By: Anon (no.delete@this.spam.com), August 13, 2022 12:54 pm
Room: Moderated Discussions
Wilco (wilco.dijkstra.delete@this.ntlworld.com) on August 13, 2022 11:12 am wrote:
> Floating point code is typically very data heavy but every ISA still uses immediate
> loads for floating point constants. Are all ISA designers stupid, or are misses
> from immediate loads not as crippling for performance as claimed?
AMD GPUs does support floating point constants on the instruction stream.
Other ISAs are far more dated and lack of support for embbeded constants may be due to very different reasons, take x86 for example, 8087 was a separate chip and supported 80 bits data on a 16 bits CPU, that's a very different problem from embbeding a 32 or 64 bits data in a wide superscalar 64 bits single chip.
> Floating point code is typically very data heavy but every ISA still uses immediate
> loads for floating point constants. Are all ISA designers stupid, or are misses
> from immediate loads not as crippling for performance as claimed?
AMD GPUs does support floating point constants on the instruction stream.
Other ISAs are far more dated and lack of support for embbeded constants may be due to very different reasons, take x86 for example, 8087 was a separate chip and supported 80 bits data on a 16 bits CPU, that's a very different problem from embbeding a 32 or 64 bits data in a wide superscalar 64 bits single chip.