Memory isolation is critical for system reliability, security, and safety. Unfortunately, read disturbance can break memory isolation in modern DRAM chips. For example, RowHammer is awell-studied read-disturb phenomenon where repeatedly opening and closing (i.e., hammering) a DRAM row many times causes bitflips in physically nearby rows. This paper experimentally demonstrates and analyzes another widespread read-disturb phenomenon, RowPress, in real DDR4 DRAM chips. RowPress breaks memory isolation by keeping a DRAM row open for a long period of time, which disturbs physically nearby rows enough to cause bitflips. We show that RowPress amplifies DRAM's vulnerability to read-disturb attacks by significantly reducing the number of row activations needed to induce a bitflip by one to two orders of magnitude under realistic conditions. In extreme cases, RowPress induces bitflips in a DRAM row when an adjacent row is activated only once. Our detailed characterization of 164 real DDR4 DRAM chips shows that RowPress 1) affects chips from all three major DRAM manufacturers, 2) gets worse as DRAM technology scales down to smaller node sizes, and 3) affects a different set of DRAM cells from RowHammer and behaves differently from RowHammer as temperature and access pattern changes. We also show that cells vulnerable to RowPress are very different from cells vulnerable to retention failures. We demonstrate in a real DDR4-based system with RowHammer protection that 1) a user-level program induces bitflips by leveraging RowPress while conventional RowHammer cannot do so, and 2) a memory controller that adaptively keeps the DRAM row open for a longer period of time based on access pattern can facilitate RowPress-based attacks. To prevent bitflips due to RowPress, we describe and analyze four potential mitigation techniques, including a new methodology that adapts existing RowHammer mitigation techniques to also mitigate RowPress with low additional performance overhead. We evaluate this methodology and demonstrate that it is effective on a variety of workloads. We open source all our code and data to facilitate future research on RowPress.

Data copy is a widely-used memory operation in many programs and operating system services. In conventional computers, data copy is often carried out by two separate read and write transactions that pass data back and forth between the memory hierarchy and processor registers. Some prior mechanisms propose to avoid this unnecessary data movement by using the shared internal bus in DRAM chip to directly copy data between two DRAM banks. While these methods exhibit superior performance, compared to conventional techniques, this technique does not allow data copy over different DRAM channels. Hence, this technique has limited benefit for the emerging 3D stacked memories (such as HMC and HBM) that contains tens of banks across multiple memory controllers. In this paper, we present Network-on-Memory (NoM), a lightweight inter-bank communication scheme that enables direct data copy within memory. NoM adopts a TDM-based circuit-switching design, where circuit setup is done by the memory controller. Compared to previous state-of-the-art approaches, NoM enables both data copy over multiple DRAM channels and concurrent copy operation. Our evaluation shows that NoM improves the performance of data-intensive workloads by 3.8X on average compare to the state-of-the-art techniques, respectively.

Modern Machine Learning (ML) training on large-scale datasets is a very time-consuming workload. It relies on the optimization algorithm Stochastic Gradient Descent (SGD) due to its effectiveness, simplicity, and generalization performance (i.e., test performance on unseen data). Processor-centric architectures (e.g., CPUs, GPUs) commonly used for modern ML training workloads based on SGD are bottlenecked by data movement between the processor and memory units due to the poor data locality in accessing large training datasets. As a result, processor-centric architectures suffer from low performance and high energy consumption while executing ML training workloads. Processing-In-Memory (PIM) is a promising solution to alleviate the data movement bottleneck by placing the computation mechanisms inside or near memory. Several prior works propose PIM techniques to accelerate ML training; however, prior works either do not consider real-world PIM systems or evaluate algorithms that are not widely used in modern ML training. Our goal is to understand the capabilities and characteristics of popular distributed SGD algorithms on real-world PIM systems to accelerate data-intensive ML training workloads. To this end, we 1) implement several representative centralized parallel SGD algorithms, i.e., based on a central node responsible for synchronization and orchestration, on the real-world general-purpose UPMEM PIM system, 2) rigorously evaluate these algorithms for ML training on large-scale datasets in terms of performance, accuracy, and scalability, 3) compare to conventional CPU and GPU baselines, and 4) discuss implications for future PIM hardware. We highlight the need for a shift to an algorithm-hardware codesign to enable decentralized parallel SGD algorithms in real-world PIM systems, which significantly reduces the communication cost and improves scalability. Our results demonstrate three major findings: 1) The general-purpose UPMEM PIM system can be a viable alternative to state-of-the-art CPUs and GPUs for many memory-bound ML training workloads, especially when operations and datatypes are natively supported by PIM hardware, 2) it is important to carefully choose the optimization algorithms that best fit PIM, and 3) the UPMEM PIM system does not scale approximately linearly with the number of nodes for many data-intensive ML training workloads. We open source all our code to facilitate future research at https://github.com/CMU-SAFARI/PIM-Opt.

Nanopore sequencers generate raw electrical signals representing the contents of a biological sequence molecule passing through the nanopore. These signals can be analyzed directly, avoiding base-calling entirely. We observe that while existing proposals for raw signal analysis typically do well in all metrics for small genomes ( e.g. , viral genomes), they all perform poorly for large genomes ( e.g. , the human genome). Our goal is to analyze raw nanopore signals in an accurate, fast, and scalable manner. To this end, we propose RawAlign, the first work to integrate fine-grained signal alignment into the state-of-the-art raw signal mapper. To enable accurate, fast, and scalable mapping with alignment, RawAlign implements three algorithmic improvements and hardware acceleration via a vectorized implementation of fine-grained alignment. Together, these significantly reduce the overhead of typically computationally expensive fine-grained alignment. Our extensive evaluations on different use cases and various datasets show RawAlign provides 1) the most accurate mapping for large genomes and 2) and on-par performance compared to RawHash (between 0.80×-1.08×), while achieving better performance than UNCALLED and Sigmap by on average (geo. mean) 2.83× and 2.06×, respectively. Availability: https://github.com/CMU-SAFARI/RawAlign.