wonderfield
Supreme [H]ardness
- Joined
- Dec 11, 2011
- Messages
- 7,396
Heh. For whatever it's worth, NVIDIA claimed that the trio wasn't paid to attend the G-Sync launch.Good point. Carmack made his millions by attending NVIDIA PR events.
On the subject of Intel and 'serious level', they're already there. Intel dedicates larger percentages of die space to the GPU with each successive generation. The reason is simple: they can't meaningfully improve IPC by increasing the pipeline depth (which they've been decreasing) and they can't do it with cache, either. Dedicating more logic to branch prediction doesn't seem to pay off either. They have little they can do with the transistor budgets they have except increasing core counts (which pays little performance dividend in most cases) and allowing more to the GPU, which pays off in a major way. Embarrassing parallelism and all that.
If Intel is willing to implement four times as much eDRAM as what their engineers deem ample, and to do that for the purpose of increasing graphics performance, you know they're pretty serious about it.