Seems people do not understand how Xbox One could get 4 x 68 gb/s from DDR3 ram configuration...
Insider: "4 x 68 = 272 which is the same bandwidth penello said is in the system.
is this possible that the system can have 4 lanes of main memory just like they have 4 lanes of eSRAM?"
100% correct :) "
The number crunching Albert used was simple he just added the 64 x 4 = 272. He did this cos he was probably told to add the memory peak bandwidth . Now the payed article only has one purpose and that is to dis - credit any information that will see the x1 in any superior aspects. And its also a personal attack. Typical sony article I would say you know were the these technique minds hang out.
Any way Albert should have explained why he added the number like he did. Also The DDR3 memory inside x1 is any thing but
pc standard issue ram. There is 4 address lanes each lane has 2GB of ddr with a max bandwidth of 68gb's . The ram is so modified that data transfers is basically the same as the ddr4 memory that will ship next year to pc. Now like the esram 4 address lanes 4 x 8 = 32. what is confusing people is you can not add these number up to equal 272gb.s as your still only getting max bandwidth of 68gbs not 272gb's. But that is if you are reading the memory number as one lane reads.
And that is were things get confusing. But like the development kits had 16 gig of system memory. It was to emulate the specific of the x1 memory lanes and you can add 68bg's cos there is 4 individual lanes. for each 2GB ddr3 memory pools. so hence the 2GB x 4 = 8GB each 2GB = 68gb's 4 x 68gb's = 272. And this is why in coherent cpu read mode is 30gb's. cos when coherent memory reads are happening in the Dram cntr they can do 4 passes . But at half speeds of max Bandwidth.
And we are not even discussing the esram and extra functions of the console .. X1 has two memory systems. Ps4 has one 8GB of gddr5 .. yes onion and garlic + - :) .. And it does have a massive bandwidth. But it also has latency. The latency is not a major problem but the cpu is crippled with much lower bandwidth to help with Latency and it helps cpu cycles. but when compared to the x1.. They may be of similar design but that's about it.. x1 is way more advanced.
Now this is why I said 100% correct..
Mistercteam Daily: a) SOC summary,re-highlighted the Gfx core for Offload
b) DX 11.1 Gfx core, the one that off load CPU & improve GPGPU
c)15SPP offload CPU & GPU core ( put the b into context)