When the notebook market has been shrouded in the haze of "lack of innovation" for many years, the new ThinkBook 16+/14+ 2024 with the Core Ultra processor seems to be a ray of sunshine penetrating the thick clouds, once again making people feel the excitement brought by "technological progress", especially with the support of the Core Ultra processor, which gives PC AI attributes, and its stronger computing power also makes it easy to complete local AI applications with a thin and light core display. So what is the AI computing power of an AI PC equipped with Intel's first-generation Core Ultra processor, and what is the experience of local AI applications?
With such a mood and question, I got the new ThinkBook 16+ 2024, without further ado, just follow me and experience it!
CPU: Intel Core Ultra 7 155H processor, 16 cores and 22 threads;
GPU:Arc Graphics 8Xe核心,128EU
NPU:Intel AI Boost,双神经计算引擎,1.4GHz
Memory: 32GB DDR5 5600MHz (dual memory slots)
SSD: 1TB PCIe 4.0 (dual M.2 bays)
Screen: 16 inches/2.5K/120Hz/100%sRGB
无线网卡:AX211 Wi-Fi 6E
Other features: ForcePad Touchpad / TGX External Graphics Expansion Interface / Intel Evo Certified
The ThinkBook 16+ 2024 has been labeled with the new Intel evo logo, which means that it is the Intel Evo flagship AI superlaptop powered by the new Intel Core Ultra processor.
All performance tests in this article are conducted in "Extreme Performance" mode, unless otherwise specified
1. Core Ultra processor, the underlying guarantee of user experience
The ThinkBook 16+ 2024 is equipped with a new generation of Intel Core Ultra processors, codenamed Meteor Lake.
Compared with the previous X86 processor, and even the 12th and 13th generation Core core design, the Core Ultra processor is a big change, and the Meteor Lake architecture is a revolutionary innovation in the CPU architecture in the past 40 years.
Computing module: The traditional CPU core, high-performance P-core and high-performance E-core are located in the module, and the computing module in the Core Ultra processor is upgraded to the Intel 4 process, which improves the performance and energy efficiency ratio.
Graphics module: The core graphics of the Core Ultra processor is built with the Xe LPG architecture, which is modified from the Xe HPG architecture of Intel Arc discrete graphics cards, and the theoretical performance per watt is doubled compared to the previous Iris Xe core graphics.
SoC module: The SoC module is a key design for the Core Ultra processor to improve its energy consumption ratio, with two ultra-low-power LPE cores, and many other functions are also integrated into the SoC module, such as the codec engine and display engine, which are often integrated into the SoC module In the past, these functions had to be processed by P-cores or E-cores, which would affect the energy consumption ratio, but with LPE cores, these functions can be digested and completed by using LPE cores in SoC modules to improve the battery life of laptops.
In addition, the NPU is also located in the SoC module, which can be specially used for the calculation of AI-related applications, improve the performance and energy efficiency ratio of AI computing, and realize the collaborative AI computing of CPU+GPU+NPU
IO module: integrated with thunderbolt, PCIe lanes, etc
Specifically, the ThinkBook 16+ 2024 Ultra 7 155H in our hands has a total of 16 cores, including 6 high-performance P-cores, 8 high-performance E-cores, 2 bottom-copying low-power LPE cores, 22 threads, a maximum turbo frequency of 4.8GHz, 24MB Intel Smart Cache, TDP 28W, and a configurable maximum power consumption of 115W
1、ThinkBook 16+ CPU功耗释放
As is customary, before we start testing the CPU's performance, let's take a look at the ThinkBook 16+'s depyretic capabilities and determine how much power it can free up for the Ultra 7 155H.
After 30 minutes, the CPU power consumption was stable at 75W, the P core frequency was 3.8GHz~4.2GHz, the E core frequency was 2.8GHz, and the LPE core was 2.5GHz, and the CPU packaging temperature was 106 degrees and 107 degrees recognized by HWinfo software, while Lenovo Computer Manager recognized it as 97 degrees.
If you adjust to the geek mode with more aggressive performance release, 30 minutes of Stress FPU baking, the CPU power consumption is stable at 80W, the P core frequency is 3.9GHz~4.3GHz, the E core frequency is 2.8GHz, the LPE core is 2.5GHz, the CPU package temperature is AIDA64, HWinfo software recognition is 105 degrees, 107 degrees, and Lenovo PC Manager recognition is 94 degrees.
Although the ThinkBook 16+ 2024 has a business-like serious appearance, the performance release is really unambiguous, and it is a top-of-the-line existence among thin and light laptops, allowing the performance of this Ultra 7 155H to be fully utilized.
2. Core Ultra 7 155H performance evaluation - the ultimate energy efficiency ratio!
Let's take a look at the performance of the Core Ultra 7 155H processor.
In CPU-Z, the single-core performance score is 745, and the multi-core performance score is 8534.2;
In CINEBENCH R23, the single-core performance score is 1816pts, and the multi-core performance score is 17925pts;
In CINEBENCH 2024, the single-core performance score is 106pts, and the multi-core performance score is 1025pts;
The ThinkBook 16+'s top-of-the-line cooling capabilities, coupled with the Core Ultra 7 155H 115W's wider range of configurable power consumption, show a significant performance improvement in both single-core and multi-core performance compared to the previous generation i7-1360P with a TDP of 28W. On top of this, the more advanced Intel 4 process, innovative modular design, and the Core Ultra processor also have an excellent energy efficiency ratio, providing better off-power performance and longer battery life.
I THEN TESTED THE SCORES OF CINEBENCH R23 AND CINEBENCH 2024 WITHOUT PLUGGING IN:
CINEBENCH R23：单核性能得分1808pts，多核性能得分17602pts；
CINEBENCH 2024：单核性能得分99pts，多核性能得分960pts；
It can be seen that in the case of not plugging in the power supply, the Core Ultra 7 155H can also give users quite strong performance, and the performance gap in the plugged state is very small, that is to say, when used off the power, the system's responsiveness, application opening speed, multitasking ability, etc. will not be greatly reduced, for users who often need to travel or go out to work, there will undoubtedly be a lot of bonuses in the use experience.
3. Core display performance evaluation - Arc core display, full power!
The core graphics card of the Core Ultra processor has ushered in a wave of great upgrades, which has been introduced earlier, the architecture has been upgraded from Xe LP to Xe LPG, and the core specifications of full blood have also come from 96EU to 128EU, and the maximum frequency of the core graphics in the Core Ultra 7 155H is 2.25GHz. In addition, Arc also supports DX12 Ultimate and XeSS technologies.
Let's start with 3DMark for benchmarking, which scored 8934 points in Fire Strike, which represents DX11 performance, and 3698 points in Tims Spy, which represents DX12 performance.
Time Spy's score exceeds 3500 points, in terms of theoretical performance, the full-blooded Arc GPU in the Core Ultra is currently the most powerful GPU, which is comparable to the discrete graphics card GTX 1650 on the notebook side, slightly lower than the GTX 1650Ti, which is enough for a certain amount of game entertainment, and can also have stronger performance in video, 3D rendering and other related work. It should be noted that the memory specification in Lenovo Xiaoxin Pro 16 is dual-channel DDR5 5600MHz, and if it is paired with LPDDR5 7467MHz memory with higher frequency, the core graphics performance will be higher.
In Dota 2, select 1920*1200 resolution, preset high image quality, select a game at the beginning of the game on the home page to watch, free perspective, the average frame rate of the whole game is: 88. In competitive online games that do not require high performance, the Arc GPU in the Core Ultra processor can handle it completely, and it can maintain a high frame rate even at high image quality.
Let's also test the efficacy of XeSS, XeSS is a new feature of Arc graphics card, Xe core graphics did not have before, its working principle is to render and pre-process the game screen at a lower resolution, and then output these pictures to high resolution through machine learning, so as to achieve the effect of not reducing the image quality and improving the frame rate performance, especially in the performance-eating triple-A game, the effect of XeSS is more significant, so that the core graphics thin and light laptop can also run smoothly 3A masterpieces.
In "Cyberpunk 2077", choose 1920*1200 resolution, preset low image quality, when XeSS is not turned on, the average frame rate of running the game's built-in BenchMark is: 34.66 frames, after turning on XeSS to the performance level, the average frame rate of the game's built-in BenchMark is: 53.36 frames, the frame rate is quite close to 60fps, and the increase is close to 55% after turning on Xess, and the effect is already quite considerable.
2. AI PC - the combination of software and hardware for a more complete AI experience
The Intel Core Ultra processor is the first time that Intel has added an NPU specifically for AI computing to the Core processor, and the concept of AI PC came into being. With the addition of NPU, the AI computing power of the entire processor has been greatly improved compared to before. Compared with pure CPU AI computing, NPU can release the computing power of the CPU and GPU after intervening in AI computing, so that the CPU and GPU can be better invested in other work without being affected. For example, when I perform AI benchmark performance tests, I can find that only the NPU is computing at a high rate, and the CPU and GPU are not involved in this work.
Next, let's use this ThinkBook 16+ 2024 equipped with Core Ultra 7 155H to experience AI applications, and also feel the power of the first-generation Core Ultra processor in this regard.
1. AIGC experience
Compared with cloud-based AIGC applications such as ChatGPT, local AIGC applications have the advantages of faster deployment and faster enablement, and more importantly, local large language models can be used without networking, and running locally can also better protect privacy, which is more secure and stable, which has inherent advantages for many industries.
So how efficient is the Core Ultra processor when running local large language models? Let's use the AIGC assistant demo version to experience it (note: the AIGC assistant demo version is for demonstration purposes only, not the software that comes with the notebook).
We use the ThinkBook 16+ 2024 to call the Baichuan2-7B model locally, and this model has the current mainstream parameter scale.
I asked three different questions to the Baichuan model: please plan a 5-day trip to Beijing, how to improve the level of Chinese men's football in 20 years, and as a brand marketing planner, help me draft an outline of the year-end summary. All three questions were answered very quickly, basically within a few seconds of asking the question, and the answers were very detailed, just like the experience we had when we had these conversations in the cloud on the web.
The powerful AI performance of the Intel Core Ultra processor allows it to support the local operation of large models with up to 20 billion parameters, and my tests were conducted without an Internet connection.
When running AI conversations, I also opened the task manager to monitor the hardware situation, the CPU and GPU will have a high load, and the memory occupation has reached about 20GB~22GB (only this one application is run after booting), it seems that the local AI large language model has a lot of pressure on memory.
Next, let's take a look at the Wensheng diagram, which theoretically requires high performance, AIGC Assistant not only has an AI dialogue function, but also has a built-in Stable Diffusion-LCM, which can be used to perform Wensheng diagrams.
After entering the keyword, select the Fast Painting mode, it takes about two or three seconds to output four pictures, which means that the average time for each picture to be generated is less than 1 second, and the quality of the raw pictures is not low.
At the same time, there is also the function of drawing a picture, for example, you can change the dragon in the picture to black, and the completion time is about 10 seconds.
In the past, functions such as Wensheng Wen and Wensheng Tu needed to be connected to the Internet to run in the cloud, or required expensive local hardware support, and may also face problems such as fees, limited times, waiting in line, privacy leaks, etc., but now with the blessing of Intel Core Ultra processors, a laptop of several thousand yuan can even complete these AI applications without being connected to the Internet.
2. AI video keying
In the past, the character keying of the video is a very cumbersome thing, which requires the operator to modify frame by frame, but now with the help of AI, this work can be done quickly, Intel and its domestic partner Jianying have developed the AI video keying function, with the help of this software, you can complete one-click AI keying, and the function has been fully adapted to the NPU unit of Core Ultra, and the user can manually switch the load to the CPU or NPU.
Let's take a look at the software's video keying feature and see how fast and power the load performs on the CPU and NPU. In Jianying, import a video with 4K resolution, 25fps, and a duration of about 1 minute, and perform video keying operations. At this point, the performance mode is Smart Mode.
First of all, we chose to load the video keying function on the NPU, and you can see that the NPU unit is working in the image below, and the whole smoothness takes about 35 seconds.
I also recorded the power consumption of the CPU package over this short period of time, and the average power consumption of the CPU package during this period was roughly 23.76W.
Moving the load to the CPU and repeating the operation of the video keying function is about 35 seconds, but the average power consumption of the CPU package during this time is 42.13w.
In other words, in this scenario, due to the blessing of NPU, compared with the CPU, the NPU can save nearly twice the power consumption in the same time.
In the past, this technology was difficult for the average individual to reach due to its complexity and high labor or computing costs, but now with the help of the Core Ultra processor, video keying has become easier and more efficient, allowing more people to easily enjoy its convenience and make their personal creations more sophisticated without lag or stuttering.
3. Video conferencing
Video conferencing is currently one of the most popular AI applications, and functions such as background blur are actually recognized by AI during video conferencing. Let's still experience two software, one is the "Lenovo Smart Meeting" that comes with the Lenovo ThinkBook 16+, and the other is the third-party software "XSplit VCam".
Lenovo Smart Meeting has a lot of features, in addition to the common background replacement, there are also tracking mode, beautification, manual face adjustment, temporary meeting away from the temporary avatar and so on.
There are three tracking modes, solo tracking, group tracking, and manual tracking. Single-person and group tracking is very easy to understand, the camera will focus after AI recognizes the person, and the manual tracking function can manually frame the frame within the camera range so that the camera only shows the frame area.
There are many options for manual face adjustment, such as eyes, face, nose, chin, etc., which is comparable to beauty software.
Finally, there is a very useful feature, temporary clones. If you get up and leave or answer the phone directly, it will inevitably seem a little embarrassing, the temporary clone function can resolve this embarrassment, it can record the user's state in real time a few seconds before turning on this function and then repeat the playback, the user does not have to turn off the camera, and it will not affect the normal progress of the meeting.
XSplit VCam does not have as many functions as Lenovo Smart Meeting, mainly background blur or background replacement, but XSplit VCam allows users to choose whether to put this load on the NPU, CPU or GPU for calculation, I also tested the difference in power consumption of the whole machine when selecting NPU or GPU after background replacement is enabled.
In the case of power failure, the performance mode selects the smart mode, and when the load is selected to be placed on the GPU, the discharge power consumption of the battery is about 24W~25.5W.
When the load is placed on the NPU, you can see that the NPU starts to work, and the discharge power consumption of the battery is about 21.5W~22.5W.
Before the NPU, functions such as background replacement/background blur were basically run on the GPU, but after the NPU that can handle AI computing more efficiently, the power consumption of the whole machine has dropped by about 2.5W~3W, compared to before, the Core Ultra processor does have a higher energy efficiency ratio in AI.
4. Summary of AI PC experience
AI artificial intelligence is the trend of the times, AI will continue to develop, it is only a matter of time before it penetrates into all walks of life and daily life, laptops as the most powerful computing power in portable devices will lead people to the AI era, and Intel Core Ultra processors have stronger AI computing power and higher energy efficiency ratio, so that AI, which was far away from the public, has become an application that can run locally on ordinary PCs, lowering the threshold for AI use and popularizing AI to the public.
In addition to the local large language model, Wensheng graph, AI video keying, and video conferencing related functions experienced in the above parts, there are also local AI applications such as AI voice changing, AI music education, and AI motion capture.
3. TGX independent display interface + ForcePad pressure-sensitive touchpad
The innovation of the Lenovo ThinkBook 16+ 2024 is not only the AI features brought by the internal core hardware, but also the design of the product, which is mainly reflected in the TGX independent graphics interface and the ForcePad pressure-sensitive touchpad.
1. TGX independent display interface
The TGX GPU expansion interface allows users to use Oculink external desktop discrete graphics cards to explore higher graphics performance. The Oculink interface is essentially a PCIe direct connection port and does not require protocol conversion, but if you want the discrete graphics card connected to the Oculink interface to use the bandwidth without loss, then you must directly connect the Oculink interface to the PCIe lane of the CPU.
The previous processor had to give priority to the use of hard disks with M.2 interfaces due to the limited PCIe lanes directly connected to the CPU, so the Oculink interface was difficult to implement, while the Core Ultra processor adopted a modular design, and the PCIe lanes in the IO module = the CPU was directly connected to the PCIe lanes, so the ThinkBook 16+ 2024 was in the Core Ultra With the blessing of 7, there are more PCIe lanes to be allocated, which can not only ensure the dual M.2 hard disk interfaces, but also separate PCIe4.0*4 lanes for the Oculink interface, with a bandwidth of 64Gbps, and there will be no loss when connecting external high-end graphics cards such as RTX 4070s.
Advanced IO Tile design
In this way, the use cases of the ThinkBook 16+ 2024 have been greatly expanded, and in the past, if we had the need to work out, play or design at the same time, then we needed at least two devices, a thin and light laptop for office work, and a desktop or gaming laptop for gaming or design, and the ThinkBook 16+ The advent of 2024 will allow us to meet these multiple needs with one computer, and it will be able to take it to the office with you when you unplug the graphics dock, and you can also solve the needs of working on the go with the excellent energy efficiency ratio of the Ultra processor.
TGX graphics card expansion interface
2、ForcePad压感触控板
The pressure-sensitive trackpad is only really useful when you actually use it, especially for heavy touchpad users like me.
Using the pressure sensor and vibration motor, any area of the global pressure-sensitive touchpad can be clicked and the pressure and vibration feedback are constant, compared with the traditional T-shaped touchpad, the use experience is equivalent to a "dimensionality reduction strike", the positioning is more accurate, the feel is more comfortable, and the improvement of use is not a little bit.
The ThinkBook 16+ 2024 brings the pressure-sensitive trackpad to the mainstream of Windows thin and light laptops, and the trackpad area is quite large, which is really rare.
4. Dual memory slots + dual M.2 hard disk bays
Lenovo ThinkBook 16+ 2024 has dual memory slots + dual M.2 hard disk bays inside the fuselage, which is directly full of expandability in the field of thin and light books!Compared with many products of the same type of onboard memory + single hard drive bay, Lenovo ThinkBook 16+ 2024 is much more flexible in the future upgradeability, on the other hand, some gaming laptops are not even as good as this ThinkBook 16+ 2024.
The 1TB PCIe 4.0 SSD comes from Yealink, with a maximum read speed of 7000MB/s and a maximum write speed of 4830MB/s, which is a full-speed PCIe 4.0*4 SSD.
5. Traditional family design language
The Lenovo ThinkBook 16+ 2024 still adopts the traditional family-style design language, which is in the same vein. The A side is still a classic contrasting design, with a mirror ThinkBook on one side, simple, atmospheric, designed, low-key, and serious.
The screen supports one-handed opening and closing, and the maximum angle is 180 degrees. Thanks to the "brim" design of the B-side camera area, the screen is more convenient to open and close, and the camera position is physically blocked.
Lenovo ThinkBook 16+ 2024 adopts an ultra-narrow bezel design, with a screen-to-body ratio of up to 91%, a size of 16 inches, a screen ratio of 16:10, a resolution of 2.5K, a refresh rate of 120Hz, a color gamut covering 100% sRGB, DC dimming, and has passed the TUV Rheinland hardware low blue light certification and TUV eyesafe certification, and the screen quality belongs to the first-class of the current notebook.
When you come to the C side, the first thing that catches your eye is the full-size arrow keys and function keys, as well as the keypad area, which is well received. The shape of the keycap is still the smile design of the Lenovo system,The feel of the keyboard is better than the previous generation,1.5mm key travel,Powerful feedback,The key feels good。 The power button is located in the top right corner of the keyboard independently of the keyboard area, and models from the Core Ultra 7 also have additional support for fingerprint recognition.
Finally, let's take a look at the interfaces, a total of 10 interfaces, which can be described as a "perfect" grand slam, on the left is a full-featured USB 3.2 Gen2 Type-C, a USB 3.2 Gen1, an HDMI 2.1, a Thunderbolt 4, a 3.5mm headphone/microphone interface, a TGX graphics card docking station interface, and on the right is an RJ45 network cable interface, a USB 3.2 Gen1 interface, an SD card reader, and a hidden USB 2.0 interface.
Sixth, write at the end
The Lenovo ThinkBook 16+ 2024, with the blessing of Core Ultra, has burst out infinite vitality for this year's notebook industry.
First of all, the core Ultra processor with innovative architecture, the CPU part has a stronger energy consumption ratio, and the performance and function of the Arc core graphics card make it possible to run 3A masterpieces smoothly even in thin and light laptops. The second is the most critical AI, Intel not only makes efforts in hardware, adding NPU units to CPUs, but also actively working with partners in software to accelerate the implementation of AI applications, and software and hardware work together to popularize AI.
The ThinkBook 16+ 2024's exciting Oculink graphics expansion interface is actually made possible by the more advanced design of the Ultra processor, which expands the use cases and makes the use of thin and light laptops more flexible, and the ThinkBook 16+ 2024 introduces a pressure-sensitive trackpad in the mainstream price point to take the touch experience to the next level.
Overall, whether it's AI computing, regular office, learning and entertainment, or the user experience of peripheral configurations, the Lenovo ThinkBook 16+ 2024 of Intel Core Ultra Edition is definitely worth choosing.