@boxrekt said:
@son-goku7523: Exactly, perfectly summarized.
I expect the next xbox to actually be a far more capable of a 8k machine but It doesn't really even matter now because again most people don't currently care about 8k or are not even considering an up 8k upgrade but the SHOCK value of saying "8k" on E3's stage, no matter how much more powerful the next xbox might be over PS5, that angle to promote it's power will be neutralized.
Like I said in the OP, "IF" MS had been the one to announce 8k on E3's stage the same group who are all ranting and bashing Mark Cermy for name dropping 8k would have went BALLISTIC hyping it.
Sony effectively took the wind out of 8k, ray-tracing, and SSD talking points that MS 100% expected to be the first one to speak on.
8 ROPS per 32 bit memory channel is a bottleneck and it would take major innovation to overcome this limitation e.g. 16 ROPS per per 32 bit memory channel design.
It's expected NAVI to catch up with 8 ROPS per 32 bit memory channel GDDR6 design.
The fastest 64 ROPS based GPU is RTX 2080 with six geometry-rasterization engines with 64 ROPS read-write units with 4MB L2cache tiled rendering and up to +1900Mhz clock speed. At 1900Mhz, 11.9 TFLOPS FP32 potential.
For comparison
GTX 1080 Ti FE has six geometry-rasterization engines with 88 ROPS read-write units with 3MB L2cache tiled rendering and up to 1800 Mhz clock speed, typical is 1700Mhz. 12.2 TFLOPS at 1700Mhz,
RTX 2070 has four geometry-rasterization engines with 64 ROPS read-write units with 4MB L2cache tiled rendering and up to +1900Mhz clock speed via over 256 bit bus GDDR6-14000
VII has four geometry-rasterization engines with 64 ROPS read-write units with 4MB L2cache tiled rendering and up to 1800Mhz clock speed. At 1800Mhz, 13.8 TFLOPS FP32 potential, typical clock speed is around 1750Mhz which is 13.4 TFLOPS.
Volta GV100 has six geometry-rasterization engines with 128 ROPS read-write units and up to ~1627 Mhz clock speed via quad stack HBM v2.
RTX 2080 Ti has six geometry-rasterization engines with 88 ROPS read-write units with 6MB L2 cache tiled rendering and up to +1900 Mhz clock speed over 352 bit bus GDDR6-14000. RTX 2080 Ti is a class above VII, RTX 2080 and RTX 2070.
PS; Volta and Turing GPUs has full async compute hardware.
Log in to comment