2
0
mirror of https://github.com/m-labs/artiq.git synced 2024-12-30 05:33:33 +08:00
artiq/doc/slides/taaccs.tex
Robert Jordens 01416bb0be copyright: claim contributions
These are contributions of >= 30% or >= 20 lines (half-automated).

I hereby resubmit all my previous contributions to the ARTIQ project
under the following terms:

    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation, either version 3 of the License, or
    (at your option) any later version.

    This program is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.

    You should have received a copy of the GNU General Public License
    along with this program.  If not, see <http://www.gnu.org/licenses/>.

Closes 

Signed-off-by: Robert Jordens <jordens@gmail.com>
2015-09-06 16:08:57 -06:00

276 lines
9.1 KiB
TeX

% Copyright (C) 2014, 2015 Robert Jordens <jordens@gmail.com>
\documentclass[final,presentation,compress]{beamer}
\usepackage[mathcal]{euler}
\usepackage{amsmath, amssymb, amsopn} %amssymb,amstext
\usepackage[cm-default]{fontspec}
\usepackage{xltxtra}
\usepackage[english]{babel}
\usepackage{multicol}
\usepackage{multimedia}
\usepackage{tikz}
\usetikzlibrary{arrows,shapes,snakes,positioning,backgrounds,decorations,graphs}
\definecolor{ethblue}{rgb}{0, 0.2, 0.3568}
\usepackage{minted}
\mode<presentation>
{
\useoutertheme{default} % simplistic
\setbeamertemplate{headline}[default] % kill the headline
\setbeamertemplate{navigation symbols}{} % no navigaton stuff in lr corner
\useinnertheme{circles}
\setbeamercolor*{palette primary}{use=structure,fg=white,bg=ethblue!70}
\setbeamercolor*{palette secondary}{use=structure,fg=white,bg=ethblue!80}
\setbeamercolor*{palette tertiary}{use=structure,fg=white,bg=ethblue!90}
\setbeamercolor*{palette quaternary}{use=structure,fg=white,bg=ethblue!100}
\setbeamercolor*{structure}{fg=ethblue!70}
\hypersetup{
}
\setbeamercovered{invisible}
}
\graphicspath{{fig//}}
\title{Real-time experiment control for quantum physics}
\author[Robert~Jördens]{{\bf Robert~Jördens}}
\institute[]{
Ion Storage Group, Time and Frequency, NIST, Boulder, CO \\
\url{rjordens@nist.gov}%
}
\begin{document}
\begin{frame}[plain]
\titlepage
\tikz[overlay,remember picture]\node[anchor=south,above=-.5cm] at (current page.south)
{\includegraphics[width=\paperwidth]{flatirons_winter_wikipedia}};
\tikz[overlay,remember picture]\node[anchor=south east, fill=white,
inner sep=.3mm] at (current page.south east) {%
\tiny Jesse Varner, AzaToth, CC-BY-SA};
\end{frame}
\begin{frame}
\includegraphics[width=\columnwidth]{jost_trap-3}
\end{frame}
\begin{frame}
\frametitle{Quantum gate sequences}
\includegraphics[width=\columnwidth]{gate_sequence}
\end{frame}
\begin{frame}
\begin{tikzpicture}[box/.style={rectangle,fill=white}]
\node[inner sep=0] {\includegraphics[width=\columnwidth]{lab}};\pause
%\draw[help lines,white] (-4, -3) grid (4, 3);
\node[box] at (-4, -2) {FPGA};
\node[box] at (3.5, 0) {ion trap};
\node[box] at (-3, 3) {$\sim$10 attenuators};
\node[box] at (2, 2) {$\sim$50 DAC};
\node[box] at (-4, 0) {$\sim$20 DDS};
\node[box] at (-2, 1) {$\sim$50 GPIO};
\node[box] at (.5, 0) {$\sim$10 motors};
\node[box] at (2, -2) {$\sim$10 power supplies};
\node[box] at (4, -3) {$\sim$10 lasers};
\end{tikzpicture}
\end{frame}
\begin{frame}
\frametitle{Physicists are not programmers:}
\footnotesize
\begin{center}
\only<1>{\includegraphics[width=\columnwidth]{labview}\\
LabVIEW: a ``visual programming language'' (a.k.a. ``high viscosity language'')}
\only<2>{\includegraphics[width=\columnwidth]{expwiz_matrix}\\
Rigid time-versus-channel matrix: inflexible (loops, conditionals?)}
\only<3>{\includegraphics[width=\columnwidth]{control_buttons}\\
Hard-coded components: not generic and opaque implementation}
\end{center}
\end{frame}
\begin{frame}
\frametitle{Enter ARTIQ}
\alert{A}dvanced \alert{R}eal-\alert{T}ime \alert{I}nfrastructure for \alert{Q}uantum physics
\footnotesize
\begin{itemize}
\item High performance --- nanosecond resolution, hundreds of ns latency
\item Expressive --- describe algorithms with few lines of code
\item Portable --- treat hardware, especially FPGA boards, as commodity
\item Modular --- separate components as much as possible
\item Flexible --- hard-code as little as possible
\end{itemize}
\end{frame}
\begin{frame}[fragile]
\frametitle{Define a simple timing language}
\footnotesize
\begin{minted}[frame=leftline]{python}
trigger.sync() # wait for trigger input
start = now_mu() # capture trigger time
for i in range(3):
delay(5*us)
dds.pulse(900*MHz, 7*us) # first pulse 5 µs after trigger
# re-reference time-line
at_mu(start + seconds_to_mu(1*ms))
dds.pulse(200*MHz, 11*us) # exactly 1 ms after trigger
\end{minted}
\begin{itemize}
\item Written in a subset of Python
\item Executed on a CPU embedded on a FPGA (the \emph{core device})
\item \verb!now_mu(), at_mu(), delay_mu(), delay()! describe time-line of an experiment
\item Exact time is kept in an internal variable
\item That variable only loosely tracks the execution time of CPU instructions
\item The value of that variable is exchanged with the RTIO fabric that
does precise timing
\end{itemize}
\end{frame}
\begin{frame}[fragile]
\frametitle{Convenient syntax additions}
\footnotesize
\begin{minted}[frame=leftline]{python}
with sequential:
with parallel:
a.pulse(100*MHz, 10*us)
b.pulse(200*MHz, 20*us)
with parallel:
c.pulse(300*MHz, 30*us)
d.pulse(400*MHz, 20*us)
\end{minted}
\begin{itemize}
\item Experiments are inherently parallel:
simultaneous laser pulses, parallel cooling of ions in different trap zones
\item \verb!parallel! and \verb!sequential! contexts with arbitrary nesting
\item \verb!a! and \verb!b! pulses both start at the same time
\item \verb!c! and \verb!d! pulses both start when \verb!a! and \verb!b! are both done
(after 20\,µs)
\item Implemented by inlining, loop-unrolling, and interleaving
\end{itemize}
\end{frame}
\begin{frame}[fragile]
\frametitle{Physical quantities, hardware granularity}
\footnotesize
\begin{minted}[frame=leftline]{python}
n = 1000
dt = 1.2345*ns
f = 345*MHz
dds.on(f, phase=0) # must round to integer tuning word
for i in range(n):
delay(dt) # must round to native cycles
dt_raw = seconds_to_mu(dt) # integer number of cycles
f_raw = dds.frequency_to_ftw(f) # integer frequency tuning word
# determine correct (to FP precision) phase
# despite accumulation of rounding errors
phi = n*mu_to_seconds(dt_raw)*dds.ftw_to_frequency(f_raw)
\end{minted}
\begin{itemize}
\item Need well defined conversion and rounding of physical quantities
(time, frequency, phase, etc.) to hardware granularity and back
\item Complicated because of calibration, offsets, cable delays,
non-linearities
\item No generic way to do it automatically and correctly
\item $\rightarrow$ need to do it explicitly where it matters
\end{itemize}
\end{frame}
\begin{frame}[fragile]
\frametitle{Invite organizing experiment components and code reuse}
\footnotesize
\begin{minted}[frame=leftline]{python}
class Experiment:
def build(self):
self.ion1 = Ion(...)
self.ion2 = Ion(...)
self.transporter = Transporter(...)
@kernel
def run(self):
with parallel:
self.ion1.cool(duration=10*us)
self.ion2.cool(frequency=...)
self.transporter.move(speed=...)
delay(100*ms)
self.ion1.detect(duration=...)
\end{minted}
\end{frame}
\begin{frame}[fragile]
\frametitle{RPC to handle distributed non-RT hardware}
\footnotesize
\begin{minted}[frame=leftline]{python}
class Experiment:
def prepare(self): # runs on the host
self.motor.move_to(20*mm) # slow RS232 motor controller
@kernel
def run(self): # runs on the RT core device
self.prepare() # converted into an RPC
\end{minted}
\begin{itemize}
\item When a kernel function calls a non-kernel function, it generates a RPC
\item The callee is executed on the host
\item Mechanism to report results and control slow devices
\item The kernel must have a loose real-time constraint (a long \verb!delay!)
or means of re-synchronization to cover communication, host, and device delays
\end{itemize}
\end{frame}
\begin{frame}
\frametitle{Kernel deployment to the core device}
\footnotesize
\begin{itemize}
\item RPC and exception mappings are generated
\item Constants and small kernels are inlined
\item Small loops are unrolled
\item Statements in parallel blocks are interleaved
\item Time is converted to RTIO clock cycles
\item The Python AST is converted to LLVM IR
\item The LLVM IR is compiled to OpenRISC machine code
\item The OpenRISC binary is sent to the core device
\item The runtime in the core device links and runs the kernel
\item The kernel calls the runtime for communication (RPC) and interfacing
with core device peripherals (RTIO, DDS)
\end{itemize}
\end{frame}
\begin{frame}
\begin{center}
\url{https://github.com/m-labs/artiq}
\end{center}
\footnotesize
\begin{itemize}
\item Fully open-source, BSD licensed
\item Ported and running on two different FPGA boards
\item Design applicable beyond ion trapping (superconducting qubits,
neutral atoms...)
\item Fastest open-source DDR3 SODIMM controller as a sub-project: 64\,Gbps
\item Interfacing with lab hardware
\item Hardware-in-the-loop unittests
\item Self-contained simulator
\item Currently $\sim$1\,µs latency and $\sim$1\,MHz event rate
\item DMA should improve that dramatically
\end{itemize}
\end{frame}
\end{document}
% vim:ts=2:sw=2:et:ai