By using this site, you agree to our Privacy Policy and our Terms of Use. Close
HoloDust on 18 June 2013
Edit Thread
ethomaz said:

HoloDust said:

No, not really, it is based on similar architecture, specially TMU and ROP, but shader units have more in common with R600 (2xxx) family. If you look at 2600Pro/XT, 360 is basically 2 of those slapped together, but at lower clock and significantly lower combined memory bandwidth. Some math and educated guesstimation are required to get to that number at the end.

Actual performance is higher of course, but considering we can't really measure how much specific customizations play role in final products, taking just vanilla numbers is a good starting point for the sake of discussion (hence, 141VP of 7850 for PS4, and 96VP of 7770 for XOne, though in reality, both solutions perform lot better).

A exemple... the RSX is supposed to be close a GeForce 7800... the Xenon is a better GPU than RSX... and the GeForce 7800 are in the 12.3 to 15.9 VPs... I'm pretty sure Xenon will be over 15 VP at that table.

In fact most forum and techs guys uses X1900 for X360 (~25VP) and GTX 7900 (~22VP) for PS3 but I think that a little exagerated... I think it is more in line with ~20VP (360) and 16VP(PS3)

A good article about that: http://br.dir.groups.yahoo.com/group/gamejutsu/message/2690

Edit - lol sorry I forget the article is in Portugue... look just the graph.

Using the 20VPs for X360:

PS4: ~7x X360
Xbone: ~5x X360

Only comparing GPU power.


RSX is gimped down 7800GTX with memory bandwidth of 7600GT.  Xenos is nowhere near X1900XT,  with lower pixel/texel rate and half the memory bandwidth. Good example of how much bandwidth is important are 5550s (speculated architecture for WiiU's GPU) with the same core clocks, but different bandwidth  - 27VP, 21.8VP and 16.5VP (last result for the card with same bandwidth as WiiU).

As I said in earlier post, due to customization in console's GPUs (like eDRAM in 360 and WiiU) they will perform better than what the math is suggesting in vanilla versions - that's why I'm using lower numbers, that way you can compare them with vanilla versions of PS4/XOne's counter parts, and not try to guesstimate their enhanced performance (which would be more practical when you want to compare them with PCs).

As for graph, you'll see it puts Xenos at nearly the same level as 4670, and 2900 Pro and 5570, if you check VPs - which is quite silly, tbh, as Xenos just doesn't have initial muscle to be nowhere near that.

Anyhow, whatever VPs you take into measurement, one way or the other, MS themselves stated that XOne= 8x360, so PS4 being 1.5x XOne, you come to what I stated at the beginning - PS4 = 12x360...just for fun, try to extrapolate VPs for XOne and PS4 using that graph, and you'll come to some really redonkulous numbers.