Anastasi On the Death of Memory SRAM and Chiplets

[Daniel Nenni]

Modern CPUs, GPUs and SoCs have a major problem. SRAM memory scaling is dead. In this video I discuss new disruptive memory technology that may solve this problem.

Timestamps
:00:00 - Major Problem with Modern Chips
09:00 - Possible Solution
11:24 - New Memory Technology Explained

 

[MKWVentures]

She is correct, Chiplets are the answer. I have a presentation at FMS in 3 weeks on how Chiplets solve memory scaling issues, Cost issues, and new memory adoption issues. PCM obviously is not the solution, and there never will be a "universal memory" but we have cost comparisons of different options and impacts to foundry and niche memory players.

 

[milesgehm]

Closing in on universal memory for large data processing

Stanford researchers have developed a new phase-change memory that could help computers process large amounts of data faster and more efficiently.

news.stanford.edu

 

[MKWVentures]

Closing in on universal memory for large data processing

Stanford researchers have developed a new phase-change memory that could help computers process large amounts of data faster and more efficiently.

news.stanford.edu

There are new memory materials every year. They take about 10 years to reach chip form. and there are physical reasons why memory is not likely to be universal. We show that in our memory product lifecycle model. We also have updates on all new memories at FMS in 3 Weeks.

 

[Fred Chen]

Novel nanocomposite-superlattices for low energy and high stability nanoscale phase-change memory - Nature Communications

Data-centric applications benefit from dense, low-power memory. Here the authors use a combination of chalcogenide superlattices and nanocomposites to achieve low switching voltage (0.7 V) and fast speed (40 ns) in 40-nm-scale phase-change memory.

www.nature.com

Closing in on universal memory for large data processing

Stanford researchers have developed a new phase-change memory that could help computers process large amounts of data faster and more efficiently.

news.stanford.edu

It looks like RESET current scaling is still a challenge for PCM. https://www.nature.com/articles/s41467-023-42792-4

 

[Daniel Nenni]

She is correct, Chiplets are the answer. I have a presentation at FMS in 3 weeks on how Chiplets solve memory scaling issues, Cost issues, and new memory adoption issues. PCM obviously is not the solution, and there never will be a "universal memory" but we have cost comparisons of different options and impacts to foundry and niche memory players.

Can you share the details here afterwards? Sounds interesting.

 

[MKWVentures]

Can you share the details here afterwards? Sounds interesting.

I will post on my website and do a follow up blog for Semiwiki. I should have it around the 15th of August

 

[Paul2]

Modern CPUs, GPUs and SoCs have a major problem. SRAM memory scaling is dead. In this video I discuss new disruptive memory technology that may solve this problem.

Timestamps
:00:00 - Major Problem with Modern Chips
09:00 - Possible Solution
11:24 - New Memory Technology Explained

SRAM specific process