Soundwave said:
JRPGfan said:
ARM vs x86? for servers running programs probably optimised for x86? and your surprised intel with their better proccessing nodes & cpu arch are able to beat a ARM cpu?
Thats "ARM Cortex A57" designed by ARM btw, not AMD.
Still just useing ARM design cpu's from them, is going to be equal to nvidia's arm cpu designs.
This is why I think AMD can make a ARM based mobile chip thats as good as Nvidia.
Everyone that licences from ARM, can use their designs for the cpu's they make.
It becomes a matter of the gpu effeciency, which nvidia admittedly has the upperhand in.
However makeing use of HSA I believe AMD could make programs suited for HSA that run better on a mobil chip than Nvidia can.
|
Mind you these are decisions Nintendo had to make like 18-24 months ago.
Nvidia had the Tegra line ... AMD had jack shit.
If this is the approach Nintendo wanted, Nvidia to me is clearly the better choice.
Not only that but they also tend to get better performance per watt than AMD and better memory bandwidth usage, which is critical to a Nintendo device that needs to be portable.
If Nvidia was willing to give Nintendo a big break on the hardware costs, then it was really a no brainer of a decision, the Tegra X1 can even run the Dolphin emulator reasonably well, which means it probably can take on Nintendo Virtual Console fairly easily since Nintendo I'm sure will be able to make a better emulator using official resources.
|
In windows running directx 11. Things are differnt if your not running windows or dx11.
Things are differnt if you design software around it.
Thats why I mention HSA.
Do you know what it is?
Do you know how GPUs can accelerate certain tasks and do them much faster than a CPU? and use much less power?
So your CPU is idle, but your GPU is working at the task and does it faster than your cpu does.
But this isnt true for everything... some tasks, your cpu does better than your GPU.
For those tasks your CPU is active, and your GPU is idle.
So far that is the grunt of it, your useing half of it (cpu or gpu) and ignoreing the other part.
HSA is differnt.
Its a design thats intelligent, and does 1 task, but uses both gpu+cpu at the same time, splits a task and lets the cpu do the parts its good at, while the gpu does the parts its good at.
The results are speedups that are insane, and useing much less power.
My point was, and still is, that AMD could design a ARM based apu, that can compete with Nvidia if they are asked too.