2
0
mirror of https://github.com/m-labs/artiq.git synced 2024-12-22 18:04:03 +08:00
artiq/doc/manual/faq.rst

77 lines
2.6 KiB
ReStructuredText

FAQ
###
How do I ...
============
prevent my first RTIO command from causing an underflow?
--------------------------------------------------------
The RTIO timestamp counter starts counting at zero at the beginning of the first kernel run on the core device. The first RTIO event is programmed with a small timestamp above zero. If the kernel needs more time than this timestamp to produce the event, an underflow will occur. You can prevent it by calling ``break_realtime`` just before programming the first event.
override the `sysclk` frequency of just one DDS?
------------------------------------------------
Override the parameter using an argument in the DDB.
organize parameters in folders?
-------------------------------
Use GUI auto-completion and filtering.
Names need to be unique.
enforce functional dependencies between parameters?
---------------------------------------------------
If you want to override a parameter ``b`` in the PDB to be ``b = 2*a``,
use wrapper experiments, overriding parameters by passing them to the
experiment's constructor.
get rid of ``DBKeys``?
----------------------
``DBKeys`` references keys in PDB, DDB and RDB.
write a generator feeding a kernel feeding an analyze function?
---------------------------------------------------------------
Like this::
def run(self):
self.parse(self.pipe(iter(range(10))))
def pipe(self, gen):
for i in gen:
r = self.do(i)
yield r
def parse(self, gen):
for i in gen:
pass
@kernel
def do(self, i):
return i
create and use variable lengths arrays in kernels?
--------------------------------------------------
Don't. Preallocate everything. Or chunk it and e.g. read 100 events per
function call, push them upstream and retry until the gate time closes.
execute multiple slow controller RPCs in parallel without losing time?
----------------------------------------------------------------------
Use ``threading.Thread``: portable, fast, simple for one-shot calls.
write part of my experiment as a coroutine/asyncio task/generator?
------------------------------------------------------------------
You can not change the API that your experiment exposes: ``__init__()``,
``build()``, ``run()`` and ``analyze()`` need to be regular functions, not
generators or asyncio coroutines. That would make reusing your own code in
sub-experiments difficult and fragile. You can however always use the
scheduler API to achieve the same (``scheduler.yield(duration=0)``)
or wrap your own generators/coroutines/tasks in regular functions that
you then expose as part of the API.