Alibaba partner announces 16 core RISC-V chip

https://www.caixinglobal.com/2019-07-25/alibaba-chip-subsidiary-launches-first-product-using-open-source-architecture-101443785.html

by rwmj • 7 years ago

294 245 7 years ago

caixinglobal.com

rwmj6• 7 years ago

There are several links available for this announcement, but none of them are that great. Alternatives: https://kr-asia.com/alibabas-pingtouge-launches-own-processo... https://www.caixinglobal.com/2019-07-25/alibaba-chip-subsidi...

yorwba2• 7 years ago

Alibaba doesn't seem to have any press releases written about this, not even in Chinese. All reporting seems to be based on what CTO Zhang Jianfeng said at the currently ongoing Alibaba Cloud Summit ( https://summit.aliyun.com/ ). Maybe they'll have more information on the specifics once the summit is over.

dang• 7 years ago

Those do seem marginally better so we swapped out https://onties.com/china/brother-pingtou-released-the-first-... for one of them.

joelthelion• 7 years ago

What's the significance of this announcement? What are the likely markets for this chip?

rwmj6• 7 years ago

I don't have any special knowledge on this, but from the announcement: According to reports, Black Iron [Xuan Tie] 910 can be used to design and produce high-performance end-to-end 5G chips, artificial intelligence and automatic driving.

So I guess they want to sell it as an alternative to ARM/nVidia/Qualcomm in those roles.

yourfate• 7 years ago

I'd assume custom server hardware for ali's cloud services.

privateSFacct2• 7 years ago

China is going to be key here.

It's not just a normal market - China may see this as essential to its ability to develop its technology. It's Made in Chine 2025 policy.

That's taken on new urgency as the west has started cutting China off from western tech - so it may be normal companies wanting some insurance in case intel / arm cut them off (trade disputes etc) AND the govt itself wanting to product its industrial base from cutoff during trade disputes.

IceWreck• 7 years ago

For one, RISC-V is fully open hardware unlike x86, amd64, arm and other architectures. More adoption is always a good thing.

majewsky• 7 years ago

RISC-V is an open ISA, but a particular RISC-V chip is not necessarily open hardware. In fact, most probably aren't.

stunt3• 7 years ago

China's market is enough for it. Also all national security stuff.

mbrumlow2• 7 years ago

Indeed. RISC architecture is gonna change everything.

snvzz2• 7 years ago

RISC-V is good.

nguoi• 7 years ago

RISC architectures have existed for about 4 decades.

mbrumlow2• 7 years ago

It's a quote from hackers the movie from 1995.

nickik• 7 years ago

I would really liked to know what extentions they will support. Will the support the Vector extention? How about some of the once that are not quite finished yet, like Bit manipulations.

snvzz2• 7 years ago

I would really liked to know what extentions they will support.

Me too.

Will the support the Vector extention?

The vector (V) extension is not standarized yet. It'll likely take two years, as per updates in recent conferences.

How about some of the once that are not quite finished yet, like Bit manipulations.

That one (X) isn't ready either, but more likely to be ready soon.

Symmetry• 7 years ago

Just because standardization isn't finished doesn't mean they won't go ahead an build the closest thing to the current consensus they can manage.

childintime2• 7 years ago

That and the power envelope

mtgx• 7 years ago

They probably won't be in a hurry to support the security extensions.

londons_explore• 7 years ago

How long before Android runs properly on RISC-V?

And will they be able to persuade app-makers to rebuild all apps for the platform - x86 Android shows how hard that can be.

buildzr• 7 years ago

Many apps will work with minimal if any changes as they don't use native stuff at all. There's also some interesting solutions for this like libhoudini that emulates ARM to make many native apps work without any effort on the app developers part.

sct202• 7 years ago

If China's doing the push, they only need their local app ecosystem to move over.

jdietrich3• 7 years ago

To a first approximation, "local app ecosystem" = "WeChat".

microcolonel6• 7 years ago

How long before Android runs properly on RISC-V?

It's mostly up to OpenJDK and V8 porting. When it comes to porting V8, I've found out the hard way that there's more to it than a generic understanding of compilers. Somebody still has to step up to the plate for a real JVM.

And will they be able to persuade app-makers to rebuild all apps for the platform - x86 Android shows how hard that can be.

Real x86 Android devices had libhoudini (DBT), which was pretty decent. If RISC-V starts with low-end devices, or with a high-visibility model that developers are pretty sure will sell well, then the native compilation target will become more popular over time.

If China's bizarre local app market is involved, chances are they'll start supporting the target as soon as it's available.

cat13• 7 years ago

https://github.com/alibaba/dragonwell8 Alibaba do has their own JVM to handle high traffic during single day sales.

defer• 7 years ago

A small correction, Android does not use OpenJDK. It uses the ART runtime (https://source.android.com/devices/tech/dalvik) which will definitly need porting.

Slightly longer answer is that _some parts_ of the OpenJDK are used. Namely, LUNI (implementation of java.lang, java.util, java.net, java.io) run on the device, but those are high-level enough that they will require either none or minimal changes for risc-v.

Compiling applications also uses OpenJDK (or the Jack compiler in platform builds) but that is an intermediate step on the host. The produced JVM bytecode gets converted to Dex bytecode during the build process.

I think the effort for Android will be two-fold:

On the software side, Bionic (android's libc), Dalvik (runtime), LLVM, external dependencies (i.e. boringssl, lib{vpx,hecv,mpeg4,etc},) definitely need work.

On the hardware side it all lies on the practical availability of a SoC. While I have no doubt we'll have something similar to android-x86 running on discrete chips, commercial devices will need a more complete cpu+gpu+dsp+modem package.

mhh__2• 7 years ago

When it comes to porting V8, I've found out the hard way that there's more to it than a generic understanding of compilers.

How so?

tjoff• 7 years ago

I haven't thought about it much, but first reaction - I get that it might be a good candidate to get volumes up but isn't it kind of self-defeating? Why would you want to run android, of all things, on an open CPU?

microcolonel6• 7 years ago

It's extensible, so they could in theory introduce something to the ISA which would give them a competitive advantage, maybe even a publicly available extension that their competitors have simply left out, e.g. one of the crypto extensions.

There's also the benefit of being more flexible with teams who work on your SoC. With RISC-V, everyone has an ISA license, so they can contract, team-up, collaborate, or do basically anything they want, in order to work on their chips.

This also means that acquisitions are a much richer market than they are with other ISAs; in part because you don't have to justify the cost in time, money, and uncertainty of licensing or planning to license ARM.

jononor2• 7 years ago

Android is the leading OS for mobile phones and tablets. It is also quite popular in embedded for devices that have a screen, mostly as a side-effect of its dominance in mobile. It can often be easier to get graphics and touch drivers/support for Android compared to Linux. It is also easier to find developers who can write a custom Android app for the UI than any of the typical embedded/Linux UI frameworks, and the tooling is better.

yjftsjthsd-h2• 7 years ago

x86 Android also shows that you can generally get around it with an emulation layer that's used as needed.

tgtweak• 7 years ago

Article explaining the relevance of this news: https://technode.com/2019/07/24/chinas-chipmakers-risc-v-san...

estomagordo• 7 years ago

Link seems terribly dead =/

jtbayly• 7 years ago

Worked for me.

exohuman• 7 years ago

Try again? I was able to see it.

chvid3• 7 years ago

This article adds good background. Also what the announcement seems to be about is a design for a chip yet to be put into production. How this will compare to arm and intel cpus in terms of cost and performance is unclear (to me).

class4behavior• 7 years ago

Just load it via an archive?

https://web.archive.org/web/20190724090600/https://technode....

https://archive.is/I5bgm

baybal22• 7 years ago

Being in Arm's position sucks now.

Would they not be so enthusiastic about sanctions, Chinese would've been way more cautious jumping on RISC-V boat.

Now instead of having RISC-V as one off research projects, which would have probably died off silently, now they are all treated as survival essential projects.

mark_l_watson• 7 years ago

I agree. I feel like we (USA) are forcing China on a path that will make them more competitive long term.

bilbo0s5• 7 years ago

I wonder if most of what we are doing is being done for political purposes? Sort of, just to look good and get more votes. And if so, you kind of have to admit that from that perspective, the calculation works out even if China surges ahead even faster.

(In fact, you might even be able to use that surge, to do more grandstanding, to get more votes.)

Not saying this is 100% what's going on, but it's undoubtedly a factor in the equation here.

privateSFacct2• 7 years ago

For sure. Third party countries may also decide less risk in buying from China (ie, China less likely to cut them off because of real or perceived issues).

adrianN• 7 years ago

And in the end competition is good for normal people, so great job?

madez6• 7 years ago

Europe could have stepped up and become competitive long before, but they are too complacent with the status quo. It's good that finally someone steps up.

sangnoir• 7 years ago

The same thing happened when China wanted to join the ISS[1] - the US Congress blocked this to avoid "tech transfer". Instead of being dejected and sad, China kickstarted its space station program, and are now collaborating with Europe.

1. https://en.wikipedia.org/wiki/Politics_of_the_International_...

addicted2• 7 years ago

The best thing for China is that this is happening at a time where the US is insistent that the brightest Chinese minds in the US go back to China, and the ones in China, stay there.

It's genius. Genius, I tell ya. But the current administration doesn't care, because the negative impacts will only be felt a decade from now.

tw04• 7 years ago

which would have probably died off silently

Given Western digital will be producing millions of risc-v chips, the "it'll die off silently" ship has long since sailed.

https://www.westerndigital.com/company/innovations/risc-v

jononor2• 7 years ago

Not many other companies or developers are involved in WDs effort, thats mostly an internal project. There are tens of obscure ISAs and CPUs that are used in similar applications, and no-one is particularly excited about those. But RISC-V seems to have serious momentum in both academia, and as base for upcoming platforms from new companies.

baybal22• 7 years ago

Just remembered that. WD had a humongous recruiting event for chip developers in 2017 in Shenzhen, but nobody was told what was that all about

childintime2• 7 years ago

It is probably no coincidence that ARM was sold to Softbank right when RISC-V was starting to make a name for itself. They saw the writing on the wall.

microcolonel6• 7 years ago

Too bad they only quoted the CoreMark, and only compared it to an in-order processor.

I'd love to see a real benchmark, and a comparison to a processor in the same ballpark in terms of power and area.

bogwog4• 7 years ago

This is both awesome and terrifying. I'll never forget this research(https://www.wired.com/2016/06/demonically-clever-backdoor-hi...) which showed how an attacker can insert a backdoor into a chip during the manufacturing process in a way that is practically impossible to catch. That makes me very concerned about using a chip completely manufactured in China.

But at the same time, RISC-V needs to happen. It's a critical step to move computing forward. It's just a damn shame that this is the way it's going to happen. Hopefully this will push RISC-V manufacturing efforts in the US before the market becomes flooded with cheap open source and most likely backdoored chips manufactured in China.

microcolonel6• 7 years ago

Nobody can afford to keep the compilers and subroutine libraries to themselves, so the more people use the ISA, the more useful it becomes. Even if you wouldn't in a million years deploy such a chip at your financial institution or on your email server; or, without a separate network, in your home.

flohofwoe• 7 years ago

Not sure how to best say it, but I think in the end it's better to not put any trust into any hardware- (or software-) provider anyway, it shouldn't matter whether your chips are built by a "good" or "bad" actor, just assume it's a bad actor when designing the systems around it. Maybe this leads to more robust systems overall, also for cases when there was no "evil intent" (see Spectre/Meltdown).

whatshisface• 7 years ago

That's not practical. If bytes anywhere in memory can trigger a backdoor, webservers won't be possible because they have to listen to arbitrary requests.

strooper• 7 years ago

Is there a way to know that the present chips are not backdoored? Or, backdoored by US? Why should the world be concern about Chinese backdoors and not US ones?

seandhi• 7 years ago

Because most of the world has flourished under Pax Americana. The world can only imagine what the Chinese as a world power will bring.

eloff5• 7 years ago

Let's not pretend the US and China are equally authoritarian please.

bogwog4• 7 years ago

Anyone who has done even a tiny bit of research into the Chinese government wouldn't need to ask that question.

Here's a preview: https://www.hrw.org/news/2011/09/22/why-are-people-disappear...

See also what's happening in Hong Kong right now.

swebs• 7 years ago

Many people consider IME a backdoor

https://en.wikipedia.org/wiki/Intel_Management_Engine

reenfranchised• 7 years ago

Put it this way— as Americans, we’re all pretty confident that our government doesn’t pay any substantial number of people to go online and propagandize our human rights abuses under the guise of just being reasonable world citizens.

(Maybe that’s just because we know that right now our government can’t scratch its ass and elbow at the same time, that we won’t face any penalty for our human rights abuses anyway, and there’s no point in paying bureaucrats to do something that ignorant yokels will go out and do for free... but that’s beside the point :)

stunt3• 7 years ago

Lets not jump into politics! Backdoors are bad and it doesn't matter who does it.

Good to see development from any country. Obviously they have enough reason to do it. Just consider sanctions. They also have to protect their own market.

Anyone that can afford it, should do it. Ultimately it is a good thing from technology prespective.

bilbo0s5• 7 years ago

Is there a way to know that the present chips are not backdoored?...

As a practical matter, that's largely irrelevant to us in the West. The world that's shaping up is a world where we won't be able to get those processors in any case.

whoevercares• 7 years ago

Given the manufacturing profit margin, it’s unimaginable if any big player will take the lead(unless the government chime in, but wait that’s how China works!)

elamje• 7 years ago

Can you explain why the RISC-V needs to happen? I am not familiar with the chip industry, so I don't understand why the RISC-V will move computing forward and x86/ARM won't move it forward.

Edit: Does it have to do with being open-source?

lsllc• 7 years ago

Any industry needs competition! Sadly the consolidation of the [server] processor market over the last decade from half a dozen or more very different designs (which had collectively driven computing for the last 30+ years) has left us with the duopoly of Intel who's not really innovating and ARM who kind of are, but mostly in mobile.

Obviously, RISC-V also needs software support; gcc supports it, Golang has added RISC-V as a target (although it's not yet implemented) and there are several projects working on LLVM (e.g. lowRISC and SiFive's port). Having played around with the Fedora-RISCV port under QEMU, it's actually surprisingly good thus far.

Having an open-source processor design with good software support that anyone can manufacture without royalties is a good thing and will hopefully spur both Intel and ARM into innovation.

tyingq2• 7 years ago

Removal of "percentage of revenue" based royalties is probably the big one. Others may cite the open source ISA, but I think royalties is the big deal. See Western Digital's plans for RISCV. Taking a 1.5 to 2 percent "off the top revenue/not-profit skim" away is a big deal.

Edit: Fwiw, I hope the open source ISA has secondary societal benefits. I'm not skeptical, just pragmatic. Follow the money.

madez6• 7 years ago

Neither AMD nor Intel offer x86 chips that give end users full access to the highest privilege levels.

tyingq2• 7 years ago

"That makes me very concerned about using a chip completely manufactured in China."

I understand the risk, but isn't that cat "already out of the bag" for ARM licensees and other ISA implementations? Is there something particularly special about this one?

ibic• 7 years ago

I don't get it - NSA has inserted many backdoors in different devices like routers and even hard disks, I don't recall China has such a record. Why are you so concerned about China but the U.S. ?

joey_bob• 7 years ago

I don't get it - CPC has inserted many backdoors in different devices like edge routers and even cell phones, I don't recall U.S. has such a record. Why are you so concerned about U.S. but the China ?

nwah1• 7 years ago

Lenovo and many other Chinese companies have been caught adding rootkits to their laptops, etc.

pkilgore• 7 years ago

This is nonsense whataboutism backed by...poor memory??

Also, just because there are concerns about Chinese back-doors does not mean the author is entirely unconcerned about US back-doors, nor does it even tangentially relate to his argument. (In fact, I think the author would agree that both are equally bad.)

At least, however, in the US there are enforceable State and Federal Constitutional protections for citizens and residents that limit the ability of the government to spy on its citizens. Even if these protections are not perfect (they are not) or sometimes abused (they are), they are objectively better than the clusterfk of privacy and rights violations the Chinese state inflicts on its citizens every day, with the consent and complicity of national technology firms.

It may not mean much if you're not a US Citizen (it shouldn't) but like it or not this is a new front of national security, and to "not get it" is to miss the point entirely.

vorpalhex2• 7 years ago

Yeah, the NSA does it, totally cool for the Chinese to do it as well..

perspective12• 7 years ago

Any government that can backdoor chips will do it. Now at least you'll get to pick who you give that access to.

yorwba2• 7 years ago

Considering that this was announced by Alibaba Cloud, there's no need to worry about hardware backdoors since you probably won't be able to buy the chips anyway. If you are hosting something in their cloud, a backdoor is unnecessary, since the authorities can just come knocking on the front door if they need anything.

duxup• 7 years ago

IIRC in the wake of Spectre, Google had a security blog that noted that even at Google with all the layers of software involved they felt that there really was no way to ensure that something in those layers isn't running something unexpected / doing something it shouldn't.

IIRC The general topic was speculative execution and such security, but I think it also made it clear that more generally all the chips, software involved ... you just can't know.

To be clear, I don't say that to dispute anything you're saying, quite the opposite.

chvid3• 7 years ago

How does this compare to the Intel X86 CPUs in terms of power consumption, price and performance?

Is this CPU a viable alternative for server and desktop computing?

rwmj6• 7 years ago

This chip exists entirely as a file on a computer, and maybe a few FPGAs, so I doubt even the company announcing this has very much idea. It seems as if they're hoping to get SMIC to manufacture this, so it won't exactly be a competitive node (although China is pouring vast amounts of resources into making their semiconductor manufacturing competitive, so give it a few years).

chvid3• 7 years ago

So far is this thing actually from being produced? (In time and cost)

gridlockd• 7 years ago

A better question would be, how does it compare to ARM?

ARM is generally more competitive in terms of price and power, less competitive in peak performance.

RISC-V would have the added price advantage of being royalty-free.

bhouston• 7 years ago

When will RISC-V move to multi-chip modules (https://en.wikipedia.org/wiki/Multi-chip_module) as AMD has done very successfully with Ryzen?

It allows for easy scaling of CPU designs as well as increasing yield by making smaller components that have overall lower defect rates.

hat seems like the logical next move for most CPU chip vendors/designers except for maybe ARM.

rwmj6• 7 years ago

"RISC-V" isn't an organization that can "move to multi-chip modules". RISC-V is a foundation which essentially publishes a couple of PDF files (the user spec and priv spec) and organizes a bunch of other stuff around reference hardware designs, simulators and the software toolchain.

All of this is BSD licensed, so anyone can pick it up and manufacture chips based on the specification or by modifying one of the reference designs. Note that the largest part of designing and building a chip is not the choice of ISA (that's probably well under 10% of it).

What you may want to know is whether any manufacturers are going to make modular RISC-V chips. None of them to my knowledge. But most of them are currently focusing on the ARM space (embedded, IoT, AI, etc.) including this particular design.

digikata• 7 years ago

The nice thing is that software and work on a RISC-V toolchain can now be much more common for vendors that elect to use RISC-V, though it remains to be seen how well the actual manufactured chip variation gets supported like instruction set extensions, peripherals, and configurations for many cores. ARM had some motivation to help ease the customization pain for tooling for their IP licensees, so I'm curious if RISC-V support will be more like ARM or more like early UNIX vendors.

monocasa5• 7 years ago

Like a lot of chip desicions, this one is based on the effects of the process node in question. So the answer is probably when a company gets to TSMC 7nm while being capable of the high end bonding on the package substrate between chips, at a per chip gate count probably close to the ryzen cpu chips. Before then, it doesn't really make sense.

tverbeure• 7 years ago

Today's RISC-V cores on a modern silicon process are (often far) less than 1mm2.

fulafel• 7 years ago

MCM is a old trick in the toolkit, eg Intel x86 and POWER chips have used it on and off over the years, as have various game consoles. The downside is the cost and complexity of packaging, and the power required for inter-module communication. Many of the past usages have been to closely couple a die from a DRAM process and another die from a logic process. It's not obvious that MCMs are becoming more popular on average in the future.

1121redblackgo• 7 years ago

The 'muddying the waters, what about the US' comments on these threads always intrigue me.

TallGuyShort2• 7 years ago

Reminds me of Adapteva's Parallela, which I bought and was really excited about, but was obviously a commercial failure. I work on software that has an above-average (but far less than HPC) need for concurrency, but even our most demanding customers seem to be content with off-the-shelf Intel, either because they don't need THAT many cores, or because they value the more sophisticated instructions (like encryption acceleration, etc.) I'm curious to know what kind of use cases y'all work on that might crave something like this.

Quequau• 7 years ago

When they first did their kickstarter they were talking about doing a follow up with 64 cores and being able to connect them together with those fancy pants connectors they had going on. I was all set to buy four... and none of that ever materialised.

I got the impression that a lot of folks had unrealistic expectations about how Linux would support the chip right from the start, which honestly wasn't all that surprising recalling the way they worded some of their promotional material.

panpanna• 7 years ago

One presenter claimed the board would pay for itself if you used it for mining...

I assume something like that would be much easier to implement than a real multicore system like linux but I never heard of anyone mining on their boards.

BubRoss• 7 years ago

Modern x64 is very difficult to beat and takes a lot more than just raw core count. The out of order execution, prefetching, multiple execution units, SIMD, cache, cache synchronization etc. are all very strong. I think the space where more weak cores can be utilized but not GPUs is pretty slim. Intel themselves even had that briefly with their high core count out of order atom core chips that didn't take off (or weren't given a chance).

I think software that can take advantage of more than a few cores without just using brute force fork-join parallelism in some of the heavy loops is rare. I don't think it has to be that way, but the problem comes down to software architecture which isn't going to be solved by leaving each application programmer to their own devices. It will take libraries that give them the means to do it without having to reason about low level synchronization.

marktangotango• 7 years ago

Modern x64 is very difficult to beat and takes a lot more than just raw core count.

Are there any alternatives on the horizon?

TallGuyShort2• 7 years ago

> the space where more weak cores can be utilized but not GPUs is pretty slim

Yeah I think that's what it is. As you said that I realized that I do actually have a few customers that can benefit massively from many cores, but they're just using MANY, MANY cores and it's all CUDA. I don't see many people going for the compromised approach of say, Xeon Phi.

icelancer2• 7 years ago

RIP. I have one of those boards from Kickstarter that's dead due to thermal destruction of the chip. They ran incredibly hot and didn't have sufficient throttling. Not that it was very useful anyway, but I liked the company's mission and attempt.

wmf• 7 years ago

Nobody wants this particular chip but there isn't money to pay for a RISC-V Manhattan Project so there will have to be an incremental evolution of several generations of RISC-V chips before they produce a usable one.

MisterTea• 7 years ago

either because they don't need THAT many cores, or because they value the more sophisticated instructions (like encryption acceleration, etc.)

The parallela was much more complex, had its own bus architecture which required an FPGA to glue the chip to the Arm cores in the FPGA SoC which ran Linux. Compare that to a simple x86 or Arm system where the tooling and software is mature and you don't have to do any extra fancy work. Just write a program and execute it. Same thing that killed the IBM/Sony Cell processor. Neat idea, PITA to program.

RantyDave• 7 years ago

It might've done well as a PCIe (or thunderbolt) thing. There are definitely large scale users of scalar math that could make use of it but you'd need to build it in 5nm and dozens of cores to be even remotely competitive.

ev0lv• 7 years ago

For all my .NET developers, if you want your .NET Core app to target RISC-V ISA, you would need to use Mono AOT and to specify the LLVM target is RISC-V.

zoobab• 7 years ago

I mirrored the PR of RISC-V foundation here:

http://www.zoobab.com/alibaba-unveils-most-powerful-risc-v-p...

Alibaba partner announces 16 core RISC-V chip

Related Stories