hckrnws

Love reading these highly detailed analyses. Short version: Zhaoxin's currently competitive with 2010/2011-era AMD and Intel, with some asterisks around RAM speed.

There is to my mind a sort of race to get up to "fast enough to host H100 competitor AI hardware" with non-US IP that makes sense to engage in. In those terms, it looks like they're maybe 2 revs away -- I'm not sure what process node the KX7000 is on, but there's some architectural work to finish up. That said, this is interesting. I assume the chips will continue to improve from Zhaoxin, unless they lose their core team.

"Zhaoxin did not specify what process node they’re using. Techpowerup and Wccftech suggests it uses an unspecified 16nm node."

That’s pretty easily verified by a reasonably competent chip engineer and a microscope, if its 16 and they need to stay on shore only, they could be a little closer in terms of time than I speculated. SMIC at 7nm is reportedly stable; not sure about yields though.

I think if I were in charge I’d probably do a final architecture spin at 16, and then shrink that one to 7 or 5 if I could get it.

If the goal is to make a CPU that is "fast enough to host H100 competitor AI hardware", then why bother with x64? Huawei could have just produced a powerful ARM chip to go with their new AI processor. After all, Nvidia GB200 also uses Arm-based CPUs (the Grace in GB).

This review is an object lesson about why there is so much more to shipping a decent processor than making a CPU core with reasonable performance (and decent is being polite given that we are talking about Bulldozer-class single-threaded perf, which most folks were beyond thrilled to abandon when Zen arrived eight years ago.)

The behavior of the memory controller is wild to see in this day and age. You really don't want to see latency that high in general, but especially not for a client processor. I'd really like to see how it behaves with a reasonably powerful GPU in a CPU-bound gaming workload relative to the competition (to simulate what one of these might see in an internet café setting, for instance).

Power efficiency also seems truly dismal according to PCWatch: https://pc.watch.impress.co.jp/docs/column/hothot/1626253.ht... . In Cinebench MT, it's consuming about the same power as a Ryzen 5 5600G while delivering about 1/3 the performance, and the idle power is much higher than the Core i3-8100/R5 5600G to boot. That's not a huge issue for desktops, but it would not make a good foundation for a mobile system.

Overall an improvement versus past Zhaoxin efforts but people shouldn't kid themselves about the quality of the overall package here. There is a long way to go.

Absolutely a lomg way to go.

Interestingly, the chip is rated to run at DDR4-3200 or DDR5, so it's strange C&C got half that.

The power issues are likely from by modern standards pre-historical clocking behavior (single P-state to my understanding)!

It does clock ramp from 800 MHz idle to 3.2 GHz under load, with 900, 1000, 1100, 1300, 1500, 1800, 2200, and 2700 MHz steps in between until it hits 3.2 GHz after 71.6 ms. Article was getting long enough so I just left it at, it reaches 3.2 GHz and stays there even though the spec sheet says it should go higher.

I remoted into the system for testing (Cheese/George had it), and he said it took 3-4 cold reboots for it to come up, and suspected memory wasn't training correctly. So I did all the testing without ever rebooting the system, because it might not come back up if I tried.

Tangential but thank you for always providing such detailed benchmarks and insights. Your work is a treasure!

Memory controllers are the biggest bottleneck (ha) to performant systems these days. The cores themselves are fine, but the memory controllers are slow and buggy.

This appears to be a new x86 design, but why? I thought there was good riscv-64 stuff out there now.

(Binary) compatibility? Not everyone runs Linux, most software can't be recompiled. And emulation tends to come with a heavy speed penalty.

They are "cooperating" with VIA. Basically, they didn't start from 0.

I wonder if Zhaoxin's VIA heritage is helping them or holding them back - because of the patents, they were the only ones allowed to try, but since x86_64 and SSE2 are both now more than 20 years old, most of the patents don't matter any more (and AVX is not far from the cutoff).

The breakaway ARM China or SpacemiT or Loongson could drop in an x86_64 frontend and might get better results.

Minor nit. Compound pinyin words shouldn’t use StudlyCaps so it should be “Lujiazui”

I think it should. Chinese language does not have the concept of word in writing, there is no space, and each character is a unit in writing. Pinyin was to mark pronunciations of each characters, it would be Lu Jia Zui, meaning 3 characters. Pinyin is not English. LuJiaZui would make it easier to pronounce if you don't know Pinyin. There is no standard way to write it though, maybe Lu-Jia-Zui, Lu'Jia'Zui, or Lu'jia'zui. The most confusing and would lead to wrong pronunciation version is Lujiazui.

What's the deal with the municipal government being a partner in this project? Is that structure common in china? Is it just them giving VIA tax breaks and things, or are they more involved than that?

Yeah. They know the chips aren’t commercially competitive so they just create artificial demand by making gov and state controlled entities buy it.

Basically an attempt to bootstrap an industry brute force style

Seems like it depends on the price point. These chips might be slow by modern standards, but if they're cheap enough then it doesn't really matter for a lot of the potential applications. I'm typing this post on a chip that is roughly in that performance bracket (an i5-3750k) that only rarely feels like the bottleneck. And this is my gaming machine.

Plus the vast majority of work computers don’t need to be particularly fast. Add a lightweight Linux distro, and that’s more than enough for paperwork.

Early into high school, I needed something to take to class, but since I already had a decent desktop at home, plus we were broke, I picked up some cheap Asus K55N; AMD A8-4500M, 4GB DDR3, etc -- nothing particularly fancy; only upgrade I did to it was removing the mechanical hard drive and swapping in my old 120GB Corsair Force GT.

I eventually upgraded, went off to university, etc. When I finally came home, I found out my mom apparently "borrowed" it, figured out how to install Ubuntu, and has been using it ever since for grading papers and what not.

No idea how much longer it will remain in use, but aside from the awful screen, ironically, honestly, I think the browser and the seemingly ever increasing resource requirements of the web will eventually be the only thing that finally causes an upgrade.

I've got a desktop with a Core2Quad Q9300, originally built in 2008. Updated the motherboard in 2010 to one with USB3, updated the GPU in 2012. Kept using that machine as my daily-driver until 2020, when I built my current desktop. The only real reason that I did that was to improve gaming performance. The machine itself still functions fine on the modern web (using an up-to-date Linux distro, of course). 8GB of RAM used to feel like a monster.

As is the Chinese web tends to be less full of complex JavaScript and CSS tricks.

It wouldn't surprise me if these processors are more than fine for municipal workers.

I'm running a well-optimized 2014 Haswell I5-4590 system with a Radeon 7800 XT in my virtual pinball cabinet. It's handling real-time 3D at 1440p 120fps at medium-high settings and VPX is pretty CPU heavy. My system is probably only a generation or two ahead of the one described (although it's true that Haswell was one of those occasional Intel generations that became legendary for outperforming and generally aging very well).

Haswell/Broadwell with embedded 128MB of eDRAM Level 4 cache are some extremely awesome hidden gems.

That's very lopsided in favor of the GPU - if VPX is more CPU intensive that the average video game you could probably swap the 7800 xt for something much cheaper and get the same performance.

Yes, I agree. I was actually planning to retire that mobo and CPU after the GFX upgrade but that damn Haswell is so good, I didn't need to. The previous GFX card (a 1080) was the bottleneck getting 120fps reliably. I really didn't expect the i5-4590 to keep up with 120fps at low latency but got surprised.

The Intel i5 4000 series was an odd duck and keeps pace still today in a lot of things. I still have one running my Plex workload kicking around.

Yeah it is pretty common. Governments invest in key area corporations to provide fund, tax breaks, regulatory aid and a bunch of other benefits, and sometimes sell its chunk of shares in a few years.

One early example is Chongqing government with Huang Qifan as mayor back in the 2010s.

Do governments allow some of their employees to be highly compensated relative to others? Would someone with real expertise in chip development work for the government at what the government is willing to pay? I think the answer is no.

The Chinese government has definitely "bought back" some top talent from the US. It's probably a small number of people.

I'm not sure why local governments would get involved although in general China has had a problem with too much investment and not enough places for it to go. It's not impossible that there are essentially local sovereign wealth funds.

They work at companies or universities, and there is a "market" to look to for the pay scale.

This initiative seems to be a private company propped up by government funds rather than direct government employment. Think Lockheed Martin not DARPA.

Yes, it's very common in China.

This is interesting! Does anyone know how China’s reliance on chips from intel and amd is in the non-AI space (so regular consumer and server loads)? I’m wondering how it was 10 and 5 years ago, now, and how we predict in the next couple of years. Surely if they’re not mostly using their own chips they will very soon right?

They push local chips for independence, but unless the west embargoes them, I don't think you will see a major leap forward.

Huawei is a whole different beast though. They have a everything from the chip design up, and by now also an operating system that has arguably both a better frontend framework and a better kernel that the Linux alternatives. When we talk about Chinese AI chips being slow we specifically talk about classic desktop chips.

Also! For normal desktop work a 2011 intel chip is plenty fast. A lot of critical systems like train booking systems are keyboard focused ancient UI systems, and they seem fine.

If Huawei had access to TSMC and free of sanctions,it would have run so many companies out of town.

Ignoring Fabs for now which is a different sets of issues.

They have JV with ARM ( ARM China ) and AMD ( Zen 1 ), IMG ( PowerVR and MIPS ) Along with investment on RISC-V. Alibaba and Huawei are all investing into RISC-V as well. Considering they dont sell CPU I wont be surprised if China one day give away RISC-V CPU design for free.

Surprisingly ARM China issue is still somewhat unresolved and ARM now has a separate subsidiary inside China.

> Considering they dont sell CPU I wont be surprised if China one day give away RISC-V CPU design for free.

Already done: https://www.cnx-software.com/2021/10/20/alibaba-open-source-... https://github.com/XUANTIE-RV

https://github.com/OpenXiangShan/XiangShan

This one is mode than twice as fast.

[flagged]

Hmmm.. it maybe free from IME! Maybe the FSF want a word with them.

No IME but whatever unknown chinese rootkit? Out of the frying pan, into the fire

I can reason about an "unknown chinese rootkit" as much as about an "unknown US rootkit".

How would use of the Kylin OS instead of Windows 11 affect the user's perception of performance?

> How would use of the Kylin OS instead of Windows 11 affect the user's perception of performance?

Windows 11 is dog slow on corporate hardware. Linux, even with bloated KDE or Gnome, is much faster.

Been interesting following Zhaoxin, but yeah, looks like there's still a mountain to climb before these chips hit the big time. Kinda wild they're still so far behind, but I get why China wants to push their own stuff anyway.

Note there are competent RISC-V architectures in China which might already be faster at emulating x86 than the KX-7000 is at running it directly.

Not just RISC-V, when it comes to performance the Loongson CPUs (with the LoongArch ISA) are likely the most competitive -- the IPC of Loongson 3A6000 is around that of 10th ~ 12th gen Intel CPUs! And 3A7000 is coming soon.

Zhaoxin is China's answer to "what if we need a drop-in x86 replacement immediately?" It does not represent the frontier of CPU development in the country.

Zhaoxin is China's answer to "what if we need a drop-in x86 replacement immediately?"

Exactly this. China isn't going to be doing HPC on these, or building domestic clouds, or wooing gamers. This is a strategic project so Chinese users can continue to access the x86 ecosystem regardless of 'supply chain issues'.[1][2] And much of that ecosystem can get by with a fraction of the compute of the average gamer, it's just fine that it hits mediocre performance numbers, for now.

[1] And, sure, some national pride. No harm in that. [2] No, "emulators!" is not the only answer.

It's also not the only answer, one Chinese company has rights to Zen 1. (which shows just how close AMD was to bankruptcy)

Rather than signaling AMD's bankruptcy, this could be seen as a strong indication of how heavily China is investing in legacy technologies. That's why I wouldn't be so quick to draw conclusions. China’s eagerness in this area doesn’t mean that the investments are efficient. What truly matters here is the ability to build a full ecosystem. They practically wiped Huawei off the global smartphone market simply by banning Google services on their devices.

The license was done in 2016, though. It does not cover any later versions, and even if it did it would be cut off due to export restrictions.

Why does anyone still need to license the x86 ISA? Haven't all / sufficient number of patents expired by now?

it seems bad investment. firstly, it's an old architecure. secondly, How will they measure the efficiency of their investments?.Instead of investing in outdated architectures, they could fund universities, or support at least two or more companies in a competitive setup — that way, something much more efficient might emerge. But when a company is state-funded in a monopolistic way, it's quite hard for it to succeed. The U.S. had a reason for supporting AMD against Intel back then.

They are doing that as well. There's a lot of investment in RISC-V: https://asiatimes.com/2025/03/china-all-in-on-risc-v-open-so...

For x86, supporting multiple companies would be more difficult, as they would need a license for the ISA (which Zhaoxin inherits from Via).

risc-v is much more reasonable for countries like china.What I want to emphasize is this: Instead of paying license fees for outdated x86 architectures, it would be much more beneficial to allocate those resources to universities. If you're paying for a license but can't generate a return on that investment quickly, then something is wrong.

You are completely misunderstanding China's goal. They want the expertise, being self-sufficient, while taking what they can so they don't start from 0.

Risc-V is arguably risky, so they have their hands on everything at the same time so they're sure they have a winner.

Markets don't matter. They want to develop the technology and have the talent to achieve it, fast.

As in the article, it doesn't have to be good. yet.