Ian Cutress’ Post

On July 16th, 10am PT, I'll be talking about PCIe 7.0 with Priyank Shukla from Synopsys Inc. We'll cover where we are the technology pipeline, how the insatiable demand for new connectivity standards is enabling high-speed SERDES, and how hardware designers are getting to grips with the technology! Watch live at https://lnkd.in/edkGE4S3 #Sponsored

Now official. Graphcore has joined SoftBank Group Corp.. No numbers, almost certainly an acquihire. Some/Most investors had already wrriten down their investments. The best exit we were all assuming for them.

So, it's official. They're in the early stages of closing down the Chemistry department at my alma mater. I penned a heartfelt letter to the VC of University of Hull and got this copy/paste boilerplate response. This is happening despite Hull being #4 (Guardian, overall rankings 2024) in the country for Chemistry. It's madness! This is where practical LCDs were invented, and formed the basis for so much of what the interaction with modern technology is all about. If any press/journo wants to story this, my DMs are open if needed. I have further details not mentioned in this response. I may have to create a video of my own.

Today's stream is sponsored by ASM! https://lnkd.in/enqm8_eZ  My Newsletter: http://www.more-moore.com Chips and Cheese: https://lnkd.in/eY2vRFWk

Going live in an hour (2pm ET) with George Cozma to cover the last month of high-octane chips, events, and what we're expecting over the next few months. NVIDIA, AMD, Intel Corporation, Qualcomm. It's going to be fun! https://lnkd.in/e68meDt3 Stream today is #sponsored by ASM!

Also quoted in El Pais' tech coverage (in Cinco Días) of Computex! :) https://lnkd.in/emSKGg6d

Modern timing is still heavily based on mechanically manufactured quartz crystals - cheap, but not fit for a new wave of AI compute. AI requires timing to be more power efficient, more accurate, and with added configurability to enable the future. In this post, we look at SiTime - a company enabling the future of timing through the use of small in-silicon structures called MEMS. We found one of SiTime's devices inside an NVIDIA networking switch at Computex only a couple of weeks ago. https://lnkd.in/etVMwT2S

Growth of Arm in the datacenter starts with availability, then adoption. With some hyperscalers rolling their own designs, everyone else has to look to the open market to see what's for sale. The only real option here is from Ampere, which recently announced a custom 3nm Arm 256-core CPU as part of the AmpereOne family. More cores, more memory channels, and a custom microarchitecture they might disclose later this year. It's a chiplet design too, showcasing where this market is going. https://lnkd.in/e6JMXCum

Does NVIDIA make chips, or does it make AI factories? It's a common thread of Jensen's keynotes, and it was probed in more detail by the press at their Computex Q&A session. Lots of questions in this interview on gaming, HBM supply, and Taiwanese R&D. https://lnkd.in/eCJZp46z