Now that artificial intelligence is beginning to pervade all debates, regardless of the intellectual level of the circles in which it is discussed, it is time that many of its concepts universally used. Gamers have been involved in processes such as smart image enhancement, as well as more creative people using the same or cleaner programs. This year has been a boom for chat models and AI, but these uses have not yet gained much support in work environments and daily life when processes are driven by customers rather than the cloud.
That’s why it was very interesting to try to review the new laptop Medion E15443 (MD62621). Basically, this is a “multimedia notebook”, a so-called mid-range laptop that is suitable for less demanding use and can work in office automation, multitasking and even image editing, audio and video. Interestingly, thanks to its Intel chipset, together with Intel Core Ultra 5 processors and Intel Arc graphics, it is presented as a machine focused on AI processes, including a special “Neural” Unit (NPU ) for AI.
Without wasting too much time in theory, this NPU is not only designed to handle client-based cloud tasks, but also to support the CPU and GPU and reduce the load. At the same time, it must allow non-high-end computers to run processes that previously required more powerful and expensive units. In fact, Intel hopes to have as many as 100 million computers of all sizes using AI accelerators by the end of next year.
Here’s a hint:
In other words, even if the Medion MD62621 is created as a laptop, it must offer better performance in user-side AI than a mid-range device. We started by testing the general use of a laptop based on a traditional view, and then finished with more specific AI tests.
The price of the laptop goes around £650, and it comes with a good 15.6-inch 16:9 HD screen, including an Intel package, 16 GB of RAM and a 512 GB SSD. It’s very thin and compact, with no effort to compete with the lightest books, at just 20.6mm and 1.8kg.
As for connectivity and other features, I really appreciate an SD card reader, which is increasingly rare, but I miss an Ethernet port, although I know it’s not a priority for devices with Wi-Fi Fi 6. In terms of hardware, the only real nitpick is the length of the power cable, which is very short.
For a little over a week, we had the pleasure of using a modern and fast computer on a variety of tasks. Its specs are well coordinated, and response times are fast, even with a fully tabbed search engine, several automated office suites, and some multimedia features that include the most some of its RAM. Its keyboard is responsive and soft, and thankfully includes a number pad. Similarly, the construction feels solid although it is not luxurious in design and there is no cooling solution.
Here’s a hint:
In general, if you are looking for a general multifunctional laptop with a 15-inch screen that is lighter than the mentioned price tag, this Medion is a real choice worth considering.
Moving on to artificial intelligence, the results were very close to what I expected. While the Intel Core Ultra 5 NPU can account for some processes by making them run faster than other similar PCs without it, it really depends on software and compatibility, and that than I expected. Some programs offload these tasks to the CPU or GPU in very specific ways, and I feel that the Intel architecture, if it is literally distributed, could take on some of the workload itself. Likewise, when there are differences in graphics performance when comparing Nvidia and AMD, I assume the same is true when relying on a graphics card, in the case This is integrated with SoC Intel.
Having liked the examples Medion used in his demos of various AI applications, I tried three of them myself. Overused generational AI aside, of all the apps shown, I chose to clean up a high-resolution photo in Adobe Photoshop, split a song into tracks in Audacity, and run a few games to see if scaling can make them work in a non-game. laptops.
To make a casual comparison, I measured the runtime of two 2018 desktops with the Medion MD62621 in the first and second positions. For the first test, I used Intel Core i7-8700K and Nvidia GeForce GTX 1080Ti 32 GB RAM (my work PC). As for the second test, I used a compact desktop with Intel Core i5-8499 @ 2.8 GHz and Nvidia GeForce GTX 1060 16 GB RAM (my home PC).
I photographed a dark interior with my Nikon Z6 in RAW. This is a cropped image with dimensions of 5405×3603 pixels (from 6048×4042 primes), shot at 300 ppp at ISO 2500, 14-bit color depth and 27.5 MB file size, just for your information. Using Adobe Camera Raw’s 50-stop “noise reduction” feature (high on purpose), the most powerful desktop took 27 seconds, the humble desktop 1:19, and the Medion 3:20.
Continuing with the song, instead of using a lossless format, I converted the Red Hot Chilli Peppers “Goodbye Angels” album from my iTunes to a 45.2 MB WAV. It was originally 9.15 MB 256k AAC M4A with a duration of 4:28. I used the OpenVINO plug-in to divide it into 4 stems: Drums, Bass, Vocals and Other, tried both CPU and GPU. A more powerful desktop took 1:18 with the CPU and failed with the GPU. His younger brother took 1:22 with the CPU, and the GPU also failed. In the end, Medion took 1 minute and 32 seconds using the CPU and 37 seconds using the GPU. To be honest, it only took 58 seconds because he was loading the model for the first time, but after that, it only took 37 seconds, as you read.
As for video games, I ran benchmarks with demanding games like Returnal and Avatar: Frontiers of Pandora, and it was clear to me that the demands on the computer were too high. None of them reach comfortable frames for playing at low settings (they both hit 15 fps), but it will be interesting to see how Ubisoft’s game improves performance by changing AMD FidelityFX Super Resolution 3 to Intel Xe Super Sampling, perhaps a logical video resolution. rather than graphics.
Overall, the results of the AI tests are very interesting and different, and they reinforce the theory that there is not enough optimization or compatibility when it comes to what NPUs can do. As far as I can tell, I get the impression that Adobe uses support on Nvidia specific graphics cards and does not know how to use the NPU properly, and in the build we see more concrete examples. Quality aside, this song shows that Medion can face and even overcome more powerful computers. Although it’s important to note that OpenVINO learning comes from Intel, so ironically there are compatibility issues if you try to use an Nvidia GPU. Finally, while we’ve seen the Medion run old games like Rise of the Tomb Raider smoothly thanks to Intel Arc expansion, that’s probably as high as a laptop like this can go. In other words, it’s not bad to play older or more challenging games.
Having said that, I still think the Medion 62621 is a balanced product, with good multitasking performance and an intelligent AI integration idea that we will definitely see updated over the next few years , not years, but months. Now, it’s not fully optimized as seen in the tests, but it’s always better to have a computer ready to go, and if I were to buy a laptop in this price range, I’d want the SoC this.
