From: Joshua Harlan Lifton Date: Wed, 28 Nov 2018 19:56:02 +0000 (-0800) Subject: Merge branch 'master' of ssh://libre-riscv.org:922/crowdsupply X-Git-Url: https://git.libre-soc.org/?p=crowdsupply.git;a=commitdiff_plain;h=5a7d34a644693c7ed2809996df0ad57fa5aec4ff;hp=3f7fcab03dd777e315978302483416af4bcac973 Merge branch 'master' of ssh://libre-riscv.org:922/crowdsupply --- diff --git a/updates/001_2018nov28_why_a_libre_soc.mdwn b/updates/001_2018nov28_why_a_libre_soc.mdwn new file mode 100644 index 0000000..caf011c --- /dev/null +++ b/updates/001_2018nov28_why_a_libre_soc.mdwn @@ -0,0 +1,153 @@ +# Why make a quad-core 64-bit SoC: surely there are enough already? + +So the year is 2018, and there does not exist a single commercial +"System-on-a-Chip" that is capable of running 3D mobile games and +playing 1080p60 video, where the end-user, including commercial +customers, cannot say that they have full control over their +devices. + +Let that sink in for a moment. + +To reiterate: there does not exist, anywhere in the world, one single popular +modern commercial portable tablet, IPTV device, netbook or smartphone, +today - in the year 2018 - where an end-user may download the full and +complete source code of the bootloader, kernel, operating system, +Video Processor library *and* 3D GPU library, *and* all the internal +and external peripherals (a more in-depth analysis was done +[here](https://www.crowdsupply.com/eoma68/micro-desktop/updates/picking-a-processor). +and two years later the situation still has not changed). + +Now, there happen to be some medium to high-end systems based on Intel +processors, from both [ThinkPenguin](https://thinkpenguin.com) +and [Purism](https://puri.sm/products/librem-13), where +both these companies go to the trouble of actually re-flashing the +BIOS, replacing it with LibreBoot or Coreboot. They both also make +sure that the WIFI firmware is libre (ruling out 802.11ac), and they also +specifically do not use an NVIDIA GPU. They also ensure that the +Intel "Management Engine" (known as an NSA backdoor spying co-processor) +is (or may be) disabled. + +However that is the medium to high end, using relatively expensive +power-sucking Intel processors. 15 to 50 watts just for the processor +is not uncommon, here. Everything else - tablets, smartphones, and +most netbooks, use ARM SoCs in order to keep power consumption well +below 10 watts and in some cases below 5 watts, and that's where it +goes to hell in a handbasket. + +It's not specifically ARM's "fault": it's just the way that it goes. +Imagine that you are a new (or even an established) Fabless Semiconductor +Company. Your "job" is to get an integrated all-in-one product out the door +with the minimum cost (where that's going to be at least USD $10m to +$30m), and the least amount of risk. In evaluating the options, you +absolutely want tried-and-tested, proven, risk-free "off-the-shelf" +peripherals (called "hard macros") which you can first test in the +biggest, most horribly-expensive FPGAs you can get hold of, then +when the engineers are happy, throw around a quarter of a million +dollars a pop at a test chip, and, finally, once that's tested +and known to be working, put down a couple of million on production masks. +That's before even actually getting chips manufactured. + +The sums of money involved are so vast that absolutely no Fabless Semi +company will take the risk of using an unknown, unproven design. They +would far rather pay USD $250,000 to an established company to license +a proprietary GPU hard macro, for example, which comes with an associated +proprietary software library, because the company that licenses that GPU +design has had multiple customers successfully tape it out. The same +story goes for the VPU: another USD $100,000 to $200,000 on license fees +is better than spending USD $10m and above, only to find that the chip +doesn't work. DDR3/4 PHY and Controller hard macro licensing: in +excess of USD $1m, even as high as $2m. These are not costs where +you can mess about. + +All of this is because these are *integrated* processors. There is no +separate VPU: it's on-board. There is no separate GPU: it's on-board. +There is no separate "Northbridge" or "Southbridge" IC: it's on-board, +all on the same die as the actual processor, as a way to save both on +space (think mobile phones) and power (driving external pins uses a huge +amount of power, which an "embedded" all-in-one design does not have). +The down-side of this all-in-one approach: one single mistake and +the entire chip, with all the investment up to that point, is junk. + +And the problem is compounded by the fact that Foundries themselves make money +only by selling wafers. They absolutely hate having their time wasted. +If you as a new Fabless Semiconductor company come to them with an design +that fails, and yet you booked a production run because you were expecting +it to succeed, now the slot's cancelled because *your chip failed*, and +they just lost tens to hundreds of millions of dollars worth of business, +if none of the Foundry's other customers happen to be ready with a mask +set and the cash lined up. + +If that happens, do you think that Foundry will ever take your calls again? + +This is just how things are. Most Fabless Semi Companies looking to +compete with other embedded tablet / smartphone / netbook / IPTV / etc. +SoC companies are taking what they can get, doing the best integration +job that they can, getting it out the door and moving on to the next +product. Samsung actually has two separate teams (one internal, +one of them is a third party called Nexell) that produce Samung-badged +SoCs on overlapping cycles, as a way to help reduce the risk. + +The problem for all of these companies is: apart from the increased speed, +they're all using ARM cores, they're all using the same GPU hard macros +licensed from the same handful of companies: there really is absolutely +nothing really significant that differentiates them from each other +(and to be frank, the end-user genuinely doesn't care what the processor +is: they just buy the end-product). + +Along comes RISC-V, which in the same geometry has a power envelope +that is a whopping 40% lower than any ARM or Intel processor available, +today. + +So this is where it gets interesting, given that power consumption is +key to the success of mobile devices. Here in Taiwan, kids as young +as eight carry around a smartphone... oh and a "power bank" that's twice +the size of the phone (anyone who used to have a Nokia 6310i, with +a standby time of over two weeks, will be laughing and crying at the same +time). + +The point of this story is: it's not enough to just go "oh I think I +will design a Libre SoC today", it has to have an actual commercial +hook: it has to have compelling reasons why it will sell. For the +Libre RISC-V SoC, those are threefold: + +* **(A) the power consumption of a RISC-V core is so much lower** + (the technical reason why is down to "Compressed" instructions, + which result in a 25% code reduction, which in turn means that + the L1 Instruction cache can be smaller, and that translates + to a huge - 40% - reduction in power) +* **(B) availability of source code significantly reduces development costs** + This is not an end-user argument, it's one for the OEMs (Original + Equipment Manufacturers). A good example is + [here](http://www.h-online.com/open/news/item/Intel-and-Valve-collaborate-to-develop-open-source-graphics-drivers-1649632.html), + where two completely independent really large companies came together to fix + bugs in their respective 3D codebases, all without requiring NDAs + or lawyers to get involved. +* **(C) Reduced royalties means reduced selling price** + Softbank recently ordered ARM to increase royalties. They believe that + they have the market cornered. What they don't realise is: end-users + don't care what processor is inside. They just want a device that + "does the job". By using libre-licensed hard macros (including + for the main on-board CPU, and the VPU, and the GPU), the royalties + are slashed literally to zero. + +There are additional justifications for going libre, even with the +hard macros for the peripherals: the costs of licensing proprietary +hard macros are enormous. The highest is for DDR3/4, which can come +to around USD $2 million for the PHY and the Controller (per 32-bit +interface). Gigabit Ethernet: USD $50k. USB2: $100k. +USB3: $500k. All of these costs are per interface instance. If as many of +these can be cut as possible, it adds up to a saving of over USD $4 million, +bringing the development cost down to only around USD $6 million. + +Sad to have to say it: being ethical isn't enough. Money talks. +Once that's accepted, it turns out that yes, there's a strategy that +happens to reduce both development cost and end-product cost, oh +and happens to be ethical and gives people back control of their devices +at the same time. + +If this is something you want to help with, join the +[mailing list](http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev) +list and get in touch. If you want to help sponsor the project +or invest in the team, contact me directly by +email. +