By using this site, you agree to our Privacy Policy and our Terms of Use. Close
fazz on 27 October 2007
Edit Thread

Ok, it seems I've been living and beileving a lie for at least 12 years... can you at least tell me WHEN triangles stopped being polygons? >_>

I'll try to explain the bandwidth of the XDR memory in a simple way, beacuse it seems that some people still believe is INSANELY faster compared to GDDR3 or GDDR4... it is faster (than GDDR3) but not insanely.

First example, GDDR3: Imagine this is a highway with medium speed cars (the arrows being the cars)...

-------------
>->->->->
>->->->->
>->->->->
>->->->->
>->->->->
>->->->->
>->->->->
>->->->->
>->->->->
-------------
<-<-<-<-<
<-<-<-<-<
<-<-<-<-<
<-<-<-<-<
<-<-<-<-<
<-<-<-<-<
<-<-<-<-<
<-<-<-<-<
<-<-<-<-<
-------------

As you can see, it has a good number of lanes (bandwidth, 128 bit) so we can have more "cars" carrying stuff (data) at the same time.

Now, let's move on to XDR memory's highway, it has faster cars but...

------------
>>>>>>>
------------
<<<<<<<
------------

... it has much less lanes (bandwidth 16 bit if I remember correctly), so our "cars" while still being much faster, have to do more "trips" to give us the data.

I know it's not exactly scientific, but it's better than the "girls and apples" example I was thinking :P

PS: And yeah, I know pixel fillrate is a part of the equation... if you turn the Xenos into a 32-Shaders/16-Vertex configuration... well, bye bye RSX :)