Rival cellphone maker Meizu congratulates Huawei on their strategic pimpslap against tech sanctions.
Chinese semiconductor industry
- Thread starter Hendrik_2000
- Start date
- Status
Rival cellphone maker Meizu congratulates Huawei on their strategic pimpslap against tech sanctions.
I'll be honest and say that I understand nothing about Chips. I do however know that China has an issue with the actual machinery needed to make semiconductors. From what I understand the Chinese ASML is called SMEE can only make lithography machines that can just make machines that are 90nm. This means that if China is fully cut off they have a full internal supply chain of just 90nm. However I have seen recent news that SMEE was able to make a 28nm device lithography machine. Does that mean China has a full internal supply chain to make 28nm chips on home territory?
云天励飞 (Intellifusion) () talks about its RISC-V based Edge AI SoC called DeepEdge10V
comes with
all fully domestically designed
and fabbed domestically with 14nm process and packaged + put on board domestically. Completely sanction proof
It supports dual Gigabit ethernet ports, multi-channel network access & access of different sensors + 8x PCle3.0 interface
Has a 64-bit RISC-V core at 1.2GHZ
looks like they support Pytoch, Onnx, Caffe and other tools
I'll plug one of my own posts here. This is a good recap of the EUV situation in China:
Chinese semiconductor industry
Banning products and distort business rules to cripple the adversary is a loosing proposition in the long term, as US is demonstrating. China has done the first real action with Micron, and mybe some more will be done, but IMO they will be "defensive" actions, to try to dissuade US from doing...
www.sinodefenceforum.com
More from weibo folks on Kirin 9000s
Is he saying the yield on Kirin 9000s is higher than Snapdragon 8+ Gen1?
Seems crazy to thnk about
If yield was in the 60s at end of last year, I think mid 70s by now is reasonable estimate.
basically, we don't have a shortage of Kirin 9000s. Plenty of orders/produced (probably 10 million+) that it can be used on other phones also.
My guess is that they've been producing this stash for a while and each batch get successfully better and higher yielding. So they accumulated a lot. Whether or not they are actually at 1.5 a month is hard to say, but I think it will need to be at that level (and probably even higher through higher yield) to cover Huawei until probably 2025 when they can really ramp up 7nm production.
I'm not sure what to make of this. He says yield for Mate 60 is higher than Mate X3. X3 uses Snapdragon 8+ Gen1华子这次麒麟5G芯片量很大很大，普通用户用不着加钱买Mate60，9月产能爬坡期上去了就放开卖。而且现在前期放货已经是很多同价位旗舰生命周期规划总数，看下单的芯片数量不一定消耗完，其它品等着吧.
芯片曝光工艺和良率数据是机密，我只能说良率比MateX3高，下单量是真的大，华子可以cover![[doge]](https://face.t.sinajs.cn/t4/appstyle/expression/ext/normal/a1/2018new_doge02_org.png "[doge]")
Is he saying the yield on Kirin 9000s is higher than Snapdragon 8+ Gen1?
Seems crazy to thnk about
If yield was in the 60s at end of last year, I think mid 70s by now is reasonable estimate.
basically, we don't have a shortage of Kirin 9000s. Plenty of orders/produced (probably 10 million+) that it can be used on other phones also.
My guess is that they've been producing this stash for a while and each batch get successfully better and higher yielding. So they accumulated a lot. Whether or not they are actually at 1.5 a month is hard to say, but I think it will need to be at that level (and probably even higher through higher yield) to cover Huawei until probably 2025 when they can really ramp up 7nm production.
Last edited:
No way to know for sure until we are a couple of months into the Mate 60’s release. From the looks of it the phone is wildly popular, so there will be no demand issue for it. If sales don’t break record then only issue is probably with supply bottleneck.
That’s because Samsung’s 4 nm node has low yields. Not sure if they’ve improved by now but I’ve heard numbers as low as 30-50%. Pretty sure Qualcomm wouldn’t be using Samsung if TSMC had spare capacity but Apple has been hogging more of TSMC’s capacity as it’s been rolling out a broader lineup of Apple Silicon chips.
If they have an instrument ready for verification testing by middle of 2024, fabs may not wait for testing to finish before adopting the instrument. They may join the verification process, since unlike the DUV instruments, which are substitutes for a technology they already have alternatives too, they will need to learn how to use a whole new technology and develop other processes around it. Furthermore, if initial testing goes well, prototype instruments may start getting immediate use in actual production flow, since unlike with DUV they don’t have an alternative instrument to provide the capabilities that these prototype instruments can do, especially since EUV doesn’t need to be employed for every patterning step and even at low utilization can offer immense capability jumps in a few critical steps.
Whether China’s EUV development actually plays out this way remains to be seen, *but* this is how EUV adoption first started for Samsung and TSMC. Both clients had earlier prototypes and pre production units from ASML on site that were mostly being used for fab process research. Field upgrades boosted the capability of those preproduction instruments until TSMC decided to start using them for their first 7nm line. The actual intended for mass production design only came later after TSMC had done a few production batches of 7 nm chips. In China’s case from what we can tell the first prototypes they will have won’t be low powered like ASML’s, so if adopted into production flow while still in testing their usefulness may not be limited to fab research. The irony here is that if development goes well Chinese fabs may be using a Chinese EUV instrument before they’ve finished the DUV adoption process for their fully indigenous mature lines.
Last edited:
- Status