AMD Ryzen 7 5825U debut: HP Envy x360 15 2022 convertible review

[Redaktion]

The 2022 Envy x360 15 may look a lot like the 2021 model, but all the under-the-hood upgrades help make it a much better system. Users are getting additional ports, a higher resolution webcam, a faster Zen 3 processor, and more.

https://www.notebookcheck.net/AMD-Ryzen-7-5825U-debut-HP-Envy-x360-15-2022-convertible-review.657382.0.html

[NikoB]

Again, a terrible tuning of RAM, although even the installed config is like from a madhouse - 8+4...

Absolutely lousy screen. It seems that I don't want to look further, despite the set of cool ports...but then the miracle begins...suddenly, the processor gives out just a f*cking(super) level of performance (37W PL1), but the miracles do not end there ... already seeing such a level, I figure that it will be quite noisy there, but what do I see in the end? Just an awesomely quiet cooling system while consuming as much as 46W! 
Just like today, the same "miracle" happened with the Thinkpad Z16, where it is quiet even at 65W consumption! But then a question arises for other testing modes - how in "Witcher 3" mode with consumption LOWER than in load average, with only 41W vs 46W, the noise, on the contrary, becomes monstrous 43dB(A)?!

Today does not seem to be the first of April, the day of fools, but I would still like to get an explanation from the authors, as it is also physically possible. Is this a review paid by the manufacturer or a "random" mistake? You must understand that the data for consumption vs noise are physically impossible with "measured" data for for another modes, right?

[RobertJasiek]

I am afraid there is no simple correlation between consumption and noise but it also depends on how (badly) each fan mode is implemented in the UEFI.

[hei]

Stop spending effort on reviewing outdated dead on arrival products thanks

[IndexingAlex]

I cannot understand WHY manufacturers like HP and Lenovo continue slap such dim, subpar screens into 2-in-1 devices. The target audiences for these are supposedly artists/creatives and business people showing demos and presentations, right? These are people who need bright screens and/or accurate color representation, for Pete's sake! Want to upgrade? Well, this year the only option you'll get is a gimmicky OLED guaranteed to halve the battery life, all but chaining you to an outlet.

I always hate buying a new laptop, but this year has an especially awful lineup. There's literally nothing with the specs and performance I want.

[RobertJasiek]

this year has an especially awful lineup. There's literally nothing with the specs and performance I want.

Consider yourself lucky to have been waiting for only one year so far! I have been waiting for 2.5 years now despite only having a few essentials: at least RTX 3070, at most 43 dB, at least 32GB RAM, at least 16:10, no PWM, ordinarily sized arrow keys, dedicated PageUp, PageDown, Home, End keys. Among the three closest candidates, one is too loud, one only has 16GB, one has 16:9 (not to mention their keyboards).

[usacomputer]

This device has to be updated to the new AMD ZEN 4 7040 Phoenix= LPDDR5 + RDNA 3 + USB 4.0 + HDMI 2.1 + artificial intelligence that will be very useful for the future Windows 12, office and the new Direct XII APUs all include artificial intelligence and everything with AMD at 4nm vs intel at 10nm that consumes more power and more heat making the life of the processor shorter
I hope to see this laptop with the new AMD ZEN 4 7040 Phoenix labeled ORANGE because it would be a success.

[NikoB]

all include artificial intelligence

Please stop misleading people. There is no "AI" and will not be in the next 100 years. People in the subject understand this. Public propaganda deliberately misleads the masses. These are ordinary neural networks, rather stupid.

Second, real effective neural networks require hundreds of terabytes of RAM. Here is the usual chip of a primitive level. But even he will need at least 128GB of RAM to do something meaningful.

Why do you think AMD increased the available RAM to 256(in the future 384)GB only in Zen4 Phoenix, while only 64 GB is available for the "top" 7945HX? It is for this reason that a decent neural network requires huge amounts of RAM.

All buyers, as it were, receive an implicit warning from AMD in advance - DO NOT buy laptops with less than 128GB of RAM if you plan to work intensively with neural networks and their chip built into the SoC.

So people naively thinking that they can use W12 on memory less than 64GB are just stupid.

Who wants to argue with me for the future?

[RobertJasiek]

As I have explained in German, there are AI (such as a specific-purpose neural net) and General AI (such as a human adult's brain). Special-purpose AI can often run on much less RAM than 64GB and needs some VRAM smaller than, say, 24GB. Some (such as a Go AI running for less than ca. 80 minutes) even need only 1GB VRAM and ca. 20GB RAM. Of course, there are counter-examples but thst is my point: it all depends on what the application is.

[NikoB]

Robert, the question is in the task level for the neural network. I will once again refer to the support for 256 (384 in the future) GB of RAM exclusively in the 7040 series. This is not done so easily! AMD understands perfectly, understands in advance that at least some decent neural network in terms of its functions can easily gobble up 128GB or more.

As I wrote earlier in another topic - now the situation with the size of RAM painfully reminds me of the end of the 90s, when memory from tens and hundreds of megabytes quickly turns into gigabytes. Now, apparently, the same thing will happen - the RAM of ordinary PCs will grow to 1TB and more in a few years. Serious workstations will be equipped with tens of terabytes for the development and operation of neural networks in enterprises and professionals. Because working with public neural networks from large DCs is one thing (and it is already paid and expensive), but your own network is completely different. While this is all in its infancy, but this is only for now...

The PCs themselves with their von Neumann architecture are already outdated morally, a new architecture for quantum computing and neural networks is urgently needed, fundamentally different.

[RobertJasiek]

Not every neural net needs a new kind of computer. Quite a few neural nets work well on TPUs (tensor) or GPUs can be fine, too. That is not to say that quantum computing or just unified memory might be useful again.