You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
That's the story I heard a the time.
I'm wondering why this is in the news now. (A) who cares what didn't happen? (Something didn't happen! News at 11!) (B) among those who care, who cares what didn't happen 20 years ago? (C) who was unaware of this?
Aviators have a saying: "Any landing you can walk away from is a good landing." McAfee, Mobileye, and now maybe Altera are all deals the company has unwound and been able to walk away from--not unscathed but not with a total loss, either. Among the company's other deals, it's hard to point any...
That MI325X is an answer to H200 and falls short of Blackwell should come as no surprise. AMD has been clear about this for some time. Is it better on inference workloads on H200? It's put up or shut up. Nvidia (and Intel) have been good about posting MLPerf results. AMD, not so much.
I'd argue...
More specifically, the new Cowos version using silicon bridges embedded in a poly/organic substrate is implicated, whereas the previous silicon interposer was more robust. It's all speculative.
