Technical Comparison Sony PS4 and Microsoft Xbox

[Betanumerical]

Interesting?

I am just quoting the official Sony/Microsoft information (press release / architecture panel / interviews) instead of some random internet rumor/leak sites.

Link to the Microsoft tech panel video: http://www.youtube.com/watch?v=tLBVHZokt1Q.
For the technical details, I recommend watching Nick Baker 25:00 -> 26:30.

Quote (Baker):
"...get high capacity and high bandwidth, so with our memory architecture we are actually achieving all of that, we are actually getting more than 200 GB/s per second across the memory subsystem".

Well the coherent bus and all of the random move engines and what not are all part of the memory sub system so it doesnt really surprise me that they can quote that number whilst at the same time saying nothing.

 

[Love_In_Rio]

Well the coherent bus and all of the random move engines and what not are all part of the memory sub system so it doesnt really surprise me that they can quote that number whilst at the same time saying nothing.

And dont forget they include in a 5 billion transistors figure 1,6 billion of memory transistors. In E3 Sony perfectly could include the GDDR5 transistors in the total system transistors and call it a day to fight back and we would be back to the prior generation 1 and 2 tflops absurd claims.

 

[Xbat]

It seems to me that ps4 is superior in almost every aspect. Using aggregate bandwidth numers is being purposely vague, 360 has better aggregate numbers than both the ps4 and Xbone.

 

[Arwin]

The HDD can realistically supply 2MB per frame.

Cheers

Yes, it seems like 60MB/s is a decent rate expected from 2.5" 500GB, which would be enough for 2MB/s

Sony in particular I think degraded the maximum capacity by encryption / filesystem inefficiencies. I am a little worried something like that could happen again, but I really don't hope so. It may end up a discernable factor though, in terms of load speeds and pop-in.

If they could get an SSD into the system somehow, that could be a huge win (in terms of performance, maybe not reliability), but financially it doesn't seem likely to happen.

 

[Rangers]

And dont forget they include in a 5 billion transistors figure 1,6 billion of memory transistors. In E3 Sony perfectly could include the GDDR5 transistors in the total system transistors and call it a day to fight back and we would be back to the prior generation 1 and 2 tflops absurd claims.

Except they're completely different, and one is on the SOC the other isn't, and all manufactures call embedded dram as transistors and none call ddr, ...nice try though.

 

[Rangers]

It seems to me that ps4 is superior in almost every aspect. Using aggregate bandwidth numers is being purposely vague, 360 has better aggregate numbers than both the ps4 and Xbone.

except cpu...

and ram amount...

and disc capacity/throughput

you know just some little things

 

[Xbat]

I did say almost every aspect

 

[Rangers]

I did say almost every aspect

those are three of the biggest, most fundamental aspects i just listed.

 

[Love_In_Rio]

I did say almost every aspect

In the most important aspects at least

 

[Xbat]

In the most important aspects at least

Yes, for me cu count and memory bandwidth are the most important in comparing these two very similar architectures. There is the esram but I feel thats just there to make up for the slow ddr3.

 

[Lycan]

except cpu...

and ram amount...

and disc capacity/throughput

you know just some little things

What are you comparing exactly, if I may ask.

 

[Betanumerical]

except cpu...

and ram amount...

and disc capacity/throughput

you know just some little things

We dont know the CPU reservations between the two
Same for ram
And we dont know what HDD/disk media they are using

so they are better in the 3 things we dont know?

 

[DJ12]

Cpu is not a major part in either console and you know it Rangers. Its all about the gpu and has been since the 360 beat the ps3 in 99% of head to heads

Both are farming computational task out to the gpu as well

 

[blakjedi]

Except that the HDD could only realistically supply about ~20MB/frame.

Sure its doable but it is no where near equal, you cannot equate a bus that can at most copy ~25MB/frame to one that can copy GB's/frame.

what does equal mean here?

All you are trying to do is provide data to the ESRAM. If the ESRAM were in the GB range then that rate would be anemic. But its in the MB range and can be easily filled with nearly zero impact on the Northbridge.

 

[Betanumerical]

what does equal mean here?

All you are trying to do is provide data to the ESRAM. If the ESRAM were in the GB range then that rate would be anemic. But its in the MB range and can be easily filled with nearly zero impact on the Northbridge.

Expect that your not really likely to get even 20MB/frame out of the HDD as noted above your more likely to get 2MB/frame, you couldnt even fill it once a frame using the bus, its not what its meant for.

You could fill 7MB/frame at most, thats whilst destroying the CPU cache and causing all kinds of trouble there too.

 

[blakjedi]

Expect that your not really likely to get even 20MB/frame out of the HDD as noted above your more likely to get 2MB/frame, you couldnt even fill it once a frame using the bus, its not what its meant for.

You could fill 7MB/frame at most, thats whilst destroying the CPU cache and causing all kinds of trouble there too.

hunh? Per frame at 30 frames per second is 60 MB. Clearly more than enough. All other counts are at bandwidth/second.

 

[Betanumerical]

hunh? Per frame at 30 frames per second is 60 MB. Clearly more than enough. All other counts are at bandwidth/second.

Im doing it per frame because you generally want your graphics to be rendered per frame and not per second and most of the assets are read and worked on per frame, not per second.

hence the per frame usage.

 

[Mianca]

Bingo.

That is also how the ESRAM can be filled.

The ddr bus has 68gb to the nb to receive that data.

The ESRAM has to use its 102gb bus then the 30GB bus to receive that data. Its really no different depending upon how the developer wants to utilize the ESRAM.

The ESRAM can be copied to from DDR, cpu or hdd theoretically.

Looks like there'll be a hell of a lot of data-shuffling going on in XBOX One. I understand the "move engines" are supposed to kind of alleviate that problem. But still: Looking at the high concept, SONY's appraoch seems way simpler and, as a result, more reliable.

I can't even begin to imagine the headache involved in checking for possible bottlenecks when tons of data keep being shuffled around between [BD], HDD, RAM, and ESRAM.

 

[sebbbi]

You could fill 7MB/frame at most, thats whilst destroying the CPU cache and causing all kinds of trouble there too.

With virtual texturing, you only need to load a few hundreds of kilobytes per frame from the HDD. And that gets you 1:1 texel / screen pixel quality (as much texture resolution as you want).

Copy & pasted from liolio's signature (thank you for keeping this link around!):
Sebbbi about virtual texturing

 

[Betanumerical]

With virtual texturing, you only need to load a few hundreds of kilobytes per frame from the HDD. And that gets you 1:1 texel / screen pixel quality (as much texture resolution as you want).

Copy & pasted from liolio's signature (thank you for keeping this link around!):
Sebbbi about virtual texturing

And youd be better off chucking this over the DDR3 right? because it makes no sense to read that over the coherent link.