By using this site, you agree to our Privacy Policy and our Terms of Use. Close
megafenix on 24 November 2013
Edit Thread
Pemalite said:
megafenix said:


how can the be an extrnalk bandwidth if the wii u edram is in the sae die as the gpu?

just look at the difference dude

gamecube edram or 1tsram embedded in the gpu gives 10gb/s

and thats only for 1 jegabyte

 

the wiiu edra is packed in the sae die with the gpu just like gaecube had its ebedded emory in the gpu

 

there is no external bus in the wii u edra with the gpu, the access its straightfowar, just ike xbox 360 rops had full access to the edram  256gb/s since they were on the sae die, the same applies to wii u gpu with the edram, the wii u gpu has full accessto the edram internal bandwidth


There is always an external bus that connects one group of functional units to another, regardless if they are on the same die or not.
Nintendo would have done the math on this and chosen the most optimal amount of bandwidth, one thing is for sure, it's going to be substantually lower depending on the type of bus they used.
If we get back to internal bandwidth... If it used a hub bus, Nintendo would then be able to optimise bandwidth for each individual unit based on need, that then reduces chip complexity and by extension power.

Ring Busses also introduce latency as data will have to travel around a ring, it's a much longer trip than a road that connects to each individual unit.

For example, AMD had a Ring bus in the Radeon 2900 and Radeon 3800 series, which had obscene amounts of bandwidth to all the individual parts of the GPU, when the Radeon 4800 series launched, AMD dropped the Ring bus and instead opted for a hub bus, the important parts got massive amounts of bandwidth, whilst the less memory hungry units got less bandwidth.
The result is, you still get an increase in performance, without any of the waste.
The chip was then connected to video memory at much lower levels of bandwidth, which then had to travel down a much smaller interconnect to reach System Ram.

As for the Xbox 360, initially the eDRAM wasn't on the same die, the eDRAM had a road that connected all the individual parts of the eDRAM with a 256GB/s bus, it still connected to everything else like the CPU and GPU with a much slower 32GB/s bus however.
When Microsoft moved the eDRAM on-die the bandwidth characteristics didn't actually change for obvious reasons.

This Diagram (from wikipedia) explains it perfectly.

 

 

nope, not always, in xbox 360 was only between the edram+rops to the gpu fraebuffer

 

 

here, just a little on this

"

The one key area of bandwidth, that has caused a fair quantity of controversy in its inclusion of specifications, is that of bandwidth available from the ROPS to the eDRAM, which stands at 256GB/s. The eDRAM is always going to be the primary location for any of the bandwidth intensive frame buffer operations and so it is specifically designed to remove the frame buffer memory bandwidth bottleneck - additionally, Z and colour access patterns tend not to be particularly optimal for traditional DRAM controllers where they are frequent read/write penalties, so by placing all of these operations in the eDRAM daughter die, aside from the system calls, this leaves the system memory bus free for texture and vertex data fetches which are both read only and are therefore highly efficient. Of course, with 10MB of frame buffer space available this isn't sufficient to fit the entire frame buffer in with 4x FSAA enabled at High Definition resolutions and we'll cover how this is handled later in the article.

"

 

 

More info on the internal bandwidth of the caches from here:

Both internal memory buffers have a sustained latency of under 5 nanoseconds. The frame and z-buffer memory is capable of 7.68 Gbytes/second of bandwidth. The texture buffer boasts an even faster bandwidth of 10.4 Gbytes/s because it's divided into 32 independent macros, each 16 bits wide for a total I/O of 512 bits. This gives each macro its own address bus, so that all 32 macros can be accessed simultaneously, said Mark-Eric Jones, vice president of marketing for Mosys. 

"


thats why 4096bits is the minium belieavable for wii u edram, and since the main ram doesnt even pack high bandwidth, the 8192bits sounds ore logical

not to ention that shinen can do 720p with double buffering with just about 7 megabytes of edram, while xbox need the full 10 megabytes of edra for the 720p with double buffering

doesnt that kind of suggest that with just 7 megas of wii u edram is the same as the 10 megas of xbox edram bandwidth?
dont forget that the actual bandwidth, despite the rops having full access t the edram bandwidth, is limited later by an external bus of 32gb/s(consider it like a latency problem or something between the rops and the gpu framebuffer), so the performance in wii u case is better since there are no external buses