Author: Wan Nan
At CES, Intel mainly displayed its DG1 graphics card, tiger Lake processor and comet lake-h processor, but the desktop version of the ten generation core comet lake-s processor was not released as scheduled. Unfortunately, the latest news said it was related to the power consumption of the 10 core processor, and it still needs time optimization.
Ten generations of Corey's desktop processor is the Comet Lake-S series, with at least 26 models fully upgraded to HT hyperthreading, the top-of-the-line Core i9-10900K upgraded to 10 core 20 threads, with a base frequency of 3.7 GHz, a full-core acceleration of 4.8 GHz, a single-core acceleration of 5.1 GHz, and TB 3.0 also lowered from the previous Corey X series to 5.2 GHz, while the new Velocity Boost acceleration mode was added at a frequency of 5.3 GHz.
In the case of Corey i9-10900K, Intel's 14nm process has not been upgraded to further boost Corey's processor performance by boosting the CPU's core and frequency, but it's costly to do so. Germany's Computerbase, citing data from mainboard manufacturers, said: The maximum load power consumption of core i9-10900k exceeds 300W。
CB's website says it's not surprising, either, because the current 8-core 5ghz processor, core i9-9900k, already has 275w power consumption at prime95, as follows:
From the previous disclosure, we know that the TDP power consumption of core i9-10900k is 125w, which is increased by at least 20-30w compared with the conventional 95w / 105W. Although the TDP power consumption does not represent the real power consumption, it is just the thermal design power consumption, but the increase of TDP means that the overall power consumption is also increased, so the full load power consumption of core i9-10900k will bring additional trouble.
Of course, the premise of the power consumption increase is that the performance has also increased a lot. The official ppt revealed that the core i9-10900k has increased the performance by up to 30%, but the single core improvement is not so obvious.
At this CES exhibition, core i9-10900k processor was not officially released, and the reason is not the main board. The main board manufacturer said that their lga1200 platform is ready, but now it seems that the processor is not fully ready. CB website said that it will be postponed to April to may, so it will not be released until Taipei Computer Exhibition.
In addition, at the CES exhibition, Intel officially announced its own high-performance GPU debut product
According to officials, Intel DG1 is its first stand-alone graphics card product for consumer-level platforms, based on a new XeGPU architecture that integrates powerful media and display engines. Designed for power optimization platforms (i.e. energy efficiency), and optimize for game and content creation.
Intel did not announce the specific performance of the DG1 graphics card, but on the spot demonstrated the effect of the DG1 graphics card running "Destiny 2" game, looks very smooth, but there is no specific picture quality settings, frames and other actual data reference.
At the same time, the Xe architecture will also be used for Intel's CPU processor, among which tiger Lake's gen12 core display is also based on the Xe architecture,There are projected to be 96 EU units,1.5 times the size of the existing Ice Lake's Gen11 kernel and three times the size of the Coffee Lake 。
Not only the scale of EU has been greatly increased, but also the architecture of Xe GPU core has been greatly improved. Therefore, it is reported that the performance of gen12 nuclear display can reach 2tflops, double the performance of 1tflops compared with gen11 nuclear display, and double the performance of gen9.5 nuclear display, which is really exciting.
With 100% increase in performance, Tiger Lake's core area control is also good, and under the same TDP power consumption and process, the area occupied by Gen12 is reported to have increased by only 25%, that is, using 125% of the area for 150% of the scale,200% of the performance, and the Xe architecture has a high PPA (performance, power consumption, area) index of scary.
Then Intel looks at AMD.
At CES 2020, amd officially launched the 7 nm sharp dragon 4000 series APU processor, as well as the RX 5600 XT graphics card and the sharp dragon threadriver 3990x processor. Among these products, 7Nm sharp dragon APU is particularly important, which is related to the success or failure of AMD in notebook market.
Before the sharp dragon 4000, amd began to try to enter the notebook market since the first generation of sharp dragon was launched in 2017. The reason is very simple, because the main global PC shipment volume is notebook and other OEM markets. DIY market alone can not guarantee that AMD's x86 share will return to the historical peak of more than 25%, while Intel's consumer PC has 70% of the revenue from the notebook market , so you can understand why Intel is giving priority to the mobile market every time it releases a new generation of processors.
AMD's Reelon 2000/3000 series of APUs aren't successful in the notebook market, mainly because the Zen architecture of the 14/12nm process doesn't work, and the performance and performance still don't surpass Intel's 14nm and 10nm processors.
Of course, the biggest advantage of the Ruilong notebook has to be mentioned, that is, the price performance ratio is extremely high. The Ruilong light and thin books of Huawei / glory, Lenovo and other companies can be bought even in the early 3000's, and the Ruilong 5 3500u 4-core CPU sold cabbage price.
With the release of the sharp dragon 4000, the situation has changed. At the same time, amd launched the sharp dragon 4000U series for light and thin books and the sharp dragon 4000H series for game books, which also attracted the support of high-end notebooks from ASUS, Dell and other companies. It is expected that more than 100 sharp dragon 4000 notebooks will be listed this year.
The root of all this is that the CPU architecture of the sharp dragon 4000 has changed, upgraded to 7Nm zen2, and AMD has given up the small chip design of desktop version and server version, and made the original core for notebook processor again, which is a great cost.
In a post-CES group interview, AMD CEO Su Zifeng also talked about the work done by Reelong 4000, noting that the processor was designed three years ago and that AMD's engineering team made a lot of efforts to optimize performance, energy efficiency and frequency.
In addition, Su Zifeng confidently said that zen2 with 7Nm technology is the best CPU core at present, and they are very proud of it.
Of course, CPU core is better than Zen2, which is the Zen3 core launched in the second half of this year. Simultaneous interpreting will increase the performance of IPC, and the increase will be about 10-15% or 17%.
However, when zen3 came out, intel was going to launch tiger Lake processor. It also used a new willow Cove core. The so-called performance improvement was also double-digit, that is, no less than 10%.