So /g/, I was just curious and decided to look this up, 16 years ago in May, AMD stocks were $92, now 16 years later, it's $2. Someone tell me, what the fuck did AMD do wrong?
How can a company put themselves through so much pain and incompetence, they do so bad on the stock market?
I never actually did research on this. So the damage was dealt quite hard as a matter of fact.. It was settled in '09. Intel paying AMD $1.25 billion. My curiosity got the better of me, we are 7 years further and AMD still suffered this much, almost unable to get their stock prices up again?
What about their GPU line? They compete with nVidia on that side, while with Intel on the CPU line. Without a source of money, surely they've had investors over the years after '09?
AMD was never a very expensive stock to begin with, before then they were like $15/share. I think a large amount of their value at that time was hype which never really materialized into anything after the Athlon 64. Intel was doing some shady things during the early 2000s which probably hindered their growth and then Intel released the Core2 series and ended up on top again and that put an end to AMD's rule. AMD actually went up again in 2006 though after they announced the acquisition of ATI.
>1000x tesselated roadblock hidden in already dense jungle
>TDP vs Wattage vs average consumption vs peak consumption
I'm sure AMD is no saint either, but it's clear when they have been outscummed.
>x86_64 would have never existed
>Itanic flops hard
>Intel forced to add 64 bit extensions to x86
>Takes a little longer, maybe you'd see the first 64 bit x86 CPUs in 2005
>'muh 3.2 GB RAM limit'
>what is PAE
Face it, Itanium was a piece of shit. Poor performance and lack of backward compatibility with x86_32 software/OSs was a fucking death sentence. All the other CPU architectures out there were even worse.
If making x86_64 was so god damn simple why didn't jewtel do it?
>All the other CPU architectures out there were even worse.
>talking shit about Alpha, MIPS, POWER and SPARC
>and lack of backward compatibility with x86_32 software/OSs was a fucking death sentence
does it even fucking matter if the damn thing was aimed at enterprises initially
also, itanium did have x86 emulation but it was too slow. it was there though
>also, itanium did have x86 emulation but it was too slow. it was there though
ARM also can emulate x86, but that shit didn't work out either. It was a completely useless feature like it is on ARM.
Itanium could have worked if it had a community to build it up, Intel and HP kept it limited to enterprise and the enterprise already had their own software(written for x86) so they didn't need it. Running x86 software through emulation is not a real solution the performance penalty meant it was fucking worthless. What company in their right mind would buy itanium when they're just going to be running all of their software on an emulator.
> Intel and HP kept it limited to enterprise and the enterprise already had their own software(written for x86) so they didn't need it.
>enterprise already had their own software(written for x86)
Hell fucking no. Back then it'd be on a real server, something with MIPS/POWER/SPARC on it.
For any serious business x86 was a total fucking joke until Opteron came out.
Itanium was a server play for Intel. It was a way to shed the history of IA32 and try a brand new architecture, a design called EPIC, for Explicitly Parallel Instruction Computing. Some of the early design inspiration was based on HPs PA-RISC architecture and they worked with HP in overall design. Intel wanted to copy what they did with IA32, have a common chip for all big servers and leverage massive economies of scale. Since you're asking what Itanium is, they obviously didn't do very well on the scale part :). Its nickname was Itanic, obviously not a name that indicates massive economic success.
As far as the chip market, it did accomplish one thing. It scared some other RISC vendors out of the market. Part of the reason for SGI dropping MIPS as a workstation chip and DEC dropping Alpha was the threat of Itanium. They figured if Intel can dominate with the badly designed IA32 architecture, what if they had a clean slate and money to back it up? They dropped out, figuring they'd port to Itanium and still sell their OS. SGI soldiered on a bit selling Itanium workstations, but their ability to be different was smashed, and they died soon after (which makes you congratulate Apple a bit being able to sell near commodity Intel laptops). HP just wanted to cut their chip costs down (they were making both Alpha and PA-RISC) and instead concentrate on moving printer ink.
As far as actually selling Itaniums in the market, it kind of landed with a thud. The new EPIC architecure made it VERY compiler dependent, and there were no good compilers in the beginning (and maybe not even now). It had the classic chicken and egg problem - no apps because no systems sold, no systems sold because of no apps. And its IA32 support sucked in the beginning. The first versions of the chip were particularly bad, though got a bit better on later generations.
Eventually, AMD released 64 bit extensions to IA32, x86_64, AMD64, whatever you want to call it. This gave decent speed at not a huge cost jump. The internal architecture was easy to write compilers for, and had very good IA32 performance. It cleaned up. Intel was forced to backtrack, and released the extensions as EM64T. It had the rights to from earlier licensing agreements with AMD regarding 486 production. Itanium would be forever relegated to a niche server product.
As far as "What has replaced them", nothing really. Itanium didn't really sell well, but it's still being produced. If you actually have Itanium, you can replace with a newer one if you like. If not, there are no emulators that I know of, you'd need to port your code to a new architecture. If it's fast enough for you, Intel Xeon (server versions of normal Intel chips), if not, probably IBM POWER. But you'd need to buy new machines.
So now even HP, one of the architects of the Itanium is now even slowly migrating away from Itanium, moving to Xeon x86_64 chips.
And you probably don't buy IA32 chips anymore, you most likely buy EM64T chips, which have great IA32 compatibility.
TL;DR: It was a big-iron server chip, that never sold well, and cheaper 64 bit Intel chips (x86_64, EM64T, whatever you want to call them) took much of its reason to be.
Daily reminder that we fucking own you intelcucks.
If it wasn't for amd64, you'd all be eating molten itanium dogshit for dinner.
It's meant to simplify CPU design. All the logic involved with branching is pushed onto the compiler for the architecture itself so none of that needs to be designed in hardware and embedded in the CPU. It also removes a lot of redundant hardware like registers because they're not actually necessary with VLIW. The result is a relatively small and simple CPU that is supposed to be very efficient. The problem is the compiler has to be incredibly complex to make up for the simplicity of the hardware itself.
Intel engaged in heavy anti-competitive practices during the short period where AMD objectively and undeniably made better CPUs than Intel. This is on the books, not some conspiracy bullshit.
>the damage was dealt quite hard
>Intel paying AMD $1.25 billion
Are you really that naive? A billion dollars for years of lost income and, equally importantly, lost reputation? That MIGHT sway things if a billion dollars was any amount of money to Intel, but it isn't. Intel kept on trucking like nothing happened, with consumers being familiar with the brand and assuming it was best, while AMD got the consolation prize of not outright making a loss while it happened.