Open Source HW in 2030 Why Architects Need It and It Needs Them - - PowerPoint PPT Presentation

▶
open source hw in 2030
SMART_READER_LITE
LIVE PREVIEW

Open Source HW in 2030 Why Architects Need It and It Needs Them - - PowerPoint PPT Presentation

Aug 16, 2023 326 likes •731 views

Open Source HW in 2030 Why Architects Need It and It Needs Them Michael Bedford Taylor UC San Diego The Fate of Computing Today Is determined by a small number of companies Ill get back We have great to you.. ideas for how we should

slide-1
SLIDE 1

Open Source HW in 2030

Why Architects Need It and It Needs Them

Michael Bedford Taylor UC San Diego

slide-2
SLIDE 2

The Fate of Computing Today

Is determined by a small number of companies…

I’ll get back to you.. We have great ideas for how we should compute!

16 nm

ISCA

slide-3
SLIDE 3

What prevents adoption of our ideas?

  • 1. Our methodology is pragmatic but broken
slide-4
SLIDE 4
  • 1. Our methodology is pragmatic but broken

performance: repeat (modify_c_simulator()) until (perf>=10% || sim_bug_in_my_favor || overtrained_on_my_10_benchmarks )

What prevents adoption of our ideas?

slide-5
SLIDE 5
  • 1. Our methodology is pragmatic but broken

performance: repeat (modify_c_simulator()) until (perf>=10% || sim_bug_in_my_favor || overtrained_on_my_10_benchmarks ) assert(it_would_really_work_in_hw)

What prevents adoption of our ideas?

slide-6
SLIDE 6
  • 1. Our methodology is pragmatic but broken

performance: repeat (modify_c_simulator()) until (perf>=10% || sim_bug_in_my_favor || overtrained_on_my_10_benchmarks ) assert(it_would_really_work_in_hw) power: assert(we_used_McPat && no_space_to_describe)

What prevents adoption of our ideas?

slide-7
SLIDE 7
  • 2. Our stuff works great … for our CPU/GPU microarch

… but theirs is different

What prevents adoption of our ideas?

slide-8
SLIDE 8
  • 3. We didn’t solve all of the important problems

“Context switching happens .. out of band ..”

What prevents adoption of our ideas?

slide-9
SLIDE 9
  • 4. “We regret to inform you that your idea was too

revolutionary for us to consider as the successor to Core2Duo in our roadmap” “This paper entirely rethinks … Our cycle-accurate trace simulator…”

What prevents adoption of our ideas?

slide-10
SLIDE 10
  • 5. Chicken-and-Egg

16 nm SoCs need huge volume to amortize costs; your emerging app that needs your accelerator is not already in use by many users; too risky to dedicate that much die area on iPhone 7 à No tech transfer

What prevents adoption of our ideas?

slide-11
SLIDE 11
  • 6. The Last Mile

Your idea is great, but probably only you have the will and patience to adapt it to their system ... and you don’t work there.

What prevents adoption of our ideas?

slide-12
SLIDE 12
  • 7. Your awesome needle in the ISCA/MICRO/… haystack

Everybody shows good results, but unbeknownst to all, yours is actually worth doing!

What prevents adoption of our ideas?

slide-13
SLIDE 13
  • 8. Smaller and smaller number of commercial architects

have less and less time to find a home for our ideas

What prevents adoption of our ideas?

slide-14
SLIDE 14

Current Tech Transfer Pipeline

ISCA 12-pages Intel 7-nm 5 million units C sim 50K LOC

repeat (modify_c_sim()) until (perf>=10% || sim_bugr || overtrained )

Maybe we need a few more intermediate points?

slide-15
SLIDE 15

Proposed Tech Transfer Pipeline

ISCA 12-pages Intel 7-nm 5M units Proto Implementation

Maybe we need a few more intermediate points?

OOO Processor Multicore GPU Community-maintained Open Source Cores (Like Linux) Integrated Into Community Open Source Quadcopter 65 nm ASIC Samsung TV 32 nm Talking Toy 90 nm You You You

This would address all of the 8 problems I showed… Reproducible, flushed out, real results, ….

slide-16
SLIDE 16
  • Good news: enrollment in undergrad Computer

Architecture: 30à400

What will the Hardware workforce look like in 15 years?

Switching gears to a different facet of open source…

slide-17
SLIDE 17
  • Good news: enrollment in undergrad Computer

Architecture: 30à400

  • Bad news: “professor, which chapter of Patterson

& Hennessy covers ?”

What will the Hardware workforce look like in 15 years?

slide-18
SLIDE 18
  • Good news: enrollment in undergrad Computer

Architecture: 30à400

  • Bad news: “professor, which chapter of Patterson

& Hennessy covers ?”

  • Students don’t want to design hardware at a

stodgy old HW company, they want to start the next Instagram!

  • Attracting the best talent is a serious problem for

the vibrancy of our HW industry

What will the Hardware workforce look like in 15 years?

slide-19
SLIDE 19
  • Good news: enrollment in undergrad Computer

Architecture: 30à400

  • Bad news: “professor, which chapter of Patterson

& Hennessy covers ?”

  • The undergrads don’t want to design hardware,

they want to work for Twitter!

What will the US Hardware workforce look like in 20 years? Now with Apps!

slide-20
SLIDE 20

HW diversity of computing devices is dwindling…

Source: Gartner Group, T. Austin 20 20 2000 4000 6000 8000 10000 12000 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Total ¡ASIC ¡Starts Year

How can open source revitalize the HW field in general?

slide-21
SLIDE 21

Can we make hardware design exponentially leaner so we can have more startups exploring more ideas?

Un Under the Pillow of Ou Our CS Undergrads…

Can we get to a “Minimum Viable Product*” with a few people years of effort? Is it possible?

  • Most basic version of your product that customers actually pay for/use
  • r in terms of research, show a ”real” design
slide-22
SLIDE 22

Costs of Latest Nodes Are Skyrocketing

20 40 60 80 100 120 140 0.5u 0.35u 0.25u 0.18u 0.13u 90nm 65nm 45nm 28nm 20nm Cost ¡to ¡Market ¡($ ¡million) Silicon ¡Technology ¡Node Mask ¡Costs S/W ¡Development ¡and ¡Testing H/W ¡Design ¡and ¡Verification Source: International Business Strategies, T. Austin 22 22

$88M $120M $500K

slide-23
SLIDE 23

Software Innovation Today

By Created by Uwe Kils (iceberg) and User:Wiska Bodo (sky). - (Work by Uwe Kils) http://www.ecoscope.com/iceberg/, CC BY-SA 3.0,

Instagram Op Open Source Linux GNU * Apache Redis Postgres/SQL Python Proprietary Code Django Memcached 500K-->13 people & $1B GCC

slide-24
SLIDE 24

Hardware: Where is the Open Source?

By Created by Uwe Kils (iceberg) and User:Wiska Bodo (sky). - (Work by Uwe Kils) http://www.ecoscope.com/iceberg/, CC BY-SA 3.0,

Calibre DRC/LVS Formality IC Compiler Design Compiler VCS DDR Phy Spice IO Pads ARM Interconnect Your Secret Sauce In Instagram Standard Cells ARM A57, A7, M4, M0… Closed Source ($$) Open Source

slide-25
SLIDE 25

From $120M to $5M: Open Source Can Address most of the Cost

20 40 60 80 100 120 140 0.5u 0.35u 0.25u 0.18u 0.13u 90nm 65nm 45nm 28nm 20nm Cost ¡to ¡Market ¡($ ¡million) Silicon ¡Technology ¡Node Mask ¡Costs S/W ¡Development ¡and ¡Testing H/W ¡Design ¡and ¡Verification Source: International Business Strategies, T. Austin 25 25

$88M $120M

$5M

slide-26
SLIDE 26

And going back a few nodes can get us from $5M to $500K for a 4X perf. Penalty (post-Dennard scaling)

20 40 60 80 100 120 140 0.5u 0.35u 0.25u 0.18u 0.13u 90nm 65nm 45nm 28nm 20nm Cost ¡to ¡Market ¡($ ¡million) Silicon ¡Technology ¡Node Mask ¡Costs S/W ¡Development ¡and ¡Testing H/W ¡Design ¡and ¡Verification Source: International Business Strategies, T. Austin 26 26

500K 500K

$120M

$5M

slide-27
SLIDE 27

How can Hardware Design Be More Like Software?

  • Open source infrastructure allows us to create systems

where we may only have to write 5% of the total code to create an entirely new product. à Leverage, not labor (and not IP $$$)

  • Open source Languages and Libraries so we don’t have to

redesign everything every time. (like STL or Python or Java Libraries)

  • Reduce the overhead of creating + testing new designs
  • Open Source CAD, Open Source Packages, Open Source

Standard Cells, Open Source Testboards, NO NDA’s.

  • IAAS clouds allow us to scale quickly from small

companies to large ones from 1 customer to 1 billion customers à Scaling ideas from the small to the big

slide-28
SLIDE 28

The Open Source HW Vision

Think GNU/Linux, but for everything HW related: Open Source CAD Tools (Like GNU) VLSI HLS, RTL to GDS … PCB Design and Simulation Tools Open Source Chip Designs (Like Linux) Out-of-order In-order GPU FPGA Open Source IP PLLs, I/O, Standard Cells, DRAM Controllers…

slide-29
SLIDE 29

Emerging open source projects

Pr Processo ssors IS ISA: RISC-V

In In-or

  • rder: Rocket, Pulpino, Leon3, OpenRISC

OOO OOO Superscalar: Boom, Fabscalar GP GPU: MIAOW, GPLGPU, Nyuzi Ma Manycore: OpenPiton Mi Microcontroller: OpenMSP430

CA CAD To Tools (imagine if Linux did not have GCC) Ve Verilog to GDS: Qflow Ve Verilog to Gate Level: Yosys La Languages: Chisel, PyMTL, myHDL, … Fr FreePDK15: Standard Cells Mo Motherboards Co Commercial: Facebook OpenCompute Pr Prototyping: UCSD Basejump

slide-30
SLIDE 30

But who will do this work?

We need people who: are idealistic have lots of free time will work for free Who might that be?

slide-31
SLIDE 31

But who will do this work?

We need people who: are idealistic have lots of free time will work for free Who might that be?

Students!

(Remember Linus Torvalds?)

slide-32
SLIDE 32

An Experiment: CSE 190

CS CSE 190: 190: Th The Open Source Hardware Movement with Pr

  • Prof. Mi

Michael Taylor: The open source software movement has blossomed over the last 30 years, and is directly responsible for the current surge in the software industry, where developers can create large startups in which only 5% of the source base is their own code. Recently, the open source hardware movement has been rapidly gaining

  • ground. In this class, we will study the development of the movement,

including progress in open-source processors (RISC-V), open-source GPUs (MIAOW), open-source FPGAs, and open-source libraries (opencores.org). In this class we will brainstorm about this movement, and students will engage in an open source hardware project of their choice to advance the state-of-the- art in open source hardware development. Prerequisites: A+ or A or A- in CSE 141L or ECE 11, or excellent knowledge of SystemVerilog, or Permission of Instructor.

slide-33
SLIDE 33

CSE 190

First month of f class has students presenting on various

  • p
  • pen sou
  • urce proj
  • jects and estimating their impor
  • rtant

an and traje ajectory. St Students then work in teams. To

  • get an A, they needed

to to have changes accepted to to an Open Source Hardware pr

  • project. (“T

“To GIT you must commit!”) ”)

slide-34
SLIDE 34

Teaching

slide-35
SLIDE 35

Research

Have your funded students use and commit to open source HW efforts during their research … instead of “rolling your own” or using your own proprietary stuff (e.g. Raw)

slide-36
SLIDE 36

Basejump: A “Base Class” for Open Source HW

Xilinx Zedboard Click!! Basejump Motherboard Basejump BGA Basejump IP Cores (Clockgen, PLL, IO, …) Basejump TSMC 28nm Flow Basejump Skeleton Your chip Your Application C-core generator (UCSD)

slide-37
SLIDE 37

Basejump: Early Adopters

DARPA CRAFT (16nm) NSF SaTC Large (Crypto) NSF SaTC Med (Security)

slide-38
SLIDE 38

Thanks!

slide-39
SLIDE 39

****