editor's blog
Subscribe Now

An Unambiguous Message for Multicore Architects

I had the good fun of co-moderating a panel with Multicore Association President Markus Levy at the Multicore Expo last week. The goal of the panel was to explore how software programmers who aren’t multicore architecture experts – and who don’t want to be – can write code that won’t have to be re-written for each multicore architecture variant that comes around. We were pushing our luck on the scheduling, since it was on the last day just after 4 PM. The exhibit booths were mostly down by then, and the chances were good that everyone would have gone home.

But a healthy crowd postponed their beer and sat in. And it was the most involved crowd I’ve seen at a panel: clearly the topic touched a nerve. Markus and I hardly had to do anything – the audience was right in there driving the conversation.

We tried hard to end on a positive note, since it was clear that there’s a lot of work to do before software programmers are relieved of the burdens of optimizing to the architecture in the way that single-core programmers currently are. Audience frustration with the current state of affairs was palpable.

But there was one comment that got more audience response than any other. When a member of the panel suggested that software programmers should be consulted during the process of designing the multicore architecture, there was a spontaneous round of applause. Clearly these guys felt like the hardware guys go off and do their architecture, and then the programmers have to work around it.

As to the positive note, well, the panelists all felt like much of this is a solvable problem, largely requiring tools, and that things will improve. Programmers may not be completely relieved of the need to consider concurrency, but they should be able to focus only on the concurrency inherent in their program, without having to worry about how that happens to match up with the computing architecture on which it will run.

Leave a Reply

featured blogs
Sep 25, 2020
[From the last episode: We looked at different ways of accessing a single bit in a memory, including the use of multiplexors.] Today we'€™re going to look more specifically at memory cells '€“ these things we'€™ve been calling bit cells. We mentioned that there are many...
Sep 25, 2020
Normally, in May, I'd have been off to Unterschleißheim, a suburb of Munich where historically we've held what used to be called CDNLive EMEA. We renamed this CadenceLIVE Europe and... [[ Click on the title to access the full blog on the Cadence Community site...
Sep 24, 2020
I just saw a video from 2012 in which Jeri Ellsworth is strolling around a Makerfaire flaunting her Commodore 64-based bass guitar....
Sep 24, 2020
Samtec works with system architects in the early stages of their design to create solutions for cable management which provide even distribution of thermal load. Using ultra-low skew twinax cable to route signals over the board is a key performance enabler as signal integrity...

Featured Video

AI SoC Chats: Memory Interface IP - DDR, LPDDR, HBM, GDDR

Sponsored by Synopsys

When building AI SoCs, how do you choose the optimal memory interface? Learn about the market trends and challenges for DDR, LPDDR, HBM, and GDDR, and how Synopsys DesignWare IP can help.

Click here for more information about DesignWare IP for Amazing AI

Featured Paper

An Introduction to Automotive LIDAR

Sponsored by Texas Instruments

This white paper is an introduction to industrial and automotive time-of-flight (ToF) light detection and ranging (LIDAR) solutions to serve next-generation autonomous systems.

Click here to download the whitepaper

Featured Chalk Talk

Introducing Google Coral

Sponsored by Mouser Electronics and Google

AI inference at the edge is exploding right now. Numerous designs that can’t use cloud processing for AI tasks need high-performance, low-power AI acceleration right in their embedded designs. Wouldn’t it be cool if those designs could have their own little Google TPU? In this episode of Chalk Talk, Amelia Dalton chats with James McKurkin of Google about the Google Coral edge TPU.

More information about Coral System on Module