India's First CPUs Are Ready For App Development (tomshardware.com) 142
The Indian Institute of Technology (IIT) Madras has released the software development kit (SDK) for its open-source Shakti processor. Shakti is based on the open-source RISC-V instruction set architecture and was funded by the Indian Ministry of Electronics and Information Technology. The institute promised that a development board will also be released soon. Tom's Hardware reports: The RISE group at IIT Madras started working on the Shakti project in 2016 with a plan to release a family of six classes of processors, each serving a different market. The group promised that the reference processors will be competitive with commercial offerings in terms of area, performance and power consumption. Now India, like China and the European Union, are showing interest in designing their own processors, rather than relying on ones designed by U.S. manufacturers. With the release of the Shakti SDK, developers can begin to develop applications for the Shakti processors, even before they're commercialized.
Hail India (Score:1, Insightful)
It's better the keep these tech in own hands. You never know when the yankees will threaten to put your companies into the blacklist.
Re: (Score:1, Insightful)
Moreover, there is much to be said in favour of standard, moderately powerful processors. Today extra speed comes at a significant cost in electrical power and heat - both of which worsen pollution and resource depletion.
Intel has just presented us with a magnificent picture of the terrible damage a company can do to itself by chasing to enthusiastically after performance gains.
Far better to accept chips with adequate speed, low cost and reasonable power and heat characteristics. Most efficiency gains come
Re: (Score:1)
What are you smoking?
Compare the speed of Ruby vs Rust code (Score:3, Insightful)
Compare the speed of execution of some Ruby code versus Rust code that does the same task.
Ruby code tends to have a lot of overhead due to having far too many pointless abstractions. More of the CPU is consumed doing relatively useless work.
Rust code tends to be ultra fast, because it has just the right amount of abstraction. It uses the CPU for real work.
You get a lot more efficient use of a CPU when using Rust instead of Ruby.
Re: Compare the speed of Ruby vs Rust code (Score:1, Interesting)
How the hell is that comment at -1? It perfectly demonstrates what user Archtech was referring to with the "Most efficiency gains come from software." statement, using a practical real world example.
Why are the best comments at /. always wrongly modded down to -1? Why even have a moderation system here if it's so broken that we have to always browse at -1 just to see all of the insightful and informative comments that have been wrongly modded down?
Re: (Score:3)
Re: (Score:1)
The performance of all software matters.
If your cloud service is running thousands of VM instances all using a massively inefficient language like Ruby or Python then you are burning a whole lot more power than id they were using C++, D, Rust etc.
Don't forget folks, using Ruby, Python, etc causes global warming, climate change and the ensuing starvation of millions. How would you like that on your conscience?
Re: (Score:2)
Re: (Score:2)
Doesn't the economics matter more?
A fast CPU is much cheaper than a highly skilled developer's time optimizing software to achieve 10% performance increase.
Re: (Score:2)
Re: (Score:2)
Re: (Score:1)
How the hell is that comment at -1? It perfectly demonstrates what user Archtech was referring to with the "Most efficiency gains come from software." statement, using a practical real world example.
Why are the best comments at /. always wrongly modded down to -1?
because, sometimes, what's insightful isn't "nice". due to a world perspective, my comments that implicitly criticise US-centric "Imperialist"-style world domination are modded down - by americans - because they're not "nice to americans". this just isn't easy to fix. people hear what they want to hear, and drown out what they don't.
Re: (Score:3, Insightful)
It is because you have a groups of moderators that automatically mod down views and people that don't fit their personal views or beliefs.
For the people, it doesn't matter if you are wrong or right you are automatically modded down because in the past you have expressed views, that while correct, don't meet with political views or correctness.
In the past when someone abused their moderation privilege it was revoked. But now there is no effort to correct the system. You just better hope your fans and
Re: (Score:2)
It was a good system, 20 years ago.
Re: (Score:3)
20 years ago it was a good system because the people running /. cared about it. A few jealous moderators like to mod me overrated no matter what my post said. Back then those moderators would have lost their mod privileges.
Re: (Score:2)
What are you smoking?
Probably Cormen, Leiserson, Rivest, and Stein.
Re: (Score:2)
What are you smoking?
It's an illusion. Normally we all write terrible, inefficiency, and bloated software. On the rare occasion where we stop working on marketing gimmicks and design a real solution we amaze ourselves with concise beautiful software. But it's wrong for us to pat ourselves on the back for doing our fucking jobs once in a while.
Re:Hail India (Score:5, Interesting)
Intel has just presented us with a magnificent picture of the terrible damage a company can do to itself by chasing to enthusiastically after performance gains.
you may be fascinated to know that the "mild register renaming" present in the Shakti Core is the same principle utilised in the CDC 6600 Mainframe, from the 1960s: Q-Tables. this capability of the 6600 is one that is not clear or well-understood. except that Professor Kamakoti, who was the lead unit test engineer working for Intel in the 80s and 90s, came up with the Q-Table system used in the Shakti Core independently (of Seymour Cray and James Thornton).
also: do you remember how the AMD Opteron processors from the early 2000s had to be quoted with a higher "equivalent" clock number than their "real" one? this was because Mitch Alsup (the creator of the 68000) was the lead design engineer, and he has made a life-long study of the CDC 6600 and is one of the few people in the world who understands it.
the reason for the higher efficiency of the AMD Opteron was down to the use of unary register encoding (a long bit-string) instead of a binary number representation, within the Dependency "hazard resolution" Matrices. this unary encoding has two highly significant consequences:
(1) the power savings are enormous. where a CAM has to activate binary XOR comparators in *every single row* of a table, an Unary Matrix activates a single AND gate (per row) to do the same job.
(2) multi-issue in an Unary Dependency Matrix is dead simple: you just activate *multiple* of those unary bits, accumulating the dependent registers *in a transitive fashion*. instruction 1 just gets the current dependencies. instruction 2 gets the current instruction dependencies *plus* those of instruction 2. instruction 3 gets the dependencies from 1, 2 and 3.
consequently, Opteron was able to do 50% more instructions per clock, at the same core frequency, despite being on a *lower geometry* than Intel, and they were still power-performant competitive!
this approach is so simple and so power-efficient it's almost comical. Intel on the other hand followed the Tomasulo Algorithm (invented by IBM in the 1970s). its core concept: a (binary) CAM, baked in to the Reorder Buffer. and it can't do multi-issue: you have to add *yet more* complexity around it, to get that.
so, now you know _why_ Intel's designs are so power-inefficient: it's just that they're based on an inferior design concept. Intel *could* do a lot better.
sorry, i've been doing a study of these things, for the Libre RISC-V SoC :) http://libre-riscv.org/3d_gpu/ [libre-riscv.org]
Thanks! (Score:1)
Wow. Thanks for this one. You made my day.
Since this is Slashdot: no, not sarcasm. Candid joy.
Re: (Score:2)
Wow. Thanks for this one. You made my day.
Since this is Slashdot: no, not sarcasm. Candid joy.
:)
Re: (Score:2)
Wow, I learnt something new today. Thanks for this. I also have a question for you regarding power efficient. In the past few years i* series has been siginificantly lower in power consumption as compared to comparable AMD processors. Is this solely due to the process used (10nm etc.,)?
Re: (Score:3)
One-hot encoding for CAMs or matrix schedulers isn't exactly new technology?
Enjoy your posts on comp.arch but this one is mostly baloney: Opteron wasn't able to do 50% more per clock than the concurrent Intel Pentium 4 Northwood and used smaller schedulers to boot. It was a development of the AMD Athlon K7 and the reason it often performed better for legacy code (read: not recompiled for the P4 with the Intel Compiler) was that it didn't penalize some relatively common instructions and instruction patterns
Re: (Score:2)
One-hot encoding for CAMs or matrix schedulers isn't exactly new technology?
it isn't: it just can't be used in a Tomasulo ROB (i tried). and no, matrix schedulers aren't new: they do however seem to be a lost art. one of the reasons for that seems to be that the academic literature on the 6600 criticises scoreboards for not having "precise" capability. as in, cancellation (rollback) "seems not to be possible" leaving the register file "damaged" on an exception, because instructions *after* the exception were "permitted" to write.
this is horseshit.
an augmentation designed by mitc
Re: (Score:2)
Just wanted to say that this is valuable historical information on how it influences CPUs that are in use today. Thanks man.
Re: (Score:2)
Thanks. Amusingly enough, I learned to program on a CDC 6600.
Re: (Score:2)
I have, and I can't say I've ever been a fan of Javascript as a language. Classic example of a language that you "have to" use but would never choose to use, if you had a choice.
Nevertheless, when I read the Spectre paper [spectreattack.com] I had my socks blown off by how out of date I was on compiler tech. Around page 8 of that PDF, look at listings 2 and 3 in that paper. No, really, go look if you don't already know what has
Re: (Score:3)
You have to trust someone at some point. Open source ideals don't work for hardware production because of the free market; hell most open source software doesn't work without government protection (copyright law). You'll do all the R&D and I'll just fab and sell the processor cheaper than you can.
Re: (Score:2)
Re: (Score:2)
Open source works perfectly well with copyright law. What's wrong about it?
Copyright law is enforced by government, hence the claim that open source software doesn't work without gov protection.
Re: (Score:2)
Copyright law is enforced by government, hence the claim that open source software doesn't work without gov protection.
Open Source (Which meant simply that you could get the source and nothing else, which is why Free Software was created) works perfectly well without government protection. Free Software (you can use the source) only works with government protection. But since the problems that copyleft addresses are caused by copyright, the fact that copyleft is powered by copyright is a strength, not a weakness.
Re: (Score:2)
Errr, don't you have that sort of mixed-up/backward? The term "open source," as applied to software, came much later than the term "free software." Many in the free software movement still feel "open source" is a bastardization of what they set out to achieve.
You also seem to be conflating the terms "open source" and "public domain."
Re: (Score:2)
Errr, don't you have that sort of mixed-up/backward? The term "open source," as applied to software, came much later than the term "free software."
https://archive.org/details/UNIX1985
See at about 13 minutes and 50 seconds.
Re: (Score:2)
1. The term used in the video is "open source code" which to me is about as relevant as "open discussion" or "open refrigerator door." "Open source software" did not become a term of art until the late 90s.
2. That video is from 1985, and Richard Stallman founded the Free Software movement in 1983.
3. You still seem mixed up about open source not requiring government protection. You can't have an open source license without the concept of copyright. Hell, even public domain works require government protection
Re: (Score:2)
The same can be said for closed source products or for anything - without government protection I can just resell the software you developed for a cheaper price.
That's only a problem for you, not for users.
You can also still sell support, and since you are the only one with the sources, you should be able to do the best job, right?
Or if you are selling beverages I can just copy your recipe, print the same sticker and use the same bottles, then sell the product at half the price.
That's why the recipe is a trade secret. A business can go so far as to have a physically separated factory which prevents anyone from knowing the complete recipe. Gas chromatographs haven't permitted the copying of Coca-Cola. If you come up with a novel recipe, it will probably resist copying.
Re: (Score:2)
Can someone please tell me why we trust closed source CPUs to run open source software? You think they don't have backdoors in Intel / AMD / Broadcom / etc silicon?
this is exactly the argument i gave to NLnet, and as a result they are funding the development of the libre riscv soc: https://hardware.slashdot.org/... [slashdot.org]
the f-cpu.org team worked on their design as well for the same reasons.
Re: (Score:1)
Too late. The RISC-V is an American (UC Berkeley) designed architecture and the foundation is based in San Francisco.
Re: (Score:1)
I'm not sure I totally agree with you. The USA has a long history and we don't seem to be the kind of country that uses its military power for selfish purposes, at least not normally.
Yes, we've been in some conflicts that where indeed self serving, but since WW2 ended, where we literally owned the majority of Western Europe, ALL of Japan and quite a bit of territory around the Pacific, with minor exceptions, we have returned all if this to its' original owners. I don't think your view of the USA as a c
Re: Hail India (Score:1)
How long have you been on /. and you call others ignorant? You're talking about bombs and nobody has any clue what your problem is why don't you explain? Because you have no explanation other than saying "it's ok right?" and purging accounts that can tell you're full of shit.
Re: (Score:3)
Well of course. You don't think we'd fight with the deliberate intention of hurting ourselves, do you?
That's not the point. For 150 years U.S. has not been engaging in wars with the objective of stealing other people's stuff or bossing them around. The spread of capitalism is the spread of freedom.
Anybody at any time has the right to free slaves, to free victims of torture, to free victims of enforced starvation. Those imposing such heino
About time. (Score:3)
Re: (Score:1)
Are all Americans inbred retards like you?
Re: (Score:1)
Re: (Score:2, Funny)
and they'll stop working at exactly 5pm every day
Socialists bastards. Don't they know you're supposed to work until you drop dead?
Re: (Score:1)
There's a big difference between working stupid hours 7 days a week and dropping the mic and walking off on the dot of 5 every day even when they're halfway through troubleshooting an important problem
Re: (Score:2)
That is a highly offensive and grossly inaccurate stereotype. You should be ashamed.
They'll obviously scam you for tech support.
Re: (Score:1)
Re: If they're anything like indian software dev.. (Score:1)
Actually it really depends on who their bosses are. Indians, culturally, are very hierarchical and mostly only concerned about the perception their superiors have of them. Their lateral or below coworkers can just go die for all they care.
If the boss has gone home and they don't see a way staying late will be perceptually advantageous they will head home right then. If the boss is still there, you better fucking believe they're going to be in the office.
This hierarchy and perception management is so deeply
Re: (Score:2)
If your boss is your hero and he really does need you, regardless of how many times you spill coffee on his papers, it will become quite evident all on its own, even if it feels like forever.
But your boss might be a bad manager, and so they might never actually realize it.
Re: (Score:1)
Re: If they're anything like indian software dev.. (Score:2, Informative)
Unless you work for yourself there is no reason to work past 5pm. You will never get more back than you lose.
Re: (Score:2)
They've already eaten most of the lunches they'll ever get. Firms are ditching companies in india for eastern europe now because the cost is the same but the code is a hell of a lot better.
first boot (Score:5, Interesting)
i was there last year when they did the first boot when the 22nm test chips came back from the Intel Foundry - the roar from the team was awesome. eighteen months of hard work, doing FPGA and other simulations, and a team that had never worked together before on such a major project, their not just first silicon, their *first ever* silicon worked *first time*. that's a stunning achievement.
to put this properly into perspective: this is India's first ever home-grown 64-bit processor, and the core design is capable of being expanded from single-core lowly "embedded" right the way to a full monster multi-core VLIW Beast. India now joins very few other countries in the world - China, USA and.. err... i really can't think of any others - that have full SOVEREIGN control over the computing devices running within their borders.
that means, for example, ooo.... that the ageing Motorola 68000s running the Fast Breeder Nuclear Reactors can be replaced with something that they trust, and Mossad will not be able to do a SCADA (Stuxnet) attack against them like they did to Iran, causing a meltdown.
when i left they were discussing doing a dual layout of a new Industrial QFP chip, using software libre ASIC layout tools *and* proprietary ones, side-by-side, to do a 180nm low-cost 64-bit variant (to be taped out at a local foundry under Sovereign Control within India's borders). the reason for using libre ASIC layout tools: the India Department of Defense does not want any opportunity for proprietary software from foreign Corporations to compromise the ASICs.
their plans over the coming years include creating commercial WIFI, 3G and LTE ICs, Server-grade multi-core multi-issue designs, mobile-class processors and much more, all developed and manufactured *in* India, upgrading their Foundries in the process, in order to make them not just for Sovereign use: they'll be designed to be price-competitive not only for use within infrastructure equipment and smartphones within their own borders, but for sale to the rest of the world as well.
so they're not just taking on ARM and Intel: they're taking on Global Foundries, Intel Foundries, TSMC and CSMC as well. the irony is: much of the core expertise that went into the development of those Foundries came from people *in india*!
this is just the beginning: it's really exciting, and i'll be cheering for them, all the way. they've started on a process that will give the Silicon ASIC industry (which is borderline cartelled) the arse-kicking it truly deserves.
Re: (Score:2)
their *first ever* silicon worked *first time*. that's a stunning achievement.
I'm sure that ARM people are chuckling now. ;)
Re: (Score:2)
their *first ever* silicon worked *first time*. that's a stunning achievement.
I'm sure that ARM people are chuckling now. ;)
they should have tried to bribe neel to go work for them with waaaay more than 6 figures :)
Re: (Score:1)
ARM should be quaking in it's boots.
It's era of rent collecting on processor designs is coming to an end.
Re: (Score:2)
Re: (Score:3)
that have full SOVEREIGN control over the computing devices running within their borders
How exactly do you define that?
The UK has, or rather had ARM, it's Japanese now. Japan has been making CPUs for decades though, some under licence and some home grown designs like the Hitachi ones. Their 8088 licenced designs were sought after back in the day, because they were cheap and faster than the Intel ones.
Re: (Score:2)
that have full SOVEREIGN control over the computing devices running within their borders
How exactly do you define that?
The UK has, or rather had ARM, it's Japanese now. Japan has been making CPUs for decades though, some under licence and some home grown designs like the Hitachi ones. Their 8088 licenced designs were sought after back in the day, because they were cheap and faster than the Intel ones.
i couldn't think of any that were still popular. i'd forgotten that Japan's industry now runs on PowerPC... except that PowerPC isn't manufactured *in* Japan itself. that was the point: there's only three countries in the world that design *and* manufacture *popular* mass-volume CPUs:
* Intel (designed in USA, fabbed in USA)
* AMD (designed in USA, fabbed with TSMC or GlobalFoundries, neither of which are in the USA)
* China (ICT Loongson aka Godson - a MIPS64 clone)
* India (Shakti Core, soon to be fabbed in
Re: (Score:2)
Russia does not really count because Baikal's home-grown core is an ARM core. it wasn't *designed* in Russia, although it was taped-out on a Russia fab.
Wrong. Baikal-T1 uses MIPS: http://www.cpu-world.com/news_... [cpu-world.com]
Though you can rightfully argue it's not fully Russian as it uses a commercial core. "Elbrus" line of CPUs, however, is fully "home-grown": https://en.wikipedia.org/wiki/... [wikipedia.org]
I guess "Elbrus' does fall under "designed in Russia", right?
Re: first boot (Score:1)
I say congrats for developing their own domestic industry.
I wish America had enough pride to design and manufacture our chips at home.
But we're so obsessed with growing profits we can't keep our own people employed in highly skilled jobs, we had to ship our jobs to India, China and elsewhere so profits could grow.
Now on slashdot we're butthurt and surprised the people we exploited have shunned our top heavy management structures and are doing the same work, but for themselves.
Re: (Score:2)
and Mossad will not be able to do a SCADA (Stuxnet) attack against them
Produced in a US fab. Mossad probably already has some prototypes to do development on.
"Upgrading their foundries" WHAT foundries? (Score:4, Informative)
Last I checked, India had one 6", 180nm fab [wikipedia.org] run by its government's space organization. Thatr's not competitive with any major foundry these days.
Could an Indian company start its own cutting-edge wafer foundry capable of turning out competitive CPUs for the world market? Sure! All they need is a few years $15-20 billion worth of investment...
Also, though Stuxnet was an amazing piece of engineering, to the best of my knowledge it never caused a meltdown, or even close to it, merely destroy some of Iran's Uranium-enrichment centrifuges. Also, I fail to see what circumstances would lead Israel to attack India's infrastructure (as opposed to Pakistan).
Re: (Score:3)
Of course, the problem is that the chip foundry isn't under their control. So the chip they received may not be quite the chip they designed.
Re: (Score:2)
i was there last year when they did the first boot when the 22nm test chips came back from the Intel Foundry
That's really awesome, I'm sure they must have been over the moon. I'm surprised more countries aren't developing their own microprocessors to be honest.
that means, for example, ooo.... that the ageing Motorola 68000s running the Fast Breeder Nuclear Reactors can be replaced with something that they trust, and Mossad will not be able to do a SCADA (Stuxnet) attack against them like they did to Iran, causing a meltdown.
Are you suggesting that India will replace SCADA with something homegrown? Do you think that will make it LESS likely to be susceptible to attacks? Because I would think the opposite would be true? Also, forgive my ignorance, but why would the Mossad want to attack India?
their plans over the coming years include creating commercial WIFI, 3G and LTE ICs, Server-grade multi-core multi-issue designs, mobile-class processors and much more, all developed and manufactured *in* India, upgrading their Foundries in the process, in order to make them not just for Sovereign use: they'll be designed to be price-competitive not only for use within infrastructure equipment and smartphones within their own borders, but for sale to the rest of the world as well.
I'm fascinated by projects like this. I think owning and developing these technolo
Re: (Score:2)
So, the Indian processor is based on an open design done by a US public university, and fabbed by an American company.
Yay, India!
in another post, i mention that, at the time i left to go back to taiwan, they were talking about using an India University Foundry, and using libre ASIC layout tools.
FPGA (Score:5, Informative)
Not obvious from the article: there are no physical Shakti processors available yet.
As of July 2019, users are required to have a Xilinx FPGA board, compile the Shakti processor schematics from Verilog source code to RTL bit-patterns and load that RTL onto the FPGA.
Re: (Score:3)
Not obvious from the article: there are no physical Shakti processors available yet.
it's not mentioned in the article: it was mentioned in the submission i did last year (i was there when they did the first boot), they do have actual 22nm silicon, from a Multi-Vendor (MVP) Test Run. it's just that they're so rare, they'll be prioritising giving them out to their sponsors, first. when the dust settles they'll be able to do a production run.
Re:FPGA (Score:4, Informative)
Apparently they do have silicon: https://www.thehindu.com/sci-t... [thehindu.com]
Made by Intel at their fab in the US. Well, I suppose maybe that doesn't meet your definition of "available", as it was only a run of 300 so presumably not many people have access to them.
Re: RISC-V CPU is *NOT COMPLETE* (Score:1)
Now is the time on sprockets when we dance
Re: (Score:3)
The open-source version of RISC-V is not complete. There are still some modules missing.
it's... "good enough". like how SVE (etc.) are enabled, multilib now provides a way to override individual functions with "more efficient" versions that detect and then "redirect" functions to use whatever accelerated hardware or faster, better instructions would do the job.
so for example, *when* xBitManip is available, video and crypto and other tasks that can be done more efficiently if bit-manipulation is available, thanks to multilib, these can be accelerated *without* significant disruption to all sof
Sometimes an item is simply "good enough: (Score:2)
US manufacturers? (Score:2)
There are no US chip manufacturers this would be competing with. Lucian Armasu is just your typical anti-American moron.
Re: (Score:2)
Seriously? None? Not Qualcomm, AMD, AppliedMicro, Broadcom, Apple ... nobody?
(Admittedly, those are all ARM licensees, so your mileage on the phrase "designed by U.S. manufacturers" may vary.)
Re: (Score:2)
I think there's a difference between manufacturers and designers.
That's what I mean by the phrase "designed by U.S. manufacturers" -- it's open to interpretation (because it's poorly worded). Who is the designer and who is the manufacturer? If a U.S. company has to do its chip fabbing in the U.S. to count as a "U.S. manufacturer," and it has to draw up its chip dies from scratch to be considered a "designer," then you could say Qualcomm -- an American company that licenses designs from ARM and fabs chips overseas -- doesn't do much of anything at all. I guess?
Re: (Score:2)
There's actually a lot of foundries in the US still chugging along that produce silicon at older/larger nodes. They won't be making the top-end CPUs for the latest computers or smartphones, but they do produce plenty of parts cheaply for embedded applications or custom ASICs. If you need a custom ASIC taped out for cheap, they'll do it for less since the foundries have long since paid for themselves. Sometimes 180nm is just fine for the job and can still run faster than the same circuit in a 22nm FPGA.
Also
Server grade not there yet (Score:3)
Many fellow /dotters are saying that this implementation in particular (and RISC-V in general) can scale to server grade processors.
While this design could escale to high performance (and I hope it does), or even supercomputing grade, in this day and age, the year of our lord 2019, a processor can not be considered server grade until it supports virtualization in some way shape or form, and RISC-V has just pre-defined the instructions to allow virtualization (we are more or less at the V0.01 spec), let alone those bein implemented in any physical processor.
https://content.riscv.org/wp-content/uploads/2017/12/Tue0942-riscv-hypervisor-waterman.pdf
Do not get me wrong, I hope RISC-V takes off, and we get a free as in speech processor (RISC-V is MIT, IIRC), and more architecture options in the market to boot (like in the hacylon days of X-86, AMD64, Itanium, Alpha, Transmeta, Cyrix, Centaur, MIPS, SPARK, ARM, POWER), but there is still a long way to go, until we get a server grade RISC-V part....
Hail IIT (Score:1)
Why present this as nationalistic nonsense? (Score:2)
Both IBM and Intel have computer development labs all over the globe. Some Intel processors came from designs done in Israel. ARM started as a research project in England. The code breaking machines at Bletchley Park started as a project in Poland before the Germans invaded.
Re: (Score:2)
Flag waving about the nationality of this work is a waste of mental resources.
if that's what was being done, i would agree.
RISC-V is about Open Source Hardware no matter where it happens.
this is not well understood: RISC-V is *not* "Open". it's *believed* to be open, because there are so many "open" implementations, and there is *access* (unrestricted access) to the specifications.
sadly, the reality is: it's turning into an abusive arrangement, and a Cartel. the innovation (extension and standards development) takes place behind *closed doors*, using ITU-style standards development practices.
try getting *actively involved* in the standards deve
Re: (Score:2)
I smell sour grapes. You, or someone you know, acted badly with respect to the RISC-V protocols and got in trouble. Now you are making slanderous charges.
Grow up. Your whining is annoying the adults present.
Obligatory joke (Score:1)
You buy some Indian tech. A lot of us do.
Now instead of Microsoft Support....
Hi, I'm Apu from India. You bought our super smart buster device. We can tell it's malfunctioning. Please allow us remote access to your controlling household device. Thanks. Now all your files are encrypted and this is ransomware.. Bla ha ha.
Compatibility? (Score:1)
"Shakti is based on the open-source RISC-V instruction set architecture "
With how much compatibility? Will binaries that run on Shakti run on RISK-V machines, and vice versa, with or without special accommodation? If with accommodation, will there be a performance penalty? The article doesn't say.
Intel iAPX432, Indian style? (Score:1)
From the article, a brief mention of "the T class, which should support object-level security".
Didn't Intel try this already? A processor that had rather fine-grained memory access permission? It had no assembler. Instead, it had an Ada compiler. And it went nowhere.
Or perhaps that is just a single point of similarity, and not especially relevant.
https://duckduckgo.com/?q=Intel+iAPX432
https://en.wikipedia.org/wiki/Intel_iAPX_432
Re: (Score:1)
Well Ammonia has been used as a refrigerant in the past
Re: (Score:2)
Do you know of any design since 1945 that wasn't based on predecessors?