News on China's scientific and technological development.

S

schrage musik

Junior Member
Registered Member
Tam said:
Because they did use the MIPS instruction set under license. Now that MIPS is practically dead all around the world except for China and Loongson. The new architecture appears to be an instruction superset of MIPS, like a new branch. So no surprise it would use MIPS instructions and libraries.
Click to expand...

Why spend all that effort resurrecting something like MIPS that everyone else has abandoned? Nobody wants MIPS processors. A high performance RISC-V processor, on the other hand, could bring in a lot of interested customers.

Another thing I keep seeing is the chinese industry just seems obsessed with the hardware components without putting much effort into the core system software required. Take for example operating systems. Not much effort into producing something like Integrity, QNX or VxWorks. (yes, I know of Huawei's Harmony) Again, nothing remotely comparable to formally verified programming languages like Ada and SPARK or parallel/distributed languages like Cray's Chapel or GPU programming frameworks like CUDA.

You're not going to be able to develop an ecosystem around your shiny new hardware unless you have outstanding tooling for the coders to make use of those capabilities. The result will be that even when China develops a supercomputer that is faster in raw benchmarks, with something like Chapel or CUDA a less powerful american supercomputer will outperform the Chinese machine in the real world.

It is like developing a fifth-gen fighter like J-20 with no fly-by-wire controls. In a dog fight, even a Gripen with FBW will outperform it.
 
ZeEa5KPul

ZeEa5KPul

Colonel
Registered Member
schrage musik said:
Again, nothing remotely comparable to formally verified programming languages like Ada and SPARK or parallel/distributed languages like Cray's Chapel or GPU programming frameworks like CUDA.
Click to expand...
Why would you do that when all of this is open source? And who the f uses Ada anymore?
 
krautmeister

krautmeister

Junior Member
Registered Member
schrage musik said:
Why spend all that effort resurrecting something like MIPS that everyone else has abandoned? Nobody wants MIPS processors. A high performance RISC-V processor, on the other hand, could bring in a lot of interested customers.

Another thing I keep seeing is the chinese industry just seems obsessed with the hardware components without putting much effort into the core system software required. Take for example operating systems. Not much effort into producing something like Integrity, QNX or VxWorks. (yes, I know of Huawei's Harmony) Again, nothing remotely comparable to formally verified programming languages like Ada and SPARK or parallel/distributed languages like Cray's Chapel or GPU programming frameworks like CUDA.

You're not going to be able to develop an ecosystem around your shiny new hardware unless you have outstanding tooling for the coders to make use of those capabilities. The result will be that even when China develops a supercomputer that is faster in raw benchmarks, with something like Chapel or CUDA a less powerful american supercomputer will outperform the Chinese machine in the real world.

It is like developing a fifth-gen fighter like J-20 with no fly-by-wire controls. In a dog fight, even a Gripen with FBW will outperform it.
Click to expand...
This has been the pattern of China IT development since forever. The only serious attempts China has made at integrated core systems software is KylinOS, which is simply a Linux distribution. They use this in a lot of government offices and the military. Before this, they used Windows in most government offices, including the military. How stupid! At least MIPS is going to be pulled from future iterations of Godson and replaced with RISC-V.

Ten years ago, there was a research project called the Godson-T. This was a manycore 64 tile processor and there were plans to develop it into 1000+ tiles that would run symmetrically in parallel. They had this fully working by 2011 and was one of the research projects that allowed China to advance to the leading edge in high performance computing. There is none of this kind of research happening on the software side except for Huawei's HarmonyOS. The lack of vision and more important, imo, the lack of confidence is the reason why China isn't further along than it is now and it's already doing gangbusters.
 
Last edited:
B

BlackWindMnt

Captain
Registered Member
From what I have seen of HarmonyOS is that it will most likely also be Linux distribution with a big company backing it. Or atleast abstract away the Linux kernel. Too be honest it's something the Linux desktop could really need.

Haven't seen any evidence that the harmony kernel is actually still alive, it will be most likely Google that will probably succeeds bringing a open-source micro kernel designed OS to the mass market.

Not sure if the many core design was actually viable for mass market. If I'm not mistaken Intel knight corner platform never really took off.

I would like to see a Chinese spin on a modern C programming language something like Zig for the developers under us.
 
T

Tam

Brigadier
Registered Member
schrage musik said:
Why spend all that effort resurrecting something like MIPS that everyone else has abandoned? Nobody wants MIPS processors. A high performance RISC-V processor, on the other hand, could bring in a lot of interested customers.

Another thing I keep seeing is the chinese industry just seems obsessed with the hardware components without putting much effort into the core system software required. Take for example operating systems. Not much effort into producing something like Integrity, QNX or VxWorks. (yes, I know of Huawei's Harmony) Again, nothing remotely comparable to formally verified programming languages like Ada and SPARK or parallel/distributed languages like Cray's Chapel or GPU programming frameworks like CUDA.

You're not going to be able to develop an ecosystem around your shiny new hardware unless you have outstanding tooling for the coders to make use of those capabilities. The result will be that even when China develops a supercomputer that is faster in raw benchmarks, with something like Chapel or CUDA a less powerful american supercomputer will outperform the Chinese machine in the real world.

It is like developing a fifth-gen fighter like J-20 with no fly-by-wire controls. In a dog fight, even a Gripen with FBW will outperform it.
Click to expand...


Its not a MIPS processor. Its a Loongson processor. Its only talking with the MIPS instruction set. Get it, instruction set. RISC V is an instruction set. The difference is the machine language it reads. Not sure if Loongarch is a new architecture, or a superset of the MIPS instruction set. Is there a better or superior RISC instruction set? That's like asking which human language used in the planet right now is superior. The appeal of ARM is not because of its technological superiority, but like X86, it has a large base of established applications, operating systems and tools for it. The appeal of RISC-V is that it is open and you don't need a license for it. MIPS, PowerPC, SPARC, ARM, RISC V are all in the same RISC banana.

Loongson has a favored audience that includes the government, the science institutes and the military. It has its own self sustaining ecosystem. If you don't think that's successful, look at the Chinese satellite networks and space probes in the sky.

China probably has their own bunch of RT/OSes.

Please, Log in or Register to view URLs content!

This one is used in automotive.
Please, Log in or Register to view URLs content!

Please, Log in or Register to view URLs content!

Harmony is probably positioning itself in the RT/OS field also.
 
G

gadgetcool5

Senior Member
Registered Member

Scientists in China ‘struggle to get instruments because of US export controls'​

An AI system was trained on a database of detailed information about most hardware used in laboratories across China and found that nearly 1,900 items (42 per cent) on the US [export ban] list were scientific instruments.

Scientific instruments are one of the few commercial products not usually made in China. None of the world’s top 20 scientific instrument suppliers are Chinese.

According to Wang’s study, the instruments on the US export control list were not isolated, but often linked to others because of shared technologies or fields of application. An X-ray machine, for instance, was banned together with an instrument to measure vaults that could be used in a chemical plant.

Please, Log in or Register to view URLs content!
 
O

Overbom

Brigadier
Registered Member
gadgetcool5 said:

Scientists in China ‘struggle to get instruments because of US export controls'​

An AI system was trained on a database of detailed information about most hardware used in laboratories across China and found that nearly 1,900 items (42 per cent) on the US [export ban] list were scientific instruments.

Scientific instruments are one of the few commercial products not usually made in China. None of the world’s top 20 scientific instrument suppliers are Chinese.

According to Wang’s study, the instruments on the US export control list were not isolated, but often linked to others because of shared technologies or fields of application. An X-ray machine, for instance, was banned together with an instrument to measure vaults that could be used in a chemical plant.

Please, Log in or Register to view URLs content!
Click to expand...
Scientific instruments are no joke. They are state of the art machines which require huge R&D resources and extremely talented people to create.

It isn't strange that China, which is a new great power, cannot manufacture such instruments yet. It will happen eventually but it will take a long time
 
You must log in or register to reply here.
Top