Why don't current GPU's use volatile but markers to improve concurrent GPU compute and stream processor performance? Because going by your logic, this PS4 advantage is inferior because it is not implimented in current GPU architectures...superclocked
Â
No not no sony make some very small changes because they wanted more compute jobs,GCN has 2 Aces 4 compute commands,across all GCN,on PS4 is 8 Aces 64 commands,Aces is what allow in order instructions GPU like GCN do out of order executions.
Â
ESRAM is not something that was only created for xbox one,ESRAM exist outside the xbox one,what sony did to the GPU doesn't.
Â
The story goes that by the time MS chose 8GB of memory because that it what they need it,GDDR5 was impossible do to density of the chips not been there yet,so there was a big chance that if MS went GDDR5 they would not have enough chips for the unit and the console would have to be even bigger do to having more memory chiops.
So MS went DDR3 which was abailable on higher density and was cheaper 2 birds one stone,but DDR is slow for today graphics,so they use ESRAM to compensate for any bandwidth problem the unit may face,the 7770 which has 1.27 TF actually a tad higher than the xbox one,only has a 72Gb/s bandwidth,so MS chose that 32MB because with the 102GB/s they would compensate.
This is the same sh** all over again an i already prove how MS say on 2005 that the xbox 360 had way superior bandwidth to the PS3,based on EDRAM which ran even faster than the xbox one, at 256Gb/s..
Â
The xbox one can have 1TB bandwidth it would not matter the xbox one GPU will never ever saturate that bus,the 7770 which has similar performance has only 72GB/s bandwidth,so i don't see why MS would need 200GGb/s with that GPU performance.
Log in to comment