CPU plateau'd or RoE really effecting price ?

[Dolby]

... or there's more to a CPU than benchmark!

Just over 3 years ago my PC died and I needed to get a new one. For the CPU, I chose a 3570k process, which was pretty decent at the time. I can't remember the price, but it was more expensive than average. CPU benchmark gets 7144 (http://www.cpubenchmark.net/cpu.php?cpu=Intel+Core+i5-3570K+@+3.40GHz)

I decided to check out new CPU and came across the Intel Core i5 4690K and checkend the benchmark - 7708 (https://www.cpubenchmark.net/cpu.php?cpu=Intel+Core+i5-4690K+@+3.50GHz). While it is more, thought in 3 years it may be twice the speed.

Performance aside, they want close to R5,000.00 for this CPU :wtf:

Has the exchanged killed component pricing?
Thought thing gets much faster, and much cheaper?

I've always you'd get the same performance PC a year later, at 30% cheaper ....

 

[cguy]

A given benchmark may not show off the full potential of a new chip. The 3570k is an Ivybridge generation processor. In terms of floating point maths, clock-for-clock a given core in the newer Haswell or Skylake processors can do 2x the maths of the 3570k. The benchmark (or application), may have to be recompiled or rewritten to full exploit the new features though.

Apart from this, there are other microarchitecture improvements that may only show up in certain applications/benchmarks (improved ILP, improved TLB caching, improved instruction latency, better ILP, off the top of my head), so don't just use one benchmark, use many if you really want to test it properly.

 

[TJ99]

Short answer without marketing BS - Yes, desktop CPU growth has slowed down dramatically. It's all about laptops and cellphones these days. My old 2600K is still more than enough for anything out there today.

Same thing happened with graphics cards but not as dramatic.

 

[MickZA]

Moore's law has died at the age of 51 after an extended illness.

Thought thing gets much faster, and much cheaper?

I've always you'd get the same performance PC a year later, at 30% cheaper ....

Afraid Moores Law failed to hold true a couple of years ago ...

Moore’s law really is dead this time

http://arstechnica.com/information-technology/2016/02/moores-law-really-is-dead-this-time/

 

[garyc]

There are only four industry verified methods of desktop benchmarking, the rest are speculative. https://en.wikipedia.org/wiki/Benchmark_%28computing%29#Industry_standard_.28audited_and_verifiable.29

It is best to use the results from one of these for comparison. SPEC is informative if you are prepared to learn how to interpret the results.