Sujet : Re: Datasheet-flation?
De : blockedofcourse (at) *nospam* foo.invalid (Don Y)
Groupes : sci.electronics.designDate : 23. Nov 2024, 23:57:26
Autres entêtes
Organisation : A noiseless patient Spider
Message-ID : <vhtml0$1ss17$2@dont-email.me>
References : 1 2
User-Agent : Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:102.0) Gecko/20100101 Thunderbird/102.2.2
On 11/23/2024 1:40 PM, Jeff Liebermann wrote:
On Sat, 23 Nov 2024 03:50:54 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:
The data "sheet" for the new processor I'm using is ~16,000 pages.
(note carefully the position of the comma separator)
I suggest you switch to a RISC processor. Fewer instructions implies
that there might be fewer pages of documentation. Let's see if that
works.
THIS documentation just addresses hardware capabilities. And,
only parts of it. E.g., the details of the FPU, MMU, crypto
processing, GPU, etc. are handled in other documents.
There are 12000 pages documenting the *registers* (peripherals)
in the device.
The RISC-V instruction set manual is 238 PDF pages:
<https://riscv.org/wp-content/uploads/2019/06/riscv-spec.pdf>
while the Intel 64 and IA-32 combined developer manual is 5,237 PDF
pages:
<https://cdrdv2-public.intel.com/835781/325462-sdm-vol-1-2abcd-3abcd-4.pdf>
Yep. More instructions means more pages.
More instructions often address more capabilities. Does the
RISC processor, above, have an MMU? Hardware support for
floating point? GPU?
This is getting ridiculous. I thought 2,000 page datasheets were
over the top, but this one is a personal record! <frown>
Nobody RTFM's the entire manual. You're not expected to read and
understand everything, just the part that is useful for whatever
you're doing.
If you are responsible for writing the OS -- and its drivers -- you
need to understand all of the peripherals in-built to the device.
"Applications" don't talk directly to UARTs, or display hardware,
or PWM controllers, or timers, or...
And, even if you OPT not to use a particular "I/O" device, you
have to understand how to ensure it can't interfere with the
operation of the other I/O devices in the machine. The whole
notion of a "spurious" interrupt -- from a device that should be
quiescent!
I have numerous books where I've only read a few
relevant chapters, skimmed a few others, and ignored the rest. When
you lookup a telephone number in a telephone directory, do you read
the entire directory? Same with modern devices, which are often full
of features that you'll never need or use.
You select a device for the capabilities that you need. You
wouldn't select a device with an FPU/MMU/GPU/etc. if you didn't
plan on using/needing those features. You wouldn't select a device
with a hardware (integer) multiplier if THAT wasn't needed.