|
Oh neat, we have a dedicated RISC-V thread now  I thought this was interesting, looks like some chinese company is going to be selling a full "PC" with an AMD GPU https://www.tomshardware.com/news/milk-v-offers-a-trio-of-risc-v-raspberry-pi-alternatives  quote:If you don't want to spec your own Pioneer, then the Pioneer Box is a ready-to-use RISC-V PC that provides between 32 and 128GB of RAM, a 1TB NVMe SSD, an Intel X520-DA2 network card, and two 10Gbps SFP ports. GPU duties fall to the AMD R5 230 — not a GPU on our list of best GPUs, but it'll get the job done. Power is provided by an MSI A350 350W PSU. Very interested to see where this goes. Not sure how much of this mfg in china, or if they're getting the chips from taiwan
|
#
?
May 25, 2023 02:35
|
|
|
|
| # ? May 28, 2024 08:25 |
|
|
Commercial products that aren't hobbyist/dev boards beginning starting to roll in. This one looks like one of those 2x2x.75" micro routers. The article says the company is trying to convince someone to port openwrt to it https://liliputing.com/this-2-inch-square-board-is-a-dual-port-ethernet-router-with-a-risc-v-processor/ Pretty minor but I'll take it for now
|
|
#
?
Jul 5, 2023 17:20
|
|
|
Why don’t they pay someone to port OpenWRT to it?
|
|
#
?
Jul 6, 2023 11:28
|
|
|
It doesn’t sound like it exists yet.
|
|
#
?
Jul 6, 2023 14:25
|
|
|
Did SiFive give up the ghost or something, because it seems like they actually had enough of a technical foothold to be able to poo poo out plans for RISC-V cores for everyone and their grandma on request five years ago but nobody has come up with a realistically useful RISC-V based appliance board yet. Two GigE ports on some mysterious prototype from Shenzhen is better than what we've seen until now but it's still basically loving nothing. We're a year out from PC Engines shutting down because they can't get parts for their frankly archaic GX-412TC based board but at least that could do 8Gbps of firewalled throughput for the 4x1GbE ports on the board just fine. Since they're giving up the ghost and the only apparent successor is fly by night Aliexpress brands making GBS threads out unconfirmed spec-adjacent RV64 boards, I have no idea what the plan is from here on out for low cost open source network edge machines. And as someone who is a network person professionally and an OSS network person personally, that concerns me. Kazinsal fucked around with this message at 11:22 on Jul 8, 2023 |
|
#
?
Jul 8, 2023 11:19
|
|
|
Kazinsal posted:Did SiFive give up the ghost or something, because it seems like they actually had enough of a technical foothold to be able to poo poo out plans for RISC-V cores for everyone and their grandma on request five years ago but nobody has come up with a realistically useful RISC-V based appliance board yet. SiFive's website is up and their corp twitter account has posted in the last few days, they seem to have a pulse. I think you expected way too much if you thought SiFive alone was going to make RISC-V a huge thing in under a decade (or maybe ever). It's a company led by the same academics broadly responsible for the flawed (imo) RISC-V ISA spec. I have no doubt they can do real things, and possibly even good things, but competing head to head with relative giant Arm Holdings was and is a tall order. I'm not just hot taking on the known deficiencies of the ISA and the tendency of academics to underestimate how difficult it is to ship things in the commercial world. It's just very hard to attack a successful incumbent in this space. Hardened Arm IP cores are available on every process node anyone wants to use, they're all debugged to hell and back, Arm's suite of non-CPU IP is extensive (which helps people bolt together SoCs fast), design technical support is excellent, and the software ecosystem is far more diverse and mature. If you're a SoC design house looking at which to use for an upcoming project, the extra cost of going with Arm buys you a lot less engineering risk, more options, and demonstrably better cores. Only thing RISC-V has going for it is price. Sometimes that's enough to win, sometimes it's not...
|
|
#
?
Jul 9, 2023 10:48
|
|
|
I wonder how RISC-V is doing in the embedded space, they have an opening if Arm continues to try to squeeze folks on licensing their cores. I’ve worked at a couple places where the next product was scheduled to have a risc-v core in it but it ended up going with arm so there must be some kind of pricing shenanigans arm plays when it looks like they are gonna lose out.
|
|
#
?
Jul 9, 2023 23:57
|
|
|
ARM has a deeply entrenched ecosystem advantage. I have zero doubt RISC-V could shake things up on the lower end/embedded sector, but some of the proclamations made have sounded a little cargo cult-ish. I’d like to think OpenPower could get a shot in the arm between Power11 down the line and LibreSOC in the next year or two, especially with the SIMD improvements cited as part of SVP64, but we will see. I have an eight core Power9 in a Blackbird motherboard, so I am a little biased in the ecosystem’s favor. Also, hearing lately about Loongson dusting off old/deprecated MIPS64 code, barely changing it, and submitting it to OSS projects with predictably bewildered responses has me all curious about where that’s going. Hasturtium fucked around with this message at 00:27 on Jul 10, 2023 |
|
#
?
Jul 10, 2023 00:08
|
|
|
It’s fun seeing MIPS just popping back up again! One of the places I worked they had MIPS cores for the longest time until moving to an Arm core (with risc-v in the serdes, interestingly) but then other projects that were already in flight would come up with mips32 on there so you’d have to have quite the array of probes to talk to everything at bringup 
|
|
#
?
Jul 10, 2023 00:16
|
|
|
Every fpga dev I know has been talking about riscv for years. I think it’ll take hold in soft cores and fpga focused SoCs, even if it fails to take off in embedded more broadly.
|
|
#
?
Jul 10, 2023 03:09
|
|
|
BobHoward posted:SiFive's website is up and their corp twitter account has posted in the last few days, they seem to have a pulse. What are the problems with the ISA
|
|
#
?
Jul 10, 2023 19:14
|
|
|
I can see china (and russia? if they ever develop a modern IC industry) going all in on RISC-V and being a leader in that space, if only to fully divorce themselves from western x86-64 and ARM, which seems like a very traditional isolationist china thing to do. I have already bought two RISC-V boards (one arduino style, M0(?) instruction set compatible, one raspberry pi style SOC) from China, no idea what country they were fabbed in. The dev pipelines for RISC-V are pretty ok already, even rust has decent support for embedded riscv, and in ten years sure seems like it'll be a good option, especially if it's a couple % cheaper.
|
|
#
?
Jul 10, 2023 19:41
|
|
|
https://www.phoronix.com/news/Debian-Official-RISC-Vquote:Debian 13 "Trixie" has been aiming for official RISC-V support and indeed it will happen: RISC-V has now been promoted to an official Debian CPU architecture. While long available as a Debian port, as of this weekend RISC-V 64-bit is now considered an official Debian architecture.
|
|
#
?
Jul 24, 2023 06:21
|
|
|
priznat posted:I wonder how RISC-V is doing in the embedded space, they have an opening if Arm continues to try to squeeze folks on licensing their cores. Well, yes, of course. Also and contrariwise places will threaten to go to RISC-V with no real intention of doing so in order to get concessions out of ARM. ARM's whole thing is to squeeze people juuuuuust enough and no more that it's worth going with them for the infrastructure/ecosystem advantage rather than trying something more out of left field.
|
|
#
?
Jul 24, 2023 16:38
|
|
|
It looks like a number of motherboards with Loongson CPUs are appearing on AliExpress and the like at sub-$400 price points. With the knowledge that the CPU is not performance-competitive with x86 and subject to various limitations and frustrations endemic to being outside of more widespread ISAs, what are these things supposed to be like?
Hasturtium fucked around with this message at 01:41 on Sep 6, 2023 |
|
#
?
Sep 6, 2023 01:37
|
|
|
VostokProgram posted:What are the problems with the ISA sorry for the delayed reply. Here's a decent writeup someone did a while ago which covers a lot of it: https://gist.github.com/erincandescent/8a10eeeea1918ee4f9d9982f7618ef68 arm64 is a much better designed 64-bit RISC ISA, IMO.
|
|
#
?
Sep 6, 2023 07:27
|
|
|
Hasturtium posted:It looks like a number of motherboards with Loongson CPUs are appearing on AliExpress and the like at sub-$400 price points. With the knowledge that the CPU is not performance-competitive with x86 and subject to various limitations and frustrations endemic to being outside of more widespread ISAs, what are these things supposed to be like? should run NetBSD pretty well
|
|
#
?
Sep 6, 2023 07:36
|
|
|
BobHoward posted:sorry for the delayed reply. Here's a decent writeup someone did a while ago which covers a lot of it: wish rjmccall would collect his RISC-V criticism like this, be good to point folks to
|
|
#
?
Sep 6, 2023 07:37
|
|
|
ploots posted:Every fpga dev I know has been talking about riscv for years. I think it’ll take hold in soft cores and fpga focused SoCs, even if it fails to take off in embedded more broadly. HDD controllers (WD) was one place they were going after it, and I'm sure various SoC designs will consider it vs. dropping an -A9 or -M4/M7 anywhere a core is needed. Didn't NVIDIA at one point have some stupid amount of -A9s in their design? Though, I'm partially convinced those guys just burned die area in the interest of speed to market.
|
|
#
?
Oct 15, 2023 18:54
|
|
|
movax posted:HDD controllers (WD) was one place they were going after it, and I'm sure various SoC designs will consider it vs. dropping an -A9 or -M4/M7 anywhere a core is needed. Didn't NVIDIA at one point have some stupid amount of -A9s in their design? Though, I'm partially convinced those guys just burned die area in the interest of speed to market. High speed PHYs have arm/RISCV cores on each lane for equalization and station keeping so they can really explode the number of cores in an SoC!
|
|
#
?
Oct 15, 2023 19:10
|
|
|
priznat posted:High speed PHYs have arm/RISCV cores on each lane for equalization and station keeping so they can really explode the number of cores in an SoC! 100G+ / PCIe 5.0 type stuff? Do they run FW that's loaded at runtime / are the algorithms now more suited for a CPU implementation / branchiness vs. using a FSM like the lord intended?
|
|
#
?
Oct 15, 2023 19:35
|
|
|
movax posted:100G+ / PCIe 5.0 type stuff? Do they run FW that's loaded at runtime / are the algorithms now more suited for a CPU implementation / branchiness vs. using a FSM like the lord intended? Yeah the really higher speed PCIe is mostly what I'm familiar with them used for, but not everyone does that. It's pretty basic algorithms but it gives them the capability for more tuning options in the wild, and doing periodic checks on the link quality and recentering the eye if needed. These are more for the diagnostics and lane maintenance than the initial link control which is still under hardware. Just having the option to do fixes on the phy stuff after the fact is a huge lifesaver!
|
|
#
?
Oct 15, 2023 20:06
|
|
|
In my experience they only let HW folks design FSM's up to a certain point of complexity then they slap a 486/M3 in there and punt it all to software.
|
|
#
?
Oct 16, 2023 19:15
|
|
|
what we need is a way to trick software people into writing something that looks like “normal” software but which is actually an FSM that can be formally verified
|
|
#
?
Oct 17, 2023 05:26
|
|
|
eschaton posted:what we need is a way to trick software people into writing something that looks like “normal” software but which is actually an FSM that can be formally verified I dream of this... replace software with a FSM that I can formally verify and is deterministic as long as the laws of physics don't change.
|
|
#
?
Oct 17, 2023 05:39
|
|
|
JawnV6 posted:In my experience they only let HW folks design FSM's up to a certain point of complexity then they slap a 486/M3 in there and punt it all to software. Or even a Cortex-M0. Those things are like 1mm square these days, cheap as chips and you can do a lot with 16k of sram or whatever. (Source: did the software side of this two jobs ago, writing, yes, a fairly complicated FSM sitting on an FPGA to orchestrate some hardware)
|
|
#
?
Oct 17, 2023 16:54
|
|
|
feedmegin posted:Or even a Cortex-M0. Those things are like 1mm square these days, cheap as chips and you can do a lot with 16k of sram or whatever. Yeah, cheaper all around. Rather than a senior RTL person and reams of testing, I can slap a team together with junior HW/SW folks and curse software gremlins when it runs over.
|
|
#
?
Oct 17, 2023 20:51
|
|
|
feedmegin posted:Or even a Cortex-M0. Those things are like 1mm square these days, cheap as chips and you can do a lot with 16k of sram or whatever. This but 1mm^2 is a massive overestimate. Arm says M0 needs 0.11mm^2 in a 180nm process (presumably TSMC, 25 years old and still commercially relevant), 0.03mm^2 at 90nm, and 0.008mm^2 at 40nm. That's why modern SoCs end up with dozens of microcontroller-class cores - they're tiny and they reduce the risks involved in letting people like me try to write complex state machines in Verilog. Sometimes it's not even logic designer mistakes, it's that the state machine implements a specification defined by an outside standards body, and they decided to change it a bit after your tapeout. Also, even when you bake a microcontroller's program into mask ROM, it's relatively cheap to patch in a stepping since you only have to change one or two metal layers. Another attraction is that mask ROM is a regular structure that's relatively easy to edit with a FIB machine (basically a write-capable electron microscope), so when you're bringing a chip up with buggy µC firmware you can test changes before finalizing metal layer changes for the next stepping. So many choices in chip design are driven by optimizing the time and cost of fixing bugs found after you get first silicon back rather than optimizing purely for gate count, power, etc.
|
|
#
?
Oct 17, 2023 22:36
|
|
|
Hasturtium posted:It looks like a number of motherboards with Loongson CPUs are appearing on AliExpress and the like at sub-$400 price points. With the knowledge that the CPU is not performance-competitive with x86 and subject to various limitations and frustrations endemic to being outside of more widespread ISAs, what are these things supposed to be like? Isn't Loongarch just MIPS with the serial numbers filed off? MIPS has been around for ages and there's no reason that it couldn't be used in a high performance design but it hasn't anytime recently. I don't know why you'd want to play with a Loongarch CPU when there's so many other exciting things going on.
|
|
#
?
Oct 18, 2023 03:03
|
|
|
Twerk from Home posted:Isn't Loongarch just MIPS with the serial numbers filed off? MIPS has been around for ages and there's no reason that it couldn't be used in a high performance design but it hasn't anytime recently. Yeah, Loongarch is essentially little-endian MIPS64 with some adjustments. I was luridly curious, but by the time I’d be done frankenputering something together, I’d be around halfway to a Mac Studio at Micro Center, and I’d get to deal with an obscure architecture where at least half the documentation is written in Chinese and Debian hasn’t even solidified support. I don’t feel bad for looking into it, but it’s not for me. Rumor has it Raptor Computing is going to make an announcement about an open Power10 motherboard this week, for release next year. I’m happy with my Power9/Blackbird, but look forward to seeing the benchmarks.
|
|
#
?
Oct 18, 2023 03:31
|
|
|
The prices on Raptor Computing stuff is eye-watering these days. The Blackbird bundle today is nearly twice what I paid for it in 2020. I understand that they have a really niche product and they've probably been hit really hard by COVID and all the supply chain fuckery that caused, but at today's prices put buying a second one completely out of the question. I'm not expecting their POWER10 stuff to be any less expensive.
Mr.Radar fucked around with this message at 04:17 on Oct 18, 2023 |
|
#
?
Oct 18, 2023 03:42
|
|
|
Sifive seem to have fired…nearly everyone https://morethanmoore.substack.com/p/the-risk-of-risc-v-whats-going-on
|
|
#
?
Oct 24, 2023 12:19
|
|
|
in a well actually posted:I remain deeply skeptical that companies that are trying to shave fractions of a cent avoiding arm licensing are going to invest enough to make riscv competitive. I’m also skeptical that any sort of common platform or toolchains are going to develop with a wildly varying “do what you want; I’m not a cop” ISA.
|
|
#
?
Oct 24, 2023 12:41
|
|
|
tbf it has worked out alright with LLVM, and with Linux, so I don't think it's insane to think an open ISA could work too. it really does seem like there's efficiency gain when companies collaborate on common infrastructure that supports the thing they actually want to sell
|
|
#
?
Oct 26, 2023 19:25
|
|
|
China seems pretty heavily invested in RISC-V. Building a chip from the ground up with a standard architecture seems like a great way to build a core competency, and then lock all their vendors into building their weapons with domestically supplied RISC-V. That's mostly speculation but uh seems like what they're doing. I can already buy RISC-V powered boards on ebay for under $10, it's not going anywhere. Also most of the RISC-V base instruction set is based on the patented ARM Cortex M0 instruction set... whose patent has expired. There's a lot of base compatibility there. Arm 2-9 are largely extensions of the original M0 set SiFive is a tech company in a historically unfriendly funding period, makes sense to cut things down to the bone, especially as a company heavily invested in R&D, until funding becomes more viable once again
|
|
#
?
Oct 27, 2023 02:13
|
|
|
I hear a lot about China’s investment in RISC-V, and good for them, but what’s up with Loongson? Is the value proposition of undead MIPS64LE with a twist still worth funding and investment?
|
|
#
?
Oct 27, 2023 02:42
|
|
|
Hasturtium posted:what’s up with Loongson? Is the value proposition of undead MIPS64LE with a twist still worth funding and investment? From Wikipedia quote:Application It's probably used in industrial (factory) controller hardware, and like, I dunno, the targeting system on a fighter jet and any long range smart missiles designed in the 1990s, or whatever. For aerospace and military you need parts availability for 30+ years typically
|
|
#
?
Oct 27, 2023 05:19
|
|
|
Hadlock posted:Also most of the RISC-V base instruction set is based on the patented ARM Cortex M0 instruction set... whose patent has expired. wat Where'd you get this insane idea from? Certainly not from reading ISA manuals; there are few points of similarity beyond the bare minimum you'd expect between two ISAs that both get classified as RISCs. (We're talking big differences in some pretty fundamental design choices here - think stuff as basic as "the number of general purpose registers" and "how decide whether to branch".) Also the stuff you wrote about China makes no sense to me.
|
|
#
?
Oct 27, 2023 12:15
|
|
|
Hadlock posted:From Wikipedia I've seen their PRs across plenty of human genetics research tools. Someone is paying for broad scale porting efforts to get all sorts of general purpose computing working well on Loongson Loongarch. It has had pretty good GCC support for a while and looks like they're trying to get Clang/LLCM support too: https://www.phoronix.com/news/LoongArch-Clang-Linux-Builds
|
|
#
?
Oct 27, 2023 15:03
|
|
|
|
| # ? May 28, 2024 08:25 |
|
|
I'm going to speculate some more about Loongson because I think that there's a lot of interesting forces in motion here. Loongarch looks like an emergency backup plan if China loses access to Intel, AMD, and big ARM processors. That might sound unlikely, but the US federal government has been updating the export rules of what is allowed to be exported to China recently, and they've changed the rules to specifically limit computing power. This was a surprise to Nvidia and AMD, who had specifically designed China-market products that were limited to meet the letter of the previous rules, which now leaves them with a bunch of parts designed for the China market that are now illegal to sell to China: https://www.techspot.com/news/100534-nvidia-stock-falls-after-us-government-restricts-export.html. It's not just GPUs. ARM has said that the new Neoverse V cores are too high performance to be legally exported to China: https://www.reuters.com/technology/export-controls-hit-chinas-access-arms-chip-designs-ft-2022-12-14/. Federal lawmakers are now singlehandedly choosing what CPUs and GPUs China is allowed to use, and I think they're expected to basically freeze the performance limit right now, so as Intel / AMD / ARM processors keep getting faster over the next few years, China just plain won't be allowed to purchase those. Nvidia apparently feels pretty burned by the rule change, which may make them less likely to try and offer a China-specific model that conforms to the new rules, just because they might get the rug yanked out from under them again. The current loophole is that "gaming products" are specifically exempt from these requirements even if they exceed the maximum performance limit: https://www.techspot.com/news/100611-nvidia-rtx-4090-looks-have-escaped-china-ban.html. This is breaking news this week, under the text of the rules it looked like the RTX 4090 would be banned in China but at the 11th hour it has been permitted. If the gaming exemption sticks around, we're going to see a hell of a lot of 4090s in China getting put into datacenters, and there would be a market for Nvidia to do a $10,000 "RTX 5090" that is a gaming product in name only. Chips and Cheese have been covering Loongarch, both older and newer products: https://chipsandcheese.com/2023/04/09/loongsons-3a5000-chinas-best-shot/. It looks like current Loongarch cutting edge is somewhere around Nehalem, but at lower clock rates. That's enough computing power to be useful still, and if you can't buy Intel, AMD, or modern ARM they need an option to keep the country running.
|
|
#
?
Oct 27, 2023 18:19
|
|