By using this site, you agree to our Privacy Policy and our Terms of Use. Close
MortienGerrux said:
Cyran said:

You comparing 32FP performance to 16FP.  It a common mistake since mobile apu advitise there 16FP performance while video cards/consoles almost always list there 32FP performance.

The Nvidia Tegra X1 FP32 performance is 512 GigaFlops.  less then half of the XB1 and PS4.  

That's the K1 in the 2014 Nvidia Shield Tablet, the X1 was released at CES 2015 and is the chip in the Nvidia Android TV box.

Nvidia showed two X1's running in a self driving car system, with a total power of 2.3 Teraflops, which is more then the PS4.

Imagine if they used that for the PS4.5 they could make it 10x smaller then the current PS4.


Source: http://wccftech.com/nvidia-drive-px-dual-chip-tegra-x1-revealed/

But again thats FP16 and not 32....

and those X1's in the android TVs use more than 20 watts of power each.

 

So your looking at 1.15 Teraflops (FP32) @40watts+  (from 2 x X1)

vs

PS4 1.84 Teraflops (FP32) @90? watts.

 

also you know that X1 memory bandwidth is much much lower right?

It cant do 176 GB/s memory bandwidth, without serious rework to the design and memory channels ect that the chip has.

Which would make the chip draw more power than it already does. (or have really bad memory bandwidth and not be able to run 1080p)

Plus the CPU inside the PS4, is faster than the CPU inside the X1 (I think).

 

 

MortienGerrux:

read the first few lines of the article you linked:


"This particular beauty is a soldered package that will be provided to cars with not one but two Tegra X1 Chips, for an impressive total of upto 2.3 TeraFlops of compute power on FP16." 
-quote.

Its FP16 (which is much easier to do than FP32).
Thats why its number is that high.

PS4 does FP32 when it says 1.84 Teraflops.