Chinese semiconductor thread II

T

tokenanalyst

Brigadier
Registered Member

Shandong Liguan Microelectronics Equipment: Accelerating the development of 12-inch liquid phase SiC crystal growth equipment​


Recently, Shandong Liguan has made significant achievements in the research and development of 12-inch silicon carbide (SiC) crystal growth furnaces. According to the company, it has successfully overcome the technical difficulties of the 12-inch physical vapor transport (PVT) resistance heating crystal growth furnace and successfully completed the delivery of the first two units of the equipment. At the same time, Shandong Liguan is stepping up research and development of 12-inch liquid phase SiC crystal growth equipment. With the key breakthrough in the commercialization of the 12-inch resistance crystal growth system, Shandong Liguan has formulated corresponding plans and is expected to achieve mass supply of 12-inch crystal growth equipment in 2025, and is committed to promoting domestic silicon carbide materials to occupy a leading position in the global supply chain and enter the first echelon.

1747251784879.png

Shandong Liguan has successfully developed an 8-inch PVT silicon carbide crystal growth furnace and has achieved mass sales. The crystal growth furnace integrates induction and resistance heating functions and is suitable for the growth of conductive and semi-insulating SiC crystals. Its innovative thermal field design improves temperature uniformity and stability by more than 30%, thereby significantly improving the yield and consistency of crystals to international mainstream standards. At the same time, the equipment is equipped with an automated control system that effectively reduces manual intervention and production costs, and has been selected by leading domestic and foreign customers.

According to public information, Shandong Liguan's product line covers a wide range of semiconductor material process equipment from the first to the fourth generation, and all of them have independent intellectual property rights, achieving complete independent control. These devices are widely used in the manufacturing of new electronic devices such as integrated circuits, power semiconductors, compound semiconductors, 5G chips, optical communications, and MEMS.

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

JPaladin32

New Member
Registered Member
Alb said:
Is this Harmony OS PC? That would be an impressive score.


View attachment 152214
Click to expand...
This is simply impossible. I mean we could discuss rumors but some sanity checks are needed first before posting these outrageous performance claims.

Some confirmed info shows a 4 big + 4 mid + 2 mid CPU topology on the HarmonyOS PC. The 4 big cores and 2 mids are the same ones from Kirin 9010 (Pura 70 Pro). The other 4 mids are improved ones over the 9010 mids. Given that the single core score on 9010 is around 1440 and the PC big cores are clocked at the same frequency, there would be at best 10% improvement on single core from larger caches, higher power envelope and improved memory bandwidth. My guess is around 1600 on single core. Multi-core is a bit difficult to predict, but given there are 10 cores, my guess is somewhere between Apple M1 and M2, which may be 9000.

So, 1600/9000 would be a much closer estimate. The scores in the screenshot are definitely made up and we should not have the wrong expectations.
 
A

Alb

Junior Member
Registered Member
JPaladin32 said:
This is simply impossible. I mean we could discuss rumors but some sanity checks are needed first before posting these outrageous performance claims.

Some confirmed info shows a 4 big + 4 mid + 2 mid CPU topology on the HarmonyOS PC. The 4 big cores and 2 mids are the same ones from Kirin 9010 (Pura 70 Pro). The other 4 mids are improved ones over the 9010 mids. Given that the single core score on 9010 is around 1440 and the PC big cores are clocked at the same frequency, there would be at best 10% improvement on single core from larger caches, higher power envelope and improved memory bandwidth. My guess is around 1600 on single core. Multi-core is a bit difficult to predict, but given there are 10 cores, my guess is somewhere between Apple M1 and M2, which may be 9000.

So, 1600/9000 would be a much closer estimate. The scores in the screenshot are definitely made up and we should not have the wrong expectations.
Click to expand...
That is why I put the question mark in my post.

Nevertheless the screenshot is taken from the geekbench official website:

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

JPaladin32

New Member
Registered Member
Alb said:
That is why I put the question mark in my post.

Nevertheless the screenshot is taken from the geekbench official website:

Please, Log in or Register to view URLs content!
Click to expand...
Actually submitting a fake entry to the Geekbench database isn't too difficult. The device label (HAD-W32) can be changed fairly easily in Linux and Android.

Okay, I clicked your Geekbench link and checked the details. This is likely a Linux virtual machine running on an Apple device. They mocked up the device label but forgot to hide the CPU implementer. ARM implementer 97 (0x61 in hexadecimal) is Apple for sure. This is an Apple device. Judging by the scores it's likely the latest Apple M4 Macbook.
 
T

tokenanalyst

Brigadier
Registered Member

Domestic high-end photoresist material developer Dongkaixin Semiconductor completes Series A financing​


Chengdu Dongkaixin Semiconductor Materials Co., Ltd. (referred to as "Dongkaixin Semiconductor") recently completed its Series A financing, and the investor was Chengxin Venture Capital.
Official data shows that Dongkaixin Semiconductor is a subsidiary of Dongcai Technology Group. It was established in May 2023 and is a high-end photoresist material developer. The company has certain technical strength in the synthesis and purification of photoresist materials and has applied for a number of patents, such as "A synthesis process of (4-tert-butylphenyl) diphenylsulfonium chloride salt" and "1-(4-alkoxynaphthyl)-1H-cyclosulfonium perfluorobutane sulfonate".

In January this year, Dongcai Technology announced that its holding subsidiary Dongkaixin Semiconductor plans to increase capital and expand shares and introduce institutional investors. After the capital increase is completed, the registered capital of Chengdu Dongkaixin will increase from 73 million yuan to 100 million yuan. The company's shareholding in Chengdu Dongkaixin will be reduced from 75.34% to 55.00%, and it will still be its controlling shareholder with actual control.

Dongcai Technology responded to investors in May this year, saying that the company, Chemax of South Korea and Zhongyi Chemical jointly invested in the establishment of Dongkaixin Semiconductor, focusing on the synthesis and purification of monomers, photoacids, resins, etc. required for high-end photoresists. At present, the production equipment has been installed and debugged, and has entered the trial production and promotion and application stage. The company will use the deep technical accumulation, rich production experience and certification channel advantages of Chemax of South Korea in the field of photoresist materials, combined with the company's organic synthesis and modification technology reserves, intelligent manufacturing experience and capital and talent advantages, to jointly carry out the research and development, production and sales of advanced photoresist materials.

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

tokenanalyst

Brigadier
Registered Member

An Improved YOLOv5 Model for Lithographic Hotspot Detection​


Abstract​

The gap between the ever-shrinking feature size of integrated circuits and lithographic manufacturing ability is causing unwanted shape deformations of printed layout patterns. The deformation region with problematic imaging, known as a hotspot (HS), should be detected and corrected before mask manufacturing. In this paper, we propose a hotspot detection method to improve the precision and recall rate of the fatal pinching and bridging error due to the poor printability of certain layout patterns by embedding a spatial attention mechanism into the YOLOv5 model. Additionally, transfer learning and pre-trained techniques are used to expedite training convergence. Simulation results outperform the depth-based or representative machine learning-based methods on the ICCAD 2012 dataset with an average recall rate of 1, a precision rate of 0.82770.8277 and an F1-score of 0.90570.9057.​

1747330029246.png

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

huahua_weiwei

Just Hatched
Registered Member
JPaladin32 said:
Actually submitting a fake entry to the Geekbench database isn't too difficult. The device label (HAD-W32) can be changed fairly easily in Linux and Android.

Okay, I clicked your Geekbench link and checked the details. This is likely a Linux virtual machine running on an Apple device. They mocked up the device label but forgot to hide the CPU implementer. ARM implementer 97 (0x61 in hexadecimal) is Apple for sure. This is an Apple device. Judging by the scores it's likely the latest Apple M4 Macbook.
Click to expand...
Why did no one run Geekbench on X90 given that it can use Windows?
 
You must log in or register to reply here.
Top