[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vr / w / wg] [i / ic] [r9k / s4s / vip / qa] [cm / hm / lgbt / y] [3 / aco / adv / an / asp / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / qst / sci / soc / sp / tg / toy / trv / tv / vp / wsg / wsr / x] [Settings] [Search] [Home]
Board
Settings Home
/g/ - Technology


Thread archived.
You cannot reply anymore.



File: RISCV.png (28 KB, 512x512)
28 KB
28 KB PNG
Support RISC-V
https://riscv.org/
https://www.electronicdesign.com/embedded-revolution/11-myths-about-risc-v-isa
https://www.sifive.com/
https://www.youtube.com/watch?v=L8jqGOgCy5M
>>
>>70968609
>Support RISC-V
It's much slower than x86. It's much slower than ARM. It's not particularly energy efficient either.
>>
RISC-V has even worse fragmentation problems than ARM.
Fuck off, shill.
>>
>>70968715
>>70968729
>t. CIAniggers
>>
>Linux is a working and open source operating system BUT IT FUCKING SUCKS
Yeah that's what happened in the 90s wasn't it? Everyone just gave up because it wasn't immediately as good as Windows.
>>
>>70968609
MIPS is open source now and 100x better.
>>
>>70968851
prove it
>>
>>70968609
x86 is fine.
Intel, however... it turns out all the performance gains they developed were at the cost of security.
>>
>>70968851
MIPS cannot be used in desktop pc's
>prove me wrong
you cant
>>
>>70969196
risc-v neither.
>>
>>70968752
You can't provide any counter arguments.

Fuck u
>>
>>70969226
I can prove you wrong by showing you video of RISC-V pc playing doom
>fuck you faggot go search it yourself
>>
>>70969309
no
>>
>>70968715
>It's much slower than x86. It's much slower than ARM. It's not particularly energy efficient either.
Not inherently, just in current implementations.
We don't judge x86's current performance off of the original 8086.
>>
File: Elbrus 2000 summary.png (764 KB, 1118x6281)
764 KB
764 KB PNG
sumimasen, Elbrus-shill desu.
>>
>>70969540
>>
File: Elbrus-8SV.png (496 KB, 891x1519)
496 KB
496 KB PNG
>>70969568
>>
>>70969488
what about x86_64? would it be faster than that?
>>
>>
>>70968729
Cite?
>>
File: server Elbrus.png (452 KB, 1599x899)
452 KB
452 KB PNG
Performance:

https://translate.google.com/translate?hl=en&sl=ru&tl=en&u=https%3A%2F%2Fhabr.com%2Fcompany%2Fsmartengines%2Fblog%2F317672%2F

https://translate.google.com/translate?hl=en&sl=ru&tl=en&u=https%3A%2F%2Fhabrahabr.ru%2Fcompany%2Fsmartengines%2Fblog%2F345758%2F&sandbox=1
>>
>>70969645
>habr
>>
>>
>>70969582
> an instruction set is an architecture
>>
>>70969599

Yeah yeah, very original, but has anyone on here even seen one with their own eyes? It's near-impossible to get one of these even in russia
>>
>>70968609
No thanks, I'll be waiting for POWER10
>>
>>70969666
Is it a good site for russian speakers? Or r*ddit?
>>
>>70969854
based
>>
>>70969854
Based and IBM pilled
>>
>>70969853
The lower core count CPUs will be for the public and will become available in a few years. MCST also already published their own linux distro for it. It is developing slowly but surely.
>>
>>70969853
Well on the software side the sources and all the compiled e2k binaries are publicly available from MCST Volga.
>>
File: mbe8c-pc_v5_top_1024x1024.jpg (318 KB, 1024x1024)
318 KB
318 KB JPG
https://www.youtube.com/watch?v=BMvnuvCcpOE
>>
There is a company that made software to allow multiple desktop setups on one Elbrus computer. This is a large cost saving for offices.
>>
https://kongoucheats.com/vid/OTQzNzQ2OTAz
>>
>>70969860
I would not say it good for any users, only for retarded ones. I read hackernews.
>>
>>70969860
http://tehnoomsk.ru/
use google translate.
>>
>>70968609
Based. I hope the x86 botnet dies asap
>>
>>70968851
>MIPS
>better than RISC-V
No
>>
>>70969935
What about cost? Importing from Russia is very complicated and expensive.
>>
>>70969540
>vliw
In what way hasn't Itanic been over this already?
>>
>>70968851
There is literally no single way in which MIPS is better than RISC-V. RISC-V, on the other hand, has multiple benefits over MIPS:
>no speshul registers for multiplication
>no dedicated link register
>more flexible branch instructions
>actually supports position-independent code in a non-retarded way
>better compression
>made to be extensible
>better immediate encodings
>will support 128-bit word size when necessary

Not to mention RISC-V has the future V extension, which will be the greatest revolution in computer architecture for the past few decades. Say good bye to GPUs.
>>
>>70970528
>Not to mention RISC-V has the future V extension, which will be the greatest revolution in computer architecture for the past few decades. Say good bye to GPUs.
>Say good bye to GPUs.
Red pill me on this.
>>
>>70970545
RV-V does everything GPUs do (vector code), only way better:
>polymorphic instructions
>variable-size register files, scaling from very small to very big implementations
>more extensible for more vector types, and even 2d- and 3d-matrices (like nvidia's tensor operations, only more regular)
But it only does it on a more CPU-like manner, using the same basic integer ISA for control flow, and being able to handle page faults. Both of those are huge. Using the same integer ISA means that one OS image can run both CPU-like (latency-optimized) RV implementations and GPU-like (throughput-optimized) RV implementations at the same time, in a completely orthogonal manner; compiling shader code will be the same as compiling any ordinary CPU code. Being able to handle page faults remedies one of the largest current drawbacks of GPUs, especially in the HPC space, namely their inability to handle datasets larger than the VRAM attached to a single GPU.

I fully expect RV-V to revolutionize HPC (and similar things, like movie ray-tracing) in the first place, and then trickle down from there for real-time graphics rendering on the desktop. Meanwhile, royalty-free RV-I will eat ARM alive in the µC space, and work its way up from there to desktops.
>>
Nothing will beat x86
It's everywhere and replacing it would be a nightmare. And x86 is way more efficient than any open source implementation
>>
>>70970611
>And x86 is way more efficient than any open source implementation
Seeing as how high-end x86 implementations aren't really improving, though, RISC-V has ample opportunity to catch up.
>>
>>70970598
Holy shit this sounds awesome! No more having to deal with shit like Nvidia/AMD drivers, and way better performance in the long run. I'm sold.
>>
>>70970652
I fully expect the V extension to be the killer feature of RV.
>>
>>70970611
This is literally only a problem for Windows. Linux, the BSDs, and even Darwin can be retargeted at a new ISA given like a year of bootstrapping. This has happened already for RISC-V.
>>
>>70970611
Replacing it honestly isn't that hard in theory, and partly in practice. In theory, all one would have to do is recompile and maybe to some extent alter programs to work with the new architecture, no different from an x86 OS deciding to support ARM or something. In practice, this has already happened. As stated in the video in the OP, most of Debian's packages are fully usable, and as the video is a tad old in the grand scheme of things, support has likely improved since then. Legit the only thing that holds us back is big fatass proprietary Windows, and all of its closed-source applications. MacOS is in a similar position as well.

Really it seems as though GNU/Linux is the based, sane, and flexible adult, while Windows and MacOS are like two drooling children with downs syndrome fucking everything up.
>>
>>70968609
There are products you can buy that use power9. You can buy a TalosII.
Where are the risc-v desktops?
I want to have a few servers with both bit right now risc-v is only used in micro controllers and sbc's.
>>
>>70969854
>>70970722
Reminder that OpenPOWER is only """Open""" to business partners.
>IBM is using the word "open" to describe this project in three ways:

>They are licensing the microprocessor technology openly to its partners. They are sharing the blueprints to their hardware and software to their partners, so they can hire IBM or other companies to manufacture processors or other related chips.
>They will collaborate openly in an open-collaboration business model where participants share technologies and innovations with each other.
>Advantages via open-source software such as the Linux operating system.
>>
>>70970760
The power9 isa yes.
Firmwares and the board schematics are open from raptor computing tho and I've heard talk that the power9 microcode will be opensourced soon.
From what is currently available with open hardware though the only option is really open sparc and using minimalistic software to close the performance gap.
I'm psyched for a high performance risc-v system but right now it doesn't exist.
>>
>>70968609
>buy my $1,000 dev boards and hype my kikestarter vaporware goy
Fuck off shill.
>>
>>70970934
Make it by your own. Instruction set is open n shit
>>
>>70970951
This thread is full of fucking idiots.
>>
>>70970934
Poorfags like you can just use QEMU for development.
>>
>>70970965
Thanks for acknowledging my existence
>>
>>70971002
Or we'll just continue using proper architectures that exist instead.
>>
>>70970720
>Legit the only thing that holds us back is big fatass proprietary Windows, and all of its closed-source applications.
I hear this a lot from webshit zoomers who forgot (or never knew) that NT used to run on three different major RISC platforms that were both widely used in the industry and actually more powerful than PCs, and it still flopped.

RISC-V has none of the clout, power or support that those systems had. Just a trendy moral narrative that works screenfetch enthusiasts into a frothy lather and some logos of cloud services megacorporations that mostly just want to shove it in their Internet of Shit toys at best while the likes of x86, ARM and POWER continue doing the real computing back at the office.

There's nothing yet to indicate this shit actually has any future until someone actually does something other than talk about it. And even when that happens, I bet you'll still gawk at the price tag since most of the fags faunching at the bit for a new architecture are either cheap-ass nerds who just want a new toy to gather dust in their drawer or consumers who think the Free Market Competition(TM) it will supposedly generate will save them money, probably on the Intel/AMD products they were already planning to continue buying while expecting the former group to pick up the slack for them.

I really like the idea of RISC-V as the start of a definitive free and open platform, but these kinds of threads are always full of such inane comments, overhyping and outright cock sucking.
>>
>>70968609
That's why I use POWER and UltraSPARC.
Fuck off with your retarded soiboi zoomer scams, this is a technology bo
>>
>>70968729
>fragmentation problems
Java related shit
>>
File: LOW_POWER_CPU.png (18 KB, 722x534)
18 KB
18 KB PNG
check this up
>>
It seems like it's getting there. Whatever it takes to free me from the espionage.
>>
>>70968835
It would help if it didn't take hours of workarounds and googling error codes to get anything working.
>>
A 45nm 1.3GHz 16.7 double-precision GFLOPS/W RISC-V processor with vector accelerators
https://ieeexplore.ieee.org/abstract/document/6942056
>>
>>70971171
>Sign in to Continue Reading
but yeah congrats on somewhat making it to 2005, this changes everything!
>>
What about PowerPC? There's affordable desktop kits available by Raptor.
>>
>>70971827
see: >>70970760
>>
>>70971083
>NT on anything but x86
>ever
Microsoft has NEVER been able to make this work right. The 90s RISC ports were a complete meme that nobody used seriously.
>>
>>70972214
NT started on the i860 architecture, and was designed from the ground up to be readily portable. It was the RISC machines themselves that were failures and "memes" that nobody wanted because they were overpriced shit that delivered nothing to justify their price tag.

Don't forget their fucking flop on ARM too, remember all the hype for RT round 2 last year that practically evaporated over night? That shit even had x86 emulation like NT on Alpha used to, and still nobody bought that garbage, because it was pointless and processor architecture is little more than a sticker on the case 99.9% of the time in purely practical terms.
>>
>>70972301
>NT started on the i860 architecture, and was designed from the ground up to be readily portable. It was the RISC machines themselves that were failures and "memes" that nobody wanted because Microsoft's singular value proposition for Windows is "it might be a turd but it'll run all your crappy, buggy, old x86 software"
FTFY. Every single time Microsoft steps outside its little bubble of x86 desktops and laptops, a unix like OS comes along and rapes it in the ass. Servers, embedded, mobile, HPC, everywhere. Hell, NT only took off as a business desktop OS because NT4 could play Win95 games so people could slack off at work without dual booting.
>>
>>70972503
>hhhuuuurrrfff duuuurrrrffff MUH GAMES!
You're a fucking idiot. Stop projecting.
>>
>>70972552
I'm entirely serious. This was a deliberate move by Microsoft to get people locked in to Windows. I remember playing AoE 2 on NT4.
>>
>>70972839
I know you're entirely serious, that's why I called you a fucking idiot.
>>
>>70968609
>literally stands for Reduced Instruction Set
Haven't the old x86 patents expired by now? New companies can and will enter the game, especially as China improves its tech sector
>>
>>70972887
They surely can enter the game, but nobody wants a Chinese 8086 or 80386 clone. That shit would only be worthwhile in embedded systems, and x86's day in that sector came and went about 25 years ago.
>>
File: high cat.webm (1.76 MB, 424x453)
1.76 MB
1.76 MB WEBM
>>70970528
>Say good bye to GPUs.
>unironically thinking the nvidia overlords would allow this to happen
lmao freeturds
>>
>>70968609
>more vulnerabilities

just because intel if a fucking mess doesnt mean jack shit
>>
>>70968609
Indeed, I'm using a POWER based machine as a main and a ARM server board.
>>
>>70970817
Are there any blob free GPUs?
>>
>>70969832
ISA - Instruction Set Architecture
>>
Fucking idiots in this thread thinking RV won't go anywhere because it's not performance-competitive with x86 yet, ignoring the fact that it's about to fucking WRECK the FPGA soft processor / embedded systems market.
>>
>>70972552
tbqh the decades worth of x86 windows games are a reason why x86 won't die any time soon
>>
>>70974390
this, and even current consoles are x86 right?
>>
>>70968851
It isn’t, you’ve to register to mipsopen first.
>>
Ultrasparc verilog is released under gpl. If someone pony up the money to tape them out they could have gpl cpus running gpl software.
>>
>>70969935
>trading American backdoors for Russian backdoors
>>
>>70969258
>What is the countless backdoors added on intel cpu's by intelligence agencies
>>
>>70973693
>unironically thinking Intel would ever allow AMD to eat into their market share
This is how smart you sound right now.
>>
>>70974390
That would only be a problem for the absolutely latest games, though. For the rest, considering the hardware they were written to run on, they could easily be emulated. DOSBos is already an emulator, even on x86.
>>
>>70968609
>more vulnerabilities
AMD SMT not affected, feelsgoodman
>>
>>70969645
>16 nm
holy shit, intel 10++++ housefirelake btfo
>>
>>70969309
Yeah, well, fuck you, here's a MIPS machine running Quake:
https://youtu.be/jgk-5l8s0RI?t=30
>>
>>70971148
t. tried Linux once in 2003
>>
>>70974390
No doubt, but it’s not the sole reason no matter how much ex-gamer kiddies like to assume it is because that’s the only thing they ever used Windows to do.
>>
File: hqdefault.jpg (62 KB, 480x360)
62 KB
62 KB JPG
>>70968729
no broblem
>>
>open thread
>Ctrl-F "mill"
>zero (0) results
>>
>>70971874
see >>70970817
>>
There seems to be a lot of interest in RISC-V right now, so I'm kind of curious to see where this ends up.
>>
File: Screenshot (997).png (42 KB, 1180x352)
42 KB
42 KB PNG
>>70968609
(I'm running libreboot)
Guess my mobo
>>
The only real problem with x86 is its deep pipeline which can mostly be overcome with thoughtful programming. I don’t see it going away anytime soon.
>>
>>70970425
http://csef.ru/en/nauka-i-obshchestvo/306/sovetskie-korni-proczessora-intel-pentium-4912

Intel can't into development. They stole all the ideas from Russia. It is only natural that their Elbrus-knockoff, the "Itanium", isn't a success.

>>70975910
Elbrus 16nm has more performance and uses less power then any Intel 10++++ CPU. Pic related
https://www.tweaktown.com/news/58180/amd-epyc-64c-128t-dual-cpu-against-intel-56c-112t-cpus/index.html
>>
>RISC-V
>not VLIW-V
Yikes
>>
>>70979810
>literally cannot buy one
Also, did they release the sources for their distro or is it still in the stage of "yeah yeah, we're gonna do it soon"?
>>
>>70969196
> not using ps2 linux as your main system
But really now i think there is some chink company selling mips boards.
>>
>>70978053
Saw it already and replied see >>70970817
I know the hardware for power9 is closed source but the firmwares are not and raptor provides board schematics.
You cant even buy a riscv system yet so *right now* it's the best alternative.
>>
>>70979943
https://mcst-volga.ru/en/download-repo/
https://translate.google.com/translate?hl=&sl=ru&tl=en&u=http%3A%2F%2Fwww.mcst.ru%2F
>>
>>70980177
And if ibm open sources the ucode that's almost as good as open sourcing the actual hardware.
Not to mention that riscv has an open sourced isa but that doesn't mean hardware implementations will also be open sourced
>>
File: ssr-2019-05-16_09.57.58.webm (2.13 MB, 1680x1050)
2.13 MB
2.13 MB WEBM
>>70978780
I can't even guess your spectre-meldown-checker version.



Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.