiguous "chunks" of cache blocks. Thus, the co-location of chunks (from different OS pages) in a row-buffer will improve the overall utilization of the row buffer contents, and consequently reduce memory energy consumption and access time. Such co-location can be achieved in many ways, notably involving a reduction in OS page size and software or hardware assisted migration of data within DRAM. We explore these mechanisms and discuss the trade-offs involved along with energy and performance improvements from each scheme. On average, for applications with room for improvement, our best performing scheme increases performance by 9% (max. 18%) and reduces memory energy consumption by 15% (max. 70%).

Micro-pages

Power consumption and DRAM latencies are serious concerns in modern chip-multiprocessor (CMP or multi-core) based compute systems. The management of the DRAM row buffer can significantly impact both power consumption and latency. Modern DRAM systems read data from cell arrays and populate a row buffer as large as 8 KB on a memory request. But only a small fraction of these bits are ever returned back to the CPU. This ends up wasting energy and time to read (and subsequently write back) bits which are used rarely. Traditionally, an open-page policy has been used for uni-processor systems and it has worked well because of spatial and temporal locality in the access stream. In future multi-core processors, the possibly independent access streams of each core are interleaved, thus destroying the available locality and significantly under-utilizing the contents of the row buffer. In this work, we attempt to improve row-buffer utilization for future multi-core systems. The schemes presented here are motivated by our observations that a large number of accesses within heavily accessed OS pages are to small, contiguous "chunks" of cache blocks. Thus, the co-location of chunks (from different OS pages) in a row-buffer will improve the overall utilization of the row buffer contents, and consequently reduce memory energy consumption and access time. Such co-location can be achieved in many ways, notably involving a reduction in OS page size and software or hardware assisted migration of data within DRAM. We explore these mechanisms and discuss the trade-offs involved along with energy and performance improvements from each scheme. On average, for applications with room for improvement, our best performing scheme increases performance by 9% (max. 18%) and reduces memory energy consumption by 15% (max. 70%).

DDR3 SDRAM with a Complete Predictor

In the arsenal of resources for improving computer memory system performance, predictors have gained an increasing role in the past few years. They enable hiding the latencies when accessing cache or main memory. In our previous work we proposed a DDR SDRAM controller with predictors that not only close the opened DRAM row but also predict the next row to be opened. In this paper we explore the possibilities of trying the same techniques on the latest type of DRAM memory, DDR3 SDRAM, with further improvements of the predictors.

PARBLO: Page-Allocation-Based DRAM Row Buffer Locality Optimization

DRAM row buffer conflicts can increase memory access latency significantly. This paper presents a new page-allocation-based optimization that works seamlessly together with some existing hardware and software optimizations to eliminate significantly more row buffer conflicts. Validation in simulation using a set of selected scientific and engineering benchmarks against a few representative memory controller optimizations shows that our method can reduce row buffer miss rates by up to 76% (with an average of 37.4%). This reduction in row buffer miss rates will be translated into performance speedups by up to 15% (with an average of 5%).

DRAM Controller with a Complate Predictor

In the arsenal of resources for computer memory system performance improvement, predictors have gained an increasing role in the past years. They can suppress the latencies when accessing cache or main memory. In paper [1] it is shown how temporal parameters of cache memory access, defined as live time, dead time and access interval could be used for prediction of data prefetching. This paper examines the feasibility of applying an analog technique on controlling of opening/closing DRAM memory rows, with various improvements. The results described herein confirm the feasibility, and allow us to propose a DRAM controller with predictors that not only close the opened DRAM row, but also predict the next row to be opened.

DRAM performance as a function of its structure and memory stream locality

This paper studies the performance of a DRAM component as a function of its structure and the locality of the memory stream. We present a method and a tool for retrieving scalar values for temporal and spatial locality and discuss caches as locality filters. Combinations of cache systems and DRAM configurations having varying number of banks are simulated, and the locality of the DRAM input memory stream is analyzed. The results show that there is a usable amount of locality in the post-cache memory stream, but it is poorly utilized by the current DRAM structures. The developed scalar metrics are found to be suitable for outlining and understanding the DRAM performance. Analyzing locality and considering possibilities to utilize it by DRAMs will become essential in the future, as the DRAM row access time becomes increasingly dominant.

Efficient use of memory bandwidth to improve network processor throughput

We consider the efficiency of packet buffers used in packet switches built using network processors (NPs). Packet buffers are typically implemented using DRAM, which provides plentiful buffering at a reasonable cost. The problem we address is that a typical NP workload may be unable to utilize the peak DRAM bandwidth. Since the bandwidth of the packet buffer is often the bottleneck in the performance of a shared-memory packet switch, inefficient use of available DRAM bandwidth further reduces the packet throughput. Specialized hardware-based schemes that alleviate the DRAM bandwith problem in high-end routers may be less applicable to NP-based systems, in which cost is an important consideration.In this paper, we propose cost-effective ways to enhance average-case DRAM bandwidth. In modern DRAMs, successive accesses falling within the same DRAM row are significantly faster than those falling across rows. If accesses to DRAM can be generated differently or reordered to take advantage of fast same-row accesses, peak DRAM bandwidth can be approached. The challenge is in exploiting this "row locality" despite the unpredictable nature of memory accesses in NPs. We propose a set of simple techniques to meet this challenge. These include locality-sensitive buffer allocation on packet input, reordering DRAM accesses to increase locality, and prefetching to reduce row miss penalty. We evaluate our techniques on cycle-accurate simulations of Intel's IXP 1200 network processor and find that they boost packet throughput on average by 42.7%, utilizing nearly the peak DRAM bandwidth, for a set of common NP applications processing a real trace.

Designing a modern memory hierarchy with hardware prefetching

In this paper, we address the severe performance gap caused by high processor clock rates and slow DRAM accesses. We show that, even with an aggressive, next-generation memory system using four Direct Rambus channels and an integrated one-megabyte level-two cache, a processor still spends over half its time stalling for L2 misses. Our experimental analysis begins with an effort to tune our baseline memory system aggressively: incorporating optimizations to reduce DRAM row buffer misses, reordering miss accesses to reduce queuing delay, and adjusting the L2 block size to match each channel organization. We show that there is a large gap between the block sizes at which performance is best and at which miss rate is minimized. Using those results, we evaluate a hardware prefetch unit integrated with the L2 cache and memory controllers. By issuing prefetches only when the Rambus channels are idle, prioritizing them to maximize DRAM row buffer hits, and giving them low replacement priority, we achieve a 65 percent speedup across 10 of the 26 SPEC2000 benchmarks, without degrading the performance of the others. With eight Rambus channels, these 10 benchmarks improve to within 10 percent of the performance of a perfect L2 cache.

Hidden double data transfer scheme for MDL design

A high-speed DRAM data transfer scheme between DRAM and logic parts in merged DRAM logic (MDL) designs is proposed with logically divided DRAM row address mapping. The proposed scheme results in a 20% faster write access and 40% faster read access. It can be used as a general design framework to maximise DRAM access speed in various MDL designs. A test chip has been fabricated by 0.16 µm DRAM technology, and the scheme has been verified in the design of a DRAM L2 cache memory.