Computing Template‑class | |||||||
|
@ User:Rando717 I removed the "2:1 <- ratio -> 1:16" info from the table. I get what you are trying to say, but it is kinda freestyle and IMHO confusing to the average reader. Furthermore this info is related to the microarchitecture and not to the individual cards in this table. However, if you really wish to have this info in the table, I think it would fit good to the processing power footnote:
"Precision performance is calculated from the base (or boost) core clock speed based on a FMA operation. For the RDNA 2 microarchitecture the half-precision processing power is double the single-precision processing power and the double-precision processing power is 1⁄16 of the single-precision processing power."... or something along those lines.
Wikiinger ( talk) 19:34, 16 April 2022 (UTC)
Computing Template‑class | |||||||
|
@ User:Rando717 I removed the "2:1 <- ratio -> 1:16" info from the table. I get what you are trying to say, but it is kinda freestyle and IMHO confusing to the average reader. Furthermore this info is related to the microarchitecture and not to the individual cards in this table. However, if you really wish to have this info in the table, I think it would fit good to the processing power footnote:
"Precision performance is calculated from the base (or boost) core clock speed based on a FMA operation. For the RDNA 2 microarchitecture the half-precision processing power is double the single-precision processing power and the double-precision processing power is 1⁄16 of the single-precision processing power."... or something along those lines.
Wikiinger ( talk) 19:34, 16 April 2022 (UTC)