• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

CPU improvements year to year

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Performance improvements are still happening, just not in traditional ways (increasing clock speed, IPCs). Instead we're seeing focuses on scheduler improvements and threading. If you look at things like the OpenPower/Power8 architecture where you've got 8 threads per core versus Intel's current offerings with only 2 threads per core you can see where things are going. This is especially true in the server space, but will also end up being true in the desktop space as more and more applications become multithreaded.

The reality is that the hardware is already TOO powerful for the software, which is why virtualization has grown so rapidly in popularity. The best way to get application performance density now is to break your hardware down and scale applications horizontally. If application developers would start using higher performance toolsets and targeting threaded/concurrent development as a priority we'd see this paradigm start to shift back in the other direction as it'd make more sense then to provide highly threaded dedicated hardware to the application.

Instead, the current market is saturated with applications built using languages like Python and Ruby which are really only ever intended to be single-threaded (there's some janky hacks to get some semblance of threading to work, but they're not perfect). I'd like to see more stuff done in Java (but with better code), Scala, Haskell, and Google Go in the future. I think some companies, Google and Facebook especially, have recognized this and are actually putting a significant amount of effort into making their software more efficiently use the hardware they already have because at some point when you hit hyperscale on your datacenters/networks/hardware, your application can't just go horizontal anymore without a significant increase in OpEx from the power usage.

Like somebody else said, as most systems become a window to the world through web applications, it's much more important to see power density and performance improvements on the server than the desktop. If you need that performance at home now, the best bet is getting the higher end enthusiast boards and installing a Xeon.

Just my thoughts.
 
Back
Top