News on China's scientific and technological development.

tphuang

Lieutenant General
Staff member
Super Moderator
VIP Professional
Registered Member
SMRs are a meme. I am increasingly convinced that the sporadic hype around nuclear energy is just another trick by oil corporations. We know they had funded hydrogen economy stuff to delay electrification. Nuclear energy is simply too expensive for base load. It also takes decades to scale up. For these reasons, it was outcompeted by fossil fuels in the late-20th century. And as soon as the decommissioning costs of reactors were understood in the early-00s, investments on new nuclear reactors decreased a lot.

View attachment 132348

SMRs mean you need to build more reactors to produce the same amount of energy. Which means they are even less economic than normal reactors. Even France is not bothering and they are the only country on the planet with a strong state backing of nuclear energy.

Please, Log in or Register to view URLs content!
it's truly amazing how out of control this SMR has gotten. If smaller reactors are cheaper than why do countries ever need to build 1.2GW plants again? They should just build 100MW plants from now on. All these fads will fade and disappoint people as usual. Nuscale is just the start of SMR failures.

It's truly funny that China has the only notable commercial SMR with pebble bed reactor and countries pretend like it doesn't exist. Whereas all these US vassals spend their time admiring the American SMRs like Nuscale that will never get anywhere.
 

gelgoog

Brigadier
Registered Member
The pebble bed reactor has more economic potential as an SMR than the light water reactors these people have been peddling.
It operates at higher temperature and is more energy efficient as a result.
The only question is how reliable the design is, since it is mechanically complex.
 
Last edited:

Rank Amateur

Junior Member
Registered Member
it's truly amazing how out of control this SMR has gotten. If smaller reactors are cheaper than why do countries ever need to build 1.2GW plants again? They should just build 100MW plants from now on. All these fads will fade and disappoint people as usual. Nuscale is just the start of SMR failures.

It's truly funny that China has the only notable commercial SMR with pebble bed reactor and countries pretend like it doesn't exist. Whereas all these US vassals spend their time admiring the American SMRs like Nuscale that will never get anywhere.

"It's truly funny that China has the only notable commercial SMR with pebble bed reactor [implementation of technology X] and countries pretend like it doesn't exist. Whereas all these US vassals spend their time admiring the American SMRs like Nuscale [technology X companies] that will never get anywhere."

It's a tale as old as time.
 

gelgoog

Brigadier
Registered Member
Recent Chinese aggression, as Ericsson's boss described it, suggests a US campaign aimed at restricting Huawei's access to high-end chips has not been successful. "The honest answer is they should comment on that," he said when asked about Huawei's ability to secure critical components. "But we see them quite aggressive in the market and it is ultimately a choice customers will have to make – how they think about network resilience."
The seething is just unreal. Cry ab it.

Huawei used to use 14nm chips and FPGAs in their base stations. Now they have SMIC 7nm and it seems like can produce them baseband chips just fine. I wonder what they did to replace the FPGAs though...
 

jli88

New Member
Registered Member
The seething is just unreal. Cry ab it.

Huawei used to use 14nm chips and FPGAs in their base stations. Now they have SMIC 7nm and it seems like can produce them baseband chips just fine. I wonder what they did to replace the FPGAs though...

Word on the market is that they designed custom ASICs and used them instead of FPGAs. I don't think the FPGA bottleneck has been resolved yet, but an FPGA will always underperform in front of a custom ASIC. Designing a custom ASIC is just hard, takes time & effort, needs some scale to be viable. Huawei was forced to put in the time and effort, and being Huawei it has enough scale to create an ASIC for everything.
 

gelgoog

Brigadier
Registered Member
Word on the market is that they designed custom ASICs and used them instead of FPGAs. I don't think the FPGA bottleneck has been resolved yet, but an FPGA will always underperform in front of a custom ASIC. Designing a custom ASIC is just hard, takes time & effort, needs some scale to be viable. Huawei was forced to put in the time and effort, and being Huawei it has enough scale to create an ASIC for everything.
Yeah I assumed as much. I said this might happen back when the ban was enacted. Unlike Ericsson or Nokia, Huawei can actually design chips. Even Samsung is kind of weak at chip design. For example they license CPU and GPU core designs, and either use them as they are or just slightly modify them with considerable difficulty. An ASIC would also be much cheaper to deploy at scale than FPGAs.

You probably can replace a 7nm FPGA with a custom 28nm ASIC. I would not be surprised if this happened.

It must have taken a lot of design effort to replace all the imported components though.

The main issue with going with the ASIC is time to market. The other companies which use FPGAs can get a product to the market faster in a short cycle, and making deep changes to the "hardware" implementation without changing the board design is actually viable.
 
Last edited:

jli88

New Member
Registered Member
Yeah I assumed as much. I said this might happen back when the ban was enacted. Unlike Ericsson or Nokia, Huawei can actually design chips. Even Samsung is kind of weak at chip design. For example they license CPU and GPU core designs, and either use them as they are or just slightly modify them with considerable difficulty. An ASIC would also be much cheaper to deploy at scale than FPGAs.

You probably can replace a 7nm FPGA with a custom 28nm ASIC. I would not be surprised if this happened.

It must have taken a lot of design effort to replace all the imported components though.

The main issue with going with the ASIC is time to market. The other companies which use FPGAs can get a product to the market faster in a short cycle, and making deep changes to the "hardware" implementation without changing the board design is actually viable.

HiSilicon is a beast. Before sanctions they were the first and only company to release a modern SoC on TSMC's latest process before Apple. Now, despite facing so many hurdles. Just think about it, your whole process of chip design is thrown into disarray, the process is not available, the process is developing/experimental, EDA tools are uncertain, etc. etc.

But despite all this, they essentially designed and produced chips in record time. Think of it, HiSilicon designs competitive professional GPUs for AI (vs Nvidia), designs mobile SoCs (vs Apple/Qualcomm), designs a zillion number of other ASICs in record time. Oh my god, do people not know what's coming for them.
 

ansy1968

Brigadier
Registered Member
Aggressive? really .. this is just business competition, the best will win

Crying baby when can't compete with Huawei and asking help from the govt and US
Well Sir you can't blame them, Karma is a bitch, they had been expecting a windfall in joining and advocating a Huawei ban instead they had awaken a beast. This episode is a microcosm of things to come, you can't stop the progress of a 1.4 billion Chauvinistic hardworking Chinese, it will only spur them on.
 
Top