Zhaoxin's KX-7000
194 points by ryandotsmith 2 months ago | 57 comments- vessenes 2 months agoLove 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.
- chasil 2 months ago"Zhaoxin did not specify what process node they’re using. Techpowerup and Wccftech suggests it uses an unspecified 16nm node."
- vessenes 2 months agoThat’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.
- vessenes 2 months ago
- undefuser 2 months agoIf 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).
- chasil 2 months ago
- jkampman 2 months agoThis 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.
- luyu_wu 2 months agoAbsolutely 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)!
- clamchowder 2 months agoIt 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.
- jkampman 2 months agoTangential but thank you for always providing such detailed benchmarks and insights. Your work is a treasure!
- jkampman 2 months ago
- clamchowder 2 months ago
- sitkack 2 months agoMemory controllers are the biggest bottleneck (ha) to performant systems these days. The cores themselves are fine, but the memory controllers are slow and buggy.
- throwaway81523 2 months agoThis appears to be a new x86 design, but why? I thought there was good riscv-64 stuff out there now.
- RetroTechie 2 months ago(Binary) compatibility? Not everyone runs Linux, most software can't be recompiled. And emulation tends to come with a heavy speed penalty.
- spookie 2 months agoThey are "cooperating" with VIA. Basically, they didn't start from 0.
- RetroTechie 2 months ago
- luyu_wu 2 months ago
- mappu 2 months agoI 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.
- hawflakes 2 months agoMinor nit. Compound pinyin words shouldn’t use StudlyCaps so it should be “Lujiazui”
- powerapple 2 months agoI 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.
- powerapple 2 months ago
- IncreasePosts 2 months agoWhat'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?
- Havoc 2 months agoYeah. 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
- jandrese 2 months agoSeems 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.
- 999900000999 2 months agoPlus 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.
- mrandish 2 months agoI'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).
- 999900000999 2 months ago
- jandrese 2 months ago
- markus_zhang 2 months agoYeah 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.
- fspeech 2 months agoDo 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.
- wmf 2 months agoThe 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.
- fspeech 2 months agoThey work at companies or universities, and there is a "market" to look to for the pay scale.
- fspeech 2 months ago
- dghlsakjg 2 months agoThis initiative seems to be a private company propped up by government funds rather than direct government employment. Think Lockheed Martin not DARPA.
- wmf 2 months ago
- jenny91 2 months agoYes, it's very common in China.
- Havoc 2 months ago
- daniel_iversen 2 months agoThis 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?
- rjzzleep 2 months agoThey 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.
- MaxPock 2 months agoIf Huawei had access to TSMC and free of sanctions,it would have run so many companies out of town.
- MaxPock 2 months ago
- ksec 2 months agoIgnoring 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.
- RetroTechie 2 months ago> 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
- camel-cdr 2 months agohttps://github.com/OpenXiangShan/XiangShan
This one is mode than twice as fast.
- camel-cdr 2 months ago
- RetroTechie 2 months ago
- daniel_iversen 2 months ago[flagged]
- rjzzleep 2 months ago
- marcodiego 2 months agoHmmm.. it maybe free from IME! Maybe the FSF want a word with them.
- haunter 2 months agoNo IME but whatever unknown chinese rootkit? Out of the frying pan, into the fire
- marcodiego 2 months agoI can reason about an "unknown chinese rootkit" as much as about an "unknown US rootkit".
- marcodiego 2 months ago
- haunter 2 months ago
- Merrill 2 months agoHow would use of the Kylin OS instead of Windows 11 affect the user's perception of performance?
- hulitu 2 months ago> 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.
- hulitu 2 months ago
- gitroom 2 months agoBeen 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.
- snvzz 2 months agoNote 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.
- orangeboats 2 months agoNot 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.
- kjs3 2 months agoZhaoxin 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.
- happycube 2 months agoIt's also not the only answer, one Chinese company has rights to Zen 1. (which shows just how close AMD was to bankruptcy)
- tempeler 2 months agoRather 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.
- tempeler 2 months ago
- kjs3 2 months ago
- orangeboats 2 months ago
- cake-rusk 2 months agoWhy does anyone still need to license the x86 ISA? Haven't all / sufficient number of patents expired by now?
- tempeler 2 months agoit 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.
- ptx 2 months agoThey 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).
- tempeler 2 months agorisc-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.
- spookie 2 months agoYou 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.
- spookie 2 months ago
- tempeler 2 months ago
- happycube 2 months agoAs in the article, it doesn't have to be good. yet.
- ptx 2 months ago