Chinese chipmaker Zhaoxin is entering the APU space with its first product, the KX-6000G CPU, which offers up to 1.5 TFLOPs of GPU power.
Chinese domestic chipmaker Zhaoxin readies KX-6000G CPU with integrated 1.5 TFLOP GPU that is as fast as NVIDIA’s decade-old GT 630
To elaborate things, Zhaoxin is a Chinese chipmaker that offers CPU and GPU IPs based on their own architectures. The manufacturer’s next-generation KX-6000 CPU lineup will replace its KX-5000 and ZX-200 CPUs that have been on the market since 2017-2018. The latest chips are expected to have 4 cores, 8 threads, clock speeds of up to 3.0GHz, support for DDR4 DRAM, PCIe 3.0 I/O, and a 16nm local architecture.
Last month, we talked about how Zhaoxin’s KH-4000 series CPUs managed to deliver single-core performance close to AMD’s first generation Ryzen CPUs based on Zen architecture. Compared to the KX line, which is Aimed primarily at the home consumer segment, the KH series is primarily targeted at the server segment and offers up to 16 cores and 16 threads. It is also based on a 16nm process node, although the architecture is not as up-to-date as the KX-6000G CPUs.
So, coming back to the Zhaoxin KX-6000G CPUs, which we can also refer to as APUs, as they use the same Zhaoxin-made ZX C1080 GPU that is featured in the Glenfly Arise-GT-10C0 dGPU that was launched in the domestic market. Chinese. Just a few days ago. Now the same discrete chip has been packaged inside Zhaoxin’s KX-6000G CPUs. The naming scheme and chip layout closely resemble AMD’s Ryzen desktop APUs, which also use a discrete GPU architecture and have the “G” tag attached.
According to Zhaoxin, the CPU includes 8 cores and 16 threads, which is possible by merging two dies in the same package. The standard version of the KX-6000 CPUs only come with the integrated C-960 GPU which offers support for VGA, HDMI and DP in resolutions up to 4K. The new integrated GPU not only offers higher performance, but also better graphics capabilities.

For starters, in 3DMark 11, the Zhaoxin integrated GPU scores around 1,000 points, which is a more than threefold increase compared to the previous GPU. It comes with a new graphics and image processing engine that is compatible with Galaxy Kirin KOS, Tongxin Software UOS, Windows and several other major Chinese operating systems. The chip also offers full support for DirectX 11, Open GL 4.5 and OpenGL 1.2 APIs while rendering a native 4K display. While a big improvement over its predecessor, the integrated chip is still on par with an NVIDIA GT 630 graphics card in the 3DMark 11 (Performance) benchmark which scores similar points. (Vmodtech GT 630 Review Score):
The Zhaoxin KX-6000G CPU was also tested in GLMark 2 where it scored 3116 points, an NVIDIA GeForce GTX 1650 scored 10516 points in the same benchmark. As for the CPU, the Zhaxoin chip gets over 15,000 MIPS in multi-thread decompression, which puts it faster than the AMD Ryzen 3 1300X and close to Intel’s Core i5-7500. We used the following TechSpot review scores to make a comparison chart:
0
4000
8000
12000
16000
20000
24000
That said, Chinese GPU manufacturers are trying their best to reach parity with AMD and NVIDIA GPUs since 2016. Jing Jiawei and Innosilicon recently teased their plans to offer GTX 1080 and Vega-class performance in the next few years. years. Zhaoxin is also working on a range of x86 processors that can rival AMD’s first-gen Ryzen CPUs, while Loongson is targeting Zen 3 performance with its upcoming CPUs by 2023.
Products mentioned in this publication.