News for hackers. Stuff that matters.
Re:Why Apple and Google are struggling to design simple soft... by _Understated_ on Wednesday March 30, 2016 @04:05AM (#11387658)
I think we will be. It has stopped.

I just upgraded my "old" Dell Precision T5500 to a X99 hand-built machine with an NVMe SSD and so on and I was rather underwhelmed: It's not as fast as a 5 year jump in architecture would have been 10 years ago (if that makes sense!), not even close.

I think Moore's law is done already. Intel have slowed their speed improvements over the last several years and are getting rid of their tick-tock.

I think it's market maturity: Things are "fast enough" (yes there are always those that push the limits, like gamers) but I reckon there are more mums and dads that just want stuff to work and will hang onto it for years than those that demand the fastest every year.

In fact, something I see in forums now is that even gamers are hanging onto their "old" 2600k chips since they are still incredibly fast.

My feelings on the OS are that the OS is something I shouldn't have to deal with much... it's merely a vessel to launch my applications (Visual Studio and SQL Server mainly) so I don't need anything else as long as I can run my applications and it's secure enough (I am still on Win 8.1 with Start8).

I am willing to go out on a limb and say that many others will be the same and from personal experience, just in my own circle, they are:

Me: Upgraded my 2010 Precision to a new machine last month Wife: Upgraded her 6 year old MAC at Christmas Dad: Upgraded his 8 year old Win XP machine last year for a £300 Dell Brother: Still rocking a first gen Macbook Pro

Re:Why Apple and Google are struggling to design simple soft... by TheOtherHobbes on Wednesday March 30, 2016 @06:07AM (#11388182)
Moore's law is still alive on GPUs, which are developing at ludicrous speed.

Of course most Office applications don't need a GPU, and all the bloat in modern operating systems doesn't help create an impression of speed on the desktop.

More the opposite - if you're saving to the cloud, it's always going to feel much slower and clunkier than saving to a local disk.

But creative and scientific applications - including AI - always need all the cycles available and then some. There are apps that could sink a 1000X CPU speed bump without blinking. They'll become more practical as they move to GPUs, but there's a significant extra cost in coding for not-really-standard GPU families than for a generic mostly-standard CPU range.