Such a shame - IBM have some good tech but it's buried beneath a crap-ton of marketing bullshit.
I was a huge fan of AIX and POWER processors back in the day and still think that Linux could have benefitted from some of the approaches back then(device handling for example).
However these days the company only pushes a smokescreen of AI this, hybrid-cloud that, blockchain the other, constant confusing renames. _Still_ "shipping more MIPS this year than last", on mainframes that no-one gives two hoots about.
As a result, they have near zero credibility to the modern generation (the last 15 years) who obtain resilience through cheap software rather than expensive hardware and expect near-instant hyper-scalability.
Yes, although the numbers don't mean what you may think they mean:
The trick here is to have one chip treat the L2's caches of other chips as its L3 cache, and also it allows one unit (group of chips) to use the caches of the other units (drawers) as their L4 cache.
I was a huge fan of AIX and POWER processors back in the day and still think that Linux could have benefitted from some of the approaches back then(device handling for example).
However these days the company only pushes a smokescreen of AI this, hybrid-cloud that, blockchain the other, constant confusing renames. _Still_ "shipping more MIPS this year than last", on mainframes that no-one gives two hoots about.
As a result, they have near zero credibility to the modern generation (the last 15 years) who obtain resilience through cheap software rather than expensive hardware and expect near-instant hyper-scalability.