AMD's Strix Successor Just Caught the M4 Pro

I've spent a bit of time testing this I've spent a bit of time testing this brand new mini PC, the Ser 10 from brand new mini PC, the Ser 10 from brand new mini PC, the Ser 10 from Beelink, [music] Beelink, [music] Beelink, [music] against last year's model, the Ser 9, against last year's model, the Ser 9, against last year's model, the Ser 9, and the Ser 8 before that. They all look and the Ser 8 before that. They all look and the Ser 8 before that. They all look kind of alike, but inside [music] kind of alike, but inside [music] kind of alike, but inside [music] they're different because the Ser 10 has they're different because the Ser 10 has they're different because the Ser 10 has AMD's new Gorgon Point chip versus the AMD's new Gorgon Point chip versus the AMD's new Gorgon Point chip versus the previous Strix Point. And after doing my previous Strix Point. And after doing my previous Strix Point. And after doing my developer and LLM AI tests, I came out developer and LLM AI tests, I came out developer and LLM AI tests, I came out with something I didn't expect. All with something I didn't expect. All with something I didn't expect. All right, so this is the Beelink Ser 10 right, so this is the Beelink Ser 10 right, so this is the Beelink Ser 10 Max. They added Max to it. It's got the Max. They added Max to it. It's got the Max. They added Max to it. It's got the Ryzen AI 9 HX470, Ryzen AI 9 HX470, Ryzen AI 9 HX470, code name Gorgon Point. But the real code name Gorgon Point. But the real code name Gorgon Point. But the real shopping question here is three shopping question here is three shopping question here is three machines, not one. [music] If you take a machines, not one. [music] If you take a machines, not one. [music] If you take a look at the Beelink site, last year's look at the Beelink site, last year's look at the Beelink site, last year's Ser 9 still available, still hot, and Ser 9 still available, still hot, and Ser 9 still available, still hot, and still on sale. Usually a couple hundred still on sale. Usually a couple hundred still on sale. Usually a couple hundred bucks cheaper. In fact, I've been using bucks cheaper. In fact, I've been using bucks cheaper. In fact, I've been using mine now for a while, for more than a mine now for a while, for more than a mine now for a while, for more than a year, as a media server. And as a side year, as a media server. And as a side year, as a media server. And as a side note, this is what it looks like if you note, this is what it looks like if you note, this is what it looks like if you just have it on a rack for over a year. just have it on a rack for over a year. just have it on a rack for over a year. It's got a little bit of dust, but it's It's got a little bit of dust, but it's It's got a little bit of dust, but it's not [music] too bad. Really solid not [music] too bad. Really solid not [music] too bad. Really solid machine. The other option is the M 4 Pro machine. The other option is the M 4 Pro machine. The other option is the M 4 Pro Mac mini, which I also have set up here.

Mac mini, which I also have set up here. Mac mini, which I also have set up here. I'm including that because the M4, the I'm including that because the M4, the I'm including that because the M4, the base M4, caps out at 24 gigs, which is base M4, caps out at 24 gigs, which is base M4, caps out at 24 gigs, which is really really really not that great for a lot of the not that great for a lot of the not that great for a lot of the workflows I'm going to show you. And workflows I'm going to show you. And workflows I'm going to show you. And also, since we're comparing it to also, since we're comparing it to also, since we're comparing it to [music] the Ser 10, which is a 32 and 64 [music] the Ser 10, which is a 32 and 64 [music] the Ser 10, which is a 32 and 64 gig configurations, then we should go to gig configurations, then we should go to gig configurations, then we should go to the M4 Pro. For the M4 Pro though, Apple the M4 Pro. For the M4 Pro though, Apple the M4 Pro. For the M4 Pro though, Apple charges some serious money for that. charges some serious money for that. charges some serious money for that. This is the basic M4 Pro, not even the This is the basic M4 Pro, not even the This is the basic M4 Pro, not even the 20-core GPU, and it's 48 [music] gigs, 20-core GPU, and it's 48 [music] gigs, 20-core GPU, and it's 48 [music] gigs, not the 64-gig version. Did they even not the 64-gig version. Did they even not the 64-gig version. Did they even have the 64? They got rid of that? Let's have the 64? They got rid of that? Let's have the 64? They got rid of that? Let's go to 1 terabyte, 10 gigs ethernet, go to 1 terabyte, 10 gigs ethernet, go to 1 terabyte, 10 gigs ethernet, suddenly you are at $2,000, and you're suddenly you are at $2,000, and you're suddenly you are at $2,000, and you're only at 48 GB of RAM. So, the Beelink only at 48 GB of RAM. So, the Beelink only at 48 GB of RAM. So, the Beelink site still makes sense. You're only site still makes sense. You're only site still makes sense. You're only going up by couple hundred bucks, and going up by couple hundred bucks, and going up by couple hundred bucks, and you're way under that. This does come you're way under that. This does come you're way under that. This does come with 1 TB of storage. So, the question with 1 TB of storage. So, the question with 1 TB of storage. So, the question isn't is Ser 10 a good upgrade? It's isn't is Ser 10 a good upgrade? It's isn't is Ser 10 a good upgrade? It's should I get a Ser 10, Ser 9, or save up should I get a Ser 10, Ser 9, or save up should I get a Ser 10, Ser 9, or save up for the M4 Pro or the upcoming M5 Pro, for the M4 Pro or the upcoming M5 Pro, for the M4 Pro or the upcoming M5 Pro, whenever that comes. But, this is a whenever that comes. But, this is a whenever that comes. But, this is a performance video, so we're going to get performance video, so we're going to get performance video, so we're going to get into that. So, these days I'm always into that. So, these days I'm always into that. So, these days I'm always flipping between models. GPT for flipping between models. GPT for flipping between models. GPT for research, Claude for coding, Nano Banana research, Claude for coding, Nano Banana research, Claude for coding, Nano Banana for image generation, VEO Kling and for image generation, VEO Kling and for image generation, VEO Kling and Runway for video, six tabs, six bills, Runway for video, six tabs, six bills, Runway for video, six tabs, six bills, and counting. Enter Chat LLM Teams. One and counting. Enter Chat LLM Teams. One and counting. Enter Chat LLM Teams. One dashboard houses every top LLM and route dashboard houses every top LLM and route dashboard houses every top LLM and route LLM picks the right [music] one. GPT LLM picks the right [music] one. GPT LLM picks the right [music] one. GPT Mini for ultra-fast answers, Claude Mini for ultra-fast answers, Claude Mini for ultra-fast answers, Claude Sonnet for coding, Gemini Pro for Sonnet for coding, Gemini Pro for Sonnet for coding, Gemini Pro for massive context. They've recently added massive context. They've recently added massive context. They've recently added Gemini 3 and GPT 5.1 the moment they Gemini 3 and GPT 5.1 the moment they Gemini 3 and GPT 5.1 the moment they dropped. Create professional dropped. Create professional dropped. Create professional presentations with graphs, charts, and presentations with graphs, charts, and presentations with graphs, charts, and deep research detailed content. Need

deep research detailed content. Need deep research detailed content. Need human-sounding copy? Humanize rewrites human-sounding copy? Humanize rewrites human-sounding copy? Humanize rewrites text to defeat AI detectors. [music] text to defeat AI detectors. [music] text to defeat AI detectors. [music] Need visuals? Pick frontier or Need visuals? Pick frontier or Need visuals? Pick frontier or open-source models. Nano Banana, open-source models. Nano Banana, open-source models. Nano Banana, Midjourney, and Flux for images, Midjourney, and Flux for images, Midjourney, and Flux for images, Magnific for upscaling, plus VEO Wan and Magnific for upscaling, plus VEO Wan and Magnific for upscaling, plus VEO Wan and Sora for video. All built in. You also Sora for video. All built in. You also Sora for video. All built in. You also get Abacus AI deep agent to pretty much get Abacus AI deep agent to pretty much get Abacus AI deep agent to pretty much do anything. Build full-stack apps, do anything. Build full-stack apps, do anything. Build full-stack apps, websites, reports with just text prompts websites, reports with just text prompts websites, reports with just text prompts and deploy them on the spot. They have and deploy them on the spot. They have and deploy them on the spot. They have Abacus AI desktop, which is the brand Abacus AI desktop, which is the brand Abacus AI desktop, which is the brand new coding editor and assistant that new coding editor and assistant that new coding editor and assistant that lets you vibe code and build lets you vibe code and build lets you vibe code and build production-ready apps. And the kicker? production-ready apps. And the kicker? production-ready apps. And the kicker? It's just $10 a month, less than one It's just $10 a month, less than one It's just $10 a month, less than one premium model. Head over to premium model. Head over to premium model. Head over to chat.abacus.ai chat.abacus.ai chat.abacus.ai or click the link below to level up with or click the link below to level up with or click the link below to level up with Chat LLM Teams. By the way, if you don't Chat LLM Teams. By the way, if you don't Chat LLM Teams. By the way, if you don't know what a Gorgon is, this is a Gorgon. know what a Gorgon is, this is a Gorgon. know what a Gorgon is, this is a Gorgon. I don't know why AMD has to go with I don't know why AMD has to go with I don't know why AMD has to go with scary things for their [music] chip scary things for their [music] chip scary things for their [music] chip names. This one will turn you into rock names. This one will turn you into rock names. This one will turn you into rock if you make eye contact, so don't do it. if you make eye contact, so don't do it. if you make eye contact, so don't do it. Some quick specs on this chip, we've got Some quick specs on this chip, we've got Some quick specs on this chip, we've got 12 cores and 24 threads, Zen 5 12 cores and 24 threads, Zen 5 12 cores and 24 threads, Zen 5 architecture plus Zen 5c, eight of architecture plus Zen 5c, eight of architecture plus Zen 5c, eight of those, and it boosts up to 5.2 GHz.

those, and it boosts up to 5.2 GHz. those, and it boosts up to 5.2 GHz. Check it out. Shared memory, half of the Check it out. Shared memory, half of the Check it out. Shared memory, half of the memory can be shared, 29 GB in my case. memory can be shared, 29 GB in my case. memory can be shared, 29 GB in my case. So, AMD's marketing material obviously So, AMD's marketing material obviously So, AMD's marketing material obviously will include the MPU. The use cases are will include the MPU. The use cases are will include the MPU. The use cases are limited, but I'll show you how we can limited, but I'll show you how we can limited, but I'll show you how we can use that later on in the video. And the use that later on in the video. And the use that later on in the video. And the iGPU is here too. This happens to be the iGPU is here too. This happens to be the iGPU is here too. This happens to be the Radeon 890M. If that sounds familiar, Radeon 890M. If that sounds familiar, Radeon 890M. If that sounds familiar, you're right. you're right. you're right. >> [laughter] >> [laughter] >> [laughter] >> It's the same one that was in the SER >> It's the same one that was in the SER >> It's the same one that was in the SER [music] 9. More on that later. The [music] 9. More on that later. The [music] 9. More on that later. The memory here, I have 64 GB on my machine, memory here, I have 64 GB on my machine, memory here, I have 64 GB on my machine, but it does come with 32. And this is but it does come with 32. And this is but it does come with 32. And this is DDR5 5600, user configurable and user DDR5 5600, user configurable and user DDR5 5600, user configurable and user upgradeable, which is different than the upgradeable, which is different than the upgradeable, which is different than the SER 9 was. You can also upgrade the SER 9 was. You can also upgrade the SER 9 was. You can also upgrade the storage up to 8 TB if you can find it storage up to 8 TB if you can find it storage up to 8 TB if you can find it and afford it. A few other life and afford it. A few other life and afford it. A few other life improvements like USB4, HDMI 2.1, improvements like USB4, HDMI 2.1, improvements like USB4, HDMI 2.1, DisplayPort 1.4, and 4K at 240 Hz triple DisplayPort 1.4, and 4K at 240 Hz triple DisplayPort 1.4, and 4K at 240 Hz triple monitor support. And monitor support. And monitor support. And 10 gigabit network ethernet. So, that's 10 gigabit network ethernet. So, that's 10 gigabit network ethernet. So, that's pretty cool considering the SER 9 only pretty cool considering the SER 9 only pretty cool considering the SER 9 only had 2.5. Now, if you're so inclined, had 2.5. Now, if you're so inclined, had 2.5. Now, if you're so inclined, there's also nice-looking orange one. Is there's also nice-looking orange one. Is there's also nice-looking orange one. Is it orange or red? I don't know, but it's it orange or red? I don't know, but it's it orange or red? I don't know, but it's got OpenClaw pre-installed on it, and got OpenClaw pre-installed on it, and got OpenClaw pre-installed on it, and they're charging 100 bucks more for it.

they're charging 100 bucks more for it. they're charging 100 bucks more for it. But, look at this option right here. 96 But, look at this option right here. 96 But, look at this option right here. 96 GB with 2 TB SSD for [music] $2300. This GB with 2 TB SSD for [music] $2300. This GB with 2 TB SSD for [music] $2300. This 96 GB for that much is actually not bad 96 GB for that much is actually not bad 96 GB for that much is actually not bad right now. If I spec the 2 TB version of right now. If I spec the 2 TB version of right now. If I spec the 2 TB version of a Mac Mini, we're at 2500 bucks, and a Mac Mini, we're at 2500 bucks, and a Mac Mini, we're at 2500 bucks, and we're half the RAM. we're half the RAM. we're half the RAM. But, it's kind of funny that they're But, it's kind of funny that they're But, it's kind of funny that they're charging 100 bucks more for the OpenClaw charging 100 bucks more for the OpenClaw charging 100 bucks more for the OpenClaw version of this. And it's definitely not version of this. And it's definitely not version of this. And it's definitely not as bad as that guy on Craigslist who was as bad as that guy on Craigslist who was as bad as that guy on Craigslist who was charging $1500 for a base Mac Mini with charging $1500 for a base Mac Mini with charging $1500 for a base Mac Mini with OpenClaw pre-installed on it. All right, let's start light. All right, let's start light. Speedometer 3 is a JavaScript benchmark. Speedometer 3 is a JavaScript benchmark. Speedometer 3 is a JavaScript benchmark. It simulates real web application It simulates real web application It simulates real web application interactions, the kind of thing you feel interactions, the kind of thing you feel interactions, the kind of thing you feel every time you open a sluggish site, every time you open a sluggish site, every time you open a sluggish site, single-threaded, and everyone's got a single-threaded, and everyone's got a single-threaded, and everyone's got a frame of reference for this one. The SER frame of reference for this one. The SER frame of reference for this one. The SER 9 hit 31.5 9 hit 31.5 9 hit 31.5 in the seven-way comparison video that I in the seven-way comparison video that I in the seven-way comparison video that I did not too long ago. That one had a did not too long ago. That one had a did not too long ago. That one had a bunch of other mini PCs included. Last bunch of other mini PCs included. Last bunch of other mini PCs included. Last year, the M4 Pro Mac Mini got 45.2 in year, the M4 Pro Mac Mini got 45.2 in year, the M4 Pro Mac Mini got 45.2 in that. The base M4 got beat the M4 Pro, that. The base M4 got beat the M4 Pro, that. The base M4 got beat the M4 Pro, got 47.8, but they're close. Apple is got 47.8, but they're close. Apple is got 47.8, but they're close. Apple is kind of hard to beat in single core kind of hard to beat in single core kind of hard to beat in single core scores right now. They've got that scores right now. They've got that scores right now. They've got that market cornered. [music] market cornered. [music] market cornered. [music] And we'll see that pattern again. Our And we'll see that pattern again. Our And we'll see that pattern again. Our speedometer is a JavaScript runtime, Web speedometer is a JavaScript runtime, Web speedometer is a JavaScript runtime, Web Tooling Benchmark is another one that I Tooling Benchmark is another one that I Tooling Benchmark is another one that I use by V8. Yeah, those are the same V8 use by V8. Yeah, those are the same V8 use by V8. Yeah, those are the same V8 people that actually write the engine.

people that actually write the engine. people that actually write the engine. This tested JavaScript tool chain like This tested JavaScript tool chain like This tested JavaScript tool chain like TypeScript, Babel, Terser, and stuff TypeScript, Babel, Terser, and stuff TypeScript, Babel, Terser, and stuff like that. Here the Sir 9 came in at 20 like that. Here the Sir 9 came in at 20 like that. Here the Sir 9 came in at 20 on the TypeScript score and 18.8 on the on the TypeScript score and 18.8 on the on the TypeScript score and 18.8 on the geometric mean. Both Mac Minis are way geometric mean. Both Mac Minis are way geometric mean. Both Mac Minis are way ahead, 35.99 ahead, 35.99 ahead, 35.99 each on TypeScript. By the way, the M4 each on TypeScript. By the way, the M4 each on TypeScript. By the way, the M4 base and the M4 Pro tie on that one. The base and the M4 Pro tie on that one. The base and the M4 Pro tie on that one. The geo mean is where they separate a little geo mean is where they separate a little geo mean is where they separate a little bit. And after running the Sir 10 test, bit. And after running the Sir 10 test, bit. And after running the Sir 10 test, it landed on 34.14. it landed on 34.14. it landed on 34.14. That's what TypeScript and 28.62 for the That's what TypeScript and 28.62 for the That's what TypeScript and 28.62 for the geo mean. Okay, hold on. Now, that's 65% geo mean. Okay, hold on. Now, that's 65% geo mean. Okay, hold on. Now, that's 65% jump on TypeScript over the Sir 9. 65. jump on TypeScript over the Sir 9. 65. jump on TypeScript over the Sir 9. 65. And the geo mean is up over 50% here, And the geo mean is up over 50% here, And the geo mean is up over 50% here, too. This is by far the biggest too. This is by far the biggest too. This is by far the biggest generation-over-generation win I've seen generation-over-generation win I've seen generation-over-generation win I've seen so far for this machine. And look at so far for this machine. And look at so far for this machine. And look at that TypeScript score. It's within 5% of that TypeScript score. It's within 5% of that TypeScript score. It's within 5% of both the Macs. By the way, I'm looking both the Macs. By the way, I'm looking both the Macs. By the way, I'm looking over here because that's where my over here because that's where my over here because that's where my computer with the chart is. So, the Sir computer with the chart is. So, the Sir computer with the chart is. So, the Sir 9 trailed the Macs by 75% on [music] 9 trailed the Macs by 75% on [music] 9 trailed the Macs by 75% on [music] this test last year. This year the gap this test last year. This year the gap this test last year. This year the gap is essentially gone. The geometric mean is essentially gone. The geometric mean is essentially gone. The geometric mean is still behind by 11 to 13%. Not a is still behind by 11 to 13%. Not a is still behind by 11 to 13%. Not a clean tie there, but this mini PC has clean tie there, but this mini PC has clean tie there, but this mini PC has finally caught up to Apple's Mac Mini on finally caught up to Apple's Mac Mini on finally caught up to Apple's Mac Mini on a real V8 tool chain benchmark. All a real V8 tool chain benchmark. All a real V8 tool chain benchmark. All right, the obligatory Geekbench right, the obligatory Geekbench right, the obligatory Geekbench benchmark. I don't usually like to dwell benchmark. I don't usually like to dwell benchmark. I don't usually like to dwell on this, but in this case, it's on this, but in this case, it's on this, but in this case, it's important. Everybody uses Geekbench to important. Everybody uses Geekbench to important. Everybody uses Geekbench to measure this stuff, so might as well measure this stuff, so might as well measure this stuff, so might as well throw it in. And last generation Sir 9 throw it in. And last generation Sir 9 throw it in. And last generation Sir 9 came in at 2889 for a single core and came in at 2889 for a single core and came in at 2889 for a single core and 14890 for multi. The M2 Pro Mac Mini,

14890 for multi. The M2 Pro Mac Mini, 14890 for multi. The M2 Pro Mac Mini, that's the one from [music] the It's the that's the one from [music] the It's the that's the one from [music] the It's the bigger one, okay? From a couple years bigger one, okay? From a couple years bigger one, okay? From a couple years ago. That one was 2,700 and 13,000. Now, ago. That one was 2,700 and 13,000. Now, ago. That one was 2,700 and 13,000. Now, the M4 Pro, that's the number to beat. the M4 Pro, that's the number to beat. the M4 Pro, that's the number to beat. It's got pretty decent score, 3945 for It's got pretty decent score, 3945 for It's got pretty decent score, 3945 for single and 15,321 single and 15,321 single and 15,321 for multi. You can see the Sierra 8 was for multi. You can see the Sierra 8 was for multi. You can see the Sierra 8 was quite a bit behind on that one. And the quite a bit behind on that one. And the quite a bit behind on that one. And the Sierra 10, that lands at 2993 single and Sierra 10, that lands at 2993 single and Sierra 10, that lands at 2993 single and 15,216 15,216 15,216 multi. Pretty big leap over last multi. Pretty big leap over last multi. Pretty big leap over last generation. And for the single-core, generation. And for the single-core, generation. And for the single-core, it's not a huge difference. However, the it's not a huge difference. However, the it's not a huge difference. However, the multi-core, compared to the M4 Pro Mac multi-core, compared to the M4 Pro Mac multi-core, compared to the M4 Pro Mac mini, that's a difference of less than mini, that's a difference of less than mini, that's a difference of less than 1%. It basically ties Apple's flagship 1%. It basically ties Apple's flagship 1%. It basically ties Apple's flagship Mac mini on multi-thread Geekbench. Mac mini on multi-thread Geekbench. Mac mini on multi-thread Geekbench. Maybe this will translate to when I do Maybe this will translate to when I do Maybe this will translate to when I do compilation tests in a bit. All right, compilation tests in a bit. All right, compilation tests in a bit. All right, it's time for the meat. Can't believe I it's time for the meat. Can't believe I it's time for the meat. Can't believe I just said that. That's just Some just said that. That's just Some just said that. That's just Some vegetarians might object, but I don't vegetarians might object, but I don't vegetarians might object, but I don't care. I I eat meat. I like it. It's care. I I eat meat. I like it. It's care. I I eat meat. I like it. It's good. Lamb is my favorite. Beef, but a good. Lamb is my favorite. Beef, but a good. Lamb is my favorite. Beef, but a good steak is really good. All right.

good steak is really good. All right. good steak is really good. All right. Python brought. Python brought. Python brought. >> [laughter] >> [laughter] >> [laughter] >> This is a test that I run to show how >> This is a test that I run to show how >> This is a test that I run to show how pushing all the cores to the max with an pushing all the cores to the max with an pushing all the cores to the max with an interpreted language like this, interpreted language like this, interpreted language like this, interpreted test like this. This is a interpreted test like this. This is a interpreted test like this. This is a Python algorithm, and it's the channel's Python algorithm, and it's the channel's Python algorithm, and it's the channel's signature move. It pegs every single signature move. It pegs every single signature move. It pegs every single core to 100%. Oh, yeah. There it goes. core to 100%. Oh, yeah. There it goes. core to 100%. Oh, yeah. There it goes. You can see that working quite nicely You can see that working quite nicely You can see that working quite nicely there. Now, the Sierra 9 already beat there. Now, the Sierra 9 already beat there. Now, the Sierra 9 already beat the M4 base on this. 28.64 the M4 base on this. 28.64 the M4 base on this. 28.64 seconds versus [music] 31.41. seconds versus [music] 31.41. seconds versus [music] 31.41. Shorter number is better in this case. Shorter number is better in this case. Shorter number is better in this case. Be like ahead of Apple silicon here, Be like ahead of Apple silicon here, Be like ahead of Apple silicon here, folks. And that's the previous folks. And that's the previous folks. And that's the previous generation. As an aside, I also ran this generation. As an aside, I also ran this generation. As an aside, I also ran this on the X Elite mini PC that came out on the X Elite mini PC that came out on the X Elite mini PC that came out Some of you might remember that. Well, Some of you might remember that. Well, Some of you might remember that. Well, the Sierra 9 tied with that machine, the Sierra 9 tied with that machine, the Sierra 9 tied with that machine, which is the first-generation X Elite which is the first-generation X Elite which is the first-generation X Elite that was the most powerful one. I want that was the most powerful one. I want that was the most powerful one. I want to see the X2 Elite in mini PCs. That's to see the X2 Elite in mini PCs. That's to see the X2 Elite in mini PCs. That's going to be a killer one. Anyway, M4 Pro going to be a killer one. Anyway, M4 Pro going to be a killer one. Anyway, M4 Pro was still the one to beat, 23.24 was still the one to beat, 23.24 was still the one to beat, 23.24 seconds on that one for this test, seconds on that one for this test, seconds on that one for this test, considerably faster. And the sear eight considerably faster. And the sear eight considerably faster. And the sear eight was quite a bit slower, 42.73 seconds on was quite a bit slower, 42.73 seconds on was quite a bit slower, 42.73 seconds on that one. So, we saw a 33% jump from that one. So, we saw a 33% jump from that one. So, we saw a 33% jump from sear eight to sear nine. That's huge.

sear eight to sear nine. That's huge. sear eight to sear nine. That's huge. How does the sear 10 do? Hmm, okay. Sear How does the sear 10 do? Hmm, okay. Sear How does the sear 10 do? Hmm, okay. Sear nine was 28.64. Sear 10 is 28.9. nine was 28.64. Sear 10 is 28.9. nine was 28.64. Sear 10 is 28.9. A nice number here. This is the fastest A nice number here. This is the fastest A nice number here. This is the fastest that it got. That's literally a quarter that it got. That's literally a quarter that it got. That's literally a quarter of a second difference. So, basically, of a second difference. So, basically, of a second difference. So, basically, kind of the same. Still beats the M4 kind of the same. Still beats the M4 kind of the same. Still beats the M4 base. So, that part of the story holds. base. So, that part of the story holds. base. So, that part of the story holds. But, generation to generation, But, generation to generation, But, generation to generation, nothing here. So, this was an nothing here. So, this was an nothing here. So, this was an interpreted multi-core test. But, before interpreted multi-core test. But, before interpreted multi-core test. But, before we get into compiled multi-core test, I we get into compiled multi-core test, I we get into compiled multi-core test, I want to back up here and show you this want to back up here and show you this want to back up here and show you this project, which is something I ran before project, which is something I ran before project, which is something I ran before on those machines as well. This is a on those machines as well. This is a on those machines as well. This is a large mono repo by Victor Savkin. This large mono repo by Victor Savkin. This large mono repo by Victor Savkin. This is NX mono repo. It includes 26,000 is NX mono repo. It includes 26,000 is NX mono repo. It includes 26,000 NX JavaScript components. There's a lot NX JavaScript components. There's a lot NX JavaScript components. There's a lot of components, but they are very small. of components, but they are very small. of components, but they are very small. So, it corresponds to a medium-size So, it corresponds to a medium-size So, it corresponds to a medium-size enterprise repository. Basically, enterprise repository. Basically, enterprise repository. Basically, JavaScript at scale. There's five JavaScript at scale. There's five JavaScript at scale. There's five Next.js apps here, 20 libraries, 250 Next.js apps here, 20 libraries, 250 Next.js apps here, 20 libraries, 250 components each. And NX provides a components each. And NX provides a components each. And NX provides a cache, but I ran this build without a cache, but I ran this build without a cache, but I ran this build without a cache, just to see what it's like. Now, cache, just to see what it's like. Now, cache, just to see what it's like. Now, before I get your hopes up, this test before I get your hopes up, this test before I get your hopes up, this test didn't work out so well because uh didn't work out so well because uh didn't work out so well because uh there's repository drifts, and I just there's repository drifts, and I just there's repository drifts, and I just don't believe the results cuz it can't don't believe the results cuz it can't don't believe the results cuz it can't be like this. Last year, sear nine, we be like this. Last year, sear nine, we be like this. Last year, sear nine, we got 14.1 seconds. The M4 Pro was 7.76.

got 14.1 seconds. The M4 Pro was 7.76. got 14.1 seconds. The M4 Pro was 7.76. The M4 base was 7.36. Yeah, I know. M4 The M4 base was 7.36. Yeah, I know. M4 The M4 base was 7.36. Yeah, I know. M4 base won again. Third test in a row. So, base won again. Third test in a row. So, base won again. Third test in a row. So, what I like about this test is it's a what I like about this test is it's a what I like about this test is it's a TypeScript tool chain. It runs through TypeScript tool chain. It runs through TypeScript tool chain. It runs through node. Lots of process spawning. Lots of node. Lots of process spawning. Lots of node. Lots of process spawning. Lots of file system hits. Apple silicon's been file system hits. Apple silicon's been file system hits. Apple silicon's been good at this for years. So, this is good at this for years. So, this is good at this for years. So, this is where I thought the sear 10 has to where I thought the sear 10 has to where I thought the sear 10 has to climb, right? Maybe not dominate, but climb, right? Maybe not dominate, but climb, right? Maybe not dominate, but climb. And sear 10 landed at 2.91 climb. And sear 10 landed at 2.91 climb. And sear 10 landed at 2.91 seconds. And I thought, come on. Five seconds. And I thought, come on. Five seconds. And I thought, come on. Five times faster than the sear nine? I don't times faster than the sear nine? I don't times faster than the sear nine? I don't think so. So, because of the way this think so. So, because of the way this think so. So, because of the way this test is set up, the NX tool itself test is set up, the NX tool itself test is set up, the NX tool itself probably got faster and optimized and probably got faster and optimized and probably got faster and optimized and [music] [music] [music] unfortunately, I'm going to have to say unfortunately, I'm going to have to say unfortunately, I'm going to have to say goodbye to this tool goodbye to this tool goodbye to this tool because because because um um um yeah, when you're doing benchmarks, you yeah, when you're doing benchmarks, you yeah, when you're doing benchmarks, you can compare things side to side, side by can compare things side to side, side by can compare things side to side, side by side, over time, but when you're doing side, over time, but when you're doing side, over time, but when you're doing real-world projects, real-world projects, real-world projects, then things tend to drift a little bit. then things tend to drift a little bit. then things tend to drift a little bit. So, it's not the same workload anymore. So, it's not the same workload anymore. So, it's not the same workload anymore. We can't really compare to the previous We can't really compare to the previous We can't really compare to the previous builds. builds. builds. However, it builds this thing in under 3 However, it builds this thing in under 3 However, it builds this thing in under 3 seconds, which is pretty incredible. And seconds, which is pretty incredible. And seconds, which is pretty incredible. And that's good for both the chip, the that's good for both the chip, the that's good for both the chip, the machine, and the mono repo, and the machine, and the mono repo, and the machine, and the mono repo, and the tool. So, good for them. We're going to tool. So, good for them. We're going to tool. So, good for them. We're going to skip this test next time, though. I want skip this test next time, though. I want skip this test next time, though. I want to move on to a benchmark that I to move on to a benchmark that I to move on to a benchmark that I designed. This is a .NET build, and this designed. This is a .NET build, and this designed. This is a .NET build, and this build basically generates 100,000 build basically generates 100,000 build basically generates 100,000 namespaces and classes, each one with namespaces and classes, each one with namespaces and classes, each one with recursive calculations and nested loops, recursive calculations and nested loops, recursive calculations and nested loops, so the compiler can't just throw them so the compiler can't just throw them so the compiler can't just throw them away and skip the work. It has to do it.

away and skip the work. It has to do it. away and skip the work. It has to do it. Same workload every time, year after Same workload every time, year after Same workload every time, year after year. This is the test where the Cier 9 year. This is the test where the Cier 9 year. This is the test where the Cier 9 beat them for base model, 91 seconds. beat them for base model, 91 seconds. beat them for base model, 91 seconds. That beat them for, which got 106.7 That beat them for, which got 106.7 That beat them for, which got 106.7 seconds. The Cier 8 was 109 seconds, and seconds. The Cier 8 was 109 seconds, and seconds. The Cier 8 was 109 seconds, and the only machine that was meaningfully the only machine that was meaningfully the only machine that was meaningfully faster was the M4 Pro at 66.4 seconds. faster was the M4 Pro at 66.4 seconds. faster was the M4 Pro at 66.4 seconds. So, this is the test for Gorgon Point. So, this is the test for Gorgon Point. So, this is the test for Gorgon Point. If Zen 5's IPC story is real, and the If Zen 5's IPC story is real, and the If Zen 5's IPC story is real, and the math behind the new generation actually math behind the new generation actually math behind the new generation actually shows up in compiled code, he's going to shows up in compiled code, he's going to shows up in compiled code, he's going to land the hardest here. And this is also land the hardest here. And this is also land the hardest here. And this is also the loudest test. I'm hearing the fans the loudest test. I'm hearing the fans the loudest test. I'm hearing the fans here. Here it is being built. Pretty here. Here it is being built. Pretty here. Here it is being built. Pretty much all the CPU cores are involved in much all the CPU cores are involved in much all the CPU cores are involved in this one, although they don't go as hard this one, although they don't go as hard this one, although they don't go as hard as that matter broad test, but still, as that matter broad test, but still, as that matter broad test, but still, it's more realistic. And we got 90.9. it's more realistic. And we got 90.9. it's more realistic. And we got 90.9. This is a difference of This is a difference of This is a difference of 1/10 of a second. 1/10 of a second. 1/10 of a second. This was supposed to show the new chip's This was supposed to show the new chip's This was supposed to show the new chip's strengths. Basically, we got the same strengths. Basically, we got the same strengths. Basically, we got the same result as last time. But, the Cier 10 result as last time. But, the Cier 10 result as last time. But, the Cier 10 still beats the base M4. So, Beelink still beats the base M4. So, Beelink still beats the base M4. So, Beelink continues to be a better choice than the continues to be a better choice than the continues to be a better choice than the base Mac mini for a compiler heavy .NET base Mac mini for a compiler heavy .NET base Mac mini for a compiler heavy .NET work. But, on the M4 Pro, Apple silicon work. But, on the M4 Pro, Apple silicon work. But, on the M4 Pro, Apple silicon leads leads leads still. Now again, the previous test was still. Now again, the previous test was still. Now again, the previous test was synthetic, so I wanted to go and to synthetic, so I wanted to go and to synthetic, so I wanted to go and to build a real thing, Umbraco. That's a build a real thing, Umbraco. That's a build a real thing, Umbraco. That's a open source CMS, it's been around for open source CMS, it's been around for open source CMS, it's been around for ages. It's a mature .NET project, open ages. It's a mature .NET project, open ages. It's a mature .NET project, open source. But again, this is a real source. But again, this is a real source. But again, this is a real project on GitHub. So again, there's project on GitHub. So again, there's project on GitHub. So again, there's going to be variation over time. It's going to be variation over time. It's going to be variation over time. It's constantly evolving, they're adding constantly evolving, they're adding constantly evolving, they're adding features, they're refactoring, pulling features, they're refactoring, pulling features, they're refactoring, pulling in new dependencies. So the number I

in new dependencies. So the number I in new dependencies. So the number I might get might not be fully repeatable might get might not be fully repeatable might get might not be fully repeatable year after year, but if you have a year after year, but if you have a year after year, but if you have a machine and you want to build it right machine and you want to build it right machine and you want to build it right now compared to my result that I'm going now compared to my result that I'm going now compared to my result that I'm going to get, to get, to get, then at least it's going to be valuable then at least it's going to be valuable then at least it's going to be valuable in that way. If you want to compile this in that way. If you want to compile this in that way. If you want to compile this yourself, you can find it on GitHub yourself, you can find it on GitHub yourself, you can find it on GitHub right over here. They're constantly right over here. They're constantly right over here. They're constantly updating it. 11 hours ago was the last updating it. 11 hours ago was the last updating it. 11 hours ago was the last update. Can't believe this thing has update. Can't believe this thing has update. Can't believe this thing has been around so long. Sir 9, 149 seconds been around so long. Sir 9, 149 seconds been around so long. Sir 9, 149 seconds for the compilation. Sir 8, 161. So gen for the compilation. Sir 8, 161. So gen for the compilation. Sir 8, 161. So gen over gen on the Umbraco was only 7%. Why over gen on the Umbraco was only 7%. Why over gen on the Umbraco was only 7%. Why is that? Well, because real builds are is that? Well, because real builds are is that? Well, because real builds are IO heavy. There's disk, there's NuGet, IO heavy. There's disk, there's NuGet, IO heavy. There's disk, there's NuGet, the file system. And again, Mac the file system. And again, Mac the file system. And again, Mac dominates this one. M4 Pro, 84 seconds. dominates this one. M4 Pro, 84 seconds. dominates this one. M4 Pro, 84 seconds. M4 base model even destroys the B link, M4 base model even destroys the B link, M4 base model even destroys the B link, 93.9 93.9 93.9 seconds there. All right, got it done. seconds there. All right, got it done. seconds there. All right, got it done. Sir 10 lands at 161 Sir 10 lands at 161 Sir 10 lands at 161 seconds. Now hold on, that's slightly seconds. Now hold on, that's slightly seconds. Now hold on, that's slightly slower slower slower than the Sir 9. What's up? And it just than the Sir 9. What's up? And it just than the Sir 9. What's up? And it just happened to match the Sir 8 from a happened to match the Sir 8 from a happened to match the Sir 8 from a couple years ago. Both 161 seconds. So couple years ago. Both 161 seconds. So couple years ago. Both 161 seconds. So again, on this one, a real .NET again, on this one, a real .NET again, on this one, a real .NET production app, full release pack, every production app, full release pack, every production app, full release pack, every project, the Sir 9 happened to be the project, the Sir 9 happened to be the project, the Sir 9 happened to be the best of the three here at 149. By the best of the three here at 149. By the best of the three here at 149. By the way, I want to flag something real quick way, I want to flag something real quick way, I want to flag something real quick here. The first time I ran this, I got here. The first time I ran this, I got here. The first time I ran this, I got 217 seconds on the Sir 10. And that was 217 seconds on the Sir 10. And that was 217 seconds on the Sir 10. And that was with Windows Defender running. So with Windows Defender running. So with Windows Defender running. So um I had to turn that off. So if you're um I had to turn that off. So if you're um I had to turn that off. So if you're doing this kind of compilation, you got doing this kind of compilation, you got doing this kind of compilation, you got Windows Defender running, you might want Windows Defender running, you might want Windows Defender running, you might want to exclude the your code directory from to exclude the your code directory from to exclude the your code directory from that.

that. that. >> [music] >> [music] >> [music] >> And >> And >> And the way I have my my systems set up, the way I have my my systems set up, the way I have my my systems set up, first of all, I automated the build, the first of all, I automated the build, the first of all, I automated the build, the deployment of this package, and I do deployment of this package, and I do deployment of this package, and I do have videos on how to set up a have videos on how to set up a have videos on how to set up a development environment on a Windows development environment on a Windows development environment on a Windows machine, and on Macs. So, check that machine, and on Macs. So, check that machine, and on Macs. So, check that out. It's on my channel. And I recently out. It's on my channel. And I recently out. It's on my channel. And I recently created a script to automate that whole created a script to automate that whole created a script to automate that whole process, too. I'd recommend going process, too. I'd recommend going process, too. I'd recommend going through the whole thing just to know through the whole thing just to know through the whole thing just to know what's getting installed, and then you what's getting installed, and then you what's getting installed, and then you can use the script on your subsequent can use the script on your subsequent can use the script on your subsequent builds if you want. It's up on my builds if you want. It's up on my builds if you want. It's up on my GitHub. I'll link to it down below, but GitHub. I'll link to it down below, but GitHub. I'll link to it down below, but I made a special video for members of I made a special video for members of I made a special video for members of the channel. Thank you to the members, the channel. Thank you to the members, the channel. Thank you to the members, by the way, for supporting the channel. by the way, for supporting the channel. by the way, for supporting the channel. I made a video on how to run it. It's I made a video on how to run it. It's I made a video on how to run it. It's all in the instructions on the GitHub, all in the instructions on the GitHub, all in the instructions on the GitHub, though. If you want to just go check it though. If you want to just go check it though. If you want to just go check it out, I'll link to it. All right, let's out, I'll link to it. All right, let's out, I'll link to it. All right, let's switch topics to another popular thing switch topics to another popular thing switch topics to another popular thing that's happening these days in 2026. Oh, that's happening these days in 2026. Oh, that's happening these days in 2026. Oh, I don't know. What could that be? I don't know. What could that be? I don't know. What could that be? AI. AI. AI. AI. Nice. AI. Nice. AI. Nice. >> [laughter] >> [laughter] >> [laughter] >> This thing has a GPU, a CPU, and an NPU. >> This thing has a GPU, a CPU, and an NPU. >> This thing has a GPU, a CPU, and an NPU. So, this little box has three chips on So, this little box has three chips on So, this little box has three chips on it, and they all can do AI. There's a it, and they all can do AI. There's a it, and they all can do AI. There's a 12-core CPU, Radeon 890M iGPU, and 12-core CPU, Radeon 890M iGPU, and 12-core CPU, Radeon 890M iGPU, and there's a 55 tops XDNA 2 NPU. AMD says there's a 55 tops XDNA 2 NPU. AMD says there's a 55 tops XDNA 2 NPU. AMD says all three together do 86 tops of AI all three together do 86 tops of AI all three together do 86 tops of AI work. Tops.

work. Tops. work. Tops. Yeah, we all know the marketing stuff. Yeah, we all know the marketing stuff. Yeah, we all know the marketing stuff. Now, out of the box on Windows, the most Now, out of the box on Windows, the most Now, out of the box on Windows, the most popular tool people use to run local popular tool people use to run local popular tool people use to run local LLMs uses one of those, but it's the LLMs uses one of those, but it's the LLMs uses one of those, but it's the wrong one. So, let's do it. I'm going to wrong one. So, let's do it. I'm going to wrong one. So, let's do it. I'm going to do Ollama run Qwen [music] 2.5 7B, and do Ollama run Qwen [music] 2.5 7B, and do Ollama run Qwen [music] 2.5 7B, and we'll do verbose, and I'll do my we'll do verbose, and I'll do my we'll do verbose, and I'll do my architecture prompt, which produces architecture prompt, which produces architecture prompt, which produces [music] a pretty long output. [music] a pretty long output. [music] a pretty long output. There it is. Results are streaming. There it is. Results are streaming. There it is. Results are streaming. Looks pretty good, but let's check task Looks pretty good, but let's check task Looks pretty good, but let's check task manager. The CPU looks pegged, but the manager. The CPU looks pegged, but the manager. The CPU looks pegged, but the GPU is not being used at all. NPU, zero. GPU is not being used at all. NPU, zero. GPU is not being used at all. NPU, zero. It doesn't even show up here. Three It doesn't even show up here. Three It doesn't even show up here. Three accelerators on this box, and [music] accelerators on this box, and [music] accelerators on this box, and [music] one of them, the regular CPU, is doing one of them, the regular CPU, is doing one of them, the regular CPU, is doing all the AI work, and we don't want that. all the AI work, and we don't want that. all the AI work, and we don't want that. That's going to be slow. Well, there is That's going to be slow. Well, there is That's going to be slow. Well, there is is Vulcan flag that you can set on the is Vulcan flag that you can set on the is Vulcan flag that you can set on the environment for Ollama that will allow environment for Ollama that will allow environment for Ollama that will allow you to use the GPU. So, on day one, the you to use the GPU. So, on day one, the you to use the GPU. So, on day one, the AMD AI chip you just bought is going AMD AI chip you just bought is going AMD AI chip you just bought is going [music] to be doing AI on the part that [music] to be doing AI on the part that [music] to be doing AI on the part that isn't the AI part, doing AI on the part isn't the AI part, doing AI on the part isn't the AI part, doing AI on the part that isn't the AI part. that isn't the AI part. that isn't the AI part. We got a result at 14.2 [music] We got a result at 14.2 [music] We got a result at 14.2 [music] tokens per second. So, to enable that on tokens per second. So, to enable that on tokens per second. So, to enable that on Windows, you just go to the Windows, you just go to the Windows, you just go to the environmental variables [music] and set environmental variables [music] and set environmental variables [music] and set the Ollama Vulcan flag to true. If there the Ollama Vulcan flag to true. If there the Ollama Vulcan flag to true. If there isn't one, just add it. And then just isn't one, just add it. And then just isn't one, just add it. And then just make sure you restart Ollama. All right, make sure you restart Ollama. All right, make sure you restart Ollama. All right, here we go.

here we go. here we go. That looks like it's going a little That looks like it's going a little That looks like it's going a little faster, doesn't it? Let's take a look at faster, doesn't it? Let's take a look at faster, doesn't it? Let's take a look at the GPU, and yes, [music] there's the the GPU, and yes, [music] there's the the GPU, and yes, [music] there's the GPU compute. It's happening on the GPU GPU compute. It's happening on the GPU GPU compute. It's happening on the GPU now, which is beautiful, 100% now, which is beautiful, 100% now, which is beautiful, 100% utilization there. So, I ran a couple of utilization there. So, I ran a couple of utilization there. So, I ran a couple of them as a test. Llama 3 0.23B, we went them as a test. Llama 3 0.23B, we went them as a test. Llama 3 0.23B, we went from 27 tokens per second to 37 and a from 27 tokens per second to 37 and a from 27 tokens per second to 37 and a half. Gwent 2.5 1.5B went from 47.5 half. Gwent 2.5 1.5B went from 47.5 half. Gwent 2.5 1.5B went from 47.5 to 68.3 tokens per second. So, there's to 68.3 tokens per second. So, there's to 68.3 tokens per second. So, there's some differences there. So, we've got some differences there. So, we've got some differences there. So, we've got the iGPU thing going, but there's still the iGPU thing going, but there's still the iGPU thing going, but there's still one chip in this box we haven't even one chip in this box we haven't even one chip in this box we haven't even touched. And that's this touched. And that's this touched. And that's this >> [music] >> [music] >> [music] >> NPU. And guess what? The old one had it, >> NPU. And guess what? The old one had it, >> NPU. And guess what? The old one had it, too. AMD's been putting this thing on too. AMD's been putting this thing on too. AMD's been putting this thing on the box for 2 years, and nobody actually the box for 2 years, and nobody actually the box for 2 years, and nobody actually benchmarks it. So, first try, I used benchmarks it. So, first try, I used benchmarks it. So, first try, I used >> [music] >> [music] >> [music] >> Lemonade Server. They got these hybrid >> Lemonade Server. They got these hybrid >> Lemonade Server. They got these hybrid recipes where the NPU does the prefill recipes where the NPU does the prefill recipes where the NPU does the prefill and the iGPU does the decode. Those are and the iGPU does the decode. Those are and the iGPU does the decode. Those are the two [music] stages of inference. It the two [music] stages of inference. It the two [music] stages of inference. It even comes with this little handy app. even comes with this little handy app. even comes with this little handy app. Let's do a high here. It's kind of hard Let's do a high here. It's kind of hard Let's do a high here. It's kind of hard to see down here, but tokens per second to see down here, but tokens per second to see down here, but tokens per second is 14.3 on this one. is 14.3 on this one. is 14.3 on this one. This is a 1 billion parameter model. We This is a 1 billion parameter model. We This is a 1 billion parameter model. We can see that it bumps the NPU just a tad can see that it bumps the NPU just a tad can see that it bumps the NPU just a tad bit, and then it uses the GPU for the bit, and then it uses the GPU for the bit, and then it uses the GPU for the decode. What if I use my longer prompt decode. What if I use my longer prompt decode. What if I use my longer prompt here? Boom. Okay. Let me see a little here? Boom. Okay. Let me see a little here? Boom. Okay. Let me see a little bit more NPU activity. That's the bit more NPU activity. That's the bit more NPU activity. That's the prefill. And the prefill happens pretty prefill. And the prefill happens pretty prefill. And the prefill happens pretty quickly. Most of the work happens on the quickly. Most of the work happens on the quickly. Most of the work happens on the GPU, though, cuz that's the decode GPU, though, cuz that's the decode GPU, though, cuz that's the decode phase. So, the NPU is kind of built for phase. So, the NPU is kind of built for phase. So, the NPU is kind of built for that stage, for the prefill. It's a that stage, for the prefill. It's a that stage, for the prefill. It's a different job, different job, [music] different job, different job, [music] different job, different job, [music] different chip. So, I ran a longer different chip. So, I ran a longer different chip. So, I ran a longer prompt, and it was a 4400 token prompt prompt, and it was a 4400 token prompt prompt, and it was a 4400 token prompt with a short reply. And that's the with a short reply. And that's the with a short reply. And that's the actual prefill workload. On Quen 7B, the

actual prefill workload. On Quen 7B, the actual prefill workload. On Quen 7B, the CPU got 255 tokens per second there. The CPU got 255 tokens per second there. The CPU got 255 tokens per second there. The Vulcan iGPU 240 and the NPU with a Vulcan iGPU 240 and the NPU with a Vulcan iGPU 240 and the NPU with a hybrid mode 631 tokens per second. hybrid mode 631 tokens per second. hybrid mode 631 tokens per second. That's 2 and 1/2 times the iGPU on the That's 2 and 1/2 times the iGPU on the That's 2 and 1/2 times the iGPU on the same chip. So, if you're streaming chat, same chip. So, if you're streaming chat, same chip. So, if you're streaming chat, use the iGPU. If you're doing rag, use the iGPU. If you're doing rag, use the iGPU. If you're doing rag, agents or coding assistance, anything agents or coding assistance, anything agents or coding assistance, anything that dumps a long context, then you want that dumps a long context, then you want that dumps a long context, then you want to use the NPU. And since this is a to use the NPU. And since this is a to use the NPU. And since this is a lemonade server, you can actually lemonade server, you can actually lemonade server, you can actually connect it to a coding agent. It's for connect it to a coding agent. It's for connect it to a coding agent. It's for another video. Perhaps you want to see another video. Perhaps you want to see another video. Perhaps you want to see that, let me know in the comments down that, let me know in the comments down that, let me know in the comments down below if you do. I'm curious myself cuz below if you do. I'm curious myself cuz below if you do. I'm curious myself cuz I haven't actually tested lemonade I haven't actually tested lemonade I haven't actually tested lemonade before on the channel. But now I want to before on the channel. But now I want to before on the channel. But now I want to show you something that this hardware show you something that this hardware show you something that this hardware does that if you just read the specs, does that if you just read the specs, does that if you just read the specs, you'd say, "No way, that's not you'd say, "No way, that's not you'd say, "No way, that's not possible." Quen 2.5 14B, that's the Q4 possible." Quen 2.5 14B, that's the Q4 possible." Quen 2.5 14B, that's the Q4 quant, over 8 GB on disk. Now, the iGPU quant, over 8 GB on disk. Now, the iGPU quant, over 8 GB on disk. Now, the iGPU on this chip, it shows you 4 gigs of on this chip, it shows you 4 gigs of on this chip, it shows you 4 gigs of memory, just 4. And I'm going to load memory, just 4. And I'm going to load memory, just 4. And I'm going to load 8.37 gigs of weights onto a 4 gig iGPU.

8.37 gigs of weights onto a 4 gig iGPU. 8.37 gigs of weights onto a 4 gig iGPU. This shouldn't work. And it's loaded. This shouldn't work. And it's loaded. This shouldn't work. And it's loaded. All 49 layers on the GPU. How? All All 49 layers on the GPU. How? All All 49 layers on the GPU. How? All right, so I'm watching this in real right, so I'm watching this in real right, so I'm watching this in real time. Get counter on the right side, GPU time. Get counter on the right side, GPU time. Get counter on the right side, GPU adapter memory. Before I loaded the adapter memory. Before I loaded the adapter memory. Before I loaded the model, it was 0.69 GB dedicated, 5.26 model, it was 0.69 GB dedicated, 5.26 model, it was 0.69 GB dedicated, 5.26 shared. And after, 3.62 dedicated, 17.36 shared. And after, 3.62 dedicated, 17.36 shared. And after, 3.62 dedicated, 17.36 shared. 20.98 GB of iGPU memory in use shared. 20.98 GB of iGPU memory in use shared. 20.98 GB of iGPU memory in use right now. On a chip, the device manager right now. On a chip, the device manager right now. On a chip, the device manager says there's only 4. So, that's UMA, says there's only 4. So, that's UMA, says there's only 4. So, that's UMA, unified memory architecture, which the unified memory architecture, which the unified memory architecture, which the M4 [music] and the Apple silicon also M4 [music] and the Apple silicon also M4 [music] and the Apple silicon also has, but this has it, too. So, when the has, but this has it, too. So, when the has, but this has it, too. So, when the iGPU asks for more memory than its iGPU asks for more memory than its iGPU asks for more memory than its allocation, Windows just quietly grabs allocation, Windows just quietly grabs allocation, Windows just quietly grabs some from system RAM and says, "Here you some from system RAM and says, "Here you some from system RAM and says, "Here you go." Hands it over and the model has no go." Hands it over and the model has no go." Hands it over and the model has no idea. So, prompt processing was about 50 idea. So, prompt processing was about 50 idea. So, prompt processing was about 50 tokens per second, generation 8.8 tokens tokens per second, generation 8.8 tokens tokens per second, generation 8.8 tokens per second at 14 billion parameter per second at 14 billion parameter per second at 14 billion parameter model. And by the way, on this machine, model. And by the way, on this machine, model. And by the way, on this machine, there's still like 25 more gigabytes of there's still like 25 more gigabytes of there's still like 25 more gigabytes of unified memory headroom left here. So, a unified memory headroom left here. So, a unified memory headroom left here. So, a 22 billion parameter model would fit.

22 billion parameter model would fit. 22 billion parameter model would fit. Probably even a 30 billion parameter Probably even a 30 billion parameter Probably even a 30 billion parameter model would fit, too, with Q4 on the model would fit, too, with Q4 on the model would fit, too, with Q4 on the iGPU. Now, remember that the SER 10 and iGPU. Now, remember that the SER 10 and iGPU. Now, remember that the SER 10 and the SER 9 have the exact same GPU in the SER 9 have the exact same GPU in the SER 9 have the exact same GPU in there. So, if you bought a SER 9 last there. So, if you bought a SER 9 last there. So, if you bought a SER 9 last year for the AI stuff, year for the AI stuff, year for the AI stuff, >> [music] >> [music] >> [music] >> your hardware is still fine. Update your >> your hardware is still fine. Update your >> your hardware is still fine. Update your software. So, quick recommendations software. So, quick recommendations software. So, quick recommendations here. You should buy the SER 10 if you here. You should buy the SER 10 if you here. You should buy the SER 10 if you care about the MPU for rag and long care about the MPU for rag and long care about the MPU for rag and long context AI work. If you want upgradeable context AI work. If you want upgradeable context AI work. If you want upgradeable RAM because SER 9 doesn't have that. Or RAM because SER 9 doesn't have that. Or RAM because SER 9 doesn't have that. Or if you want that 10 gig ethernet port. if you want that 10 gig ethernet port. if you want that 10 gig ethernet port. You should probably skip it if you You should probably skip it if you You should probably skip it if you already have the SER 9 and your main already have the SER 9 and your main already have the SER 9 and your main workload is the iGPU LLM path. Save your workload is the iGPU LLM path. Save your workload is the iGPU LLM path. Save your money until the next generation, which money until the next generation, which money until the next generation, which is Medusa, comes out in 2027. Another is Medusa, comes out in 2027. Another is Medusa, comes out in 2027. Another scary creature that I wouldn't want to scary creature that I wouldn't want to scary creature that I wouldn't want to mess with. Now, besides these, there's mess with. Now, besides these, there's mess with. Now, besides these, there's actually a bunch of other mini PCs that actually a bunch of other mini PCs that actually a bunch of other mini PCs that you can consider you can consider you can consider and I made a video up here and up here. and I made a video up here and up here. and I made a video up here and up here. You can check those out. You [music] You can check those out. You [music] You can check those out. You [music] might be interested in those. Thanks for might be interested in those. Thanks for might be interested in those. Thanks for watching and I'll see you next time.