@IanCutress@intel@SambaNovaAI Yeah, that bit was quite confusing. It felt like Tan was building up to something important, but it didn't even make it into Intel's post-keynote press release.
For the night owls out there, we have one more Computex keynote today. Intel CEO Lip-Bu Tan will be taking the stage to discuss the next era of AI, spanning the edge, clients, and DC. Come join @ServeTheHome for our live blog coverage of the keynote!
https://t.co/gHasGnbUUt
Day 2 of Computex is upon us. Come join @ServeTheHome at 7:30pm PT/10:30pm ET for our live blog coverage of the first keynote of the day, where Marvell will be outlining how the future of data center AI will depend on connectivity.
https://t.co/7PDfdThyPb
For our second Computex keynote of the day (night?) at @ServeTheHome we have Qualcomm. CEO Cristiano Amon will be presenting a keynote titled "The Year of Agents." Come join us for our full live blog coverage, starting at 11pm PT/2am ET/2pm CST/06:00 UTC
https://t.co/toihLXGwaH
@aschilling Technically AMD isn't even promising Zen 6 for AM5 in their Computex announcement. We all know it's happening regardless, but it's notable (and perhaps a bit frustrating) that AMD's support commitment is so vague
Computex season is upon us once more. Come join @ServeTheHome at 8pm PT/11pm ET/11am CST/03:00 UTC for our live blog coverage of the first keynote of the show, NVIDIA's, where the company is promising "a new era of PC." It should prove x-citing!
https://t.co/46Ebdb400T
On the one hand, I am envious that Patrick was able to work on a 8-way Blackwell Ultra server. On the other hand, I am very glad that I didn't have to lift or listen to said Blackwell Ultra server!
Kudos to one of our readers for catching this. Someone at Silicon Motion had some fun with their slides for the new SM2524XT SSD controller. "Psychic Spiritual Decoding" error correction, anyone?๐
(It should be "Group Page RAID" or similar)
@DanMatte I am not a huge fan of iOS 26. But it's decent. More importantly, Apple doesn't distribute versions of iOS 18 with the latest security updates for 26-capable devices, so it's not really a choice.=\
AMD has released a PCIe version of its Instinct MI350 accelerator, the MI350P. Their first Instinct PCIe card in over 4 years, it's neat enough. But underneath the hood it has an extra surprise: a brand-new chip with a 4 XCD + 1 IOD configuration.
https://t.co/BOWpOHArVF
After nearly a decade, SPEC has released a new version of its industry-defining CPU benchmark, SPEC CPU 2026. We take a look at what is new, as well as the new benchmark performance on a modern AMD, Intel, and NVIDIA CPU trio https://t.co/H7vy0ZuRui
We've been working on this one for a while, so I'm thrilled it's finally out. With "just" 200GbE, the scalability of a GB10 cluster varies. But it's a fun way to experiment with a Blackwell cluster for relatively cheap.
With NVIDIA's incorporation of Groq LPUs into the Vera Rubin rackscale platform, it marks a major shift in the future of inference hardware in the NVIDIA ecosystem. There's a lot of ground to cover, so please check out my full write-up over at STH:
https://t.co/sv4G3XADey
NVIDIA's GTC 2026 is upon us! Come join @ServeTheHome for our live blog coverage of the opening keynote and the many new AI technologies to be announced. The keynote kicks off in just under 2 hours, at 11am PT/18:00, so be sure to check back then.
https://t.co/elwcYDJK7R