antiterror13
Brigadier
And SK Hynix has started. "New LPDDR4 mobile DRAM runs at 4,266Mbps and consumes 20% less power than its predecessor, SK Hynix said. "
Chinese semiconductor industry
- Thread starter Hendrik_2000
- Start date
- Status
And SK Hynix has started. "New LPDDR4 mobile DRAM runs at 4,266Mbps and consumes 20% less power than its predecessor, SK Hynix said. "
As usual from Chinese publication, you have to read it between the lines, either you believed it or not the fact is SMIC N+1 (8nm) had successfully trial-produced and is in mass production, while N+2 all we had is leaks such as this one and silence from SMIC. Maybe next year we may have confirmation as the Domestic 7nm line have been develop.
Recently, some media reported that SMIC is about to trial-produce 7nm chips this month, and strive to achieve mass production targets in October this year. It is understood that SMIC's trial production of 7nm chips is not using advanced EUV lithography machines, but plans to trial production of 7nm chips through the DUV process. At present, semiconductor manufacturers that have achieved 7nm chip production in the world basically use EUV lithography machines for production. However, because Chinese semiconductor companies lack such advanced lithography machines, SMIC can only achieve this through DUV lithography machines. The goal of mass production of 7nm chips.
As the largest and most technologically advanced integrated circuit manufacturer in mainland China, SMIC's position in the domestic semiconductor field is very high. The industry has always regarded it as the core force to break through the localization of semiconductors. Trial production of the 7nm process has attracted expectations from all walks of life. If SMIC really achieves mass production of the 7nm process, its significance is very significant. It will not only greatly enhance SMIC's position in the international semiconductor field, but also It also means that the core shortage dilemma that Huawei is facing is expected to be resolved.
Although the current chip technology has developed to 3nm and 2nm, and even some semiconductor manufacturers have prepared to advance to the process below 1nm, that is, the Amy process, in fact, in all advanced processes, the process that has achieved mass production has only reached 5nm. And there are only a handful of semiconductor manufacturers in the world. Even the semiconductor giant Intel is still stuck at 7nm and has not broken through 5nm. If SMIC successfully mass-produces 7nm chips, it may not be able to compare with established foundries such as TSMC and Samsung in terms of yield and capacity, but it should not be difficult to meet domestic market demand.
As far as the current chip market is concerned, the market demand for the 7nm process is very large. Many of Huawei's Kirin chips are based on the 7nm process, such as Kirin 990 and Kirin 990 5G. As the United States cut the supply of Huawei's chips, Huawei's 7nm chips are actually in a state of being out of supply. Therefore, SMIC's mass production of 7nm chips is of great significance to Huawei. This is related to whether Huawei's chip crisis can be effectively alleviated. It is worth mentioning that, as the domestic chip industry is showing a blowout development, in addition to Huawei, many domestic companies' chip orders have flocked to SMIC. It is conceivable that after SMIC achieves mass production of 7nm chips, Its orders will be full again, and its production capacity may even enter a state of full load.
I have to admit that the development speed of SMIC has been getting faster and faster in recent years, especially after entering 2021. In February and March of this year, SMIC frequently reported good news, allowing China's semiconductor industry to see the realization of semiconductors. The hope of localization. Although it started late, the strength of SMIC is undeniable. Take its latest breakthrough 14nm process. With continuous research and development, the yield rate of SMIC's 14nm process has increased to about 95%, and it can be compared with TSMC. Samsung Electronics is comparable. I believe that this time SMIC can create surprises again, break through the 7nm process, and help the development of "China Core".
Read the latest global technology and analyze industry trends in dept
SMIC is about to trial-produce 7nm chips. Will Huawei's core shortage problem be solved?
2021-10-11 13:58 HKTRecently, some media reported that SMIC is about to trial-produce 7nm chips this month, and strive to achieve mass production targets in October this year. It is understood that SMIC's trial production of 7nm chips is not using advanced EUV lithography machines, but plans to trial production of 7nm chips through the DUV process. At present, semiconductor manufacturers that have achieved 7nm chip production in the world basically use EUV lithography machines for production. However, because Chinese semiconductor companies lack such advanced lithography machines, SMIC can only achieve this through DUV lithography machines. The goal of mass production of 7nm chips.
As the largest and most technologically advanced integrated circuit manufacturer in mainland China, SMIC's position in the domestic semiconductor field is very high. The industry has always regarded it as the core force to break through the localization of semiconductors. Trial production of the 7nm process has attracted expectations from all walks of life. If SMIC really achieves mass production of the 7nm process, its significance is very significant. It will not only greatly enhance SMIC's position in the international semiconductor field, but also It also means that the core shortage dilemma that Huawei is facing is expected to be resolved.
Although the current chip technology has developed to 3nm and 2nm, and even some semiconductor manufacturers have prepared to advance to the process below 1nm, that is, the Amy process, in fact, in all advanced processes, the process that has achieved mass production has only reached 5nm. And there are only a handful of semiconductor manufacturers in the world. Even the semiconductor giant Intel is still stuck at 7nm and has not broken through 5nm. If SMIC successfully mass-produces 7nm chips, it may not be able to compare with established foundries such as TSMC and Samsung in terms of yield and capacity, but it should not be difficult to meet domestic market demand.
As far as the current chip market is concerned, the market demand for the 7nm process is very large. Many of Huawei's Kirin chips are based on the 7nm process, such as Kirin 990 and Kirin 990 5G. As the United States cut the supply of Huawei's chips, Huawei's 7nm chips are actually in a state of being out of supply. Therefore, SMIC's mass production of 7nm chips is of great significance to Huawei. This is related to whether Huawei's chip crisis can be effectively alleviated. It is worth mentioning that, as the domestic chip industry is showing a blowout development, in addition to Huawei, many domestic companies' chip orders have flocked to SMIC. It is conceivable that after SMIC achieves mass production of 7nm chips, Its orders will be full again, and its production capacity may even enter a state of full load.
I have to admit that the development speed of SMIC has been getting faster and faster in recent years, especially after entering 2021. In February and March of this year, SMIC frequently reported good news, allowing China's semiconductor industry to see the realization of semiconductors. The hope of localization. Although it started late, the strength of SMIC is undeniable. Take its latest breakthrough 14nm process. With continuous research and development, the yield rate of SMIC's 14nm process has increased to about 95%, and it can be compared with TSMC. Samsung Electronics is comparable. I believe that this time SMIC can create surprises again, break through the 7nm process, and help the development of "China Core".
Read the latest global technology and analyze industry trends in dept
I must correct you. Even among non Chinese manufacturers, AMD has used this sort of 3D packaging with their chiplet design.
What, you want to do 3D RAM using multiple wafers? That is not what 3D NAND is about for the most part.
3D NAND does multiple layers on a wafer. Even YMTC, one of the few who uses multiple wafers, only uses two wafers, and it isn't supposed to ever use more than two wafers. Using multiple wafers would increase cost and aligning wafers isn't easy either.
The whole deal with making the transistors smaller is you reduce power consumption, and not just size, but also increase production rate. With the same wafer you can have more chips or more memory in them if the wafer output is the same. Right now EUV machines have slower throughput on a per pass basis but the situation for EUV keeps improving.
AMD's chiplets are nothing like that since AMD Zen is basically a multiple chip module design. It is just the modern incarnation of the same technologies used in an older Pentium Pro or POWER4 processor.
SMIC 7nm won't solve Huawei's problems unless they have a 100% Chinese production line. SMIC had to stop manufacturing Huawei's 14nm mobile processors due to US pressure already. The US threatened to cut SMIC's access to consumable materials and parts and SMIC conceded. Unless SMIC or some other company moves to Chinese tools or the lock of the US on the semi industry is broken this won't change.
3D NAND does multiple layers on a wafer. Even YMTC, one of the few who uses multiple wafers, only uses two wafers, and it isn't supposed to ever use more than two wafers. Using multiple wafers would increase cost and aligning wafers isn't easy either.
The whole deal with making the transistors smaller is you reduce power consumption, and not just size, but also increase production rate. With the same wafer you can have more chips or more memory in them if the wafer output is the same. Right now EUV machines have slower throughput on a per pass basis but the situation for EUV keeps improving.
AMD's chiplets are nothing like that since AMD Zen is basically a multiple chip module design. It is just the modern incarnation of the same technologies used in an older Pentium Pro or POWER4 processor.
SMIC 7nm won't solve Huawei's problems unless they have a 100% Chinese production line. SMIC had to stop manufacturing Huawei's 14nm mobile processors due to US pressure already. The US threatened to cut SMIC's access to consumable materials and parts and SMIC conceded. Unless SMIC or some other company moves to Chinese tools or the lock of the US on the semi industry is broken this won't change.
AMD chiplet design is, at it's core, basically splitting the functions of a full SoC up into different chips and packaging them together. They do this with a 3D package where chips are connected by through silicon vias for vertical stack and microbumps for horizontal interconnects (speculatively, possibly on lithography defined interconnects).
YMTC does direct wafer bonding of an IO I with a storage IC. But there's nothing that says only NAND can use the packaging strategy of wafer bonding an IO circuit to a repeating storage circuit. There's also nothing saying that the IO circuit can't itself have a through silicon via that allows for packaging of the entire IO/storage IC together with a logic IC.
Agreed that SMIC must switch to domestic tools but I do not think EUV is the most urgent. There are tons of tools, each representing a vital unit operation, for a semiconductor process line. EUV is just 1. If you look only for the high end, and not completeness, it is a mistake.
AMD has no 3D package like that yet. HBM DRAM uses similar packaging technology. However it isn't cheap.
You still have to manufacture more wafers. Which means more tools and materials and time and $$$.
It is all predicated on the fact that leading edge chip manufacture is currently limited by costly and slow current generation EUV machines or that certain parts of the product do not need higher detail features. If you look at ASML's roadmap however it is expected that EUV machine tools will continue to increase in output rate.
You still have to manufacture more wafers. Which means more tools and materials and time and $$$.
It is all predicated on the fact that leading edge chip manufacture is currently limited by costly and slow current generation EUV machines or that certain parts of the product do not need higher detail features. If you look at ASML's roadmap however it is expected that EUV machine tools will continue to increase in output rate.
Last edited:
I wonder if a country like INDIA can defeat China in semiconductor since they have access to ASML technology. This will make China's technology obsolete.
Zhang Zhongmou, the founder of TSMC, once said: Without TSMC's foundry, mainland China would not be able to produce a piece of high-end chips without the efforts of the whole country. The Dutch president once said that even if the lithography machine drawings were given to China, the Chinese would not be able to create top-level lithography machines.
Where would India get the capital and the products to put the chips into? We are talking tens of billions investment just for one set of fabs.
You would need to continue updating the fabs as time went by and tools costs rise much faster than inflation.
China can certainly manufacture high end chips. SMIC is only stalled by tools embargos and limitations like that. HiSilicon was, before the embargo, one of the top chip design companies. Also, if the Dutch President said that he is quite dumb. All evidences point that China will be able to make their own tools even without drawings. It is just a matter of time and effort.
You would need to continue updating the fabs as time went by and tools costs rise much faster than inflation.
China can certainly manufacture high end chips. SMIC is only stalled by tools embargos and limitations like that. HiSilicon was, before the embargo, one of the top chip design companies. Also, if the Dutch President said that he is quite dumb. All evidences point that China will be able to make their own tools even without drawings. It is just a matter of time and effort.
They get American and TSMC help. Since they can access to ASML, the progress will be much faster. China will left behind if they don't solve EUV issue. At that time, DUV will be outdated.
- Status