Memory ordering

Abstract: Emerging byte-addressable, non-volatile memory technologies offer performance within an order of magnitude of DRAM, prompting their inclusion in the processor memory subsystem. However, such load/store accessible Persistent Memory (PM) has implications on system design, both hardware and software. In this paper, we explore system software support to enable low-overhead PM access by new and legacy applications. To this end, we implement PMFS, a light-weight POSIX file system that exploits PM's byte-addressability to avoid overheads of block-oriented storage and enable direct PM access by applications (with memory-mapped I/O). PMFS exploits the processor's paging and memory ordering features for optimizations such as fine-grained logging (for consistency) and transparent large page support (for faster memory-mapped I/O). To provide strong consistency guarantees, PMFS requires only a simple hardware primitive that provides software enforceable guarantees of durability and ordering of stores to PM. Finally, PMFS uses the processor's existing features to protect PM from stray writes, thereby improving reliability.Using a hardware emulator, we evaluate PMFS's performance with several workloads over a range of PM performance characteristics. PMFS shows significant (up to an order of magnitude) gains over traditional file systems (such as ext4) on a RAMDISK-like PM block device, demonstrating the benefits of optimizing system software for PM.

Abstract: This paper makes the case that pin bandwidth will be a critical consideration for future microprocessors. We show that many of the techniques used to tolerate growing memory latencies do so at the expense of increased bandwidth requirements. Using a decomposition of execution time, we show that for modern processors that employ aggressive memory latency tolerance techniques, wasted cycles due to insufficient bandwidth generally exceed those due to raw memory latencies. Given the importance of maximizing memory bandwidth, we calculate effective pin bandwidth, then estimate optimal effective pin bandwidth. We measure these quantities by determining the amount by which both caches and minimal-traffic caches filter accesses to the lower levels of the memory hierarchy. We see that there is a gap that can exceed two orders of magnitude between the total memory traffic generated by caches and the minimal-traffic caches---implying that the potential exists to increase effective pin bandwidth substantially. We decompose this traffic gap into four factors, and show they contribute quite differently to traffic reduction for different benchmarks. We conclude that, in the short term, pin bandwidth limitations will make more complex on-chip caches cost-effective. For example, flexible caches may allow individual applications to choose from a range of caching policies. In the long term, we predict that off-chip accesses will be so expensive that all system memory will reside on one or more processor chips.

Abstract: This article presents a technique for the efficient compiler management of software-exposed heterogeneous memory. In many lower-end embedded chips, often used in microcontrollers and DSP processors, heterogeneous memory units such as scratch-pad SRAM, internal DRAM, external DRAM, and ROM are visible directly to the software, without automatic management by a hardware caching mechanism. Instead, the memory units are mapped to different portions of the address space. Caches are avoided due to their cost and power consumption, and because they make it difficult to guarantee real-time performance. For this important class of embedded chips, the allocation of data to different memory units to maximize performance is the responsibility of the software.Current practice typically leaves it to the programmer to partition the data among different memory units. We present a compiler strategy that automatically partitions the data among the memory units. We show that this strategy is optimal, relative to the profile run, among all static partitions for global and stack data. For the first time, our allocation scheme for stacks distributes the stack among multiple memory units. For global and stack data, the scheme is provably equal to or better than any other compiler scheme or set of programmer annotations. Results from our benchmarks show a 44.2p reduction in runtime from using our distributed stack strategy vs. using a unified stack, and a further 11.8p reduction in runtime from using a linear optimization strategy for allocation vs. a simpler greedy strategy; both in the case of the SRAM size being 20p of the total data size. For some programs, less than 5p of data in SRAM achieves a similar speedup.

Abstract: Non-blocking caches and prefetehing caches are two techniques for hiding memory latency by exploiting the overlap of processor computations with data accesses. A nonblocking cache allows execution to proceed concurrently with cache misses as long as dependency constraints are observed, thus exploiting post-miss operations, A prefetching cache generates prefetch requests to bring data in the cache before it is actually needed, thus allowing overlap with premiss computations. In this paper, we evaluate the effectiveness of these two hardware-based schemes. We propose a hybrid design based on the combination of these approaches. We also consider compiler-based optimization to enhance the effectiveness of non-blocking caches. Results from instruction level simulations on the SPEC benchmarks show that the hardware prefetching caches generally outperform nonblocking caches. Also, the relative effectiveness of nonblocklng caches is more adversely affected by an increase in memory latency than that of prefetching caches,, However, the performance of non-blocking caches can be improved substantially by compiler optimizations such as instruction scheduling and register renaming. The hybrid design cm be very effective in reducing the memory latency penalty for many applications.