Last week, AMD unveiled its latest Ryzen 8000 series APUs, codenamed Hawk PointNvidia will discontinue the RTX 3050 8GB version and replace it with the RTX 3050 6GB version in the futureLaunch of TIPLUS7100 Ti600 SSD 4TB SSDMicrosoft may release Windows 12 in June next year, and there may be AI performance requirements.
AMD officially announced the Ryzen 8040 8045 series APUs to upgrade AI performance
AMD officially unveiled its latest Ryzen 8000 series APUs, codenamed Hawk Point, at its "Advancing AI" event this morning. The chips are currently shipping to laptop manufacturers, and the first laptops with these chips will be available to consumers in the first quarter of 2024. The biggest upgrade of this generation of APUs is a dramatically faster neural processing unit (NPU) designed to accelerate AI applications. As can be seen from the naming, the Ryzen 8000 series APU has not brought new changes in the CPU and GPU architecture, and still uses the ZEN4 architecture and RDNA3 architecture. But this series of APUs uses the faster Ryzen AI NPU, which has improved its NPU performance to 16Tops despite being based on the same first-generation XDNA architecture. In addition, AMD supports the X-512 instruction set and the AI accelerator equipped with the core display in the AI field. In AMD's benchmarks,"hawk point"In terms of visual models and llama2 than Ryzen 7040"phoenix"The series has been increased by 40%.
At the hardware level, in addition to the faster xDNA NPUs,"hawk point"Basically, it is"phoenix"CPU frequencies have been boosted and power management has been optimized to ensure faster NPUs meet the typical TDP range of U-class and P-class mobile processors. The chip is manufactured using TSMC's 4nm EUV foundry node. Built-in based on"zen 4"Microarchitecture with 8 cores and 16 threads of CPU. Each core comes with 1MB of dedicated L2 cache, and the eight cores share 16MB** of cache. The IGPU is still based on the Radeon 700M series, which is based on the RDNA3 graphics architecture and contains up to 12 compute units (768 stream processors, AI accelerators and 12 ray accelerators).
For the first time in the series, nine processors were introduced, seven of which were equipped with faster 16tops Ryzen AI NPUs, but with a slight increase in CPU frequency compared to their predecessors in the Ryzen 7040 series. Among them, AMD also brought a new Ryzen 8040HS series with a TDP of 20W-30W, which is benchmarked against Intel's original P series in terms of power consumption, and the standard pressure product has become the Ryzen 8045HS series, which still has a higher TDP. In addition, the two lowest-end APUs do not come with a discrete NPU. AMD will also introduce a new sticker for the Ryzen 8040 series, which will be featured on NPU-equipped SKUs"ryzen ai"word, not for NPU-equipped SKUs"ai"Words.
With the Ryzen 8040 series"hawkpoint"With the launch of the processor, AMD has also officially launched the RyzenAI software platform, which allows software developers to build and optimize their AI-accelerated applications and software features with RyzenAI hardware on the Ryzen8040 and 7040 series processors with XDNANPU. RyzenAI accelerators support AI models in industry-standard formats such as PyTorch, TensorFlow, and OnnX.
Spicy Cool Comment:It is rumored that in addition to the NPU performance improvement, some problems with the Ryzen 7040 series have been fixed, and the performance improvement is not there, so I hope that ** is best not improved.
Nvidia will push the RTX 3050 6GB desktop graphics card to replace the upcoming 8GB version
Previously, it was reported that Nvidia is about to launch a new version of the RTX 3050 graphics card, with the video memory reduced to 6GB 96bit, and the core of the notebook version will be used. According to the latest news from Bobantang, the RTX 3050 8GB version will be discontinued, and it will be adjusted to the RTX 3050 6GB version in the future.
According to Bobantang, the RTX 3050 8G version will be discontinued soon, and it will be adjusted to the RTX 3050 6G version for market sales. According to AIC brand manufacturers, the RTX 3050 8G version model will digest the inventory cut-off, and the RTX 3050 6G version series is expected to be available and arrive in January. Combined with the previous revelations, the RTX 3050 6GB version of the desktop graphics card may use the notebook's GA107-325 GPU, CUDA core specifications are the same as the RTX 3050 8GB version, both are 2560, but the video memory will be reduced to 6GB 96bit, power consumption will also be reduced, there may be a version without independent power supply.
Spicy Cool Comment:Nvidia has its own understanding of knife techniques, and the market will also have a suitable reaction to Nvidia.
Zhitai launched the TiPlus7100 Ti600 SSD 4TB version, priced from 1299 yuan
Zhitai has previously announced that it will launch a 4TB large-capacity SSD product. At present, Zhizhi has officially announced that it has brought two new 4TB solid-state products, namely the 4TB version of Zhizhi TIPLUS7100 and the 4TB version of Zhizhi TI600, which are currently on sale on domestic platforms, Zhizhi TIPLUS7100 4TB version is priced at 1799 yuan, and Zhizhi TI600 4TB version is priced at 1299 yuan.
The Zhizhi Ti600 4TB hard drive uses YMTC original QLC particles and supports PCIe 40 Standard, Sequential Read Speed 7000MB S, Sequential Write Speed 6000MB S, Random Read Speed 900K IOPS, Random Write Speed 800K IOPS, Endurance Class 1600 TBW, MTBF 1.5 Million Hours, Single-sided Particle Design, Five-Year Warranty.
The TiPlus7100 4TB version uses YMTC's original TLC particles and supports PCIe 40 Standard, Sequential Read Speed 7000MB S, Sequential Write Speed 6000MB S, Random Read Speed 900K IOPS, Random Write Speed 800K IOPS, Endurance Class 2400 TBW, MTBF 1.5 Million Hours, Single-sided Particle Design, Five-Year Warranty.
Spicy Cool Comment:In fact, there are not many 4T solid-state products on the market that use long-term particles, but we still hope to use our own original 4T disk, which can be said to be a normal level in the original disk, and friends who need it can consider starting.
Microsoft may release Windows 12 in June next year, and there may be AI performance requirements.
According to previous news, Microsoft is expected to adjust the development plan of Windows, adjust the system upgrade cycle, plan to release a major version of Windows every three years, regular feature updates four times a year, and the next generation of Windows 12 operating system may be released in 2024, which will introduce a new user interface, support for new hardware, and support for new APIs. There has also been news from multiple sources this year that Microsoft is expected to release a new Windows 12 next year.
According to Taiwanese media sources, Microsoft will choose to launch Windows 12 in June 2024, which is expected to trigger a new wave of artificial intelligence (AI) PC sales. Leaders of industry leaders, including ACER, Quanta, MSI and Gigabyte, expressed their excitement about the "first year of AI PCs" and the business opportunities it brings. Some people in the industry believe that artificial intelligence will not only drive the sales of PC products and services, but also stimulate the development of automotive electronic equipment and contribute to the continuous and accelerated development of the industry.
In fact, major manufacturers are also preparing for new operating systems, such as Intel's upcoming Meteor Lake processors, which are already ready for new operating systems with new architectures and designs. It is rumored that Microsoft is also developing a self-developed chip with ARM architecture, which will be combined with the future Windows 12 operating system to improve the performance and efficiency of the device to bring a better user experience. Judging by current trends, hardware AI processing support could make Windows 12 one of the minimum spec requirements.
Spicy Cool Comment:In fact, the main concern is users who use older thin and light laptops, because even if there is no independent NPU, as long as the computing power of the CPU and GPU on the desktop is sufficient, it can still meet the needs of AI.