Bert Macklin
Newcomer
I'm not so sure its supposed to refer to Xenos. If you check the document its using similar phrasing for both GPUs. One is "64ALU" and the other is "48ALU". Also on the Yukon slide it refers to the CPUs with a similar data sequence. "*Quantity & Architecture Type* @ *Frequency*"
In that GCN paper I linked it contrasts older architectures with new GCN stuff.
"The shader arrays in earlier generations of AMD GPUs consisted of a number of SIMD engines, each of which consisted of up to 16 ALUs. ... In GCN, each CU includes 4 separate SIMD units for vector processing. ... The instruction buffering, registers and vector ALUs are private for each of the 4 SIMDs to sustain high performance and utilization"
I just don't know if this is some strange shorthand or GCN nomenclature mix up. It just seems like a weird descriptor for GPUs. Anyways, just wondering if anyone can explain this.
In that GCN paper I linked it contrasts older architectures with new GCN stuff.
"The shader arrays in earlier generations of AMD GPUs consisted of a number of SIMD engines, each of which consisted of up to 16 ALUs. ... In GCN, each CU includes 4 separate SIMD units for vector processing. ... The instruction buffering, registers and vector ALUs are private for each of the 4 SIMDs to sustain high performance and utilization"
I just don't know if this is some strange shorthand or GCN nomenclature mix up. It just seems like a weird descriptor for GPUs. Anyways, just wondering if anyone can explain this.