5 ms·
I remember my time at the OSDL when we got our first itanium servers in. I recall that even though we were "partnered" with Intel, it already felt like an orph
by jerrysievert 6y ago
I remember my time at the OSDL when we got our first itanium servers in. I recall that even though we were "partnered" with Intel, it already felt like an orphaned architecture. once the mainstream press started calling it "itanic", it seemed even more doomed.
I'm mostly sad to remember that the itanium was supposed to be a "replacement" for the DEC alpha, and Compaq selling the IP to Intel. very sad day for computing.
- spamizbad 6y agoIt's really a shame IA64 accelerated the deaths of Alpha and PA-RISC. Not sure they could have survived x86-64 or ARM64 but they still had legs when Itanium hit the scene.
- jerrysievert 6y agoI'm not entirely convinced that x86-64 would have come on the scene in its current form had itanium not been invested into so heavily by intel (and been such a failure). remember, intel seemed hesitant to extend the x86 architecture, and x86-64 came out of AMD.
- lizknope 6y agoLinux used to refer to the arch as amd64 and then Intel copied it a few years later. AMD licensed the EV6 bus from DEC and there was some discussion of having motherboards that could accept either an AMD K7 or DEC Alpha and just flash the BIOS. That might have made the Alpha last longer by having more low cost motherboards
- rstuart4133 6y agoIt's still referred to as amd64 in Debian.
- acomjean 6y agoI don't think anything could have survived the rise of X86. Most x86 chips were 32 bit at the time and you need Alpha/Sparc/PA-Risc, or some fancy custom chips to work in 64 bit. I wonder if the AMD/Intel competetion pushed x86 to get as fast as it got surpassing all those custom chips? HP was involved in Itanium and was discontinuing there own chip lines. We were talking about moving to it, for the "superdome" line of machines. I was working with PA-RISC at the time, and it seemed like this was the next chip. But it wasn't compatible with PA-RISC but had an x86 compatibility mode.. It was a weird chip, and required really good compilers. I thought it was long dead. It was interesting though.. https://news.ycombinator.com/item?id=14518758 https://news.ycombinator.com/item?id=14518758
- rodgerd 6y agoMy impression was that HP essentially convinced Intel to build next-gen PA-RISC for them; for HP it was mostly a success in that it kept HP Unix and VMS running.
- mhh__ 6y agoARMs rise seems to be almost miraculous considering that I struggle to find any contemporary discussion of it as a serious ISA from back in the day.
- klelatti 6y agoBack in the day to be considered 'serious' Arm would have needed to compete on performance and that meant competing with Intel which they (primarily Robin Saxby) worked out was a losing strategy - so they didn't. Then getting AArch64 onto the same CPUs as AArch32 meant in one bound they could cast off the legacy and start to compete on performance.
- mhh__ 6y agoI was more thinking along the lines of academia than market share - if you read a book like the dragon book or Muchnick's text, they clearly thought we'd all be on Alpha or some other RISC machine rather than ARM - even for teaching.
- klelatti 6y agoUnderstood. I suspect that Arm was a bit too off the mainstream for teaching (why talk about Thumb / Jazelle etc?). Plus probably better to teach for an architecture that had the performance cachet and where you could get some desktop machines.
- mrpippy 6y agoMIPS as well. SGI announced the Itanium switch and dialed back MIPS investment in 1998, but Itanium wasn't really usable until Itanium 2 in ~2003.
- soco 6y agoWhenever I hear Itanium comes up the old pang of anger that it killed Alpha. Sorry can't help it.