Hybrid Memory Cube Research Articles

Optically Disaggregated Data Centers With Minimal Remote Memory Latency: Technologies, Architectures, and Resource Allocation [Invited

Disaggregated rack-scale data centers have been proposed as the only promising avenue to break the barrier of the fixed CPU-to-memory proportionality caused by main-tray direct-attached conventional/traditional server-centric systems. However, memory disaggregation has stringent network requirements in terms of latency, energy efficiency, bandwidth, and bandwidth density. This paper identifies all the requirements and key performance indicators of a network to disaggregate IT resources while summarizing the progress and importance of optical interconnects. Crucially, it proposes a rack-and-cluster scale architecture, which supports the disaggregation of CPU, memory, storage, and/or accelerator blocks. Optical circuit switching forms the core of this architecture, whereas the end-points (IT resources) are equipped with on-chip programmable hybrid electrical packet/circuit switches. This architecture offers dynamically reconfigurable physical topology to form virtual ones, each embedded with a set of functions. It analyzes the latency overhead of disaggregated DDR4 (parallel) and the proposed hybrid memory cube (serial) memory elements on the conventional and the proposed architecture. A set of resource allocation algorithms are introduced to (1) optimally select disaggregated IT resources with the lowest possible latency, (2) pool them together by means of a virtual network interconnect, and (3) compose virtual disaggregated servers. Simulation findings show up to a 34% resource utilization increase over traditional data centers while highlighting the importance of the placement and locality among compute, memory, and storage resources. In particular, the network-aware locality-based resource allocation algorithm achieves as low as 15 ns, 95 ns, and 315 ns memory transaction round-trip latency on 63%, 22%, and 15% of the allocated virtual machines (VMs) accordingly while utilizing 100% of the CPU resources. Furthermore, a formulation to parameterize and evaluate the additional financial costs endured by disaggregation is reported. It is shown that the more diverse the VM requests are, the higher the net financial gain is. Finally, an experiment was carried out using silicon photonic midboard optics and an optical circuit switch, which demonstrates forward error correction free 10−12 bit error rate performance on up to five-tier scale-out networks.

HMC-MAC: Processing-in Memory Architecture for Multiply-Accumulate Operations with Hybrid Memory Cube

Many studies focus on implementing processing-in memory (PIM) on the logic die of the hybrid memory cube (HMC) architecture. The multiply-accumulate (MAC) operation is heavily used in digital signal processing (DSP) systems. In this paper, a novel PIM architecture called HMC-MAC that implements the MAC operation in the HMC is proposed. The vault controllers of the conventional HMC are working independently to maximize the parallelism, and HMC-MAC is based on the conventional HMC without modifying the architecture much. Therefore, a large number of MAC operations can be processed in parallel. In HMC-MAC, the MAC operation can be carried out simultaneously with as much as 128 KB data. The correctness on HMC-MAC is verified by simulations, and its performance is better than the conventional CPU-based MAC operation when the MAC operation is consecutively executed at least six times

CAIRO

Three-dimensional (3D)-stacking technology and the memory-wall problem have popularized processing-in-memory (PIM) concepts again, which offers the benefits of bandwidth and energy savings by offloading computations to functional units inside the memory. Several memory vendors have also started to integrate computation logics into the memory, such as Hybrid Memory Cube (HMC), the latest version of which supports up to 18 in-memory atomic instructions. Although industry prototypes have motivated studies for investigating efficient methods and architectures for PIM, researchers have not proposed a systematic way for identifying the benefits of instruction-level PIM offloading . As a result, compiler support for recognizing offloading candidates and utilizing instruction-level PIM offloading is unavailable. In this article, we analyze the advantages of instruction-level PIM offloading in the context of HMC-atomic instructions for graph-computing applications and propose CAIRO , a compiler-assisted technique and decision model for enabling instruction-level offloading of PIM without any burden on programmers. To develop CAIRO, we analyzed how instruction offloading enables performance gain in both CPU and GPU workloads. Our studies show that performance gain from bandwidth savings, the ratio of number of cache misses to total cache accesses, and the overhead of host atomic instructions are the key factors in selecting an offloading candidate. Based on our analytical models, we characterize the properties of beneficial and nonbeneficial candidates for offloading. We evaluate CAIRO with 27 multithreaded CPU and 36 GPU benchmarks. In our evaluation, CAIRO not only doubles the speedup for a set of PIM-beneficial workloads by exploiting HMC-atomic instructions but also prevents slowdown caused by incorrect offloading decisions for other workloads.

Heterogeneous HMC+DDRx Memory Management for Performance-Temperature Tradeoffs

Three-dimensional DRAMs (3D-DRAMs) are emerging as a promising solution to address the memory wall problem in computer systems. However, high fabrication cost per bit and thermal issues are the main reasons that prevent architects from using 3D-DRAM alone as the main memory building block. In this article, we address this issue by proposing a heterogeneous memory system that combines a double data rate (DDRx) DRAM with an emerging 3D hybrid memory cube (HMC) technology. Bandwidth and temperature management are the challenging issues for this heterogeneous memory architecture. To address these challenges, first we introduce a memory page allocation policy for the heterogeneous memory system to maximize performance. Then, using the proposed policy, we introduce a temperature-aware algorithm that dynamically distributes the requested bandwidth between HMC and DDRx DRAM to reduce the thermal hotspot while maintaining high performance. We take into account the impact of both core count and HMC channel count on performance while using the proposed policies. The results show that the proposed memory page allocation policy can utilize the memory bandwidth close to 99% of the ideal bandwidth utilization. Moreover, our temperate-aware bandwidth adaptation reduces the average steady-state temperature of the HMC hotspot across various workloads by 4.5 K while incurring 2.5% performance overhead.

Neurostream: Scalable and Energy Efficient Deep Learning with Smart Memory Cubes

High-performance computing systems are moving towards 2.5D and 3D memory hierarchies, based on High Bandwidth Memory (HBM) and Hybrid Memory Cube (HMC) to mitigate the main memory bottlenecks. This trend is also creating new opportunities to revisit near-memory computation. In this paper, we propose a flexible processor-in-memory (PIM) solution for scalable and energy-efficient execution of deep convolutional networks (ConvNets), one of the fastest-growing workloads for servers and high-end embedded systems. Our codesign approach consists of a network of Smart Memory Cubes (modular extensions to the standard HMC) each augmented with a many-core PIM platform called NeuroCluster. NeuroClusters have a modular design based on NeuroStream coprocessors (for Convolution-intensive computations) and general-purpose RISCV cores. In addition, a DRAM-friendly tiling mechanism and a scalable computation paradigm are presented to efficiently harness this computational capability with a very low programming effort. NeuroCluster occupies only 8% of the total logic-base (LoB) die area in a standard HMC and achieves an average performance of 240 GFLOPS for complete execution of full-featured state-of-the-art (SoA) ConvNets within a power budget of 2.5W. Overall 11 W is consumed in a single SMC device, with 22.5 GFLOPS/W energy-efficiency which is 3.5X better than the best GPU implementations in similar technologies. The minor increase in system-level power and the negligible area increase make our PIM system a cost-effective and energy efficient solution, easily scalable to 955 GFLOPS with a small network of just four SMCs.

HMC-Sim-2.0: A co-design infrastructure for exploring custom memory cube operations

Abstract The recent advent of stacked memory devices has led to a resurgence of research associated with the fundamental memory hierarchy and associated memory pipeline. The bandwidth advantages provided by stacked logic and DRAM devices have inspired research associated with eliminating the bandwidth bottlenecks associated with many applications in high performance computing. These augmented memory subsystems stand to change the landscape of high performance computing algorithm optimization. In addition to the two aforementioned focus areas, a third area of research is emerging to explore augmenting the stacked memory logic layer with additional operations. This first generation of Hybrid Memory Cube (HMC) devices provided rudimentary atomic memory operations. The Gen2 Hybrid Memory Cube devices provide more expressive atomic memory operations that include primitive integer arithmetic operations. Despite the inclusion of more expressive arithmetic operations, many users have expressed interest in more complex and potentially orthogonal custom memory cube , or CMC, operations in future revisions of the Hybrid Memory Cube specification. This work presents recent development associated with the HMC-Sim Hybrid Memory Cube simulation framework that provides users a powerful infrastructure to experiment and research augmented custom memory cube, or CMC, operations within the current Gen2 Hybrid Memory Cube device infrastructure. The goal of this approach is to provide computer architects the ability to experiment with augmentations to future memory devices in the scope of co-designing the future of scalable high performance computing instruments. We provide an overview of extending the original HMC-Sim simulation infrastructure to include support for CMC operations with requiring users to modify the core simulation code base. In addition, we provide a sample series of CMC operations that implement near-memory mutexes and demonstrate their efficacy using central locking and barrier synchronization algorithms traditionally found in parallel programming models and runtime libraries.

Improving 3D DRAM Fault Tolerance Through Weak Cell Aware Error Correction

Although the emerging 3D DRAM products can significantly improve the computing system performance, the relatively high cost is one of the most critical issues that prevent their wide real-life adoption. Intuitively, a strong memory fault tolerance can be leveraged to reduce the fabrication cost of DRAM dies, and the total cost will reduce if the fabrication cost saving can off-set the cost overhead of memory fault tolerance. Nevertheless, such a simple concept can be a practically viable option only for 3D DRAM because: (1) The stacked logic die can solely implement memory fault tolerance inside 3D DRAM chips, obviating any changes on the host CPUs and CPU-DRAM interfaces. (2) With the total ownership of both the logic die and DRAM dies inside 3D DRAM chips, DRAM manufacturers can fully exploit the potential to truly minimize the 3D DRAM bit cost. Following this intuition, we developed a 3D DRAM fault tolerance design strategy. It can achieve a very strong tolerance to weak DRAM cells at very small redundancy and latency overhead. The key is to cohesively leverage the detectability of weak cells and runtime configurability of error correction code (ECC) decoding. In addition, this design strategy can gracefully embrace the inaccuracy of weak cell detection (e.g., weak cell miss-detection and false-detection). We carried out thorough mathematical analysis, and the results show that, under the redundancy overhead of 1:8 (same as today’s ECC DIMM), this design strategy can tolerate the weak cell rate of as high as $10^{-4}$ and 6 $\times 10^{-5}$ if 100 and 90 percent of all the weak cells are known in prior. Using Micron’s hybrid memory cube (HMC) 3D DRAM chips as the test vehicle, we evaluated the implementation cost and the results show that it only consumes less than 0.4 mm2 (45 nm node) on the logic die. Using CPU and DRAM simulators, we further carried out simulations over a variety of computing benchmarks and the results show that this design solution only incurs less than 2 percent performance degradation on average.

Two-level main memory co-design: Multi-threaded algorithmic primitives, analysis, and simulation

A challenge in computer architecture is that processors often cannot be fed data from DRAM as fast as CPUs can consume it. Therefore, many applications are memory-bandwidth bound. With this motivation and the realization that traditional architectures (with all DRAM reachable only via bus) are insufficient to feed groups of modern processing units, vendors have introduced a variety of non-DDR 3D memory technologies (Hybrid Memory Cube (HMC),Wide I/O 2, High Bandwidth Memory (HBM)). These offer higher bandwidth and lower power by stacking DRAM chips on the processor or nearby on a silicon interposer. We will call these solutions “near-memory,” and if user-addressable, “scratchpad.” High-performance systems on the market now offer two levels of main memory: near-memory on package and traditional DRAM further away. In the near term we expect the latencies near-memory and DRAM to be similar. Thus, it is natural to think of near-memory as another module on the DRAM level of the memory hierarchy. Vendors are expected to offer modes in which the near memory is used as cache, but we believe that this will be inefficient.In this paper, we explore the design space for a user-controlled multi-level main memory. Our work identifies situations in which rewriting application kernels can provide significant performance gains when using near-memory. We present algorithms designed for two-level main memory, using divide-and-conquer to partition computations and streaming to exploit data locality. We consider algorithms for the fundamental application of sorting and for the data analysis kernel k-means. Our algorithms asymptotically reduce memory-block transfers under certain architectural parameter settings. We use and extend Sandia National Laboratories’ SST simulation capability to demonstrate the relationship between increased bandwidth and improved algorithmic performance. Memory access counts from simulations corroborate predicted performance improvements for our sorting algorithm. In contrast, the k-means algorithm is generally CPU bound and does not improve when using near-memory except under extreme conditions. These conditions require large instances that rule out SST simulation, but we demonstrate improvements by running on a customized machine with high and low bandwidth memory. These case studies in co-design serve as positive and cautionary templates, respectively, for the major task of optimizing the computational kernels of many fundamental applications for two-level main memory systems.

CasHMC: A Cycle-Accurate Simulator for Hybrid Memory Cube

3D-stacked DRAM has been actively studied to overcome the limits of conventional DRAM. The Hybrid Memory Cube (HMC) is a type of 3D-stacked DRAM that has drawn great attention because of its usability for server systems and processing-in-memory (PIM) architecture. Since HMC is not directly stacked on the processor die where the central processing units (CPUs) and graphic processing units (GPUs) are integrated, HMC has to be linked to other processor components through high speed serial links. Therefore, the communication bandwidth and latency should be carefully estimated to evaluate the performance of HMC. However, most existing HMC simulators employ only simple HMC modeling. In this paper, we propose a cycle-accurate simulator for hybrid memory cube called CasHMC. It provides a cycle-by-cycle simulation of every module in an HMC and generates analysis results including a bandwidth graph and statistical data. Furthermore, CasHMC is implemented in C++ as a single wrapped object that includes an HMC controller, communication links, and HMC memory. Instantiating this single wrapped object facilitates simultaneous simulation in parallel with other simulators that generate memory access patterns such as a processor simulator or a memory trace generator.

DRAMSpec: A High-Level DRAM Timing, Power and Area Exploration Tool

In systems ranging from mobile devices to servers, DRAM has a big impact on performance and contributes a significant part of the total consumed power. The performance and power of the system depends on the architecture of the DRAM chip, the design of the memory controller and the access patterns received by the memory controller. Thus, evaluating the impact of DRAM design decisions requires a holistic approach that includes an appropriate model of the DRAM bank, a realistic controller and DRAM power model, and a representative workload, which requires a full system simulator running a complete software stack. In this paper, we introduce DRAMSpec, a high-level DRAM bank/chip modeling tool. Our contribution is to move the DRAM modeling abstraction level from the circuit level to the DRAM bank and by the integration in full system simulators we allow system or processor designers (non-DRAM experts) to tune future DRAM architectures for their target applications. We demonstrate the merits of DRAMSpec by exploring the influence of DRAM row-buffer (page) size and the number of banks on performance and power of a server application (memcached). Our new DRAM design offers a 16% DRAM performance improvement and 13% DRAM energy saving compared to standard comodity DDR3 devices. Additionally, we demonstrate how our tool is able to aid in evaluating novel DRAM architectures, such as the Hybrid Memory Cube (HMC), for which no DRAM datasheets are available. Finally, we highlight the DRAM technology scaling for a specific HMC architecture and we quantify the impact on latency and power.

Research on Optical Transmitter and Receiver Module Used for High-Speed Interconnection between CPU and Memory

ABSTRACTWith the development of the multicore processor, the bandwidth and capacity of the memory, rather than the memory area, are the key factors in server performance. At present, however, the new architectures, such as fully buffered DIMM (FBDIMM), hybrid memory cube (HMC), and high bandwidth memory (HBM), cannot be commercially applied in the server. Therefore, a new architecture for the server is proposed. CPU and memory are separated onto different boards, and optical interconnection is used for the communication between them. Each optical module corresponds to each dual inline memory module (DIMM) with 64 channels. Compared to the previous technology, not only can the architecture realize high-capacity and wide-bandwidth memory, it also can reduce power consumption and cost, and be compatible with the existing dynamic random access memory (DRAM). In this article, the proposed module with system-in-package (SiP) integration is demonstrated. In the optical module, the silicon photonic chip is included, which is a promising technology to be applied in the next-generation data exchanging centers. And due to the bandwidth–distance performance of the optical interconnection, SerDes chips are introduced to convert the 64-bit data at 800 Mbps from/to 4-channel data at 12.8 Gbps after/before they are transmitted though optical fiber. All the devices are packaged on cheap organic substrates. To ensure the performance of the whole system, several optimization efforts have been performed on the two modules. High-speed interconnection traces have been designed and simulated with electromagnetic simulation software. Steady-state thermal characteristics of the transceiver module have been evaluated by ANSYS APLD based on finite-element methodology (FEM). Heat sinks are placed at the hotspot area to ensure the reliability of all working chips. Finally, this transceiver system based on silicon photonics is measured, and the eye diagrams of data and clock signals are verified.

(Invited) Semiconductor Packaging Technology to Progress the IoT Era

In the world economy there is a wave of the booming economy and the recession, the wave of the booming economy is going by the industry of technology innovation. Until now, the invention of the steam engine, transportation of development, has been repeated waves of good / recession of the economy by the invention of the electrical / communication. Now, the wave of major technological innovation has been visited by the progress of the Internet and the computer. It is a technique called IoT (Internet of Things) is connected and all of the places and things in the Internet, to accumulate an enormous amount of information, a wave of AI (Artificial Inteligence) is going to processing. Server for IoT era is referred to as a sensor and the edge can be determined that data to collect the information, also AI server to handle Big Data is needed. Such sensors or servers, it is important to semiconductor development for processing data in high-density memory and fast. On the other hand due to the limit of the semiconductor of the fine or photolithography, it is no longer the case is Moore's Law. As a result, technology of packaging the device in a high density is important. DRAM is, TSV (Through Via Silicon) HBM using (High Band Width Memory) and HMC (Hybrid Memory Cube) has been mass-produced is already a memory. System connecting the GPU (Graphic Processor Unit) using the Si interposer called 2.5D is still being produced. However, TSV Si Interposer called these 3D is in the advance in the future IoT because of the high cost, it is difficult to be used in general to modularize the large number of sensors. Development has been the development of new packaging technology, such as that in FOWLP. I want to discuss the outlook and its Overview.

Three-Dimensional Dynamic Random Access Memories Using Through-Silicon-Vias

This paper describes orthogonal scaling of dynamic-random-access-memories (DRAMs) using through-silicon-vias (TSVs). We review 3D DRAMs including DDR3, wide I/O mobile DRAM (WIDE I/O), and more recently, the hybrid-memory cube (HMC) and high-bandwidth memory (HBM) targeted for high-performance computing systems. We then cover embedded 3D DRAM for high-performance cache memories, reviewing an early cache prototype employing face-to-face 3D stacking which confirmed negligible performance and retention degradation using 32 nm server and ASIC embedded DRAM macros. A second cache system prototype based on POWER7 was developed to confirm feasibility of stacking $\mu {\rm P}$ and high density cache memory, with $> 2~{\rm GHz}$ operation. For test and assembly, a micro-electro-mechanical-system (MEMS) probe-card with an integrated active silicon chip, realized a 50 $\mu{\rm m}$ pitch micro-probing at-speed-active-test for known-good-die (KGD) sorting. Finally, oxide wafer bonding with Cu TSV demonstrated wafer-scale 3D integration, with TSV diameters as small as 1 $\mu{\rm m}$ . The paper concludes with comments on the challenges for future 3D DRAMs.

Logic-Base Interconnect Design for Near Memory Computing in the Smart Memory Cube

Hybrid memory cube (HMC) has promised to improve bandwidth, power consumption, and density for the next-generation main memory systems. In addition, 3-D integration gives a second shot for revisiting near memory computation to fill the gap between processors and memories. In this paper, we study the required infrastructure inside the HMC to support near memory computation in a modular and flexible fashion. We propose a fully backward compatible extension to the standard HMC called the smart memory cube, and design a high bandwidth, low latency, and Advanced eXtensible Interface-4.0 compatible logic base (LoB) interconnect to serve the huge bandwidth demand by the HMCs serial links, and to provide extra bandwidth to a generic processor-in-memory (PIM) device embedded in the LoB. This interconnect features a novel address scrambling mechanism for the reduction in the vault/bank conflicts and robust operation even in the presence of pathological traffic patterns. Our cycle accurate simulation results demonstrate that this interconnect can easily meet the demands of the latest HMC specifications (up to 205 GB/s read bandwidth with 4 serial links and 32 memory vaults for injected random traffic). It further shown that the default addressing scheme of the HMC (low interleaving) is not reliable enough and operates poorly in the presence of specific traffic patterns from real applications. This is while the proposed scrambling mechanism operates robustly even in those cases. The interference between the PIM traffic and the main links is shown to be negligible when the number of PIM ports is limited to 2, requesting up to 64 GB/s without pushing the system into saturation. Finally, logic synthesis with Synopsys Design Compiler confirms that our interconnect is implementable and effective in terms of power, area, and timing (power consumption less than 5 mW up to 1 GHz and area less than 0.4 mm2).

Low-Power Hybrid Memory Cubes With Link Power Management and Two-Level Prefetching

The hybrid memory cube (HMC) is a 3-D-stacked DRAM architecture designed for substantially improved memory bandwidth. In particular, its I/O interface achieves up to 320 GB/s of external bandwidth through high-speed serial links. However, it comes at the cost of large static power of off-chip links, which dominates total power consumption of HMCs. In this paper, we propose an adaptive mechanism to partially disable off-chip links of HMCs to reduce the energy consumption of the off-chip links. In order to determine the number of the links to be disabled upon application loads, we develop a simple hardware module called link delay monitor to simulate all different link configurations at the same time and find the largest number of the links to be disabled while satisfying the given performance constraint. We also present two-level prefetching with in-HMC prefetch buffers to further improve the efficiency of our link power management scheme in the presence of prefetching. Evaluations show that our scheme reduces the energy consumption of HMCs by 52% on average with a negligible performance degradation.

Design challenges in Interposer based 3-D Memory Logic Interface

Further improvements in system performance are often limited by the achievable bandwidth between processor and memory. In this paper we look at interposer-based and stacked solutions to integrate processor and 3D memory into a high performance system. The comparison is made for different technological decisions, design problems faced for choosing a certain 3D memory type from Wide IO/1–2, High bandwidth memory (HBM) and Hybrid Memory Cube (HMC). Logic die size, metal layers and material of interposer affected by routing requirements of memory systems are discussed.

Single-chip photonic transceiver based on bulk-silicon, as a chip-level photonic I/O platform for optical interconnects.

When silicon photonic integrated circuits (PICs), defined for transmitting and receiving optical data, are successfully monolithic-integrated into major silicon electronic chips as chip-level optical I/Os (inputs/outputs), it will bring innovative changes in data computing and communications. Here, we propose new photonic integration scheme, a single-chip optical transceiver based on a monolithic-integrated vertical photonic I/O device set including light source on bulk-silicon. This scheme can solve the major issues which impede practical implementation of silicon-based chip-level optical interconnects. We demonstrated a prototype of a single-chip photonic transceiver with monolithic-integrated vertical-illumination type Ge-on-Si photodetectors and VCSELs-on-Si on the same bulk-silicon substrate operating up to 50 Gb/s and 20 Gb/s, respectively. The prototype realized 20 Gb/s low-power chip-level optical interconnects for λ ~ 850 nm between fabricated chips. This approach can have a significant impact on practical electronic-photonic integration in high performance computers (HPC), cpu-memory interface, hybrid memory cube, and LAN, SAN, data center and network applications.

Active Memory Cube: A processing-in-memory architecture for exascale systems

Many studies point to the difficulty of scaling existing computer architectures to meet the needs of an exascale system (i.e., capable of executing $10^{18} $ floating-point operations per second), consuming no more than 20 MW in power, by around the year 2020. This paper outlines a new architecture, the Active Memory Cube, which reduces the energy of computation significantly by performing computation in the memory module, rather than moving data through large memory hierarchies to the processor core. The architecture leverages a commercially demonstrated 3D memory stack called the Hybrid Memory Cube, placing sophisticated computational elements on the logic layer below its stack of dynamic random-access memory (DRAM) dies. The paper also describes an Active Memory Cube tuned to the requirements of a scientific exascale system. The computational elements have a vector architecture and are capable of performing a comprehensive set of floating-point and integer instructions, predicated operations, and gather-scatter accesses across memory in the Cube. The paper outlines the software infrastructure used to develop applications and to evaluate the architecture, and describes results of experiments on application kernels, along with performance and power projections.

Cu electroplating chemistry and process enabling rapid TSV filling with long bath life

Through-silicon via (TSV) filling with electroplated Cu with short plating times and long bath life remains a challenge within the semiconductor and electronic packaging industries. TSVs are the primary enabler for advanced 3D architectures such as hybrid memory cubes, wide IO DDR memory, as well as chip interconnections through interposers. The TSV needs to be completely filled with void-free Cu for high reliability. Organic additives are used to achieve a bottom-up, superconformal filling profile, required for void-free filling in high aspect ratio vias. The development and selection of the proper organic additives, optimized plating process, and implementation of advanced bath control techniques enable short TSV electroplating times, while also allowing for robust operation and high bath stability. Dow's INTERLINK™ 9200 Cu TSV plating bath with the Applied Materials TSPlus plating system allows 10×100 μm vias to be filled in less than 1 hour and 5×50 μm vias filled in under 20 minutes while having a bath life greater than 40 A hr L-1 in continuous operation. This combination also exhibits low overburden for stress management, CMP cost reductions, and high reliability.

HMC-SIM: A Simulation Framework for Hybrid Memory Cube Devices

The recent advent of stacked die memory and logic technologies has lead to a resurgence in research associated with fundamental architectural techniques. Many architecture research projects begin with ample simulation of the target theoretical functions and approach. However, the logical and physical nature three-dimensional stacked devices, such as the Hybrid Memory Cube (HMC) specification, fundamentally do not align with traditional memory simulation techniques. As such, there currently exists a chasm in the capabilities of modern architectural simulation frameworks. This work introduces a new simulation framework developed specifically for the Hybrid Memory Cube specification. We present a set of novel techniques implemented on an associated development framework that provide an infrastructure to flexibly simulate one or more Hybrid Memory Cube stacked die memory devices attached to an arbitrary core processor. The goal of this development infrastructure is to provide architectural simulation frameworks the ability to begin migrating current banked DRAM memory models to stacked HMC-based designs without a reduction in simulation fidelity or functionality. In addition to the core architecture, this work also presents a series of memory workload test results that represent unit stride and random access memory patterns using the HMC-Sim infrastructure. These evaluations confirm that HMC-Sim can provide insightful guidance in designing and developing highly efficient systems, algorithms, and applications, considering the next-generation three-dimensional stacked memory devices.