AMD has agreed to license its Radeon graphics technology with Samsung in a multi-year partnership. The deal brings the American company back into the mobile GPU market for the first time in over ten years. It also spells the end to the Mali GPU partnership between ARM and Samsung that started in 2014.
https://www.notebookcheck.net/Avengers-Assemble-AMD-and-Samsung-unite-to-bring-Radeon-graphics-to-Exynos-chipsets.422983.0.html
This is somewhat expected, it's been a while since Samsung says they're going to make their own GPU for Exynos.
But it'd be impractical for them to build on their own, or for them to buy a company in this field. So they need a partner, an experienced one, so their GPUs can compete.
So the most obvious choice for them is AMD because they're already in "partnership" in HSA Foundation, and both aren't competing in any field.
They can't go partnering with the others, like ARM (because they don't want to depend on them, that's why Samsung's Mongoose exist), Qualcomm (because it's their rival obviously), or Nvidia (because both are kinda rival too and Nvidia is "do things on their own" kinda company).
It's a win-win situation for Samsung and AMD.
I don't recall even the latest Radeons being known for efficiency...I can also imagine performance taking a hit on all those apps optimized for Adreno.
@S.Yu
Desktop cards were indeed not known for being efficient. Though its partly due to AMD pushing desktop parts above the frequency curve the arch was most efficient at (which is between 800-900Mhz ) Another trick they have is Vulkan/DX12 becoming more mainstream among developers. Old Adreno graphics are still running revisions of openGLES. New ones have vulkan support but AMD still has the advantage of already supporting vulkan since the first GCN arch. They also have smaller benefits like supporting FP16 already.
AMD producing mobile graphics is not nearly as hopeless as Intel trying to compete with ARM cpu's
Quote from: Concatenate on June 03, 2019, 23:28:37
@S.Yu
Desktop cards were indeed not known for being efficient. Though its partly due to AMD pushing desktop parts above the frequency curve the arch was most efficient at (which is between 800-900Mhz ) Another trick they have is Vulkan/DX12 becoming more mainstream among developers. Old Adreno graphics are still running revisions of openGLES. New ones have vulkan support but AMD still has the advantage of already supporting vulkan since the first GCN arch. They also have smaller benefits like supporting FP16 already.
AMD producing mobile graphics is not nearly as hopeless as Intel trying to compete with ARM cpu's
Well I sort of saw it the other way, they lack notebook options because their designs lack the efficiency to be competitive in that power envelope, but I don't have a strong opinion on this, we'll have to wait and see.