Stories
Slash Boxes
Comments

SoylentNews is people

posted by martyb on Sunday July 08 2018, @12:41AM   Printer-friendly
from the chip-off-the-old...chip? dept.

China Finds Zen: Begins Production Of x86 Processors Based On AMD's IP

Chinese-designed "Dhyana" x86 processors based on AMD's Zen microarchitecture are beginning to surface from Chinese chip producer Hygon. The processors come as the fruit of AMD's x86 IP licensing agreements with its China-based partners and break the decades-long stranglehold on x86 held by the triumvirate of Intel, AMD and VIA Technologies. Details are also emerging that outline how AMD has managed to stay within the boundaries of the x86 licensing agreements but still allow Chinese-controlled interests to design and sell processors based on the Zen design.

AMD's official statements indicate the company does not sell its final chip designs to its China-based partners. Instead, AMD allows them to design their own processors tailored for the Chinese server market. But the China-produced Hygon "Dhyana" processors are so similar to AMD's EPYC processors that Linux kernel developers have listed vendor IDs and family series numbers as the only difference. In fact, Linux maintainers have simply ported over the EPYC support codes to the Dhyana processor and note that they have successfully run the same patches on AMD's EPYC processors, implying there is little to no differentiation between the chips.

The new chips are surfacing against the backdrop of the trade war between the US and China that could escalate quickly, likely reinforcing China's long-held opinion that a lack of native processor production could be a strategic liability. Today's wars are won with chips, and their strategic importance certainly isn't lost on those in the halls of power. In fact, the Obama administration blocked Intel from selling Xeon processors to China in 2015 over concerns the chips were fueling the country's nuclear programs, and subsequent actions by the US have largely prevented China from achieving the technical know-how and equipment to develop its own chips through acquisitions and mergers.

That makes it even more surprising that AMD has managed to establish a franchise that allows Chinese processor vendors to develop and sell x86 processors in spite of US regulations and the licensing restrictions with Intel, but now more information is coming to light about how AMD pulled off the feat.

Related: Intel Launches New Chips in China as US Bans Sales to Supercomputing Centers
Intel Hints at Patent Fight With Microsoft and Qualcomm Over x86 Emulation
Data Centers Consider Intel's Rivals
Tencent Chairman Pledges to Advance China Chip Industry After ZTE "Wake-Up" Call


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
  • (Score: 2, Informative) by Anonymous Coward on Sunday July 08 2018, @01:26AM (17 children)

    by Anonymous Coward on Sunday July 08 2018, @01:26AM (#704029)

    Needs to die.

    • (Score: -1, Flamebait) by Ethanol-fueled on Sunday July 08 2018, @01:35AM (4 children)

      by Ethanol-fueled (2792) on Sunday July 08 2018, @01:35AM (#704033) Homepage

      Well, it's not like these cheatin' Chinks could ever roll something that required decimal precision, These Goddamn Chinks are still doing their PH.Ds on 8-bit microcontrollers and poaching whatever higher knowledge they have through "academic solicitation."

      • (Score: 0) by Anonymous Coward on Sunday July 08 2018, @04:04AM (2 children)

        by Anonymous Coward on Sunday July 08 2018, @04:04AM (#704093)

        Go look at some scientific journals. You'll see a lot of authors with Chinese names.

          • (Score: 1, Insightful) by Anonymous Coward on Monday July 09 2018, @06:09AM

            by Anonymous Coward on Monday July 09 2018, @06:09AM (#704447)

            I have to wonder what's up with the Chinese here. As of this writing there is only one Chinese scientific* Nobel laureate (Tu Youyou, Medicine 2015). There are two other laureates who had Chinese citizenship at the time they did the work which led to their prizes: Chen-Ning Yang and Tsung-Dao Lee. However, when they did the work which won them the 1957 Physics prize they were both in the United States (Yang did not return to China until 2015, and Lee was naturalised an American citizen in 1962). Two other Nobel laureates were once Chinese nationals, but again neither of them did their prize-winning work when they were still in China. Daniel C. Tsui moved to the United States in 1958 and did his work that won him the 1998 Physics Prize in the 1980s, and Charles K. Kao did the work which won him the 2009 Physics Prize in the 1960s while he was working in England. So that leaves only Tu Youyou as the sole Chinese Nobel Prize winner who did all the prize-winning work under the auspices of China. So it's clearly not a question of race, but more one of society. There is something badly wrong with the scientific establishment in China that is responsible for their paltry results thus far. It's not just a matter of their form of government either: the Soviet Union still somehow managed at least eight scientific Nobel laureates with a government every bit as repressive as China's had ever been. Probably more the priorities of their government and society.

            _____________
            * Specifically excluding non-scientific prizes like Peace or Literature

      • (Score: -1, Flamebait) by Anonymous Coward on Sunday July 08 2018, @04:25PM

        by Anonymous Coward on Sunday July 08 2018, @04:25PM (#704252)

        You win stupid comment of the day. And you're gay, too.

    • (Score: 2) by requerdanos on Sunday July 08 2018, @01:56AM (10 children)

      by requerdanos (5997) Subscriber Badge on Sunday July 08 2018, @01:56AM (#704048) Journal

      [x86 architecture] Needs to die.

      Anyone believing this should buy and use non-x86 processors exclusively, and encourage the same.

      It usually means that you will have to use some form of free software unixlike OS, those being the ones that cross-compile more easily, but it's quite doable. ARM, PowerPC, RISC V, or DEC Alpha, Sparc, etc., take your pick.

      As long as you don't need any proprietary software that you can't recompile for your architecture of choice, you are good to go.

      • (Score: 0) by Anonymous Coward on Sunday July 08 2018, @02:10AM (2 children)

        by Anonymous Coward on Sunday July 08 2018, @02:10AM (#704050)

        Everything i have now is ARM ( or fpga ). I am not a hypocrite, and i eat my own dog food.

        • (Score: 2) by requerdanos on Sunday July 08 2018, @01:28PM (1 child)

          by requerdanos (5997) Subscriber Badge on Sunday July 08 2018, @01:28PM (#704188) Journal

          Everything i have now is ARM ( or fpga ). I am not a hypocrite

          I apologize for the implication; I was really more encouraging others and pointing out that free software provides a ready path than suggesting that you were not serious. It's really possible to do pretty much anything you would do otherwise, without using x86 for anything. The cost-per-speed, cost-per-flexibility, and learning curve maps are different, but Linux (GNU/ and Android flavors) and its software base run as happily on ARM as anywhere else.

          As for myself, I use x86 machines because I am seduced by the cheap+fast aspect, but at my desk I do at least have as many ARM systems as x86 systems on my KVM. They compare almost entirely favorably. The ARM systems can do pretty much anything the x86 ones can do (just slower).

          One thing I really appreciate is the power efficiency. I replaced an aging Pentium 4 server a few years ago with one of those single-board Olinuxino ARM computers outfitted with a nice fast SSD. The P4 ate hundreds of watts, whereas the Olinuxino is happy with 5 Volts at 1 amp. And the migration wasn't much more than copy over /home and parts of /etc.

          Given the advances over the last decade, the biggest advantages x86 chips have are momentum and huge installed base; software compatibility is no longer an x86-only thing.

          • (Score: 0) by Anonymous Coward on Sunday July 08 2018, @10:10PM

            by Anonymous Coward on Sunday July 08 2018, @10:10PM (#704337)

            Ya, with raw horsepower, x86 still wins, but i have also found that 'usability' is fine for tasks 99% of real people do, as long as you are not trying to pretend a PI was designed to be a desktop or something and go with 'desktop' class ARM. I agree too, that its hard to beat the power requirements.

            Was a huge fan of PowerPC, which could honestly compete on performance, until that ( consumer ) market dried up. Here is hoping RISC-V isn't a pipe dream, or never gets out of 'embedded' silicon.

      • (Score: 0) by Anonymous Coward on Sunday July 08 2018, @02:13AM

        by Anonymous Coward on Sunday July 08 2018, @02:13AM (#704052)

        Oh, and for the above comment.. It was posted using a Jetson TX2.

      • (Score: 0) by Anonymous Coward on Sunday July 08 2018, @06:41AM

        by Anonymous Coward on Sunday July 08 2018, @06:41AM (#704124)

        Works for me. All arm based now, but the 68k and ppc were joys to program (asm).

      • (Score: 2) by bitstream on Thursday July 12 2018, @03:42PM (4 children)

        by bitstream (6144) on Thursday July 12 2018, @03:42PM (#706239) Journal

        As long as large investments fueled by de-facto monopoly into the x86 architecture to make it competitive in the instructions per US$. It's explainable that people go for the best bang for the buck. What people can do is to ensure that software works on other architectures with ease.

        Maybe the fact that the Chinese have a hard time to get access to x86 and can thus more easy produce chips for other architectures may be the market opener.

        Haven't seen any end consumer real price competitor to x86 for a desktop/server yet.

        • (Score: 2) by requerdanos on Thursday July 12 2018, @08:27PM (3 children)

          by requerdanos (5997) Subscriber Badge on Thursday July 12 2018, @08:27PM (#706347) Journal

          Haven't seen any end consumer real price competitor to x86 for a desktop/server yet.

          Although in the tablet space, ARM is really demonstrating, I'd say, that x86 is definitely not needed for a widely-accepted computing device.

          • (Score: 2) by bitstream on Thursday July 12 2018, @11:30PM (2 children)

            by bitstream (6144) on Thursday July 12 2018, @11:30PM (#706402) Journal

            Any tablet that is strong enough to do desktop? and cheap. However connectivity would suffer (USB sucks).

            Then it's the server 19" side. Still no real competitor on both performance, price and availability.

            • (Score: 3, Informative) by requerdanos on Friday July 13 2018, @07:07PM (1 child)

              by requerdanos (5997) Subscriber Badge on Friday July 13 2018, @07:07PM (#706741) Journal

              Still no real [desktop/server] competitor on both performance, price and availability.

              I find this (you may or may not care to know) very frustrating.

              I would cheerfully use a (server or desktop) processor that was half (or a quarter) as fast per core as long as it had plenty of cores and my compiles and media encodes finished in about the same amount of time.

              I don't need an x86 for that in theory (free software mostly cheerfully recompiles wherever you need it to), but I am not finding the motherboards that make it true in practice. I'd like to.

              • (Score: 2) by bitstream on Saturday July 14 2018, @05:58PM

                by bitstream (6144) on Saturday July 14 2018, @05:58PM (#707234) Journal

                Do we smell a kickstarter or movement? Economical MIPS/PPC/ARM motherboard for server and desktop?

    • (Score: 2) by DannyB on Monday July 09 2018, @02:02PM

      by DannyB (5839) Subscriber Badge on Monday July 09 2018, @02:02PM (#704553) Journal

      Don't kill the x86 architecture just yet!

      Not until Intel Management Engine, or something like it, can be made mandatory for other architectures. Compromise baked right into the hardware.

      --
      People today are educated enough to repeat what they are taught but not to question what they are taught.
  • (Score: 4, Interesting) by requerdanos on Sunday July 08 2018, @01:38AM (7 children)

    by requerdanos (5997) Subscriber Badge on Sunday July 08 2018, @01:38AM (#704035) Journal

    AMD has managed to establish a franchise that allows Chinese processor vendors to develop and sell x86 processors in spite of US regulations and the licensing restrictions with Intel

    TFA [tomshardware.com] says that "According to the agreement, the final products can only be sold within China's borders", but the Chinese people who trade with the west, as a general rule, are pretty inventive at doing so and I would be very surprised if people outside China could not purchase Dhyana chips at will faster than you can say "Shenzhen". Perhaps at lower prices than their AMD-branded counterparts (but probably not much less).

    • (Score: 0) by Anonymous Coward on Sunday July 08 2018, @02:57AM

      by Anonymous Coward on Sunday July 08 2018, @02:57AM (#704068)

      All I care is if they licensed the PSP as well.

    • (Score: 0) by Anonymous Coward on Sunday July 08 2018, @02:49PM (2 children)

      by Anonymous Coward on Sunday July 08 2018, @02:49PM (#704218)

      Does it really take a whole lot of creativity to sell it to a third party local retailer that then goes on to sell the chips on ebay to international customers?

      • (Score: 2) by requerdanos on Sunday July 08 2018, @04:55PM (1 child)

        by requerdanos (5997) Subscriber Badge on Sunday July 08 2018, @04:55PM (#704260) Journal

        Does it really take a whole lot of creativity

        I don't know. I don't know how the enforcement of the requirement works, and don't know the details; just what was in TFA.

        But these chips' very existence is the result of some very creative legal maneuvering to satisfy the law and contract requirements to the letter. It would not surprise me if further creativity were employed to also create a grey market for buyers outside China.

        Something I am really looking forward to, in fact, as an expected side benefit of this deal, are Chinese x399 boards in the $99 range that make deploying Epyc and Epyc-like chips affordable for broke hobbyists (like myself). Major-brand high-quality x399 boards are going for around $300 and up [newegg.com]. My current primary workstation is a 10-core 20-thread Xeon that I was able to buy only because they are going for ~$199 as B-grade re-certified product, and there are ~$99 Chinese iX79 boards to go with.

        Takes pretty good creativity to make "compatible" motherboards and sustain a market for them, and I don't just hope for it, I fully expect it.

    • (Score: 2) by takyon on Sunday July 08 2018, @02:49PM (1 child)

      by takyon (881) <reversethis-{gro ... s} {ta} {noykat}> on Sunday July 08 2018, @02:49PM (#704219) Journal

      How cheap can they go? Ryzen 5 performance at $50?

      However the article implies that these will be like Epyc processors, which currently have 8-32 cores and are expected to hit 48 or 64 cores on the 7nm node. Epyc can be pretty expensive so there is a lot of room for China to undercut AMD.

      If it's possible to remove or disable any hardware spying, you could have a very interesting chip if it makes its way across the ocean. Except the Chinese companies might add their own spyware.

      --
      [SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
      • (Score: 2) by requerdanos on Sunday July 08 2018, @04:55PM

        by requerdanos (5997) Subscriber Badge on Sunday July 08 2018, @04:55PM (#704261) Journal

        If it's possible to remove or disable any hardware spying, you could have a very interesting chip

        That would be a game-changer for the x86 space--a very welcome one.

    • (Score: 2) by driverless on Monday July 09 2018, @12:14PM

      by driverless (4770) on Monday July 09 2018, @12:14PM (#704508)

      the Chinese people who trade with the west, as a general rule, are pretty inventive at doing so and I would be very surprised if people outside China could not purchase Dhyana chips at will faster than you can say "Shenzhen"

      Or, as any number of Western big brands have found out, chips that look, act, and perform exactly identically to the Dhyana's suddenly turn up, but they're not being obviously made at any licensed manufacturing facility or sold by any vendor that stays around for long enough to be prosecuted.

  • (Score: 3, Funny) by Anonymous Coward on Sunday July 08 2018, @04:47AM (3 children)

    by Anonymous Coward on Sunday July 08 2018, @04:47AM (#704104)

    blocked Intel from selling Xeon processors to China in 2015 over concerns the chips were fueling the country's nuclear programs

    Bullshit. You might have needed mainframe grade hardware and software to do nuclear blast simulations in the 70s. But nowadays the average GPU does the job. Hell, an off-the-top game engine has all the physics code you need. Just apply the ragdoll physics to buildings, ground and trees, import terrain maps, place a few mock buildings and trees, and drop a huge bomb and enjoy the simulation.

    Same math.

    China could still get its compute needs from AMD, ARM and IBM just fine. So the whole thing was meant to slow down China's cloud services (at the behest of Google and Amazon) and at the expense of Intel's bottom line. And of course, much like Trump's trade war moves, it failed miserably.

    • (Score: 1, Interesting) by Anonymous Coward on Sunday July 08 2018, @06:23AM (2 children)

      by Anonymous Coward on Sunday July 08 2018, @06:23AM (#704120)

      That's... that's not what's being simulated. Were you serious? Someone modded you interesting and not funny, though...

      • (Score: 0) by Anonymous Coward on Sunday July 08 2018, @11:20AM (1 child)

        by Anonymous Coward on Sunday July 08 2018, @11:20AM (#704176)

        That's... that's not what's being simulated.

        Yield, penetration and dissipation figures, as well as the related simulation models and results were the only research results pentagon bothered classifying from being published in research papers. Some raw data was also classified naturally but that's device specific. They released casualty estimates as part of the NATO war games as well as discussed wind and cityscape effects on dissipation post 9/11 referencing internal simulations. But overall the stuff that's being classified is the stuff designed to tell armies were to drop the bombs for maximum damage.

        • (Score: 2) by bob_super on Monday July 09 2018, @06:53AM

          by bob_super (1357) on Monday July 09 2018, @06:53AM (#704452)

          > But overall the stuff that's being classified is the stuff designed to tell armies were to drop the bombs for maximum damage.

          Did you fail to notice that everyone is talking about new warhead designs, but the big guys stopped having big boom parties to test them?

  • (Score: 2) by deimios on Sunday July 08 2018, @06:37AM (2 children)

    by deimios (201) Subscriber Badge on Sunday July 08 2018, @06:37AM (#704123) Journal

    Are you telling me that the country who is producing the vast majority of electronics can't produce the same electronics for themselves? (Physically speaking, since international IP law is the first thing to die once the real trade war starts).

    • (Score: 0) by Anonymous Coward on Sunday July 08 2018, @06:55AM (1 child)

      by Anonymous Coward on Sunday July 08 2018, @06:55AM (#704126)

      I worked for a major processor vendor out of college (early 90's). The president of the company called on all of the employees to write to congress to open up trade with china because it was viewed as a billion person market. Some guy near me did just that and had the letter along with a reply from a senator, Clinton (Bill) and the president of the engineering company framed outside of his cubical.

      The company folded a few years after that, unrelated to competition from China. They never got their billion person market and looks like they never would have. I always wondered if that guy regretted writing the letter.

      • (Score: 1, Insightful) by Anonymous Coward on Monday July 09 2018, @06:23PM

        by Anonymous Coward on Monday July 09 2018, @06:23PM (#704651)

        Dumb MBA fucks continued pushing that line about "a billion customers", long after their forced partners in China were forking off with the intellectual property to a new factory they set up down the street, and its output eating into third country markets.

        Nothing in China is real. Everything is set up to 1) keep the Communist Party in power and 2) to extract assets from foreigners to benefit (ethnic) Chinese.

(1)