There have been products in the past that use server GPU architecture in an APU. For example, Bristol Ridge and Carrizo.
NOTE: I'm interested in APUs, NOT external GPUs like the MI60 or MI50.
NOTE: I'm talking about Vega20 the GPU architecture (NOT Vega 20 the Radeon iGPU):
Because the AMD Embedded APUs are being targeted for industrial/medical applications, I thought it might be possible that the Vega20 architecture (emphasis on architecture) is used in those APUs. Is this correct? Can I get a list of Epyc or Ryzen APUs that use the Vega20 architecture?
My R5 2400G has Vega graphics but not quite 20 CU however. The Vega is based on the latest GCN revision.
Following Vega AMD has redesigned their GPU slightly and are calling it RDNA
AMD's marketing department might be the source of confusion on what Vega20 is supposed to mean.
My understanding is that Vega 20 is *both* a server GPU architecture *and* a Radeon iGPU product with 20 compute units. Very confusing!
What I'm really interested in knowing is if there a iGPU with a FP64 (64 bit floating point) rate better than 1/16?
I thought maybe V1000/R1000 might be based on the Vega20 *architecture*.
See the specs of the MI60 for "GPU Architecture Vega20"
that card is about 2:1 FP64 vs FP32 but that is a straight compute card
that card also offers FP16 and INT8 for other workloads
at 13.3 TFLOPS SP: that is over 3 times more powerful than my GTX 1060
Yes, the super expensive MI60 can do FP64 at 1/2 rate. I'm interested in APUs. I wanted to know if any of the APU use the Vega20 architecture.
I can't find any information about the GPU architecture of V1000/R1000 from that Wikipedia article.
From the responses I'm getting, I have to assume the answer is that Vega20 is not in any APUs.