|
Twerk from Home posted: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. It also means that whatever 'secure enclave' or management engine they put in it will defacto include keys from the various state and secret police, since it's a product made in china, for china, paid for by the Chinese government.
|
# ? Oct 27, 2023 23:44 |
|
|
# ? May 30, 2024 12:21 |
|
They’ve been banned from the best parts for like a decade because they let their nuke guys on their 2014 era world-leading intel supercomputer. They’ve got a few different indigenous super architectures. Rumor is they’ve held off from publishing the results from their current top system on the top 500 list to avoid a political shitstorm. This is also why the US is cracking down on asml exporting their best (and next to best) litho machines to China. This is the same reason Japan continues to put billions into building the full tech stack to support their own domestic machines like Fugaku.
|
# ? Oct 28, 2023 00:19 |
|
Why did I only find this thread like less than ~30 minutes ago? Y'all are my people. I'm home.
|
# ? Oct 30, 2023 09:46 |
|
NewFatMike posted:Sifive seem to have fired…nearly everyone https://www.sifive.com/blog/the-road-ahead-- Random gobbldygook PR thing BobHoward posted:wat It's specific to arm3. I'll have to dig it up later today
|
# ? Oct 30, 2023 10:36 |
|
Hadlock posted:It's specific to arm3. I'll have to dig it up later today It's going to be a waste of time. Two universities in Northern California, UC Berkeley and Stanford, started the big RISC revolution in the early 1980s. RISC-V is named the way it is because its creators see it as the "fifth major RISC ISA design from UC Berkeley" (see the footnote at the bottom of the "Introduction" page). ARM got its start eight timezones away on another continent. Its designers faced severe constraints which led them to create something that stood apart from the rest of the 1980s RISCs. Early ARM was such an oddball that it hasn't had much influence on modern ISA designs, including Arm's own: the 64-bit Arm ISA is a nearly clean break from its own past. Do you start to see why I'm reacting like this? The ISAs are observably quite different, nobody sane would even want to crib from ARMv3 (excpet perhaps in extremely narrow ways), the backgrounds of RISC-V's architects wouldn't lead them to take much from ARM, and said architects have confirmed that their work most closely derives from the Northern California RISC tradition.
|
# ? Oct 30, 2023 13:44 |
|
I'm reminded of the MIPS ISA protection where there was like... a barrel shifter instruction that was the only patented one? I'm sorta fuzzy on the details so Bob feel free to jump in here, I thought it was some Loongson ancestor where they stole the R3000 and re-implemented it apart from the shifter.
|
# ? Oct 31, 2023 22:57 |
|
I think I heard basically the same story about Loongson and MIPS, but am equally fuzzy. If wikipedia's article on Loongson is to be trusted they did get some form of official licensing at some point, but are now off doing their own thing again, where "their own thing" still appears to be MIPS with the serial numbers filed off. I've never gotten the impression that Loongson processors are widely used, even in China. It feels like one of those cases where the Chinese government is throwing money at trying to develop a domestic version of a key technology, but the outfit they're pumping money into is maybe a bit more about graft than doing anything effective.
|
# ? Nov 1, 2023 11:32 |
|
Yeah there’s at least like three or four different indigenous cpu / gpu programs and loongson is not one of the more successful ones. I always enjoyed SiCortex, the last gasp of MIPS in the US. https://en.m.wikipedia.org/wiki/SiCortex . Don’t know if they sold any of them, though.
|
# ? Nov 1, 2023 18:25 |
|
BobHoward posted:Do you start to see why I'm reacting like this? The ISAs are observably quite different, nobody sane would even want to crib from ARMv3 (excpet perhaps in extremely narrow ways), the backgrounds of RISC-V's architects wouldn't lead them to take much from ARM, and said architects have confirmed that their work most closely derives from the Northern California RISC tradition. Yeah I definitely get what you're saying Google history thing is running really slow for queries going back more than a year but it relates to the "Amber Core" which is ... I think... an open source implementation of the ARM v2a instruction set. I'll look at it more later. edit: maybe this is what I am misremembering quote:Infrequent RV32G Instructions https://www2.eecs.berkeley.edu/Pubs/TechRpts/2016/EECS-2016-6.pdf edit more: NO it was this https://www.wch-ic.com/products/CH32V003.html quote:Features I think it was their gobbldygook v2a that put me on to the amber thing. Here is my post from the arduino thread where I'm describing the processor and at the very bottom you can see my edit where I descend into ARM v2a madness Hadlock posted:Paid way too much for a RISC-V based, USB-C powered, uh, arduino-esque device: "CH32V003 RISC-V MCU offers 2KB SRAM, 16KB flash" so yeah you're right and i'm very wrong Hadlock fucked around with this message at 07:11 on Nov 2, 2023 |
# ? Nov 2, 2023 06:46 |
|
To be fair ARM naming conventions especially old ones are pretty confusing. ARMv3 (the ISA) was implemented in ARM6 (the concrete processor design) - from 1991. I can believe its out of patent but it's also old AF - used in https://en.wikipedia.org/wiki/Risc_PC for example. If you actually mean ARM3 (the processor design in the Archimedes) the big addition there is a multiply instruction and it still has 26-bit not 32-bit addressing. Meanwhile the Cortex-M0 is a much more recent design that only does Thumb, not classic ARM (so mostly 16 bit instructions, very compact but a bit of a bitch to write a compiler for). It literally can't run ARM6 machine code and vice versa. It''s a low end microcontroller so a v different space from what someone would normally be aiming for with the above (but those specs do suggest 'low end microcontroller'). Oh from that article - 'The only thing it omits from ARMv3 is support for the Thumb instruction set.'. Err. Not sure Thumb was in ARMv3. feedmegin fucked around with this message at 15:23 on Nov 2, 2023 |
# ? Nov 2, 2023 15:12 |
|
hm i have a lemote mini pc with a loongson 2f cpu, does that make me cool
|
# ? Nov 5, 2023 00:44 |
|
Well, I hope y'all are ready to destroy your RISC-V boards because they're going to be illegal pretty soon:quote:Some lawmakers - including two Republican House of Representatives committee chairmen, Republican Senator Marco Rubio and Democratic Senator Mark Warner - are urging Biden's administration to take action regarding RISC-V, citing national security grounds. https://www.reuters.com/technology/us-china-tech-war-risc-v-chip-technology-emerges-new-battleground-2023-10-06/ The cleanest way to make sure that the US doesn't help make RISC-V better for China to use is just to prohibit Americans from working on RISC-V. After all, we've got ARM, which is fully under the thumb of the US government and ready to restrict Chinese access.
|
# ? Nov 6, 2023 20:34 |
|
Is that proposal likely to go anywhere? Seems like a pretty hard case for national security to trump the first amendment for someone writing down how to fold proteins or whatever as an ISA extension, but I guess it’s possible. What was the government’s reaction to the proposal?
|
# ? Nov 6, 2023 20:57 |
|
Of course Mark loving Warner is supporting this stupidity. What with being completely bought by defense contractors and priding himself on being a Moderate Democrat™, he'd support burying an entire standard just to fail at suppressing advancement in another country.
SpaceDrake fucked around with this message at 21:01 on Nov 6, 2023 |
# ? Nov 6, 2023 20:58 |
|
Subjunctive posted:Is that proposal likely to go anywhere? Seems like a pretty hard case for national security to trump the first amendment for someone writing down how to fold proteins or whatever as an ISA extension, but I guess it’s possible. What was the government’s reaction to the proposal? I’d assume it’d be like 40bit encryption which I’m sure is something you’d prefer to forget.
|
# ? Nov 6, 2023 21:24 |
|
ugh
|
# ? Nov 6, 2023 21:24 |
|
For those that didn’t live through it: https://en.wikipedia.org/wiki/Export_of_cryptography_from_the_United_States?wprov=sfti1
|
# ? Nov 6, 2023 21:26 |
|
There's some lobbyist shenanigans going on behind the scenes, probably; Intel bought up a big stake in ARM right before they went public, and investors are trying to protect their investment in ARM My guess is that in 15 years BRIC will run on RISC-V and NATO will be on x86-64/ARM What a stupid slap flight. An ICBM will run on an at mega 328, and fighter jets etc are built in such laughably small numbers anyone can import the thousand or so high end industrial cpu needed for those systems. There's what 850 F-35 jets on the planet? How many gaming laptops can you fit in one suitcase? 20? 10,000 laptops in a single shipping container?
|
# ? Nov 6, 2023 22:25 |
|
hobbesmaster posted:For those that didn’t live through it: https://en.wikipedia.org/wiki/Export_of_cryptography_from_the_United_States?wprov=sfti1
|
# ? Nov 6, 2023 22:25 |
|
https://www.hpcwire.com/2023/11/08/china-deploys-massive-risc-v-server-in-commercial-cloud/ ‘Massive’
|
# ? Nov 8, 2023 20:37 |
|
in a well actually posted:https://www.hpcwire.com/2023/11/08/china-deploys-massive-risc-v-server-in-commercial-cloud/ 48 nodes, that's more than 1 rack full of 1U nodes! Imagine if someday the Chinese were able to afford a dozen racks full of computers.
|
# ? Nov 8, 2023 20:54 |
|
*slaps rack enclosure* You can automate so many lights with this bad boy
|
# ? Nov 8, 2023 20:56 |
|
in a well actually posted:https://www.hpcwire.com/2023/11/08/china-deploys-massive-risc-v-server-in-commercial-cloud/ Some really (really really) crude benchmarks show the sophon sg2042 at about 1/4 to 1/3 the speed of a 16 core, 32 thread amd ryzen 3950x. ZIP decompression isn't a very good test, but it's something. Even 1/10th performance is "good enough" for the vast majority of daily tasks https://forum.sophgo.com/t/risc-v-public-test-platform-released-7-zip-test/263 If that is developed and manufactured 100% in-country then I think the cat is out of the bag on containing China's CPU ambitions. I would expect BRIC are all focused on developing their own microprocessor industries at this point. If we were going to build walls to protect this technology, we should have been doing it in the 1980s. We're about 40 years too late. 48 1U systems isn't a lot, but like everything in the semiconductor world, once you get your yield up, you can print chips forever on that process very reliably
|
# ? Nov 8, 2023 21:16 |
|
Twerk from Home posted:48 nodes, that's more than 1 rack full of 1U nodes! 12U of Bergamo.
|
# ? Nov 8, 2023 23:44 |
|
Twerk from Home posted:Well, I hope y'all are ready to destroy your RISC-V boards because they're going to be illegal pretty soon: Communist China AKA UC BERKELEY !!!!!!
|
# ? Nov 9, 2023 04:46 |
|
POSIX with Chinese Characteristics
|
# ? Nov 10, 2023 09:17 |
|
Nordic Semiconductor is getting into RISC-V https://blog.nordicsemi.com/getconnected/why-nordic-is-getting-involved-in-risc-v Published a week ago
|
# ? Nov 10, 2023 23:46 |
|
Did I miss some major developments, or is the author of this architecture book perhaps unreasonably generous in his assessment of RISC-V’s market success?
|
# ? Nov 10, 2023 23:54 |
|
If by inroads, they mean someone is talking about it in those cases? Sure. The euros have been talking about building a leadership class super using one since shortly after arm got acquired. Actual deployments? lol, lmao
|
# ? Nov 11, 2023 00:01 |
|
I mean “begun to make” can do a lot of work but it feels like RISC-V is still trying to catch Itanium in terms of actual production usage.
|
# ? Nov 11, 2023 00:03 |
|
espressif are using it in all their new designs but they've always been ISA contrarians, previously they used xtensa
|
# ? Nov 11, 2023 00:10 |
|
what we really need are open source VHDL for a T805 Transputer, C004 link switch, and C011 link interface
|
# ? Nov 13, 2023 00:09 |
|
Subjunctive posted:Did I miss some major developments, or is the author of this architecture book perhaps unreasonably generous in his assessment of RISC-V’s market success? The security processor in google’s pixel probably means they can count all of those by vague technicality.
|
# ? Nov 13, 2023 04:28 |
|
likewise nvidia is switching to RISC-V for their GPUs management engine so technically it will be used for ML etc, though not actually doing any heavy lifting
|
# ? Nov 13, 2023 12:18 |
|
Subjunctive posted:I mean “begun to make” can do a lot of work but it feels like RISC-V is still trying to catch Itanium in terms of actual production usage. That's about where I am, I think. Lots of people are talking about it, not many people are shipping it. You can't go on Mouser and get a bunch of RISC-V chips. There's some SBCs I guess but nowhere near ARM.
|
# ? Nov 13, 2023 16:16 |
|
In other news Raptor Computing - they of the open source and increasingly inaccessibly priced Talos II and Blackbird motherboards for IBM Power9 chips - has announced that their next boards will use Solid Silicon's S1 Power ISA v3.1 CPUs. Details on who Solid is are light, but it appears the S1 is essentially a version of Power10 with the controversial closed-source bits replaced with new open ones. More info here. I'm happy enough with my Power9 that I'm skipping this one, but if there's any thread on these dead gay forums for news about OpenPower, I figured this was it. also i'm finally going to get a mac studio yay
|
# ? Nov 14, 2023 03:09 |
|
Dave Plummer did a video about current IBM mainframes, a lot of it is probably already well-known by folks here but some interesting footage at 12:25 of the physical vibration and shock testing done to them. https://www.youtube.com/watch?v=ouAG4vXFORc Also check out the comments section if you want to see current IBM employees buzzing about seeing their work represented and former mainframe programmers getting nostalgic.
|
# ? Nov 18, 2023 22:29 |
|
https://chipsandcheese.com/2023/11/20/chinas-newish-sw26010-pro-supercomputer-at-sc23/ The things that made TaihuLight hard to use? We’ve doubled down on them! (A correction: it isn’t on the main top500; hpl-mxp is a separate, mixed precision benchmark.)
|
# ? Nov 26, 2023 16:14 |
|
I guess this would be the right place for Zhaoxin discussion even though its VIA's x86 license? The long rumored and leaked ZX-F / KX-7000 is now official: https://www.zhaoxin.com/news_view.aspx?nid=2&typeid=273&id=1625 I imagine they are reusing already existing Intel packaging for cost savings. Why reinvent the wheel when you can copy/use something that already exists, the Chinese way! The performance target is apparently a Ryzen 7 1700. Supposedly this is going to be on a 7nm proccess, and it looks like they have finally ditched the ancient S3 GPU they have integrated into their last few products, as its boasting DX12 and h265 decode.
|
# ? Dec 12, 2023 23:42 |
|
|
# ? May 30, 2024 12:21 |
|
Interestingly enough, no SMT. Machine translation of the spec sheet:
|
# ? Dec 13, 2023 02:51 |