Re: Is Intel exceptionally unsuccessful as an architecture designer?

On Thu, 19 Sep 2024 00:29:09 +0000, MitchAlsup1 wrote:

On Wed, 18 Sep 2024 22:54:33 +0000, Lawrence D'Oliveiro wrote:
 

On Wed, 18 Sep 2024 16:23:01 +0000, MitchAlsup1 wrote:
>

On the other hand, and this is where the deprecation of the CPUs come
in, The engines consuming the data are bandwidth machines {GPUs and
Inference engines} which are quite insensitive to latency (they are
not not latency bound machines like CPUs).
>
When doing GPUs, a memory access taking 400 cycles would hardly
degrade the overall GPU performance--while it would KILL any typical
CPU architecture.

>
But if it’s supposed to be for “interactive” use, it’s still going to
take those 400 memory-cycle times to return a response.

 
That is why you use the CPU for human interactions and bandwidth
engines for the muscle.

But then those bandwidth engines become the interactivity bottleneck,
don’t they?

Unless you use them only for precomputing stuff in some kind of batch mode
for later use, rather than doing processing on-demand.