Check out our Unboxing and Overview of ASUS ProArt PZ13 here: geni.us/PZ13_Unboxing [Ultra portable creator PC] Full ProArt PZ13 Playlist here: geni.us/PZ13_Playlist
A,I, IS ALL HYPE, Dont NEED IT IN MY WORKFLOW.. .JUST A POWERFULL PC IS ENOUGH FOR DESIGN AND FLUID DYNAMICS...DONT NEED ANY A.I. SUGGESTIONS OR FOOLISHNESS ...
The "AtomMan X7 Ti mini PC" appears to be really Cool looking with great features ...however it's too expensive. I'd rather spend that money on a custom built Desktop PC that has access to cheap parts available everywhere & is easily repairable if/when a Hardware problem happens. 🙂
can you elaborate on the components you chose in your custom built that makes for a cheaper and at least as powerful machine (not forgetting the case the fan the alim the usb-c / 3 the lan... etc and how long it takes to assemble) ?
@@danielgriffith8911 Mini-PC's use mostly proprietary Hardware ...so when the Warranty is over & there is a Hardware problem, the user/owner will be paying really "BIG Bucks trying to get it fixed. I've been building & fixing PC's for 40 years & 99% of all PC'ers would rather have a custom Desktop build.🙂
One thing to consider is the actually upgradable memory for gpu because it will use ram. And it will allow you to run huge models. Slow yes, at least you can run it.
I asked myself the same question and had to do a little digging. The CPU (Intel Core Ultra 9 Processor 185H) has an on-die NPU, supporting: OpenVINO, WindowsML, ONNX RT So it can accelerate Machine Learning on those three frameworks. For general purpose AI stuff, you're still better off with an nVidia GPU for 99% of what's out there now (which you will not get with this). The most viable AI use case for this would be running Llama 3+ locally. OpenVINO can boost that. Mixing Llama with Obsidian and HomeAssistant does kind of turn you into a motherfucking sorcerer, lol. That is an insanely powerful combination. But it's also a very niche use-case for more hardcore DIY'ers in the automation space right now. That's not going to be on most people's radar for another year or two, until matter-spec devices become more ubiquitous. The other use-case that comes to mind is for people running AI shopping agents. Basically custom AI bots that track your purchases and prices (aggregated from thousands of sources) and basically does all the "extreme couponing" stuff for you via online vendors. People are making a killing with that right now, but I don't think there's a lot of longevity in that pursuit. If you do that sort of thing, this will probably pay for itself, but that's on the far fringes of my domain of knowledge in terms of setup. But those also use Llama, so it could theoretically benefit from acceleration. I just don't think shaving a few seconds off of those tasks would make a big difference for most who engage in that. They do explicitly mention Topaz Labs GigaPixel AI... there are better free online upscalers right now. I don't think that's compelling for most right now. I think the major push here is for Microsoft Co-Pilot acceleration inside of Office. Having Excel evaluate data for example. I think that's how 99% of people will interact with this in the short term. The NPU is a good thing to have, but the people who need it are going to go a lot bigger than this, unless they have a very specific need for a small form-factor, or low price. The price however is good. The bare-bones config (minus RAM and SSD) sells for just $50 over the processor's MSRP. There's value here for sure. But I think the AI features are mentioned mostly for the marketing buzz. They're on the chip anyway, so they will mention them. But for 99.999% of people, it shouldn't be a significant factor in whether or not they buy this. As mini PCs go however, this is a reasonably priced power-house. Sadly, they don't mention the full USB-4 spec. USB-4 is a catch-all term for a wide range of optional spec features, and you kind of need to know what flavor of USB-4 it is. The manufacturer doesn't even specify in their spec sheet. That is annoying, lol. Again, not a deal-breaker for most people, but if you're someone who needs to know, guess they expect you to just roll the dice, lol.
To be fair, any chip that uses intel Core Ultra xx can be called "ai chip" because that's how Intel market it. But i do see this as a very capable AI server over local network, because: 1. it has an OCCULINK port, so you can connect highend eGPUs like RTX 4090 without having bandwidth limitation. And you may just run Ollama, Stable Diffusion on it, use it as an AI server for other computers or even phone/ipad on local network to access the services using just browser. Almost all local AI stuffs requires good GPU only, an above average CPU will be enough. the one inside is good enough and is efficient. 2. this has 5Ghz ethernet. If you have proper network switch, this can definitely be used for NAS purpose too.
I would like to know which of these miniPcs is the best at running Resolve in your experience. I've watched almos all your videos on minipcs and this is the only one you didn't test resolve like the others. I would love to see a battle of minipcs running resolve. Your channel is the absolute best. Thanks a lot for all the work you do.
i dont do davinci, nor any other video eiditing software, but as far as i know youd need a dedicated gpu for video eiditing, no? id look at thinkstation p3 ultra, which you can configure with up to a2000. but, not with an i9 since that config has two coolers, one above pcie slot. in short, p3 ultra has a pcie slot for a dedicated gpu. it can get expensive, but you could find it on sale, or just config it for yourself over at lenovo website. i snagged one two months ago, basic config… i513500, 16gb ram, no gpu, 512gb nvme, 3 year warranty but it was only 800$ which was like 150$ less than lenovo price (i did get it through a 3rd party supplier, and managed to bump the warranty to 5 years, but through them, 3 is still with lenovo directly). also, temps are not great even without a dedicated gpu crammed in there. like, at idle ram is above 40c and so is nvme. cpu jumps up to and above 90 and the whole thing then starts to sound like a turbine engine, but… it is small factor and i can upgrade with a dedicated gpu, including a2000 since i5 (and up to i7, 12, 13 and i believe 14th gen now) has only one cooler. cheers.
Have this thing now a couple of weeks and it's a fine little PC. Fast, not to much fan-noise, energy efficient. But the AI part is not there. No copilot or other fancy stuff. I don't miss it that much, but it isn't there. Which is a bit strange, because of the Ultra9 processor.
He was shitting on the RMA they outsourced to some weird country I forget. ASUS howeber like all other hardware manufacters are going to eventually make their way into your ecosystem because in the PC building world, very few devices are simply one brand do it all. Like Apple for example, now doing its own sillicon, has basically regained dictatorship control over the machines. Which is silly, because, no one on a apple machine is going to do the kind of heresy to their form factor that PC and Linux users call innovative building... or adding a part or button instead of just cutting things off until youve got a mac pro that has pcie slots but nothing to put in them
This is an amazing bit of kit, its the first one I think is worth buying since the "Fit PC 2". Can you let me know if it comes with 11 Home or Pro, I will never ever by anything with Home on it.
So as a guy who is actually into running his own local LLM's, what's the "AI" part of this AI product? Does the iGPU have tensor cores? Is there an "NPU"? Or is it just the new marketing hype?
@@fushumang1716 The usage of LLMs and Stable Diffusion is limited by VRAM, not RAM. You can run quantized small LLMs such as phi3 (smallest version) at decent speeds in normal RAM. You may push it to 7b or 8B but anything bigger it gets painful quickly. And while you may be fine even if 1-3 token/s generation speed, the time until the system even starts generating text quickly gets into the minutes area as the chat gets longer.
I would like to see how this run on Linux. I wonder if those pc status display as well as the fingerprint reader and facial recognition will still work? 🤔
Would be super interested in seeing one more use case added as a "part 2" to this video: for bootable ZFS arrays on Proxmox, I believe the installer partitions the underlying physical disk using LVM (and setting up boot partitions for every disk) then creates the ROOT ZFS array by passing the remaining space on every disk as an LVM to ZFS to build the array. The big question is, does this storage architecture cause a "worst of all worlds" scenario where all VMs/LXCs stored on the bootable ZFS array encounter both whatever performance hit might exist from LVM, plus that from ZFS ? Or does the resulting ZFS array perform similarly to a ZFS array created natively directly on the physical disks ? Thanks as always for answering the questions that everyone wants to know, but nobody wants to test. Your channel is a treasure. Keep up the amazing work !
I would definitely purchase this mini PC if they did a slight redesign to move the power and all inputs to the rear instead of the sides. It would be more of an annoyance than a benefit with its current design. I know the PC may become wider but it would be more appealing to the consumer while in use. Great channel btw 👍
are those graphene heat spreaders on the Crucial RAM ? (referring to the grey color stickers) can this support 96 gb ddr5 crucial memory (48gb x2 sticks) ?
Interesting product but frankly, MTL/1XXh just isn't very good. I'm eagerly waiting for Lunar Lake to produce things like that, should be amazing when the first laptops come out.
I'm afraid to tell you that the PC you have displayed has spyware adware already installed ....its called Windows, and it screenshots your screen every 2 seconds. A feature from recall in all windows pc's.The only on turned roge that I know off.
Hello, unrelated to the video but some 2 years ago you guys tested the 7900 XT for creators and quickly dropped it cause it kept having bugs. Have you tried it since(or 7800 XT or XTX)? Has it gotten any better, or is it still problematic? Just curious to see if the support got better or not.
much better, but if you just compare stability, then Nvidia is still better. But definitely not sucking anymore, definitely worth checking the Radeon cards out.
Hi there! I’m currently using After Effects for video editing and am looking to get a mini PC. I’m stuck between choosing the Ryzen AI 9 HX 370 or the Intel Ultra 9 185H processor. Which one do you think would provide a smoother experience when working on the timeline? Also, which processor would give faster render times? I’m particularly concerned about avoiding any timeline lag. I’d really appreciate your advice-thanks!
lol for that price you can build an itx pc that's only a few times bigger but 100x more powerful without being too power consuming. still, I hope this tech keeps improving. we should be getting tiny pc's like this today.
It has an intel core ultra 9 in it. The only minipc I’ve found out there. An intel u9 has an advanced npu in it. Neural processing unit. It’s a tiny chip inside the u9 specifically made for using neural networks. I’m not sure how it performs compared to other chips. But, there’s not much software that can make optimal use of the NPU atm. So idk if it’s worth the buy. Probably not atm. But maybe in the future…
Yip. Agree. This dude has not researched what AI actually means. Does it have a local prebuilt LLM/GPT? Nope...you still need to connect to the cloud for full AI which any pc can do. The intel AI hardware is just really a marketing gimmick for now.
Can someone recommend something like this i can actually use to run a big llm at home? And powerful enough to run music softwares for recording and production. Please i need good trustworthy recommendations. Thanks guys
I use AI everyday; and from the data I've come across it would seem that the CPU+GPU+NPU would provide enough performance when working with large language models. If you purchase their GPU dock which can support AMD 7900 up to Nvidia RTX 4090 GPUS, then you end up with a very nice AI PC that takes up very little space. This type of PC seems better suited to those that want a simple set and forget machine. For power users that want onboard AI, it seems we will have to wait for powerful NPU chips to be implemented on high performance desktop motherboards.
it got power CPU, it only needs 1600Mz graphics and 16G ram , all go on AA games, pubg, residentevil, ... expensive traditional now goodbye, teckcompanies listens to me
What does "AI" actually mean here? If you mean machine learning then say so. Artificial Intelligence is something very different. NPU is a marketing term and uses a very simple 'neural pathways' template. All of these loose terms and names Apart from the largely useless screen, this is not all that different from a fairly old Fujitsu mini office machine I have. It can be stuffed with RAM, use fast NAND storage to a big quantity and I could even replace the Intel CPU if I chose. It has no Thunderbolt of course but doesn't thermally throttle even on punishing tests but the item you show seems to double as a room heater. Such a device will have limited sales at such a silly price point and even were the price to halved I would be suspicious of such heat and throttling. Maybe the one under text was faulty.
While I think it's a cool Mini-PC, I feel your are using *_misleading Clickbait_* by calling it "Best AI Mini-PC" Good GPU is required for AI, and this does not have a good GPU. Let's not bring AI into this discussion just to glam it up - instead just call it a nice Mini-PC.
Check out our Unboxing and Overview of ASUS ProArt PZ13 here: geni.us/PZ13_Unboxing [Ultra portable creator PC]
Full ProArt PZ13 Playlist here: geni.us/PZ13_Playlist
TechNotice I like how you are combining software and hardware lookup.And that is what most people like.Keep it up bro.
A,I, IS ALL HYPE, Dont NEED IT IN MY WORKFLOW.. .JUST A POWERFULL PC IS ENOUGH FOR DESIGN AND FLUID DYNAMICS...DONT NEED ANY A.I. SUGGESTIONS OR FOOLISHNESS ...
The "AtomMan X7 Ti mini PC" appears to be really Cool looking with great features ...however it's too expensive. I'd rather spend that money on a custom built Desktop PC that has access to cheap parts available everywhere & is easily repairable if/when a Hardware problem happens. 🙂
can you elaborate on the components you chose in your custom built that makes for a cheaper and at least as powerful machine (not forgetting the case the fan the alim the usb-c / 3 the lan... etc and how long it takes to assemble) ?
you are NOT gonna build something cheaper that has Oculink...
@@danielgriffith8911 Mini-PC's use mostly proprietary Hardware ...so when the Warranty is over & there is a Hardware problem, the user/owner will be paying really "BIG Bucks trying to get it fixed. I've been building & fixing PC's for 40 years & 99% of all PC'ers would rather have a custom Desktop build.🙂
@@JayMane-z7c Miniforum MS-A1 comes with an unsoldered chip and barebones.
@@JayMane-z7c Mini vs Tower is another topic, really. Its like comparing a laptop to a tower.
Integrating camera and visual display are really innovative concepts.
I think the temperature indicator refers to the pc’s temperature and not your weather temperature 😊😊😊
Good video. You showed one of the BEST reasons to buy Minisforum. Their COOLING technology is the best in MiniPCs.
BEST AI Mini PC - What AI exactly? Looks like the title was totally missed in the review 😁
I was wondering the same thing.
One thing to consider is the actually upgradable memory for gpu because it will use ram. And it will allow you to run huge models. Slow yes, at least you can run it.
I asked myself the same question and had to do a little digging.
The CPU (Intel Core Ultra 9 Processor 185H) has an on-die NPU, supporting:
OpenVINO, WindowsML, ONNX RT
So it can accelerate Machine Learning on those three frameworks.
For general purpose AI stuff, you're still better off with an nVidia GPU for 99% of what's out there now (which you will not get with this).
The most viable AI use case for this would be running Llama 3+ locally. OpenVINO can boost that. Mixing Llama with Obsidian and HomeAssistant does kind of turn you into a motherfucking sorcerer, lol. That is an insanely powerful combination. But it's also a very niche use-case for more hardcore DIY'ers in the automation space right now. That's not going to be on most people's radar for another year or two, until matter-spec devices become more ubiquitous.
The other use-case that comes to mind is for people running AI shopping agents. Basically custom AI bots that track your purchases and prices (aggregated from thousands of sources) and basically does all the "extreme couponing" stuff for you via online vendors. People are making a killing with that right now, but I don't think there's a lot of longevity in that pursuit. If you do that sort of thing, this will probably pay for itself, but that's on the far fringes of my domain of knowledge in terms of setup. But those also use Llama, so it could theoretically benefit from acceleration. I just don't think shaving a few seconds off of those tasks would make a big difference for most who engage in that.
They do explicitly mention Topaz Labs GigaPixel AI... there are better free online upscalers right now. I don't think that's compelling for most right now.
I think the major push here is for Microsoft Co-Pilot acceleration inside of Office. Having Excel evaluate data for example. I think that's how 99% of people will interact with this in the short term.
The NPU is a good thing to have, but the people who need it are going to go a lot bigger than this, unless they have a very specific need for a small form-factor, or low price.
The price however is good. The bare-bones config (minus RAM and SSD) sells for just $50 over the processor's MSRP. There's value here for sure. But I think the AI features are mentioned mostly for the marketing buzz. They're on the chip anyway, so they will mention them. But for 99.999% of people, it shouldn't be a significant factor in whether or not they buy this. As mini PCs go however, this is a reasonably priced power-house.
Sadly, they don't mention the full USB-4 spec. USB-4 is a catch-all term for a wide range of optional spec features, and you kind of need to know what flavor of USB-4 it is. The manufacturer doesn't even specify in their spec sheet. That is annoying, lol. Again, not a deal-breaker for most people, but if you're someone who needs to know, guess they expect you to just roll the dice, lol.
To be fair, any chip that uses intel Core Ultra xx can be called "ai chip" because that's how Intel market it. But i do see this as a very capable AI server over local network, because:
1. it has an OCCULINK port, so you can connect highend eGPUs like RTX 4090 without having bandwidth limitation. And you may just run Ollama, Stable Diffusion on it, use it as an AI server for other computers or even phone/ipad on local network to access the services using just browser. Almost all local AI stuffs requires good GPU only, an above average CPU will be enough. the one inside is good enough and is efficient.
2. this has 5Ghz ethernet. If you have proper network switch, this can definitely be used for NAS purpose too.
AI made the PC
Wish those cables were on the back. Such a messy look having the wires come out the side.
I would like to know which of these miniPcs is the best at running Resolve in your experience. I've watched almos all your videos on minipcs and this is the only one you didn't test resolve like the others. I would love to see a battle of minipcs running resolve. Your channel is the absolute best. Thanks a lot for all the work you do.
i dont do davinci, nor any other video eiditing software, but as far as i know youd need a dedicated gpu for video eiditing, no? id look at thinkstation p3 ultra, which you can configure with up to a2000. but, not with an i9 since that config has two coolers, one above pcie slot. in short, p3 ultra has a pcie slot for a dedicated gpu. it can get expensive, but you could find it on sale, or just config it for yourself over at lenovo website. i snagged one two months ago, basic config… i513500, 16gb ram, no gpu, 512gb nvme, 3 year warranty but it was only 800$ which was like 150$ less than lenovo price (i did get it through a 3rd party supplier, and managed to bump the warranty to 5 years, but through them, 3 is still with lenovo directly). also, temps are not great even without a dedicated gpu crammed in there. like, at idle ram is above 40c and so is nvme. cpu jumps up to and above 90 and the whole thing then starts to sound like a turbine engine, but… it is small factor and i can upgrade with a dedicated gpu, including a2000 since i5 (and up to i7, 12, 13 and i believe 14th gen now) has only one cooler. cheers.
Great job. I wanted something like that. Ready to go out of the box, just connect the LAN, monitor, keyboard and mouse and you are ready to go.
Have this thing now a couple of weeks and it's a fine little PC. Fast, not to much fan-noise, energy efficient. But the AI part is not there. No copilot or other fancy stuff. I don't miss it that much, but it isn't there. Which is a bit strange, because of the Ultra9 processor.
ASUS ProArt? didn´t you made a video months ago about their Motherboard and terrible Customer Service? now you recomend them?
money talks more :)
Good observation!
This guy has absolutely no morals. Zero morals. 0% morality.
Try to live in this world with just morals... this guy must have something to eat..
He was shitting on the RMA they outsourced to some weird country I forget. ASUS howeber like all other hardware manufacters are going to eventually make their way into your ecosystem because in the PC building world, very few devices are simply one brand do it all. Like Apple for example, now doing its own sillicon, has basically regained dictatorship control over the machines. Which is silly, because, no one on a apple machine is going to do the kind of heresy to their form factor that PC and Linux users call innovative building... or adding a part or button instead of just cutting things off until youve got a mac pro that has pcie slots but nothing to put in them
The future of PC
Looks very good piece of kit Lauri Pesur. Love your work. :)
They may have missed the mark on cable management but cool concept
It is very warm in that room.
This is an amazing bit of kit, its the first one I think is worth buying since the "Fit PC 2". Can you let me know if it comes with 11 Home or Pro, I will never ever by anything with Home on it.
11 pro
So as a guy who is actually into running his own local LLM's, what's the "AI" part of this AI product?
Does the iGPU have tensor cores? Is there an "NPU"? Or is it just the new marketing hype?
Predominantly marketing hype I reckon
It's got an NPU, coz of the Intel Core Ultra 9 185h CPU :)
@@theTechNotice Unless it can run a 70b LLM at Q4 and work with Flux at reasonable speeds, it's doesn't deserve the label "AI".
can the RAM be upgraded? I saw GMTek minipc ran LLMs on 96GB
@@fushumang1716 The usage of LLMs and Stable Diffusion is limited by VRAM, not RAM. You can run quantized small LLMs such as phi3 (smallest version) at decent speeds in normal RAM. You may push it to 7b or 8B but anything bigger it gets painful quickly. And while you may be fine even if 1-3 token/s generation speed, the time until the system even starts generating text quickly gets into the minutes area as the chat gets longer.
I would like to see how this run on Linux. I wonder if those pc status display as well as the fingerprint reader and facial recognition will still work? 🤔
Would be super interested in seeing one more use case added as a "part 2" to this video: for bootable ZFS arrays on Proxmox, I believe the installer partitions the underlying physical disk using LVM (and setting up boot partitions for every disk) then creates the ROOT ZFS array by passing the remaining space on every disk as an LVM to ZFS to build the array. The big question is, does this storage architecture cause a "worst of all worlds" scenario where all VMs/LXCs stored on the bootable ZFS array encounter both whatever performance hit might exist from LVM, plus that from ZFS ? Or does the resulting ZFS array perform similarly to a ZFS array created natively directly on the physical disks ?
Thanks as always for answering the questions that everyone wants to know, but nobody wants to test. Your channel is a treasure. Keep up the amazing work !
I would definitely purchase this mini PC if they did a slight redesign to move the power and all inputs to the rear instead of the sides. It would be more of an annoyance than a benefit with its current design. I know the PC may become wider but it would be more appealing to the consumer while in use. Great channel btw 👍
Awesome device - can't wait for an AMD version
are those graphene heat spreaders on the Crucial RAM ? (referring to the grey color stickers)
can this support 96 gb ddr5 crucial memory (48gb x2 sticks) ?
Interesting product but frankly, MTL/1XXh just isn't very good.
I'm eagerly waiting for Lunar Lake to produce things like that, should be amazing when the first laptops come out.
Really cool. It reminds me of CASE and TARS in the movie Interstellar. It would be nice if in the future It can speak and communicate with humans.
& walk...
I love your videos. I need a pc or Mac can edit 8K video. Which specs ? Mac and Pc. Thanks
Military Grade = Lowest Contractor Bid = Crap😜😜
What is the use of mini pc?
I was about to watch this video but the asus ad totally killed it for me.
Yea totally he should do this for free lol.
Asus + Windows. Wow. You got a good one. Good luck with it.
Nice, would be interesting if you also got the oculink dock, a power supply and a graphics card to see how well it works in an egpu set up.
Yes, I think the Oculink capability is the hidden gem with this machine. Really want to see how easy and effective this is in practice.
Storage?
yeah...but how does it game? thats the only pertinent question
Amazing. I think I want one.
imagine a lunar lake version of this. that would be a dream
Luna Lake and an exposed PCIe port, like the GTI Ultra series from Beelink
I could see this being cool as an "Alexa alternative" Running Linux and a privatize AI assistant. Any info on that being possible/practical?
Could it run Amplitube 5?
What makes it “AI”?
I'd be interested if it was linux, windows nope. Amiga o/s might be fun on it.
Neat unit. Horrible cable input positions
The big question is how many monitors of what resolution can it drive?
So finger print and face recognition? Gee that sounds like alot of e waste if it doesn't have a factory reset mode lowers resale value?
Woh great video however the coloring of it looks weird.
The Audio option its because you have a display connected using the HDMI Port. HDMI it's audio and video.
I'm afraid to tell you that the PC you have displayed has spyware adware already installed ....its called Windows, and it screenshots your screen every 2 seconds. A feature from recall in all windows pc's.The only on turned roge that I know off.
Hello, unrelated to the video but some 2 years ago you guys tested the 7900 XT for creators and quickly dropped it cause it kept having bugs.
Have you tried it since(or 7800 XT or XTX)? Has it gotten any better, or is it still problematic?
Just curious to see if the support got better or not.
much better, but if you just compare stability, then Nvidia is still better. But definitely not sucking anymore, definitely worth checking the Radeon cards out.
@@theTechNotice Thanks!
Hi there! I’m currently using After Effects for video editing and am looking to get a mini PC.
I’m stuck between choosing the Ryzen AI 9 HX 370 or the Intel Ultra 9 185H processor. Which one do you think would provide a smoother experience when working on the timeline? Also, which processor would give faster render times? I’m particularly concerned about avoiding any timeline lag. I’d really appreciate your advice-thanks!
What is the cost of that mini PC 🤔🤔🤔🤔🤔
Aesthetically the cables/ports should come out of the back! It looked so good until all the cables were plugged into the side 🤦🏻♂️
Interesting to see if thin clients will ever pick up popularity mainstream.
lol for that price you can build an itx pc that's only a few times bigger but 100x more powerful without being too power consuming. still, I hope this tech keeps improving. we should be getting tiny pc's like this today.
Nice review, PC manufacturers are just forcing us to AI which we dont need,
Is this $200?
"Keep looking directly onto the camera"
Proceeds to look at other things in the room.
MinisForum is a well-known manufacturer from Shenzhen, China, specializing in compact
It's more powerful than a TR 1950x with 32 cores. Wuot?
Can I make music on this thing?
What is an AI pc? How is it different from a normal PC with a web browser?
When you find out, let me know;)
AMD Version Needed ASAP💯....
You can make a fancy router out of it :)
Thank you.
I use AI in my workflow for LARGE technical document development and for editing video.
Thank you
why everything has to be AI stamp on it?
who else was rolling their eyes when he said "military grade"?
now we need a minecraft test on it
Microphone a little crappie camera not bad
Could you add an external video card to it?
what do you mean AI Mini pc dont just use AI to make the vid sound more flamboyant comeon dude
It has an intel core ultra 9 in it. The only minipc I’ve found out there. An intel u9 has an advanced npu in it. Neural processing unit. It’s a tiny chip inside the u9 specifically made for using neural networks. I’m not sure how it performs compared to other chips. But, there’s not much software that can make optimal use of the NPU atm. So idk if it’s worth the buy. Probably not atm. But maybe in the future…
People call pc’s with a npu AI pc’s
Yip. Agree. This dude has not researched what AI actually means. Does it have a local prebuilt LLM/GPT? Nope...you still need to connect to the cloud for full AI which any pc can do. The intel AI hardware is just really a marketing gimmick for now.
@@naidol aaa i see the agro spirit got you to.
small yet powerfull
Where's the AI? It's not even a Copilot+ Spec device... which means it's not even certified to run Windows AI features!
What we need is the hardware capacity the cooler capacity....The rest is nonsense if youre not a gamer or sowftware developer....
iGPU part is too weak compared to Ryzen 9 AI 370 HX.
Cables should have came out the back. That is an OCD nightmare
3:57 That's a Hong Kong SAR flag for Traditional Chinese
experiment for us with a recomended external graphics card
"Language Settings" - 5th one down 'Don't know what it is?' It is the HK Chinese setting
Can someone recommend something like this i can actually use to run a big llm at home? And powerful enough to run music softwares for recording and production. Please i need good trustworthy recommendations. Thanks guys
I notice tech, I tech notice and I say no, 'tis take too much . :)
I use AI everyday; and from the data I've come across it would seem that the CPU+GPU+NPU would provide enough performance when working with large language models. If you purchase their GPU dock which can support AMD 7900 up to Nvidia RTX 4090 GPUS, then you end up with a very nice AI PC that takes up very little space. This type of PC seems better suited to those that want a simple set and forget machine. For power users that want onboard AI, it seems we will have to wait for powerful NPU chips to be implemented on high performance desktop motherboards.
All of the chords coming off the side looks very messy
it got power CPU, it only needs 1600Mz graphics and 16G ram , all go on AA games, pubg, residentevil, ... expensive traditional now goodbye, teckcompanies listens to me
I would never use that. It’s too small. I love big pcs with a lot of screen size and lots of cool LEDs. However, that may be good for travel.
I see these as email and browsing machines.
What does "AI" actually mean here? If you mean machine learning then say so. Artificial Intelligence is something very different. NPU is a marketing term and uses a very simple 'neural pathways' template. All of these loose terms and names
Apart from the largely useless screen, this is not all that different from a fairly old Fujitsu mini office machine I have. It can be stuffed with RAM, use fast NAND storage to a big quantity and I could even replace the Intel CPU if I chose. It has no Thunderbolt of course but doesn't thermally throttle even on punishing tests but the item you show seems to double as a room heater.
Such a device will have limited sales at such a silly price point and even were the price to halved I would be suspicious of such heat and throttling.
Maybe the one under text was faulty.
I have a problem with the word AI being appended to anything.
As a general rule, I do not watch videos that have the word AI in them.
That looks like the flag of Hong Kong.
Nice.
The „country“ is Hong Kong.
Hong Kong
This PC is interesting because it's very interesting... That's interesting huh?
now we need a gta 6 test on it
He always gets so excited doing his cash cow reviews. This has 2.5 stars on Amazon from certified buyers. The worst of all the Minsforum PC's.
Nice little PC, but all those cables poking out the side are waaaay too ugly. Pass.
for the price, I buy a m4 macbook air next year
While I think it's a cool Mini-PC, I feel your are using *_misleading Clickbait_* by calling it "Best AI Mini-PC"
Good GPU is required for AI, and this does not have a good GPU.
Let's not bring AI into this discussion just to glam it up - instead just call it a nice Mini-PC.
Hong kong flag, cantonese language
Why are you messing with color grading? You look orange and red.
Yeah £800, of mini pc😂. Why would anyone buy this at that price, unless they have money to waste. Nice piece of kit but not worth buying.
Sad that level of hardware has to run total POS (to date) W11..
no AI found