Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
China Technology

Made in China 8-Core x86 CPU Released (techspot.com) 101

Posted by msmash from the how-about-that dept.
AmiMoJo writes: Zhaoxin, a fabless chip maker based in Shanghai, has produced a homegrown x86 CPU line that's apparently ready for the DIY scene. The Zhaoxin KaiXian KX-6000 series of processors were originally shown off in 2018, but since then we had heard little about them. Now it seems that the KX-U6780A will come to market this quarter, as listed on Chinese retail site Taobao with a March release date. For the uninitiated, Zhaoxin is a joint venture between VIA Technologies and the Shanghai Municipal Government. Zhaoxin's current CPU designs have origins in Centaur Technology, a company acquired by VIA in 1999. The VIA Nano Isaiah core design, built by Centaur, would serve as the architecture for Zhaoxin's first CPUs.

The Isaiah design was Centaur's first superscalar CPU capable of out-of-order execution. This is what seemed to pave the way for Zhaoxin's in-house designed LuJiaZui cores, as they too are built around a superscalar, out of order architecture. LuJiaZui seems to be an iterative migration from the Wudaokou architecture, but also supports modern instruction set extensions such as AVX and SSE4.2, which is an important evolution for China and its domestic CPU goals. The KaiXian KX-U6880A appears to be the series flagship, with the slightly lower clocked KX-U6780A slotting in just beneath it. All KX-6000 series chips are based on the LuJiaZui architecture, boasting eight cores and eight threads.
This discussion has been archived. No new comments can be posted.

Made in China 8-Core x86 CPU Released More

Made in China 8-Core x86 CPU Released

Comments Filter:

  • Supposedly it is as fast as a Core i5-7400 (http://www.globaltimes.cn/content/1155069.shtml) which would not be bad for this 16nm TSMC built CPU. Of course no way to verify such claims currently.

  • Keep in mind... (Score:3, Funny)

    by Anonymous Coward on Monday February 03, 2020 @12:32PM (#59685496)

    It comes with the Corona virus pre-installed...

  • And the GPU is... (Score:5, Informative)

    by williamyf ( 227051 ) on Monday February 03, 2020 @12:40PM (#59685548)

    Is based on the S3 Chrome 640 architecture, with DirectX 11. Slightly slower than an intel iGPU.

    Yes, VIA also owns S3, and even resurrected the brand for a short while. A merger of S3 and Via technology has been the foundation of all their GPUs, whether they are mobo dGPUs, iGPUs on the northbridge, or lately, iGPUs on the core itself.

    And for the record, the chips come from the lineage of Cyrix, basicaly VIA acquired Cyrix and Centaur (at the time, a subsidiary of IDT) , and in the end, the Cyrix technology was the better one, but the centaur name remained.

    I am happy that Zhaoxin/VIA gets back in the limelight, as people tend to forget that there are three vendors of X86/x64 technology. Hope they do well. More competition is always welcome...
     

    • How much of the VIA tech was inherited from their purchase of the x86 tech from National Semiconductor?

      I can remember back when their were quite a few x86 vendors....like Transmeta, and IBM.
      https://en.wikipedia.org/wiki/... [wikipedia.org]

      • Re:And the GPU is... (Score:5, Informative)

        by williamyf ( 227051 ) on Monday February 03, 2020 @01:17PM (#59685808)

        How much of the VIA tech was inherited from their purchase of the x86 tech from National Semiconductor?

        I can remember back when their were quite a few x86 vendors....like Transmeta, and IBM.
        https://en.wikipedia.org/wiki/... [wikipedia.org]

        Considering that Cyrix formed the basis of most if not all of national semiconductor's X86 technology, I would said that: a lot. But actually, the licenses and patents that came from Cyrix were vital for VIA to get use of "modern X86 (Pentium 2 ISA and sockets onwards)" in front of a litigious Intel. But the Centaur tech was the thing that kept the processors viable for a while. I made a mistake on my previous comment by saying that the processors were centaur branded but cyrix inside

        In the 90's there were plenty of companies trying to compete with intel (in the 80s there were even more!). You had AMD, Intel, Centaur/Winchip, IBM (recycling stuff from cyrix, Nextgen and others), nextgen, DP&M, Rise and Cyrix making X86 processors. Transmeta apperaed in the very late 90's and early 00's.

        Fun fact #1. The first AMD-64 chip was a transmeta chip. Since transmeta chips were VLIW emulating a target ISA that was in RAM, AMD made an alliance with transmeta to implement their AMD-64 ISA for testing and SW development purposes ahead of actual silicon shipments.

        Fun fact #2. Actually, when AMD's K5 processor proved to be an utter failure, AMD acquired a company called NextGen, and the K6 and latter processors were based on that technology.

        Then that market dwindled in the vine, with only AMD, Intel and Via making x86-64, and a few niche others making x86-32 only chips for specialist markets. Similar to mechanical hard drives, where we only have Seagate, westerndigital (hitachis is a whole owned subsidiary of WD) and Toshiba.

      • Re: (Score:2)

        by sconeu ( 64226 )

        Had a NextGen Nx586. They had a variant that was 486 pin-compatible, if I recall correctly?

        • Had a NextGen Nx586. They had a variant that was 486 pin-compatible, if I recall correctly?

          Probably not. AMD, Cyrix, Rise and winchip had 486 pin compatibles. But Nextgen had its own bus, with the processor soldered to the mobo

    • IDT WinChip in my i-Opener. A loss leader computer that was ruined by Linux tinkerers like myself.

      • IDT WinChip in my i-Opener. A loss leader computer that was ruined by Linux tinkerers like myself.

        Cue the CueCat; If companies make boneheaded desicions, is not our fault. They were using QNX, could have gone with some cheap and cheerful RISC microprocessor supported by QNX, instead of going with a X86 compatible processor.

        • Oh I'm not shedding tears, or rather I'm not surprised. I can't even put out a candy bowl at work without an adult putting both hands in and stuffing their pockets with free candy. But I wonder if the product was helpful to people who maybe couldn't afford or didn't want a full PC? And yet somehow could afford overpriced dial-up service. Seems like getting a used PC and a cheap/free dial-up like NetZero is a better option for a tight budget.

          I think a contemporary of that IDT C6 would have been a StrongARM (

          • Re: (Score:2)

            by Bert64 ( 520050 )

            There were many more RISC chips around in those days, the SA-1100 was aimed at embedded use but there were numerous other RISC processors, including very high performance ones like Alpha.

            • Best bang for buck in 1999-2000 was an x86 PC. Not an Alpha. (those were about $1500/chip when they came out!)

              Corel Netwinder was either the same year or a year after the i-Opener and much higher performance with a 275MHz StrongARM, but almost certainly much more expensive to build.

      • What a fun box that was. I beat the heck out of mine, just as an experiment, then as a gadget for a teenager. Pizza Key!

    • I had a Cyrix MII [wikipedia.org] back in the Socket 7 days. It was better supported under Linux than Windows. Ran my file server for several years.
    • All of those chips were crap. I owned Citrix and Via because I was poor and needed/wanted something. But they ran really fucking hot -and- slow. If they're vastly improved all these years later without stealing from Intel or AMD, more power to em but the "DNA" source is trash if that's what these are based on.

      • All of those chips were crap. I owned Citrix and Via because I was poor and needed/wanted something. But they ran really fucking hot -and- slow. If they're vastly improved all these years later without stealing from Intel or AMD, more power to em but the "DNA" source is trash if that's what these are based on.

        The last Cyrix chip I used was a 486-80 Mhz. From then on, either Intel or AMD. So, I can not comment. But having said that, I made a mistake in the OP, the VIA chips technology come from centaur, not Cyrix. The cyrix tech was scaped by the time the C3 hit the market. What was relevant from Cyrix was patents and Cross license agreements...

        Having said that, I agree with you that more competition is better, and hope that the chips are what they say in the tin.

        But, to give you a glimmer of hope in VIA's redemp

        • I can't recall if mine was 486 clone or 586 clone. Maybe I had a 386 and a 586 later? Anyway, yes, competition is great. If AMD hadn't got their shit back together again we'd still be on 2 ghz burst speed 4 core Intels and going gaga over 2.4 ghz and ddr4 memory support for 16 gigs ram, but only on enterprise chips, mind you. Can't use that really good stuff for the masses. Ddr2/3 is good enough for them.

    • Re: (Score:2)

      by AmiMoJo ( 196126 )

      I'm looking forward to these hitting the market. Currently the preferred platform for PFSense is an Intel x86 box but these could get the price down significantly.

  • Is the backdoor instruction set documented? (Score:3, Insightful)

    by Rick Schumann ( 4662797 ) on Monday February 03, 2020 @12:46PM (#59685584) Journal
    Or does the rest of the world get to figure them out on their own?

    • aah, you beat me to this comment. i wouldn't buy a system based on it unless i were building a honeypot for spies.

    • Re:Is the backdoor instruction set documented? (Score:5, Informative)

      by williamyf ( 227051 ) on Monday February 03, 2020 @01:28PM (#59685876)

      Or does the rest of the world get to figure them out on their own?

      The backdoors in these processors are as well documented as the NSA backdoors that Edward Snowden comented, or the AMD fTPM backdoors, or the intel SGX backdoors.

      Now is up to each one of us to decide from which state we get our backdoors, and how much are we willing to pay for the processors that have said backdoors.

      Since for me (and most of my customers) in LatAm is indiferent if Russia, China or the Five Eyes spy on me/(us), we go for the most cost effective solution, and since we know we are being spyed on, we use the savings to implement security in depth.

      JM2C, YMMV

      • Some days I wish I was still using a 6MHz Z80 running CP/M on a dumbterminal.

        • Some days I wish I was still using a 6MHz Z80 running CP/M on a dumbterminal.

          You can always run Contiki OS on it, and get cooperative multitasking, a GUI and IPv6 on your Z80

          • I don't have any of my old Morrow Designs stuff anymore so it's a moot point but out of curiosity I went and looked at that and I didn't see anything that said it was ported to Z80 code.

    • Re: (Score:2)

      by AmiMoJo ( 196126 )

      If such backdoors exist they will be extremely well hidden. Imagine if an NDA backdoor were found in Intel chips. It could destroy them, especially if it couldn't be patched.

  • As the patents expire... (Score:5, Interesting)

    by DMJC ( 682799 ) on Monday February 03, 2020 @12:52PM (#59685614)
    X86-64 is hitting public domain this year. Over the next few years the patents on most of the 64-bit technology in use today will hit public domain and Intel/AMD will no longer have a monopoly over manufacturing rights for CPUs that most consumers need. Interesting times should ensue.

    • Re: (Score:2)

      by aitikin ( 909209 )
      I was under the impression the pertinent [espacenet.com] applicable [espacenet.com] patents [espacenet.com] were filed in '05, making them enforceable under US law until 2025 [wikipedia.org] unless there was an early termination for some reason?

      • I was under the impression the pertinent [espacenet.com] applicable [espacenet.com] patents [espacenet.com] were filed in '05, making them enforceable under US law until 2025 [wikipedia.org] unless there was an early termination for some reason?

        Just reading from your links, it seems they where all provisionally filed 9 aug 2000 and then filed 2 april 2001. Tbe 2004-2005 dates are the date of patent (granted). Afaik the general rule is (since 1995) that the patent expires 20y after the earliest filing date. Were the patents extended because the uspto didn't examine them in time (they all seem to qualify for the processing extension) and do you know where can I find that information? Any insight appreciated.

    • X86-64 is hitting public domain this year. Over the next few years the patents on most of the 64-bit technology in use today will hit public domain and Intel/AMD will no longer have a monopoly over manufacturing rights for CPUs that most consumers need. Interesting times should ensue.

      Maybe the patenst on the initial X86-32 and X86-64, but the patents on SSE1,2,3,4 32, SSE1.2.3.4 64, AVX 256,512, AMD virtualization extensions, intel Virtualization extensions, and other more recent technologies have a long way to go.

      Also, the knowhow to make these processors is so complex, that new entrants have a strugle ahead of them.

      For example, the compatibility with real mode and 286 mode is such a bitch, that is easier for intel, amd and Via to keep it there, than to risk removing it and dealing wit

      • Re: (Score:2)

        by amorsen ( 7485 )

        For example, the compatibility with real mode and 286 mode is such a bitch

        Is it really? I can see difficulties for 286 mode, but chips could leave that out and no one would notice. What is difficult about real mode?

        Life would be a lot easier if they could leave out 32-bit mode with segments and virtual 8086 and all that.

        • That's the Apple kind of strategy. So now none of the legacy Mac software will work at all.

          • To be completely honest, having gone through Apple's shifts from Motorola 0X0 => PowerPC, and then PowerPC => Intel, I really didn't cry once the shifts had been made from one version of software to another. Leaving the legacy versions meant paying some money to the developer/publisher, but it was usually worth it because the performance and feature improvements made a huge difference. Of course, this was back in the time before subscription software models, and your initial buy-in price for packages

          • Re: (Score:2)

            by Bert64 ( 520050 )

            There are plenty of processor emulators out there, which you can use to emulate an obsolete processor and install an obsolete os to run all that legacy software.
            Often this kind of emulator is used for games or for educational purposes, because software you might actually need usually has a modern replacement.

            But the percentage of users who actually do this is very low, so better not to bloat up every user's system with junk that most of the users will never use.

            • Often this kind of emulator is used for games or for educational purposes, because software you might actually need usually has a modern replacement.

              Businesses don't like having to buy all their software over and over. The kind of 'obsolete' you describe is marketing dude hype.

      • Re: (Score:1)

        by AHuxley ( 892839 )
        Re "vendors for the near future"
        Not if a nation has the skills and ability to design an approved and tested new OS for the new hardware.
        Good for office work, city work, digital "paper work", games, the gov approved internet, work at home...

      • Actually, SSE 2 was included in the Athlon 64 (and have become an assumed minimum level of hardware SIMD support for all AM64 parts.)Therefore I'm fairly sure the specifications are falling out of patent protection.

        Also, the basics falling out of patent protection mean
        ARM64 on Windows 10 can finally have an x64 emulator! I's expensive to build a new competitive x86 processor, but now building an unlicensed AMD64 emulator for other CPU platforms suddenly becomes viable.

        • Actually, SSE2 debuted in the Pentium 4 in 2001, so SSE2 chips have been on the market now for almost 20 years.

          Perhaps more interestingly, the Athlon XP which also came out in 2001 lacks SSE2, which definitely puts at the top of the list of "most powerful x86's CPU that lack SSE2".

  • I hope that they kept the CPU ID as CENTAURHAULS. I miss that one.

  • Does this CPU have anything like Intel/AMD hidden management core [wikipedia.org]?

  • Why Shanghai Municipality invests into chip maker?

    Re:
    https://en.wikipedia.org/wiki/... [wikipedia.org]

  • how meny pci-e lanes?

  • I'm sure the NSA will be pleased that their backdoor was copied along with the CPU design from intel

  • VIA was king when Pi was just a math symbol (Score:3)

    by Jzanu ( 668651 ) on Monday February 03, 2020 @05:48PM (#59686982)
    VIA always was better for low power (actual low power) PC functions. Need a true homebrew network interface for your small office without spending 50k? Get a VIA System using a PC clone a slotket adapter, then add all the network cards that the PCI/ISA bus supports. Sure now there are other/better alternatives now, but not as much in 1999.
  • good for your advanced math and open source encryption attempts..
  • If the referenced product release price is for the CPU alone or the CPU and motherboard?

Slashdot Top Deals

"An idealist is one who, on noticing that a rose smells better than a cabbage, concludes that it will also make better soup." - H.L. Mencken

Close