>instead of going for low-end home/small business server/server appliance segment, go for ultra-high end "scale out" datacenter market
>get BTFO immediately since corporate customers aren't going to make the $50M minimum commitment required to even try these things in data center setting
Why do companies keep doing this? This is the exact opposite of Intel's tried-and-true formula for server CPU market domination (start with SMB and "edge" servers then work up the chain).
Other urls found in this thread:
bloomberg.com
twitter.com
Good, another one bites the dust.
>instead of going for low-end home/small business server/server appliance segment, go for ultra-high end "scale out" datacenter market
They all think they can pander to hyperscale better than both x86 vendors already do.
>trying to push ARM servers with no software ecosystem without dropping multi-billions
Unregulated capitalism is the biggest problem on the face of the earth right now.
This is just another example of a product that was superior, but failed to survive due to being beaten dead with a stick by oligopolies.
>This is just another example of a product that was superior, but failed to survive due to being beaten dead with a stick by oligopolies.
It offers literally nothing over x86 solutions.
Intel can kill it with scaled-up Atom, even.
Yes it does, the research being spent benefits everybody in the ARM world. It is RISC and not CISC and therefore much easier to program on low level.
There are reasons that it ceased to exist, yes. But those reasons are the problem.
This sadly applies to way too many areas in america-driven areas, not only tech.
>Yes it does, the research being spent benefits everybody in the ARM world.
The fuck? Centriq is proprietary Qualcomm IP.
>It is RISC and not CISC and therefore much easier to program on low level.
What the fuck are you talking about, even.
You mean like NEON vs AVX? Both are necessary evil.
Most semiconductor companies are terrible like this. They literally don't want to sell to any company worth less than several billion dollars. They will happily blow billions on product lines dying rather than letting peasants touch them.
>This is the exact opposite of Intel's tried-and-true formula for server CPU market domination (start with SMB and "edge" servers then work up the chain).
The problem with this assumption is that you're imagining these chips as some kind of new Pentium Pro, when they're much more like the entry-level RISC chips the Pro and friends were totally obliterating. Expensive, poorly supported, relatively proprietary and ultimately less versatile. A deep-pocketed enterprise niche was the only chance this chip ever had, it offers nothing to those entry-level markets that propelled Intel into server dominance.
The RISC vs. CISC debate has been meaningless for decades now and nobody writes in assembler on PCs and servers anymore, stop living in the '80s and fetishizing meaningless labels you barely understand going for ARM of all things, the least RISC of the RISCs and the x86 of the embedded market.
There's nothing "superior" about this proprietary trash beyond not having a logo you don't like on it.
ARM instruction set is objectively cleaner than x86.
Nobody gives a fuck.
Maybe AArch64 but then again, who cares about the ISA when it's the implementation that matters?
I wish the first pinned post on the catalog was the definitions of "literally" and "objectively", not some faggy freetard worship thing. It would actually be useful.
Jow Forums is a dedicated freetard worship board, just deal with it at this point.
RISC-V will probably kill any ARM server ambitions, ARM has completely failed to push ARM arch outside of MCUs and mobile chips.
RISC-V is a dedicated Internet of Shit ISA, no one would use it for anything high-performance.
I figured that out within seconds of arriving here. Doesn't mean I have to like it.
It's always the year of RISC-V, just like the year of Linux desktop. Lots of talk, nothing ever happens.
>Doesn't mean I have to like it.
Well duh, freetards are running rampant here, just get used to it.
>It's always the year of RISC-V, just like the year of Linux desktop. Lots of talk, nothing ever happens.
Acktually it's seeing some mild adoption for internet of shit.
And WD tapped it for controllers.
That's it.
Physical RISC-V chips are in mass production now as opposed to just existing on paper.
They're all low-power MCUs designed to run IoT workloads, but they are still physical chips that you can actually use.
citation needed
I can see RISC-V always having a place given its niche and being the first really viable option for it. People who take it seriously as a real competitor to established desktop/server architectures and even embedded architectures are fucking delusional, though, no matter how much they run their mouths to the contrary.
Still doesn't translate to much in the direction that Jow Forums hype spewers will it to.
>Still doesn't translate to much in the direction that Jow Forums hype spewers will it to.
But, Bluetooth dragon dildos are going to be using RISC-V chips now, user?
>People who take it seriously as a real competitor to established desktop/server architectures and even embedded architectures are fucking delusional
That's Jow Forums for you.
Not like you can discuss anything interesting here besides consumer-grade hardware.
With security holes in them, of course.
trying to compress this image and maintain fidelity was a learning experience
>RISC-V is a dedicated Internet of Shit ISA, no one would use it for anything high-performance.
Jow Forums said the same about ARM years ago yet hear we are discussing ARM servers. Someone is always going to be trying to make non-x86 server chips as long as the x86 tax is so high, high end ARM or Power9 CPUs sell for less than $2000, top Intel and AMD are $8000 to $4000, they don't cost anything more to make it's purely thanks to the x86 monopoly.
If there was a market for ARM servers AMD would have released K12.
what ARM servers? this shit was vaporware from beginning to timely end unless you were a megacorporation with deep pockets and a CIO that sucked as much anti-x86 cock as you apparently do to think that fucking POWER is about the "x86 tax" let alone a viable answer to it
You don't know what you are talking about. Cease posting.
>Jow Forums said the same about ARM years ago yet hear we are discussing ARM servers.
We're discussing what, another vendor that bailed out of DOA market?
>Someone is always going to be trying to make non-x86 server chips as long as the x86 tax is so high
AMD's single socket options are cheap as fuck.
>Power9
Imagine being this kind of a delusional gaynigger to think it's viable for anyone but vendorlocked shops.
So, you really believe that Qualcomm, offering nothing more than a new sticker to run the same old software on, had a game-changer on the level of P6 that was going to make any measurable impact on the market if it was just a little cheaper? Or you just think shitty entry-level RISC workstations and servers in the '90s were good? Or you still believe in the RISC vs. CISC delusion?
What part of your hype-driven constructed reality do I not fit nicely into?
>"Use spectre and spectre-ng CPUs on your servers!" anonymous spouted
>"this is all I've got, might as well keep shouting about it until someone cares!" said the Qualcomm marketer to himself quietly
get rekt
