HMN admin, Handmade Math contributor, high school robotics mentor/enthusiast, web developer, etc.
I want to hear more about point 1!
However, I wonder about graphical workloads. Is there anything comparable to RISC-V upcoming in the GPU space in terms of openness/simplicity? If not, how feasible is it to get by with just software rendering in the short-term? Regarding funding, do you have any specific potential sources in mind yet?
HMN admin, Handmade Math contributor, high school robotics mentor/enthusiast, web developer, etc.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568
bvisness: | https://handmade.network/forums/jobs-board/t/7186-simple_computer_company#22150 | Handmade.Network | Simple Computer Company | Jobs board | handmade.network Forums | Computers should be simplified (whole stack)There aren't good laptops on the marketNo company see… oliver: thats pretty cool dotbmp: I think "simple computer" is a pretty marketable concept bvisness: yeah, I think Chromebooks are a good example of that dotbmp: to consumers it looks like chromebooks to devs it looks like heaven | https://www.sifive.com/blog/incredibly-scalable-high-performance-risc-v-core-ip | https://www.sifive.com/share.png | Incredibly Scalable High-Performance RISC-V Core IP - SiFive | Introducing the new SiFive U8-Series Core IP hold on how the fuck did they get 7nm manufacturing...? or is this just IP ratchetfreak: or it's a smaller die dotbmp: anyway to sort of continue the conversation from mollyrocket the other day, chromebooks are a good example of not really needing backcompat for new platforms granted chromebooks have the web and thus all of web shit but browsers are a relatively small problem just plop webkit into your platform or cringe chromium bvisness: curious what OS this would be Chromebook gets off easy because everybody knows how to use a web browser dotbmp: not an existing OS bvisness: good luck running software though dotbmp: indeed bvisness: people are gonna want it to be useful dotbmp: my take is that new OS right off the bat is not the correct approach new platform, sure as a portable layer on top of existing OSes and then once you have a market, you build down to the hardware ratchetfreak: but picking an existing OS as the kernel API ties you to that OS dotbmp: not exactly you can abstract over it of course you have to be careful not to let existing OS design decisions influence your own to the greatest degree possible ratchetfreak: that's not how public facing apis work bvisness: I almost feel like you'd have better luck virtualizing the whole thing, but that is not exactly a very Handmade OS approach dotbmp: I mean yeah you could always do an OS on top of a VM | https://urbit.org/ | Urbit - Your personal server | Urbit is a simple, durable system for running a personal server that you own 100% and forever. like this but yeah I'm not super thrilled about that idea bvisness: that urbit thing is some serious galaxy brain action I do actually think there is room for some real innovation in what personal computing looks like given the broad availability of the internet, but I don't ever really want to demote computers to dumb terminals I want to benefit from strong connectivity when it's there, and still have a robust computing experience when it's not dotbmp: I think "home servers" would be a sweet spot Given the tiny computing demands of most households bvisness: that is definitely a thing I'm interested in like, here's a raspberry pi in a slick case that you can plug into your router dotbmp: It would be less expensive to buy a server box you plop in a closet, than two desktops, four laptops, five tablets, whatever Instead, those things could be thin clients You could upgrade it as needed Like, why have a game console... just buy a "console upgrade kit" or whatever and install it to your server And play from any connected display in your house AsafGartner: Home server with physical cartridges per app, so they can bring their own memory/storage/co-processors. So they don't have to compete with other apps that you have installed. dotbmp: Bwahaha bvisness: I don't think the in-home streaming situation is actually good enough in general to support something like that maybe it's just implementation, but I couldn't get my Steam Link to stream well over wifi in the same room as my computer and router until I went to 5GHz that is AsafGartner: Imagine if Google Photos was a cartridge with an SSD and an AI core for automatic tagging. bvisness: maybe just interference in my building, but still dotbmp: Put some USB 4c cables through the walls :^) AsafGartner: No more accidentally letting other people download your videos. dotbmp: Imagine if no Google Photos AsafGartner: But how would you share baby photos? dotbmp: If you have an always-on home server you can just tunnel home on your phone bvisness: I think online photo features are really useful and I wish companies weren't scraping them all for data! dotbmp: Sharing is more of a concern that does involve centralized services (in the absence of a revolutionary distributed system) But it could be much more selective than "online by default" After all people don't want to share most of their photos Well Most photos are taken to be shared, but as far as photos taken "offline" rather than as part of an app And most are not final takes, either If you take 30 photos at a family event you probably want to share 4-5 bvisness: A good photo app is one of the apps I want to make someday dotbmp: And nudes are becoming a massive problem in terms of cloud services Imagine opening your Google Photos and having someone's amateur sex tape on your phone hahaha What a farce bvisness: all the photo services are very proprietary and closed-off and I rue the day when everybody loses all their family photos because Google shut something down dotbmp: The trust is the bigger problem imo If people were actually nervous about these services failing them it would be less bad But we take this stuff for granted bvisness: I want the benefits of my photos being available on all my devices, but I also want to be able to just find a jpg file and email it to somebody things like the Apple photo library (on Mac) are so egregious about this you have to find the export button or whatever ctp: i would bet theres a market for a gitlab-like 'host your own cloud service' type thing for personal email+photos+docs+sharing AsafGartner: | https://owncloud.org/ | ownCloud | ✔ ownCloud - The leading OpenSource Cloud Collaboration Platform. | Access & share your files, calendars, contacts, mail & more from any device; on your terms. bvisness: it has to be easy to set up though, and that's where I think it usually breaks down dotbmp: Well let's take the home server thing Let's say it becomes popular Or even marginally popular Maybe there are 30K people with them in the US They're always-on You keep your photos on them on redundant storage with backups You could have opt-in peer-to-peer backup "coalitions" Just a bunch of people who agree to share some space on their storage for and some CPU time/bandwidth to do distributed redundancy through peer-to-peer streaming bvisness: some home backup programs have done that but they are inevitably frustrating (not because it's inherently frustrating, but because it's made poorly) dotbmp: Right, this would have to be a first-class feature (When I say home server I mean with a new software platform etc) ctp: shouldnt depend on peer-to-peer backups imo. "can anyone seed [picutres of my kids - siterip] im at 99% AsafGartner: I wouldn't mind a regular backup service. You can encrypt your files locally on backblaze for example. dotbmp: @ctp this is a reductive view I'm not talking about using bittorrent you'd opt in to a group of thousands of people doing this ctp: personally I would never depend on a peer to peer system for backups. your backup can be encrypted and stored on normal remote services. bvisness: I don't think I share the same concerns but I also think that I generally wouldn't care about peer-to-peer dotbmp: I literally said You keep your photos on them on redundant storage with backups ctp: has any peer to peer system reached the level of reliability required to use as a backup? not talking about cryptocurrencies dotbmp: these aren't mutually exclusive concepts bvisness: I don't think remote centralized stuff has to go away, especially for the backup case dotbmp: I disagree kind of it's fine most of the time ctp: backups are a reliability problem, not a privacy problem or a centraliztion problem. dotbmp: but when it's bad, it's really bad (see: the Fappening) a combination of local and remote backups is a good balance if your drive fails, you have a backup locally if your house burns down and your server is destroyed, you have a remote backup ctp: nobody objects to remote backups, just peer to peer backups dotbmp: if Company X has some massive Node.js dependency problem that fucks all their backups on shard #113 or whatever you have your local data bvisness: I think peer to peer backups are overly optimistic, especially given the crap state of home internet connections in the US and the social issues of getting enough people on board dotbmp: I don't see how it's not a privacy problem though prevalence of incidents is low but severity is high bvisness: No there are for sure privacy problems, I don’t disagree there dotbmp: backups are a reliability problem, not a privacy problem or a centraliztion problem. responding to this ctp: if you are uploading an encrypted copy of your stuff to a server, they literally cant read it unless your key is broken dotbmp: good thing you have that option with the common services :stuck_out_tongue: bvisness: I wouldn’t want to set up this home server thing only for my bandwidth to be forever dominated by other people’s data You need to be fairly generous to opt into a peer-to-peer scheme ctp: peer to peer backups would a require storage factor (tenfold, etc) as a function of the uptime and drop-in-drop-out rate of the network, and im pretty sure realistically you would require a massive factor. dotbmp: it's not about what's available, it's about what's convenient, and that's Google Photos and iCloud ctp: if remote encrypted backups were a single click in user software, the market would be a reliability-dominated competition. dotbmp: even a single click is too much work people back up their photos by default it has to be a default setting bvisness: My ideal would just be to have a centralized backup scheme that is all encrypted the whole way through, and with multiple providers to choose from ctp: ^ exactly dotbmp: @bvisness the privacy concern means having the files encrypted before being sent bvisness: Yeah I want that Ryan: The value of convenience to consumers is super underestimated dotbmp: It's everything And it's why new software/OS paradigms that market their technical capabilities always fail (Looking at you, Linux) Users don't give a single fuck You can make SuperAwesomeDevOS which is a developer's dream and it's dead in the water for the consumer market if it looks like Windows NT's inbred cousin bvisness: Also if stuff doesn’t run on it dotbmp: Or if you have to use a terminal for anything No bvisness: Extremely that!! dotbmp: All you need is a web browser Really ChromeOS :smile: Ryan: For the average consumer this is true, though if you are targeting certain working markets you need support for existing software e.g. no artist is going to abandon Maya just to use a slightly less awful OS bvisness: I think ChromeOS was a smart move dotbmp: Right, but if you're trying to break into the consumer market going for professionals seems like a bad starting point Ryan: Yeah that's a fair proposition, unless one particular professional space is easier to tackle... Like spreadsheets or financial software or whatever (no data on this, just an example) dotbmp: The entire content creation ecosystem is a much harder conversion than a web browser Or more generally average consumer software Right I think you go for average jane consumer first, and tackle 80% solutions for low-hanging fruit bvisness: Unfortunately the consumer space is also where you have to compete with the big guns dotbmp: So is the professional space It's all big guns bvisness: I guess dotbmp: The professional space is a lot harder to move Lot of sunk cost You're not gonna get the US government on HandmadeOS in the next twenty years Ryan: Not without force... dotbmp: Haha Ryan: #HandmadeCoup dotbmp: HandmadeAutocracy bvisness: I think the home tech space is probably the easier market at the moment dotbmp: I think the "home server" thing overlaps a lot with IoT which is why it's very interesting to me Ryan: It does seem more tractable though significant consideration should probably be afforded to various small business markets dotbmp: IoT is ridiculous Ryan: I actually have my own plan of building a home and having my own personal server, IoT style, so it's an actually smart home And it doesn't rely on anything external dotbmp: Let me flick a switch on my wall which calls out to a server in California which calls back to my $200 hub which turns off my light AsafGartner: You'd have to wrap the home server in something else, like a game console. I don't think people would want to buy a box that doesn't seem to do anything. dotbmp: I already covered game console also, it's not just a home automation server you'd use it to power thin clients Ryan: Maybe media in general. Like if you have a library of movies on the server you'd be able to stream to anywhere in the house dotbmp: most people just need to browse the web, you could power a family of five off an underpowered server Ryan: Or games, Stadia-style, except locally where the speed of light is fast enough, and you can have an actual wire dotbmp: exactly TVs are already just thin clients Ryan: Very interesting idea dotbmp: I mean if you're thinking in terms of grand ideas with minimal consideration for how to get there, we need digital wiring e.g. we should have USB-C plugs in the walls instead of three-prong electric sockets (or a better standard than USB if we're being utopian) Ryan: #HandmadeUtopia dotbmp: Indeed Ryan: #MarchOnIoT dotbmp: Right, so say you have USB-C or Thunderbolt or whatever all through your house Just plug a display into that Couple plugs if it needs extra bandwidth WiFi 6 is looking pretty good for handling multiple devices AsafGartner: I was just going to say, you should use wireless for data. dotbmp: Depending on the thing Not for displays AsafGartner: Have some local wireless tech in the ceiling of every room. dotbmp: But so, maybe you just plug in a bluetooth multiplexer by your couch AsafGartner: Something that doesn't go through walls that well, to avoid interference. dotbmp: And that's how you connect controllers And your TV USB-Cs to the server, which has a gaming bundle AsafGartner: Yeah, you'd only want very few cables. dotbmp: The biggest practical problem obviously is acquiring the economies of scale to make this happen And I'm sure the big boys are working on this already, but in a way that we would find repellant (without the home server) But the cool thing is if the home is wired for data including like, light sockets, you don't need a wifi adapter and an arm chip in every light bulb Ryan: You mean you don't like updating firmware on your lightbulbs? Come on dotbmp: You can have colored LED smart lights that just screw in and work Fucking Kill me Ryan: :smile: I like your thinking dotbmp, let's make it happen dotbmp: Yeah I'd love to be able to work on something like this Ryan: I meant the tech thing Not you being killed dotbmp: Hahaha HA :rofl: I think this would be economically viable for families aside from the upfront costs, which you have to figure, if something like this were to happen it would be a decade-long process to get a substantial segment of the market caught up but if a Chromebook can be bought for $200 how cheap could a thin laptop be Rory: The problem with data over power is noise. There were Ethernet adapters which used the power main. They didn’t work well. Electrical wiring is good for delivering power not for transmission. The wiring is generally low quality. dotbmp: just needs a display, wifi, data/charging plug, keyboard, mouse, and a little microcontroller to power the display @Rory this all being hypothetical it would not use the existing copper wiring so the question would be if you were getting the stuff in the absolute most bulk manner (like giant rolls of cabling or whatever) how cheaply could you wire a house with something like USB while being able to transmit enough power to power fridges and stuff Rory: The challenge is updating building codes for a higher standard 110/220v main power. It’s possible nobody will do it until their hand is forced. dotbmp: hmmm right so part of the economies of scale problem is being able to buy lobbying power :joy: plus yeah if you're the company actually doing the rewiring you have massive legal considerations Rory: Yes. If you can guarantee different standards manufacturers have less risk doing what you say. They know their devices work. dotbmp: but I assume you'd leave that to a distributed assortment of contractors Rory: And the building standards we have today are just happenstance of legacy. Going back to Edison vs Tesla. One way is getting codes changed. Other way is make it so ubiquitous that codes follow. The second isn’t likely these days with physical things. Like, the us still uses imperial measurements for everything. Changing power technologies is star trek level. dotbmp: haha well imperial -> metric would be a bigger task anyway my take is all of this is talk, the natural starting point is a new platform (gotta make the millions somehow :^) ) Ryan: When doing science: "The metric system is a more logically-structured and well-thought out system for defining measurements." All other times: ":flag_us: :eagle: :red_circle: :white_circle: :blue_circle: It's about 1000 feet until the turn" dotbmp: which you'd distribute as a sort of shared runtime for windows/mac/linux/ios/android apps, users wouldn't know they're using this platform then when you have a substantial dev/user base you put it on top of a microkernel and some segment of the market will use your OS, but it will be interoperable with apps using the platform, more importantly so there will be a network-effect-utilizing ease of transition for the most heavily invested users and devs it leaves a bridge back to "legacy" platforms you'd also have to pick your battles it's very unlikely you'd get 80% of graphic design work done on your platform within 10 years or even 20% so you let designers use windows and macos and work on making the content creation bridge better instead rather than trying to make a photoshop/illustrator killer I think a new paradigm is reasonably doable if you focus on a really strong core ecosystem around what users actually care about messaging/social media, calendars and scheduling, interoperability with other devices, media distribution and sharing Ryan: Hmmm that is interesting dotbmp. I see the argument but am worried about abstraction over the underlying ecosystem leaking Like how many paradigms will de facto transfer over, even if we don't want them to? dotbmp: yeah it's a big concern and if you go for a no-compromises position, you're sacrificing a lot of efficiency Ryan: It would maybe be important to understand the vision for the custom hardware (and everything on it that would come later, like an OS) very early so that abstraction over existing infrastructure is done more responsibly dotbmp: yeah, and striking a balance between idealism and practicality like, casey's standardized ring buffer I/O idea is great imo but realistically you have to support USB even if you had the ability to push a new, simplified standard, you need legacy support I think it's probably reasonable to use RISC-V as a target architecture it's not perfect but it's very simple compared to the alternatives, open, about 1/10th the IP cost of ARM, and obviously a big up and comer gaining a lot of investment right now and as long as you have the CPU architecture in mind, that's pretty much most of the hardware consideration in terms of system design is miotatsu here haha cause I would hate to reiterate this stuff on the forums :rofl: bvisness: probably not better start copy-pasting :) dotbmp: :rip:
Re: RISC-V and graphics, no, as of present there is no equivalent to either SoC graphics like we see in smartphones with Tegra or add-on graphics cards like x86. However, of course RISC-V supports PCI-e so while proprietary concerns make this a nightmare, technically drivers for existing GPUs are a distinct possibility - and I'm sure it will happen eventually, probably for non-consumer cards first like Quadro, deep-learning, compute, crypto mining solutions.
It's also perfectly reasonable, in my opinion, to imagine fully programmable RISC-V graphics extensions that introduce "graphics cores" and extended instructions for graphics pipelines. SoC/integrated graphics, but fully programmable. Of course, someone would have to develop the IP, and without action on part of the RISC-V committee, the extensions themselves. RISC-V has vector extensions in the works.
Ultimately I think this would be a forward-looking and entrepreneurial platform, looking at the RISC-V environment 5 or 10 years down the road, as well as seeking to influence the development of that environment toward its own interests. Maybe that's too ambitious, but the entire premise requires more than a little will to power to entertain.
2. There aren't any good ones on the market.
2. There aren't any good ones on the market.
While I tend to agree with that (but I never had a laptop myself), does the target audience agree with that ? And what is the target audience exactly ?
When I talk with friends about how computer or operating systems are not good in my opinion, they generally don't share my views and are "happy-ish" with how things are at the moment. While we agree that there are problems, my friends think most issues are just an annoyance on the moment, and after years of having them it's just normal and inevitable to them.
Solving those problems would be nice for them but not if it means they need to change their habits which would be needed with new computer / OS (I'm guessing here since there is no way of knowing for real). e.g. They would love Microsoft to stop forcing updates and shitty software, but they still want Windows.
In short, I'm not sure that most people want a big change. So defining the target audience and the target audience "needs" should be a early step. Even if the target audience (at first) would be "developer", developer need to ship/test software for Windows, so the laptop would need to allow that.
the product would be so far ahead of the competition that it speaks for itself. It should be one of those things where you sit down and start to use it and just instantly "get it," that you'd need this in your life. No need for marketing, branding, advertising, etc.
It is important to note here that the kind of simplicity we are shooting for here is not some clean-slate that throws away legacy. Simplicity does not have to introduce limitations, done right it EXPANDS capability rather than limits.
This seems overly optimistic to me.
I don't necessarily mean to rebuke you, but where does the simplicity "come from" then? RISC-V is able to be simple because it is a new ISA, in other words because it breaks backwards compatibility. There will surely be some plainly unnecessary cruft in the software and hardware we'd be replacing or paring down, but almost certainly not enough to bring about the revolutionary level of simplification you're counting on without also substantially cutting and/or changing features/behaviors, in my opinion.
Nah, some really basic things go a huge way for this, like booting as fast as possible, a keyboard and touchpad that feel really good and work flawlessly, little details in the design that show that care was really put into every aspect of it, that kind of thing. I think making a laptop FEEL miles ahead of the competition is actually one of the most trivial aspects of it.
we wouldn't be solving the browser situation but there is no reason you can't run chrome on that. The same kind of thing applies to other many things, X11 being the first that springs to my mind. You just pull in community provided nasty-linux-legacy-layer package, install a traditional linux distro, MS Windows, or whatever else and you're off to the races.
For keyboard+trackpad feel macbooks are already pretty much perfect, although the newest ones of course have other hardware problems like the trackbar, and everything being USB-C - But that's a compatibility problem, and again, maintaining compatibility with all the existing standards (old USB, HDMI/displayport, 3.5mm) is exactly where the complexity starts to creep in...
In the end, the hardware itself from a *user's* point of view, doesn't need much simplification. It's what's happening on the inside that could be simplified dramatically, and those simplifications can then result in much better software. But they do need to result in substantially better software, otherwise the end user has no reason to care. And even once much better software is there, the claim that there is "no need for marketing, branding, advertising, etc." is extremely dubious at best.
This is sounding more and more like there is no defined target audience.
If it doesn't run a web browser and a familiar-enough desktop environment out of the box, most laptop users are already out.
Creative professionals will need to continue using whatever platform they're already on, unless a better alternative to their entire toolset is available on the new platform, which sounds like it is not at all on the roadmap. They might be okay with having to manually install windows (which probably isn't happening soon, or at least we can't count on it unless/until microsoft announces risc-v support) or *maybe* linux, but at that point what simplicity are they benefiting from?
Developers and power users would be more willing to adopt a new platform, which I guess makes them the target audience purely by process of elimination, but developers likely still have to ship code to existing platforms (which I thought was one of the ideas behind starting with linux and paring it down, but now it sounds like that's no longer the plan?), so they will need a good reason to migrate. Maybe a rock-solid editor/compiler/debugger ecosystem could be that reason?
From a business perspective I think there needs to be a much more specific plan for what users we're trying to attract and how we would attract them, rather than just handwaving the question by saying users will just intuitively know that it's good - sure that's a good goal, but how exactly will they know, and why exactly will it be good for them? There needs to be a specific concrete reason that users provably actually care about.