No abstract

One of the main challenges for modern processors is the data transfer between processor and memory. Such data movement implies high latency and high energy consumption. In this context, Near-Data Processing (NDP) proposals have started to gain acceptance as an accelerator device. Such proposals alleviate the memory bottleneck by moving instructions to data whereabouts. The first proposals date back to the 1990s, but it was only in the 2010s that we could observe an increase in papers addressing NDP. It occurred together with the appearance of 3D-stacked chips with logic and memory stacked layers. This survey presents a brief history of these accelerators, focusing on the applications domains migrated to near-data and the proposed architectures. We also introduce a new taxonomy to classify such architectural proposals according to their data distance.

No abstract

No abstract

Processor-in-Memory (PIM) architectures, such as the Hybrid Memory Cube (HMC), are emerging nowadays as a solution for processing large amount of data directly inside the memory. In this area, several researchers are proposing and evaluating new instructions and new PIM architectures. For such evaluations, trace-driven simulators, as the Simulator of Non-Uniform Cache Architectures (SiNUCA), are commonly used in order to model these new proposed systems. Such simulators provide fast prototyping of new architectures, while it requires the researcher to write simulation traces manually when evaluating new Instruction Set Architecture (ISA) proposals, which is an time consuming and error prone task. In this work, we propose a methodology for fast generation of simulation traces focused on HMC architecture, which consists on a high-level Intrinsics-HMC library and a modification inside the trace-generator tool from SiNUCA. Our proposal enables the researchers to write high level code in C/C++ languages using our library, which mimics the behavior of HMC instructions. These codes can be compiled and executed in traditional x86 architectures for verification. After ensure the code is correct and working, the user can use our modified version of SiNUCA-Tracer to translate HMC functions into HMC instructions know by the simulator, providing a convenient solution to generate traces and fast simulations of new PIM architectures. Results using the proposed technique applied on database application kernels show the correct translation and simulation of new HMC instructions using SiNUCA.