troyan's latest activity

https://www.eurogamer.net/digitalfoundry-2024-the-big-horizon-forbidden-west-pc-tech-interview So, i guess the decompression engine...

Nobody said something else. But geometry doesnt make a frame alone. Robocop is a great example how UE5 is unable to provide proper...

Alan Wake 2 looks much better than Robocop. Even without Raytracing the lighting system is more stable. Sometimes a compromis is better...

This performance number doesnt even make any sense. Raytracing needs shading, too. So 4x would be maybe only tracing rays and doing...

https://developer.nvidia.com/blog/nvidia-gb200-nvl72-delivers-trillion-parameter-llm-training-and-real-time-inference/ Maybe nVidia...

Going by AMD's slide only three of the IODs can access the HBM from two others with full speed. 1/4 of MI300X is like the Grace-Hopper...

Its 1.2 TB/s/dir. A little bit off from nVidia's 5 TB/s/dir:

B100 doubled the HBM3 memory from H100 and has 2.4x more bandwidth. It is basically a H100 NVL with 8TB/s instead of 600GB/s (NVLink).

I thought everyone knows this because nVidia has written and spoke about it four years ago... I remember that the interconnect speed in...

Technical brief: https://nvdam.widen.net/s/xqt56dflgh/nvidia-blackwell-architecture-technical-brief Going by this Blackwell at 700W...

nVidia has released an update for RTXGI: https://developer.nvidia.com/rtx/ray-tracing/rtxgi

So, basically nVidia has done Ampere/Hopper just with two dies. Impressive. Oh and delivering 10 PFlops FP8 performance within 208...

From their PR: https://nvidianews.nvidia.com/news/nvidia-blackwell-platform-arrives-to-power-a-new-era-of-computing

No, two massive dies. :D 104 billions transistors per chip.

Reddit has another slide: 1.8 TB/s offchip bandwidth. Four years ago A100 supported 2.02TB/s with HBM2. Be able to connect 72 B100s...