Movatterモバイル変換

Hardware/Software Co-design:The Coming Golden AgeBryan CantrillOxide Computer Company

The hardware/software divide● The shift to public cloud computing over the last ﬁfteen years hasallowed software and hardware to become disconnected● On the one hand, this can be empowering: a SaaS offering can be builtwith no real understanding of the hardware beneath it● But there’s a risk of taking software-centric thinking too far -- anddrawing the mistaken conclusion that hardware is irrelevant (or worse)● This overshot in thinking is epitomized by Marc Andreessen’s 2011essay, “Why Software is Eating the World”

Revisiting Andreessen● Certainly, the essay makes an important observation on the importanceof software in essentially every domain:

Revisiting Andreessen● And the effect of Moore’s Law + open source + public cloud computinghas indisputably lowered the cost of delivering software:

Revisiting Andreessen● But the essay errs in fetishizing software, mistakenly viewing extantindustries as likely to be disrupted by SaaS alone:

Revisiting Andreessen● Software is important -- but the essay conﬂates software companieswith companies that in fact integrate software and hardware● Companies that Andreessen cited that have thrived -- Amazon, Google,etc. -- have very signiﬁcant hardware components!● Many software-only companies that are cited have disappointed:Zynga, Rovio, Groupon, LivingSocial, Foursquare● Andreessen is dismissive of Apple (up 15X) -- and entirely ignorescompanies like NVIDIA (57X), AMD (14X), or even Intel (3X)!

Revisiting another famous essay

So… Moore’s Law?● In his 1965 paper, there is no Moore’s Law per se — just a bunch ofincredibly astute and prescient observations● The term “Moore’s Law” would be coined by Carver Mead in 1971 aspart of his work on determining ultimate physical limits● Moore updated the law in 1975 to be a doubling of transistor densityevery two years (Denard scaling would be outlined in detail in 1974)● For many years, Moore’s Law could be inferred to be doublings oftransistor density, speed, and economics

Moore’s Law: Good old days?● The 1980s and early 1990s were great for Moore’s Law — so much sothat computers needed a “turbo button” to counteract its effects (!!)● But even in those halcyon years, Moore’s Law was leaving DRAMbehind: memory was becoming denser but no faster● An increasing number of workloads began hitting the memory wall● Caching was necessary but insufﬁcient...

Moore’s Law: The memory wall● By the mid-1990s, it had become clear that symmetric multiprocessingwas the path to deliver throughput on multi-threaded workloads● ...but SMP did nothing for single-threaded performance● Deep pipelining and VLIW were — largely — failed experiments● For single-threaded workloads, microprocessors turned to out-of-orderand speculative execution to hide memory latency● Even in simpler times, scaling with Moore’s Law was a challenge!

Moore’s Law: Architectural shifts● Denard scaling ended in ~2006 due to current leakage…● ...but by then chip multiprocessing was clearly the trajectory● CMP was enhanced by simultaneous multithreading (SMT), whichoffered up to another factor of two on throughput● Thanks to the earlier software work on SMP, CMP/SMT was less of asoftware performance apocalypse than some feared — but more of asecurity apocalypse than anyone anticipated!● And “dark silicon” greatly limits CMP!

Moore’s Law: Deceleration● In August 2018, GlobalFoundries suddenly stopped 7nm development,citing economics -- it was simply too expensive to stay competitive● GlobalFoundries’ departure left TSMC and Samsung on 7nm -- and Intelon 14nm, struggling to get to 10nm● Intel’s Cannon Lake was three years late and an unmitigated disaster --and for Ice Lake/Cascade Lake, Intel is intermixing 14nm and 10nm● Moving to 3nm/5nm requires moving beyond FinFETs to GAAFETs --and to EUV photolithography; new nodes are very expensive!

Aside: Process nodes● You may well wonder: when a process node is “7nm” or “5nm”, whatexactly is seven nanometers or ﬁve nanometers long? (And, um, how bigis a silicon atom anyway?)● Answer to the second question: ~210 picometers!● Answer to the ﬁrst question: nothing! Unbelievably, the name of theprocess node no longer measures anything at all (!!) -- it is merely arough expression of transistor density (and implication of process)● E.g. 7nm ≈ 100MTr/mm2(but there are lots of caveats)

Moore’s Law● Increased transistor density is continuing to be possible, but at a greatlyslowed pace -- and at outsized cost● Economically, Moore’s Law is indisputably ending● But is there another way of looking at it?

Another essay, further back in time...

Wright’s Law● In 1936, Theodore Wright studied the costs of aircraft manufacturing,ﬁnding that the cost dropped with experience● Over time, when volume doubled, unit costs dropped by 10-15%● This phenomenon has been observed in other technological domains● In 2013, Jessika Trancik et al. found Wright’s Law to hold betterpredictive power for transistor cost than Moore’s Law!● Wright’s Law seems to hold, especially for older process nodes

Aside: A contemporary weighs in on Jevons?

Back to computing!● Andreessen’s 2011 piece, while containing some truisms, is overlysoftware-centric and misses hardware’s role entirely● Moore’s Law -- while prescient! -- is indisputably slowing● Wright’s Law, however, may still be holding for transistors -- especiallyat older processing nodes (22nm, 40nm, 90nm, etc.)● The Jevons Paradox has proven again and again to apply to computing:when general purpose computation is cheaper, we ﬁnd more to do● We can expect more computation in more places

Compute everywhere?● More computation doesn’t just mean computers in new places (à la IoT),it means CPUs present where we once thought of components● E.g., open 32-bit CPUs replacing hidden, closed 8-bit microcontrollers● We are already seeing CPUs on the NIC (SmartNIC), CPUs next to ﬂash(e.g., open-channel SSD) and on the spindle (e.g. WD’s SweRV)● New opportunities for hardware/software co-design: keep hardwaresimple and put more sophistication into software and/or soft logic● There are several trends acting as accelerants for this shift...

Open instruction sets● X86 and ARM -- the two market victors -- are both encumbered byhistory and licensing● RISC-V is an attempt to learn from the ISA mistakes of the past, in avessel that is entirely open and -- with open implementations● RISC-V is very promising, but there remain many gaps to close● To succeed, RISC-V must focus as much on the SoC as the ISA -- whileremaining entirely open!

Open FPGAs● FPGA bitstreams have historically been entirely proprietary -- and oneis therefore dependent upon proprietary tools to generate them● The Lattice iCE40 bitstream format was reverse engineered in 2015 byClaire Wolf, and can be entirely synthesized with an open toolchain!● While Xilinx (AMD) and Alterra (Intel) retain proprietary components(e.g., for timing models), newcomers like QuickLogic are entirely open● See, e.g., SymbiFlow, Verilog to Routing (VTR), Yosys, OpenFPGA, andthe (new!) Open Source FPGA Foundation

Open HDLs● Hardware description languages have traditionally been dominated byVerilog and (later) SystemVerilog● Compilers have been historically proprietary -- and the languagesthemselves are error prone● In recent years we have seen a wave of new, open HDLs, e.g.: Chisel,nMigen, Bluespec, SpinalHDL, Mamba (PyMTL 3), HardCaml● Of these, one is particularly noteworthy...

Open HDL: Bluespec● Bluespec is a high-level HDL that takes its inspiration from formalspeciﬁcation languages -- and strongly typed languages like Haskell● Bluespec uses the expressiveness of the language to move away fromindividual signals -- and to atomic rules and interfaces● This allows for the compiler to do the hard work of connecting modulesand proving correctness, greatly reducing veriﬁcation time!● In the words of Oxide engineer Arjen Roodselaar, “Bluespec is toSystemVerilog what Rust is to assembly”

Open HDL: Bluespec● Bluespec was proprietary for 20 years; open sourced in early 2020!● We at Oxide feel that Bluespec is a profoundly transformativetechnology -- but not one that is broadly understood or appreciated!● More details:○ https://github.com/B-Lang-org/Documentation○ https://github.com/B-Lang-org/bsc○ https://github.com/oxidecomputer/cobalt

Open source EDA● Proprietary software has historically dominated EDA…● Open source alternatives have existed for years -- but one in particular,KiCad, has enjoyed sufﬁciently broad sponsorship to close the gaps withprofessional-grade software● The maturity of KiCad coupled with the rise of quick turn PCBmanufacturing/assembly has allowed for astonishing speed:○ From conception to manufacturer in hours○ From manufacturer to shipping board in days

Board economics● Single board computers are very accessible!○ An STM32 Nucleo-144 board with 400 MHz Cortex M7 CPU + 2MB of ﬂash + 1 MB of RAM + all I/O peripherals for less than $30○ A BeagleBone Black -- with 1 GHz Cortex A8 CPU + 4 GB of ﬂash +512 MB DDR3 + HDMI for less than $60!● All documentation available online and without NDA -- and theBeagleBone Black is (nearly) entirely open● The BeagleBone Black can also be used as a logic analyzer via sigrok

Open source ﬁrmware● The software that runs closest to the hardware is increasingly open,with drivers nearly (nearly!) always open● Increasingly, we are seeing the ﬁrmware of unseen parts of the systembecome open as well, viz. the Open Source Firmware Conference● This trend is slower in the 7nm SoCs -- but it’s happening!● However, even in putatively open architectures, there generally stillremains proprietary software in the form of boot ROMs -- and thisproprietary software remains a problem!

Embedded Rust● Rust has proven to be a revolution for systems software: rich typesystem, algebraic types, ownership model allow for fast, correct code● Slightly more surprising has been Rust’s ability to get small -- whichcoupled with its lack of a runtime lets it ﬁt everywhere!● With its safety and expressive power, Rust represents a quantum leapover C -- and without losing performance or sacriﬁcing size● We at Oxide are working on a de novo Rust operating system for theembedded use case that we will (naturally?) open source; stay tuned!

A new Golden Age!● Thanks to Moore’s Law, Wright’s Law and the rise of open source, it iseasier to build hardware than ever before!● We are going to see computers in many more places, posing challengesto us all to develop reliable, secure, high performing systems● Software remains essential, but we must not think of it in isolation; wemust co-design the hardware and the software in our systems!● The systems are open, the communities are welcoming! Let’s build!

Movatterモバイル変換

Change Language

Hardware/software Co-design: The Coming Golden Age

Embed presentation

Recommended

More Related Content

Similar to Hardware/software Co-design: The Coming Golden Age

More from bcantrill

Recently uploaded

Hardware/software Co-design: The Coming Golden Age