Replies to post #136627 on Intel Corp (INTC)

[Dmcq]

So much about benchmarking. One can prove whatever he wants by selecting the corresponding benchmarks. In the end, what matters is real life performance. Something Intel is doing fine with, especially at high work loads.

Have a look at the popularity figures on the right of those benchmarks from Futuremark. Not so good a message there.

[simplytom]

@unkwn

So much about benchmarking.

Worth reading - Futuremark found out that Apple A7 processor had a weak memory controller:
http://www.futuremark.com/pressreleases/understanding-3dmark-results-from-the-apple-iphone-5s-and-ipad-air

Links and Nodes
In simple terms, Bullet's btSoftBody object includes an array of Nodes and Links that describe the physical properties of the object. We found that when the node and link information was written to memory so that the A7 CPU could access it in a sequential fashion, as in our test app, we saw an increase in speed. The increase was higher when the A7 could run up the memory, and a bit lower when it ran down.

The way Bullet places the data in memory is more random. The A7 could not read from the memory as quickly when it had to jump back and forth to find what it needed.

To understand this better, it can help to think of device memory as a book.