Intel loses its Lustre – Chipzilla bins own-brand HPC file system Intel's decided to stop offering its own version of the Lustre file system – the code beloved of high-performance computing types because it's handy for managing exabyte-scale storage spanning colossal Linux clusters. Trish Damkroger, an Intel veep and general manager for technical computing initiatives – that's HPC in Intel- …
I doubt Ryzen has anything to do with it. It is nice, but not exactly the second coming of Athlon 64...
More likely they have finally decided to stop listening to the idiots at Gartner who keep predicting the PC sales decline has bottomed out and will soon reverse, and realized it is a terminal decline because 1) smartphones are used for more and more tasks by people and 2) performance growth on PCs is so slow that even a 10 year old PC is perfectly usable for almost anyone given an SSD and no malware. Not that anyone will upgrade a 10 year PC to an SSD, but the ones that are bought today with an SSD may still be in use in 2030.
> while SSD can breathe new life into a PC, it's an inherent part of electrinics that the efficiency wanes over (usage) time
Most electronics do not age per se (though apparently semiconductors will all eventually fail).
Windows PCs definitely get slower with time due to ever-increasing software bloat, HDDs get fragmented, and flash memory will wear out.
However in my experience an old PC with an SSD is just fine. There is very little difference in responsiveness between work's new HP Elitebook (W10) and my personal Thinkpad W500 (Vista + SSD). My previous ThinkPad T41 with SSD + LInux runs just as well, thanks to Penguin power.
Not so sure about that.
It could be argued that young people will end up yearning for bigger screens and more practical form factors than smartphones and tiny tablets, and that a huge part of the world population never had a chance to use a PC so they don't know what they are missing yet.
If they want to give VR or augmented reality a try, this will happen on PCs, not on mobile devices that don't have the processing power.
On the business side, many people will still need computers - or have you ever tried to write a presentation or a document longer than a page on a phone ?
Finally, while it's true you can perfectly use 10 years old computers, should you want to spare energy and heat, you'll purchase something newer, and very cheap at that.
What's true is that if Intel can't increase the performance of its desktop/notebook processors fast enough, people will keep their computers much longer, and the era of replacing a computer every 2-3 years is gone.
I thought there'd already been some price cuts, though if Ryzen takes off in a big way there'd certainly have to be more if they want to maintain their share of the PC market pie.
Trouble is that it’s a shrinking pie, and it's becoming difficult to maintain (never mind expand) sales, especially with AMD taking a slice. Windows 10 hasn't done them any favours either... Chasing the HPC market isn't very profitable (there's not many super computers), but it looks like support for AI might be the thing to flog to the likes of Facebook, Google, etc.
Even that looks doubtful. Google are doing their own chip, and I strongly suspect this AI thing will prove to be short lived; it's not going to be good enough to be of sufficient benefit to be worth it.
At the end of this whole thing Intel have got to recognise that most people don't need a hugely powerful CPU in their lives, and the same is true of a lot of servers too.
Chasing the HPC market isn't very profitable (there's not many super computers), but it looks like support for AI might be the thing to flog to the likes of Facebook, Google, etc.
The market for "Hyperscale" is already bigger than for HPC; I guess it's sensible for Intel to chase that market (it's painful to say this as an HPC geek).
All the claims and numbers telling that the PC market is in trouble and the decline keeps going are just lies. PC hardware components, even very expensive Nvidia GPUs are being sold like hot-cakes and motherboard manufacturers keep releasing dozens of models every year. If there really was any decline as the claims and numbers tell the public then hardware manufacturers would shrink their products range and release the bare minimum to keep going. It is just that obvious. As long as hardware manufacturers don't shrink their catalogs of products it means that the PC market decline is just a fake and multinationals managers are getting away with the money they claim to be losing.
Latest (not) News
Intel files for Chapter 11
Ok, that's a lie but there has to come a time when the R&D effort needed to keep X86 alive will exceed the income from CPU's. Then Intel will really be in trouble. It has an ARM license but apparently there is no money to be made with them. so what next for Intel? Are they really a one trick pony?
PC sales may not be doing as well as they were several years ago, but Intel has over sources of income - they have nearly 100% of the server market.. including all those fluffy cloud SaaS/IaaS offering from AWS, Azure, Google, Apple, Rackspace and every other server you have in your Data Center.
In addition, they supply 90% of CPU's to every laptop and desktop supplier, including Apple, Dell, HP, Lenovo, Acer, ASUS....
So in a nutshell, Intel are not going anywhere!
My take is, the PC market is not shrinking - It has just matured and/or stabalised...
Intel hates being seen as "just a chip maker". Over and over again we've seen it invest in projects that it thinks are more prestigious and that allow it to talk about more than speeds'n'feeds. Unfortunately, it's the chips that deliver the revenue, and eventually somebody asks the question "how does this project drive more chip revenue?".
And if there isn't a really good direct connection, that Prestige Project can start counting down the days...
Intel realized that there was a limited amount of humans creating data to drive scale up/out and refresh of the cloud data centers. IoT served, partially, to create more data to drive hyperscale (Intel's cash cow).
IoT isn't hitting needed adoption levels and a saturation of hyperscale data centers is looming. Less scale-out, less refresh.
Time to cut costs.
Intel realized the number of humans creating data center/cloud content is finite. Only so many first world teenagers taking selfies and posting to social media. IoT was partially promoted to bring about non-human data generators to keep the cloud bloated with useless data and drive hyperscale build up, build out and refresh. IoT has failed to gain the adoption hoped for and self driving cars are farther over the horizon that had been hoped. Hyperscale is going to fizzle out for a while. We might have one or two architecture generations with limited adoption as refresh isn't necessary plus ARM and AMD will get more market share. Intel is battoning down the hatches and preparing for rough seas.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
