China-Made Zhaoxin CPU Hits Retail Market

32

KaiXian KX-U6780A (Image credit: 二斤自制/YouTube)
Zhiye Liu
News Editor, RAM Reviewer & SSD Technician

Zhiye Liu is a news editor, memory reviewer, and SSD tester at Tom’s Hardware. Although he loves everything that’s hardware, he has a soft spot for CPUs, GPUs, and RAM.

  • artk2219
    I would love to see a comparison vs AMD's FX line.
    Reply
  • NightHawkRMX
    Hmm, even though this is a Chinese CPU, it probably has fewer security vulnerabilities than Intel at this point.
    Reply
  • GoatGuy
    NightHawkRMX said:
    Hmm, even though this is a Chinese CPU, it probably has fewer security vulnerabilities than Intel at this point.

    While technically true, one can be almost completely certain that it also has additional Chinese Government Back-Door vulnerabilities. Like 99 out of 100 chances.
    Just saying,
    GoatGuy
    Reply
  • CrisCo71
    That price is laughable - it's B450 w/Ryzen 5 3600, 16GB RAM and an Nvidia 1660 territory...
    Reply
  • Giroro
    NightHawkRMX said:
    Hmm, even though this is a Chinese CPU, it probably has fewer security vulnerabilities than Intel at this point.

    In Chinese, security vulnerability is spelt
    辅助功能增强
    Reply
  • cryoburner
    artk2219 said:
    I would love to see a comparison vs AMD's FX line.
    Well, checking some numbers online, it looks like 2012's 8-thread FX-8350 at stock clocks typically scores close to 1300 at Cinebench R20 multi-core, and around 1500 overclocked, compared to 845 for this processor, making that older FX chip around 50% faster at stock. The margins in CPU-Z's single-thread test look similar. The FX processor is also clocked around 50% higher though, so that difference may be mostly down to clock-rates rather than IPC.

    As for current parts, it looks like AMD's 4-core, 8-thread 3400G Zen+ APU scores close to 2000 at Cinebench R20 multicore, with a similar difference in CPU-Z single core, making it close to 150% faster. Of course, there's also the 6-core, 12-thread 1600 AF (basically a slightly underclocked 2600) available in the US market for just $85 now, which scores over 2800 in R20 multicore. And the Zen2-based 3600 scores around 3700 points in that test, well over four times as fast at that multithreaded workload, and around 500 in CPU-Z single-threaded, around three times as fast.
    Reply
  • InvalidError
    NightHawkRMX said:
    Hmm, even though this is a Chinese CPU, it probably has fewer security vulnerabilities than Intel at this point.
    It is highly unlikely that AMD's CPUs are exempt of any such flaw, the question is whether they will be discovered within the architecture's useful life if ever. Intel's biggest problem here is that it has been reusing mostly the same architecture for nearly 10 years, so most flaws have a pretty good probability of affecting a large chunk of all of Intel's CPUs from the past decade. That's a ridiculously large attack surface area on chips that account for ~80% of the market for hackers to work and cash in on.
    Reply
  • bit_user
    InvalidError said:
    It is highly unlikely that AMD's CPUs are exempt of any such flaw,
    Leaving aside your claim, why are you even talking about AMD?

    This is made in partnership with VIA, meaning it uses Centaur's x86 cores. Nothing to do with AMD.

    Edit: I'm not sure about the Centaur part, as Anandtech and WikiChip both describe its LuJiaZui cores as an in-house design. Wikichip says of its predecessor:

    WuDaoKou is largely a brand new architecture designed by Zhaoxin. This is a departure from earlier microarchitectures such as ZhangJiang which were a lightly modified version of VIA Technologies (Centaur) architecture.

    So, it sounds like the previous & current core designs might be truly 100% Chinese designs.
    Reply
  • InvalidError
    bit_user said:
    Leaving aside your claim, why are you even talking about AMD?

    This is made in partnership with VIA, meaning it uses Centaur's x86 cores. Nothing to do with AMD.
    My comment was regarding the Intel part of the comment I was replying to, though it is generally pretty safe to assume that any highly complex chip is going to have some number of bugs that manage to sneak by engineers, especially when they involve side-channel type exposure which can be extremely difficult to foresee at the design stage.
    Reply
  • bit_user
    Do we know anything about its iGPU?

    Edit: Wikichip mentions S3, since I guess VIA bought its IP. If it has anything to do with S3, it's probably unrelated to their graphics chipsets from the old days. Anyway, it seems the graphics are on-die.

    This is actually the entry from its predecessor: https://en.wikichip.org/wiki/zhaoxin/microarchitectures/wudaokou
    Edit 2: It's rumored to be the VIA C-960, which is derived from S3's Chrome series.
    Reply