1 \documentclass[slidestop
]{beamer
}
2 \usepackage{beamerthemesplit
}
8 \title{Data-Dependent-Fail-First
}
9 \author{Luke Kenneth Casson Leighton and Shriya Sharma
}
16 \huge{The Libre-SOC Hybrid
3D CPU
}\\
18 \Large{Data-Dependent-Fail-First
}\\
23 \large{Sponsored by NLnet's PET Programme
}\\
30 \frame{\frametitle{Why another SoC?
}
33 \item Intel Management Engine, Apple QA issues, Spectre
\vspace{6pt
}
34 \item Endless proprietary drivers, "simplest" solution: \\
35 License proprietary hard macros (with proprietary firmware)\\
36 Adversely affects product development cost\\
37 due to opaque driver bugs (Samsung S3C6410 / S5P100)
39 \item Alternative: Intel and Valve-Steam collaboration\\
40 "Most productive business meeting ever!"\\
41 https://tinyurl.com/valve-steam-intel
43 \item Because for
30 years I Always Wanted To Design A CPU
45 \item Ultimately it is a strategic
\textit{business
} objective to
46 develop entirely Libre hardware, firmware and drivers.
52 \frame{\frametitle{How can you help?
}
57 \item Start here! https://libre-soc.org \\
58 Mailing lists https://lists.libre-soc.org \\
59 IRC Freenode libre-soc \\
60 etc. etc. (it's a Libre project, go figure) \\
62 \item Can I get paid? Yes! NLnet funded\\
63 See https://libre-soc.org/nlnet/\#faq \\
65 \item Also profit-sharing in any commercial ventures \\
67 \item How many opportunities to develop Libre SoCs exist,\\
68 and actually get paid for it?
70 \item I'm not a developer, how can I help?\\
71 - Plenty of research needed, artwork, website \\
72 - Help find customers and OEMs willing to commit (LOI)
78 \frame{\frametitle{What goes into a typical SoC?
}
81 \item 15 to
20mm BGA package:
2.5 to
5 watt power consumption\\
82 heat sink normally not required (simplifies overall design)
84 \item Fully-integrated peripherals (not Northbridge/Southbridge)\\
85 USB, HDMI, RGB/TTL, SD/MMC, I2C, UART, SPI, GPIO etc. etc.
87 \item Built-in GPU (shared memory bus,
3rd party licensed)
\vspace{3pt
}
88 \item Built-in VPU (likewise, proprietary)
\vspace{3pt
}
89 \item Target price between \$
2.50 and \$
30 depending on market\\
90 Radically different from IBM POWER9 Core (
200 Watt)
92 \item We're doing the same, just with a hybrid architecture.\\
99 %%\frame{\frametitle{Simple SBC-style SoC}
102 %%\includegraphics[width=0.9\textwidth]{shakti_libre_soc.jpg}
110 \begin{frame
}[fragile
]
111 \frametitle{Simple-V ADD in a nutshell
}
114 function op
\_add(rd, rs1, rs2, predr) # add not VADD!
115 int i, id=
0, irs1=
0, irs2=
0;
116 for (i =
0; i < VL; i++)
117 if (ireg
[predr
] &
1<<i) # predication uses intregs
118 ireg
[rd+id
] <= ireg
[rs1+irs1
] + ireg
[rs2+irs2
];
119 if (reg
\_is\_vectorised[rd
] ) \
{ id +=
1; \
}
120 if (reg
\_is\_vectorised[rs1
]) \
{ irs1 +=
1; \
}
121 if (reg
\_is\_vectorised[rs2
]) \
{ irs2 +=
1; \
}
125 \item Above is oversimplified: Reg. indirection left out (for clarity).
126 \item SIMD slightly more complex (case above is elwidth = default)
127 \item Scalar-scalar and scalar-vector and vector-vector now all in one
128 \item OoO may choose to push ADDs into instr. queue (v. busy!)
132 \frame{\frametitle{Additional Simple-V features
}
135 \item "fail-on-first" (POWER9 VSX strncpy segfaults on boundary!)
136 \item "Twin Predication" (covers VSPLAT, VGATHER, VSCATTER, VINDEX etc.)
137 \item SVP64: extensive "tag" (Vector context) augmentation
138 \item "Context propagation": a VLIW-like context. Allows contexts
139 to be repeatedly applied.
140 Effectively a "hardware compression algorithm" for ISAs.
141 \item Ultimate goal: cut down I-Cache usage, cuts down on power
142 \item Typical GPU has its own I-Cache and small shaders.\\
143 \textit{We are a Hybrid CPU/GPU: I-Cache is not separate!
}
144 \item Needs to go through OpenPOWER Foundation `approval'
148 \frame{\frametitle{maxloc
}
153 \frame{\frametitle{Pospopcount
}
158 \frame{\frametitle{strncpy
}
163 \frame{\frametitle{linked-list walking
}
168 \frame{\frametitle{Summary
}
171 \item Goal is to create a mass-volume low-power embedded SoC suitable
172 for use in netbooks, chromebooks, tablets, smartphones, IoT SBCs.
173 \item No way we could implement a project of this magnitude without
174 nmigen (being able to use python OO to HDL)
175 \item Collaboration with OpenPOWER Foundation and Members absolutely
176 essential. No short-cuts. Standards to be developed and ratified
177 so that everyone benefits.
178 \item Riding the wave of huge stability of OpenPOWER ecosystem
179 \item Greatly simplified open
3D and Video drivers reduces product
180 development costs for customers
181 \item It also happens to be fascinating, deeply rewarding technically
182 challenging, and funded by NLnet
190 {\Huge The end
\vspace{12pt
}\\
191 Thank you
\vspace{12pt
}\\
192 Questions?
\vspace{12pt
}
197 \item Discussion: http://lists.libre-soc.org
198 \item Freenode IRC \#libre-soc
199 \item http://libre-soc.org/
200 \item http://nlnet.nl/PET
201 \item https://libre-soc.org/nlnet/\#faq