D
Deleted member 24525
Guest
Do you mean the lithography machines being used in commercial fabs, or the machines themselves being in serial production?Yes, the original interview said it will take 2-3 years to be used on production lines.
Do you mean the lithography machines being used in commercial fabs, or the machines themselves being in serial production?Yes, the original interview said it will take 2-3 years to be used on production lines.
I think he meant commercial fabs. The original interview was not clear - the word being used was 产线 and not 试验线. But also confirmed that 14nm-7nm lithography machines are under development. I think the SMEE quote was based on his understanding that Huawei hired a lot of people from SMEE.Do you mean the lithography machines being used in commercial fabs, or the machines themselves being in serial production?
somehow, Dylan Patel has become the expert here. And they also made this unsubstantiated claim that you need 3x H800 to train task in the same time vs H100. I don't know where these things come from. But one thing obvious to me is that Huawei has found a way to get their AI platform to do AI training in Chinese and it's without Nvidia products.In a paper in March, Huawei researchers demonstrated how they could use such techniques to train its latest-generation large language model using only the company’s Ascend chips and without Nvidia chips. Despite some shortcomings, the model, known as PanGu-Σ, reached state-of-the-art performance on a few Chinese-language tasks, including reading comprehension and grammar challenges, the researchers wrote in the paper.
Dylan Patel, chief analyst at semiconductor research and consulting firm SemiAnalysis, said Chinese researchers’ pain points will only exacerbate without access to the new Nvidia H100, which includes an extra performance-boosting feature especially helpful for training ChatGPT-like models.
But a paper last year from Baidu and Peng Cheng Laboratory, a Shenzhen-based research institute, showed researchers were training large language models in a way that would make the feature unnecessary. Mr. Patel said it looked promising even though the research was in its early stages.
“If it works well, they can effectively circumvent the sanctions,” he said.
I think let's not take his commentary too seriously. There is the havok insider comments and official news & patents we get which are a lot more trustworthy.I think he meant commercial fabs. The original interview was not clear - the word being used was 产线 and not 试验线. But also confirmed that 14nm-7nm lithography machines are under development. I think the SMEE quote was based on his understanding that Huawei hired a lot of people from SMEE.
A leaked transcript of a recent interview with CITIC securities equity research MD.
Lots of information (has now been deleted unfortunately), but some takeaways:
1. Big stimulus expected in Q2 2023.
2. MIIT is now in charge of China's semiconductor projects instead of MOST
3. A high-level working group on ICT industry was founded in April and is headed by Vice Premier and MIIT senior officials - selection of expert team and enterprises began in Q4 2022
4. Big Fund will not be abolished but will prioritize technological breakthroughs over market returns
5. Projects will be enterprise-focused and led by enterprises instead of academia
6. 20% of SMIC's equipment is now domestic and is being used in production; YMTC's domestic equipment share is 30%
7. YMTC does not need a high-end lithography machine. Company is expected to build a de-Americanized, domestic equipment line in 2024 with a 1000-wafer production (per month?) capacity for testing purposes
8. Shenyang Fortune and Kunshan Kinglai, apparently, are semiconductor parts suppliers to Lam, AMAT, and TEL (50% of their revenue comes from US clients)
9. Huawei is in charge of lithography machine development and 28nm machine will take 2-3 years to develop
10. No need to worry about semiconductor materials and gases as there are many domestic alternatives
11. AI data centers are all required to use Chinese chips; Domestic GPU makes up 3-5% for companies like Baidu and Alibaba
12. Compatibility issues with LoongArch is being resolved - domestic companies developing compatible software and foreign software can be accessed by emulator
13. All levels of government and SOEs will be required to use domestic equipment and CPU in the next few years so huge boost for domestic companies
14. MOF plans to establish a RMB 500 billion fund dedicated to equipment and materials but recipients will be classified for obvious reasons.
15. New strategy will focus on M&A and integration to create state champions
16. Chinese equipment companies allow fabs to trial-run equipment for free vs. Overseas equipment companies that require payment upfront
Take it with a grain of salt but more exciting news is coming for sure...
Look like this money is more allocated towards research and development rather than capacity building to solve technical bottlenecks in the supply chain.
1.75 billion is peanut imo, especially for nearly 200 firms. No wonder why some firms like SMEE has been said that they lack funding for R&D and scaling up production.
On Saturday, Chinese computer manufacturer PowerLeader launched a 'new' processor and compact desktop PC. During the press conference, PowerLeader talked about its first generation Powerstar P3-01105 CPU, featuring the "storm core" architecture, described as "extremely high performance," x86 compatible, and offering great support for Windows. And so it should, as it looks very much like a rebranded Intel Core i3-10105(F) Comet Lake CPU with 4C / 8T
In response to harukaze5719's Tweet about this Powerstar P3-01105 CPU, another Twitterer pointed out that what we are looking at is almost certainly an Intel Core i3-10105(F) Comet Lake CPU. The following clues convincingly point to this:
- Physical lugged heatspreader design, and other physical characteristics
- Physical substrate design is identical, as far as we can see
- Silk screen print format on the IHS is the same
- The PowerLeader processor name slightly jumbles Intel's: compare "10105" and "01105"
- They are both marked as capable of a "3.70GHZ" base clock
- Last but not least, the QR code on the upper right of the Powerstar P3-01105 PCB is said to match Intel's.
So, why would PowerLeader rebrand an Intel Core i3-10105? All we can find about the underlying reason for this sleight-of-hand with branding is that Chinese tech firms can benefit from attractive subsidies by developing and launching "domestic products."
这种方法的成本可能很高:You先生说,如果一家美国公司需要 1,000 个 H100 来训练一个大型语言模型,那么一家中国公司可能需要 3,000 个或更多的 H800 才能达到同样的效果
This approach can be costly: If an American company needs 1,000 H100s to train a large language model, a Chinese company might need 3,000 or more H800s to achieve the same effect, Mr You said
It seems that H800 is 3 times less powerful than H100
A leaked transcript of a recent interview with CITIC securities equity research MD.
Lots of information (has now been deleted unfortunately), but some takeaways:
1. Big stimulus expected in Q2 2023.
2. MIIT is now in charge of China's semiconductor projects instead of MOST
3. A high-level working group on ICT industry was founded in April and is headed by Vice Premier and MIIT senior officials - selection of expert team and enterprises began in Q4 2022
4. Big Fund will not be abolished but will prioritize technological breakthroughs over market returns
5. Projects will be enterprise-focused and led by enterprises instead of academia
6. 20% of SMIC's equipment is now domestic and is being used in production; YMTC's domestic equipment share is 30%
7. YMTC does not need a high-end lithography machine. Company is expected to build a de-Americanized, domestic equipment line in 2024 with a 1000-wafer production (per month?) capacity for testing purposes
8. Shenyang Fortune and Kunshan Kinglai, apparently, are semiconductor parts suppliers to Lam, AMAT, and TEL (50% of their revenue comes from US clients)
9. Huawei is in charge of lithography machine development and 28nm machine will take 2-3 years to develop
10. No need to worry about semiconductor materials and gases as there are many domestic alternatives
11. AI data centers are all required to use Chinese chips; Domestic GPU makes up 3-5% for companies like Baidu and Alibaba
12. Compatibility issues with LoongArch is being resolved - domestic companies developing compatible software and foreign software can be accessed by emulator
13. All levels of government and SOEs will be required to use domestic equipment and CPU in the next few years so huge boost for domestic companies
14. MOF plans to establish a RMB 500 billion fund dedicated to equipment and materials but recipients will be classified for obvious reasons.
15. New strategy will focus on M&A and integration to create state champions
16. Chinese equipment companies allow fabs to trial-run equipment for free vs. Overseas equipment companies that require payment upfront
Take it with a grain of salt but more exciting news is coming for sure...
光刻机是最“卡脖⼦”的设备,因为我国⽬前最多只能做到90nm的光刻机,⽽且90nm还没有真的⽴在这⾥⾯。现在光刻机的进展⽅⾯,已经不是上海微电⼦在牵头了, 是华为在牵头。华为挖了很多上海微的⼈去搞光刻机。根据我们最新的了解,⼤概需要2~3年的时间能够做出28nm的光刻机,送到产线⾥⾯。除此之外,现在也在研发更高端,14nm和7nm的光刻机。
The lithography machine is the most "stuck" equipment, because our country can only produce 90nm lithography machines at most, and 90nm has not really been established here. In terms of the development of lithography machines, SMEE is no longer taking the lead, but Huawei. Huawei recruited a lot of people from Shanghai Weibo to develop lithography machines. According to our latest knowledge, it will take about 2 to 3 years to make a 28nm lithography machine and send it to the production line. In addition, higher-end, 14nm and 7nm lithography machines are also being developed.