Caporegime
- Joined
- 9 Nov 2009
- Posts
- 25,909
- Location
- Planet Earth
Raja can share all the secretes he likes but Intel cannot use AMD's IP, all Intel have is nVidia's old obsolete IP, i'm sure Raja convinced Intel he can do this and that with it but even with AMD's IP portfolio his graphics tech was lack lustre.
They seemed to have forgotten that Jim Keller when he joined Tesla,not only hired some ex-AMD people,but it apparently lead to this:
https://www.cnbc.com/2017/09/20/tesla-building-an-ai-chip-for-its-cars-with-amd.html
AMD collaborating with Tesla on making a new chip. This probably indicates its more likely Intel will work closer with AMD,not only in licensing IP,but developing some aspects of the physical GPU development.
After all if Intel already had a decent uarch already they wouldn't be using a custom AMD chip for their high end laptop chip,so it looks like a different direction to what they were doing before.
If anything,as the chap over at HardOCP said,Intel has let go of a lot of its graphics related engineers.
So that indicates to me,they intend to either:
1.)License and use AMD base technology to implement their GPUs. AMD developes the base uarch and Intel implements their own chips based on it different to AMD.
2.)Intel buys RTG so gets all the engineers and IP.
3.)Intel buys part of RTG and shares in R and D costs.
4.)AMD developes the graphics parts of Intel CPUs.
5.)Intel developes their own base technlogy and design.
The last one has not worked as well as they probably expected,hence why the sudden change it seems.
