add slides
[libreriscv.git] / pinmux / pinmux_chennai_2018.tex
1 \documentclass[slidestop]{beamer}
2 \usepackage{beamerthemesplit}
3 \usepackage{graphics}
4 \usepackage{pstricks}
5
6 \title{Pin Multiplexer}
7 \author{Rishabh Jain}
8 \author{Luke Kenneth Casson Leighton}
9
10
11 \begin{document}
12
13 \frame{
14 \begin{center}
15 \huge{Pin Multiplexer}\\
16 \vspace{32pt}
17 \Large{Auto-generating documentation, code \\
18 and resources for a Pinmux}\\
19 \vspace{24pt}
20 \Large{[proposed for] Chennai 9th RISC-V Workshop}\\
21 \vspace{16pt}
22 \large{\today}
23 \end{center}
24 }
25
26
27 \frame{\frametitle{Credits and Acknowledgements}
28
29 \begin{itemize}
30 \item TODO\vspace{10pt}
31 \end{itemize}
32 }
33
34
35 \frame{\frametitle{Glossary}
36
37 \begin{itemize}
38 \item GPIO: general-purpose reconfigureable I/O (Input/Output).
39 \item Pin: an I/O pad. May be driven (input) or may drive (output).
40 \item FN: term for a single-wire "function", such as UART\_TX,
41 I2C\_SDA, SDMMC\_D0 etc. may be an input, output or both
42 (bi-directional case: two wires are always allocated, one
43 for input to the function and one for output from the function).
44 \item Bus: a group of bi-directional functions (SDMMC D0 to D3)
45 where the direction is ganged and under the Bus's control
46 \item Input Priority Muxer: a multiplexer with N selector
47 wires and N associated inputs. The lowest (highest?) indexed
48 "selector" enabled results in its
49 input being routed to the output.
50 \item Output Demuxer: a one-to-many "redirector" where a single
51 input is "routed" to any one output, based
52 on a selector.
53 \end{itemize}
54 }
55
56
57 \frame{\frametitle{Why, How and What is a Pinmux?}
58
59 \begin{itemize}
60 \item Why? To save cost, increase yield, and to target multiple
61 markets with the same design, thereby increasing uptake
62 and consequently taking advantage of volume pricing.\vspace{4pt}
63 \\
64 Summary: it's all about making more money!\vspace{4pt}
65 \item How? By multiplexing many more functions (100 to 1,200) than there
66 are actual available pins (48 to 500), the required chip package
67 is far less costly and the chip more desirable\vspace{4pt}
68 \item What? A many-to-many dynamically-configureable router of
69 I/O functions to I/O pins\vspace{4pt}
70 \item \bf{Note: actual muxing is deceptively simple, but like
71 a DRAM cell it's actually about the ancillaries / extras}
72 \end{itemize}
73 }
74
75
76 \frame{\frametitle{Associated Extras}
77
78 \begin{itemize}
79 \item Design Specification
80 \item Scenario analysis (whether the chip will fit "markets")
81 \item Documentation: Summary sheet, Technical Reference Manual.
82 \item Test suites
83 \item Control Interface (AXI4 / Wishbone / TileLink / other)
84 \item Simulation
85 \item Linux kernel drivers, DTB, libopencm3, Arduino libraries etc.
86 \end{itemize}
87 Example context:
88 \begin{itemize}
89 \item Shakti M-Class has 160 pins with a 99.5\% full 4-way mux
90 \item Almost 640-way routing, 6 "scenarios" (7th TBD),
91 100+ page Manual needed,
92 \bf{17,500 lines of auto-generated code}
93 \end{itemize}
94 }
95
96
97 \frame{
98 \vspace{30pt}
99 \begin{center}
100 {\Huge
101 ALL of these\vspace{20pt}\\
102 can be\vspace{20pt}\\
103 auto-generated\vspace{30pt}
104 }
105 \\
106 (from the Design Specification, after Scenario Analysis)
107 \end{center}
108
109 }
110
111 \frame{\frametitle{Reduce workload, reduce duplication, reduce risk and cost}
112
113 \begin{itemize}
114 \item Auto-generate everything: documentation, code, libraries etc.
115 \vspace{10pt}
116 \item Standardise: similar to PLIC, propose GPIO and Pinmux
117 \vspace{10pt}
118 \item Standardise format of configuration registers:
119 saves code duplication effort (multiple software environments)
120 \vspace{10pt}
121 \item Add support for multiple code formats: Chisel3 (SiFive IOF),
122 BSV (Bluespec), Verilog, VHDL, MyHDL.
123 \vspace{10pt}
124 \item Multiple auto-generated code-formats permits cross-validation:\\
125 auto-generated test suite in one HDL can validate a muxer
126 generated for a different target HDL.
127 \vspace{10pt}
128 \end{itemize}
129 }
130
131
132 \frame{\frametitle{Design Spec and Scenario Analysis}
133
134 \begin{itemize}
135 \item Analyse the target markets that the chip will sell in\\
136 (multiple markets increases sales volume, reduces chip cost)
137 \vspace{4pt}
138 \item Create a formal (python-based) specification for the pinmux
139 \vspace{4pt}
140 \item Add scenarios then check that they meet the requirements\\
141 { \bf (before spending money on hardware engineers!) }
142 \vspace{4pt}
143 \item Scenarios represent target markets: ICs to be connected\\
144 (GPS, NAND IC, WIFI etc. May require draft schematics
145 drawn up, or client-supplied schematics analysed).
146 \vspace{4pt}
147 \item Analyse the scenarios: if pins are missing, alter and repeat.\\
148 \vspace{4pt}
149 \item Eventually the pinmux meets all requirements...\\
150 { \bf without spending USD \$5-50m to find out it doesn't!}
151 \end{itemize}
152 }
153
154
155 \frame{\frametitle{Muxer cases to handle (One/Many to One/Many) etc.}
156
157 \begin{itemize}
158 \item One FN outputs to Many Pins: no problem\\
159 (weird configuration by end-user, but no damage to ASIC)
160 \item One Pin to Many FN inputs: no problem\\
161 (weird configuration by end-user, but no damage to ASIC)
162 \item Many Pins to One FN input: {\bf Priority Mux needed}\\
163 No priority mux: Pin1 = HI, Pin0 = LO, ASIC is damaged
164 \item Many FN outputs simultaneously to one Pin: {\bf does not occur}\\
165 (not desirable and not possible, as part of the pinmux design)
166 \item Some FNs (I2C\_SDA, SD\_D0..3) are I/O Buses\\
167 Bi-directional control of the Pin must be handed to the
168 FN
169 \item Nice to have: Bus sets pintype, signal strength etc.\\
170 e.g. selecting SD/MMC doesn't need manual pin-config.\\
171 \bf{caveat: get that wrong and the ASIC can't be sold}
172 \end{itemize}
173 }
174
175
176 \frame{\frametitle{Pin Configuration, input and output}
177
178 In/out: {\bf Note: these all require multiplexing }
179 \begin{itemize}
180 \item Output-Enable (aka Input disable): switches pad to In or Out
181 \item Output (actually an input wire controlling pin's level, HI/LO)
182 \item Input (actually an output wire set based on pin's driven level)
183 \end{itemize}
184 Characteristics: {\bf Note: these do not require multiplexing }
185 \begin{itemize}
186 \item Output current level: 10mA / 20mA / 30mA / 40mA
187 \item Input hysteresis: low / middle / high. Stops signal noise
188 \item Pin characteristics: CMOS Push-Push / Open-Drain
189 \item Pull-up enable: built-in 10k (50k?) resistor
190 \item Pull-down enable: built-in 10k (50k?) resistor
191 \item Muxing and IRQ Edge-detection not part of the I/O pin
192 \end{itemize}
193 }
194
195
196 \frame{\frametitle{Standard GPIO 4-way in/out Mux and I/O pad}
197 \begin{center}
198 \includegraphics[height=2.5in]{../shakti/m_class/mygpiomux.jpg}\\
199 {\bf 4-in, 4-out, pullup/down, hysteresis, edge-detection (EINT)}
200 \end{center}
201 }
202
203 \frame{\frametitle{Separating Pin Configuration, input and output}
204
205 \begin{itemize}
206 \item Standard Mux design {\bf cannot deal with many-to-one inputs}\\
207 (SiFive IOF source code from Freedom U310 cannot, either)
208 \vspace{4pt}
209 \item I/O pad configuration conflated with In-Muxer conflated with
210 Out-Muxer conflated with GPIO conflated with EINT.
211 \vspace{4pt}
212 \end{itemize}
213 {\bf IMPORTANT to separate all of these out:
214 \vspace{4pt}}
215 \begin{itemize}
216 \item EINTs to be totally separate FNs. managed by RISC-V PLIC\\
217 (If every GPIO was an EINT it would mean 100+ IRQs)
218 \vspace{4pt}
219 \item GPIO In/Out/Direction treated just like any other FN\\
220 (but happen to have AXI4 - or other - memory-mapping)
221 \vspace{4pt}
222 \item Pad configuration separated and given one-to-one Registers\\
223 (SRAMs set by AXI4 to control mux, pullup, current etc.)
224 \end{itemize}
225 }
226
227 \frame{\frametitle{Register-to-pad "control" settings}
228 \begin{center}
229 \includegraphics[height=2.5in]{reg_gpio_cap_ctrl.jpg}\\
230 {\bf pullup/down, hysteresis, current, edge-detection}
231 \end{center}
232 }
233
234
235 \frame{\frametitle{In/Out muxing, direction control: GPIO just a FN}
236 \begin{center}
237 \includegraphics[height=2.5in]{reg_gpio_fn_ctrl.jpg}\\
238 {\bf Note: function can control I/O direction}
239 \end{center}
240 }
241
242
243 \frame{\frametitle{GPIO (only): Simplified I/O pad Diagram (FN only)}
244 \begin{center}
245 \includegraphics[height=2.5in]{reg_gpio_pinblock.jpg}\\
246 {\bf 3 wires: IN, OUT, OUTEN (also = !INEN) }
247 \end{center}
248 }
249
250
251 \frame{\frametitle{Output (and OUTEN) Wiring. 2 pins, 2 GPIO, 2 Fns}
252 \begin{center}
253 \includegraphics[height=2.5in]{reg_gpio_out_wiring.jpg}\\
254 {\bf Reg0 for Pin0, Reg1 for Pin1, Output and OUTEN same mux }
255 \end{center}
256 }
257
258
259 \frame{\frametitle{Input Selection and Priority Muxing}
260 \begin{center}
261 \includegraphics[height=0.75in]{reg_gpio_comparator.jpg}\\
262 {\bf Muxer enables input selection}\\
263 \vspace{10pt}
264 \includegraphics[height=1.25in]{reg_gpio_in_prioritymux.jpg}\\
265 {\bf However multiple inputs must be prioritised }
266 \end{center}
267 }
268
269
270 \frame{\frametitle{Input Mux Wiring}
271 \begin{center}
272 \includegraphics[height=2.5in]{reg_gpio_in_wiring.jpg}\\
273 {\bf Pin Mux selection vals NOT same as FN selection vals}
274 \end{center}
275 }
276
277
278 \frame{\frametitle{Summary}
279
280 \begin{itemize}
281 \item TODO
282 \end{itemize}
283 }
284
285
286 \frame{
287 \begin{center}
288 {\Huge The end\vspace{20pt}\\
289 Thank you\vspace{20pt}\\
290 Questions?\vspace{20pt}
291 }
292 \end{center}
293
294 \begin{itemize}
295 \item http://libre-riscv.org/shakti/m\_class/pinmux/
296 \end{itemize}
297 }
298
299
300 \end{document}