Ann: 2022 AGM - Chair Address and CEO Presentation, page-5

I would say the question of "speed" is too ambigious, though that's not to say it's not a legimtiate question.

I know with flash for example, there's blocks which consist of multiple pages, and typically to make a change, you have to write out an entire page, which means you need to read the entire page first, otherwise you'd lose all the other data in the page. So in that case, modification takes longer than writing fresh data, so if you asking how fast flash is, even then it depends on exactly what operation you're performing and the state of the memory. Furthermore, that behaviour isn't necessarily fundamental to how flash has to operate, but is likely a design compromise to balance performance in different scenarios and make the most of flash memories more fundamental characteristic. So it's very hard to put a number on speed in a general sense - you need to have specific real world scenario's to make any kind of meaningful comparison.

I would say it's perhaps too early to know how "fast" Weebit's ReRAM is. Weebit's memory structure and the controllers used to access it will be continualy improved, so even if the fundamental bit-level speed is now as fast as it'll ever be, the overall speed in real-world scenarios may increase drastically over the next few years as it's optimised for the real world workloads it's used for.

Any comparison you make now has to focus on very low-level operations, like the latency to read and write a single bit, and I beleive we already have that information, it's just hard to know what that will look in a real world scenario.