Displaying posts in the categoryComputer Architecture
Computer Architecture articles
Memory
In-memory compute (IMC) does the arithmetic inside the memory array instead of moving data out to a processor. For AI inference, the weights stay in place, the input activations drive the wordlines, and each bitline adds up the products, so the multiply-accumulate happens where the data lives. It matters because moving a number costs far more energy than computing with it: in published 45 nm figures, a 32-bit DRAM read costs about 640 pJ against…
by
Murugavel Ganesan
August 27, 2026
Verilog
RTL design and micro architecture interviews test three things: whether you can describe hardware correctly in Verilog or SystemVerilog, whether you understand timing, clocking and reset well enough to build something that works on silicon, and whether you can reason about trade-offs in performance, power and area. The 50 questions below cover all three, from counters and FIFOs to pipelines, caches and clock domain crossing, and each one now com…
by
Murugavel Ganesan
January 03, 2025
Video
Virtual memory is a memory management technique in which every program sees its own private, contiguous address space, while the operating system and the processor's memory management unit (MMU) map those virtual addresses onto physical RAM in fixed-size blocks called pages. It lets many programs run at once without overwriting each other, lets a program use more memory than is physically free by keeping some pages on disk, and avoids the fr…
by
Murugavel Ganesan
January 09, 2024
Video
RISC-V is an open Instruction Set Architecture (ISA) based on Reduced Instruction Set Computing (RISC) principles, which favour a small, simple and regular instruction set. Anyone can implement it without paying an ISA licence fee, and it is modular and extensible, so designers can pick only the extensions they need and add their own custom instructions. That is the main difference from x86, which is controlled by Intel and AMD, and from Arm, wh…
by
Murugavel Ganesan
December 24, 2023
GPUs and Accelerators
FPGA based customised acceleration means moving the hot loop of an application out of software and into a hardware circuit built for that one job on a field-programmable gate array. Because the circuit is shaped around the algorithm, with its own parallel processing elements, on-chip memories and number formats, it can beat a general-purpose processor on throughput, latency and energy per operation for the right workloads. The trade-off is engin…
by
Murugavel Ganesan
December 12, 2023
Processors
In January 2021 Glenn Hinton, a former Intel Senior Fellow best known as a lead architect of the Nehalem CPU core, announced that he was coming out of about three years of retirement to rejoin Intel. He said on LinkedIn that Pat Gelsinger's return as CEO was a main reason, and that he would work on a new high-performance CPU project. Intel's 2021 proxy statement later listed him as rejoining as a Senior Fellow. This post was first writte…
by
Murugavel Ganesan
January 21, 2021
News
Intel dominates the business of PC processors. But as consumers shift increasingly to tablets and smartphones, the company is trying to quickly move to serve those devices. On a visit to Bangalore last week, Justin Rattner, CTO of Intel, talked exclusively to TOI on a range of issues, from the nature of work its Bangalore labs are being asked to do, to some really futuristic, almost sci-fi, stuff. More at this link from Times of India!
by
Murugavel Ganesan
December 14, 2010
SoC Design
There are plenty of really good, proven processor cores on the market today. But if you have more than simple control tasks, perhaps you've considered using a processor that you can customize. This paper discusses 10 good reasons why you should consider customizing your core in your next SoC design.
by
Murugavel Ganesan
December 02, 2010
Tutorials
A cluster computer is a set of CPU nodes that are used to solve any problem over a network. The way in which this cooperation is accomplished among the computers to solve a problem is called Cluster Computing. In Grid computing, the idea is very similar to Cluster Computing however they are used for solving large problems. Clusters are usually homogeneous and Grids are heterogeneous. Homogeneous is where all CPU nodes have the same Hardware conf…
Processors
A paper (PDF) to be presented later this month at the IEEE International Parallel and Distributed Processing Symposium in Atlanta describes a new approach to memory management that allows software applications to run up to 20% faster on multicore processors. Yan Solihin, associate professor of electrical and computer engineering at NCSU and co-author of the paper, says that using the technique is just a matter of linking to a library in…
Processors
A CPU has a memory unit with 32-bit instructions and a register file with 32 registers. The instruction set consists of 95 different operations. The instruction consists of an opcode, a two register file addresses, and an immediate operand part. Each instruction is stored in one word of memory. (a) How many bits are needed for the opcode part of the instruction? (b) How many bits are needed to indicate the registers? (c) How many bits remain…
Processors
The author ( Mark Smotherman, Associate Professor, School of Computing, Clemson University ) makes an effort to list the key computer architects mainly to recognize their work . The listing comprises of instruction set architecture (ISA) and its architect(s), followed by implementations of that ISA and the associated microarchitect(s)/designer(s). The processors that are listing have been available for sale commercially, and in most instances, h…
GPUs and Accelerators
OpenFPGA is an emerging effort to foster and accelerate the adoption and incorporation of reconfigurable computing based computing solutions in high-performance computing and enterprise application environments. OpenFPGA will foster shared and open efforts to address challenges of portability, interoperability and intra-application communication for FPGA and reconfigurable applications in high-performance and enterprise computing environment…
by
Aarthi Chinappa
April 21, 2009
Processors
A list of the Top 500 performing supercomputers is available at: Top 500 . Since 1993, the list has been updated twice a year to indicate the best performance on the Linpack benchmark. Except for one machine which uses Microsoft Windows, all of the top 500 machines employ either the Linux or Unix operating system.
by
Aarthi Chinappa
April 21, 2009
SoC Design
Multiple microprocessor cores which communicate via an on-chip network are increasingly found in integrated circuit designs. These systems offer unprecedented levels of performance, while still remaining highly programmable and consuming modest amounts of power. In addition to high-end CPUs, they are highly suitable for high-performance embedded applications including multimedia, digital signal processing and networking. Dr. Steve G…
by
Aarthi Chinappa
April 21, 2009
Tutorials
Semiconductor memories are characterized according to the following properties: Serial or random access, Volatile or nonvolatile, Static or dynamic, Destructive or nondestructive readout. Serial access memories are those in which data are accessed in a fixed, predetermined sequence. Magnetic tape units are an example of serial access. To read a record it is necessary to read the entire tape up to the point where the desired data exists. By way o…
by
The Editor's Desk
January 10, 2009
Tutorials
Memories are pervasive in digital computing. Consider, the personal computer which has a main memory, video memory, translation ROMs, shadow ROMs, scratchpad memory, hard disk, floppy disk, CDROM, and various other kinds of storage distributed throughout. In addition, the die that contains the microprocessor also contains one or more levels of cache. A typical PC is depicted in the block diagram below. It is basically a memory hierarchy connecte…
by
The Editor's Desk
January 10, 2009
Low Power Design
This classic VLSI and computer architecture interview question asks whether three marketing metrics, MIPS/mm 2 , MIPS/Watt and Watts/cm 3 , are sensible for customers choosing a system. The short answer: MIPS/Watt is reasonable (bigger is better, most relevant to cell phones), Watts/cm 3 is reasonable (smaller is better, most relevant to compute servers), and MIPS/mm 2 is mainly a designer's metric rather than a customer's one. The int…
by
Murugavel Ganesan
September 17, 2008