Framework 13 Pro Shows Why Apple Solders Your Memory

The ultimate developer laptop. Huh? The ultimate developer laptop. Huh? Really? This, of course, is the Really? This, of course, is the Really? This, of course, is the Framework Laptop 13 Pro. I've been a Framework Laptop 13 Pro. I've been a Framework Laptop 13 Pro. I've been a pretty happy developer on my MacBook pretty happy developer on my MacBook pretty happy developer on my MacBook Pros for the last 13 years or so. And Pros for the last 13 years or so. And Pros for the last 13 years or so. And when Framework announced what they're when Framework announced what they're when Framework announced what they're building is essentially what Macs building is essentially what Macs building is essentially what Macs couldn't accomplish, a modular laptop couldn't accomplish, a modular laptop couldn't accomplish, a modular laptop that runs Linux, that's as good as a that runs Linux, that's as good as a that runs Linux, that's as good as a MacBook Pro. The MacBook Pro for Linux. MacBook Pro. The MacBook Pro for Linux. MacBook Pro. The MacBook Pro for Linux. So, as soon as they announced it, I So, as soon as they announced it, I So, as soon as they announced it, I ordered it. I'm kind of glad I did ordered it. I'm kind of glad I did ordered it. I'm kind of glad I did because prices went up. And while I was because prices went up. And while I was because prices went up. And while I was waiting, anticipating, waiting, anticipating, waiting, anticipating, uh, I'm not really going to rhyme in the uh, I'm not really going to rhyme in the uh, I'm not really going to rhyme in the rest of this video, I promise. I wanted rest of this video, I promise. I wanted rest of this video, I promise. I wanted to know if this will actually be able to to know if this will actually be able to to know if this will actually be able to be something I can use as a Linux be something I can use as a Linux be something I can use as a Linux laptop, as a developer, and I'm sure all laptop, as a developer, and I'm sure all laptop, as a developer, and I'm sure all of you want to know, too. I don't expect of you want to know, too. I don't expect of you want to know, too. I don't expect one AI prompt to build an entire project one AI prompt to build an entire project one AI prompt to build an entire project for me. In reality, I'm constantly for me. In reality, I'm constantly for me. In reality, I'm constantly moving between models depending on the moving between models depending on the moving between models depending on the job. GPT for research, Claude for job. GPT for research, Claude for job. GPT for research, Claude for coding, Gemini for massive context, nano coding, Gemini for massive context, nano coding, Gemini for massive context, nano banana, midjourney, flux for images, and banana, midjourney, flux for images, and banana, midjourney, flux for images, and then I've got seed dance and cling for then I've got seed dance and cling for then I've got seed dance and cling for video. That's why chat lm by Abacus Aai video. That's why chat lm by Abacus Aai video. That's why chat lm by Abacus Aai makes sense. It brings day one support makes sense. It brings day one support makes sense. It brings day one support for the latest GPT, Claw, Gemini, Gro, for the latest GPT, Claw, Gemini, Gro, for the latest GPT, Claw, Gemini, Gro, Deepseek, and more in one place the Deepseek, and more in one place the Deepseek, and more in one place the moment they drop. Pick any model from moment they drop. Pick any model from moment they drop. Pick any model from the interface or let Route LLM the interface or let Route LLM the interface or let Route LLM automatically choose the best model for automatically choose the best model for automatically choose the best model for each prompt. Create professional each prompt. Create professional each prompt. Create professional presentations with graphs and charts and presentations with graphs and charts and presentations with graphs and 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. Need text to defeat AI detectors. Need text to defeat AI detectors. Need visuals? Pick frontier or open-source visuals? Pick frontier or open-source visuals? Pick frontier or open-source models. And when you need more than models. And when you need more than models. And when you need more than chat, Abacus AI agent can help build chat, Abacus AI agent can help build chat, Abacus AI agent can help build complex apps and websites, connect

complex apps and websites, connect complex apps and websites, connect payments, or run 247 agents that keep payments, or run 247 agents that keep payments, or run 247 agents that keep working through longer tasks. The best working through longer tasks. The best working through longer tasks. The best part is app hosting, back-end database, part is app hosting, back-end database, part is app hosting, back-end database, and O support comes with the and O support comes with the and O support comes with the subscription. All that starts at just subscription. All that starts at just subscription. All that starts at just $10 a month. Way cheaper than paying for $10 a month. Way cheaper than paying for $10 a month. Way cheaper than paying for all those subscriptions separately. all those subscriptions separately. all those subscriptions separately. Check out chat llm.abacus.ai Check out chat llm.abacus.ai Check out chat llm.abacus.ai or click the link below. or click the link below. or click the link below. Wow, it's thin. Oh, that's just the Wow, it's thin. Oh, that's just the Wow, it's thin. Oh, that's just the bezel. You get a little screwdriver bezel. You get a little screwdriver bezel. You get a little screwdriver modules for USBA, DP, HDMI, Ethernet, modules for USBA, DP, HDMI, Ethernet, modules for USBA, DP, HDMI, Ethernet, USBC, and micro SD. I got them all, USBC, and micro SD. I got them all, USBC, and micro SD. I got them all, baby. SSD, LP cam. This is pretty baby. SSD, LP cam. This is pretty baby. SSD, LP cam. This is pretty interesting. This is the battery, interesting. This is the battery, interesting. This is the battery, speakers, fan. Made with love by the speakers, fan. Made with love by the speakers, fan. Made with love by the framework team. All right. 1 TBTE drive. framework team. All right. 1 TBTE drive. framework team. All right. 1 TBTE drive. Place the gold. That's the memory. Just Place the gold. That's the memory. Just Place the gold. That's the memory. Just now we need this. That's metal. I guess now we need this. That's metal. I guess now we need this. That's metal. I guess I just popped that in, huh? Wait a I just popped that in, huh? Wait a I just popped that in, huh? Wait a minute. I'm not done yet. Wait. We need minute. I'm not done yet. Wait. We need minute. I'm not done yet. Wait. We need this thing, too. The bezel. There is a this thing, too. The bezel. There is a this thing, too. The bezel. There is a little bit of a give in the lid. But the little bit of a give in the lid. But the little bit of a give in the lid. But the fact that it's metal gives me a little fact that it's metal gives me a little fact that it's metal gives me a little bit more confidence. And it weighs bit more confidence. And it weighs bit more confidence. And it weighs pretty much just like a MacBook Air.

pretty much just like a MacBook Air. pretty much just like a MacBook Air. I've got Fedora installed. And I need to I've got Fedora installed. And I need to I've got Fedora installed. And I need to update it. Huh? Oh, update it. Huh? Oh, update it. Huh? Oh, >> Ethernet. This is the 10 gig one. and >> Ethernet. This is the 10 gig one. and >> Ethernet. This is the 10 gig one. and I'm online and we've got an Intel chip I'm online and we've got an Intel chip I'm online and we've got an Intel chip that wasn't supported on that version of that wasn't supported on that version of that wasn't supported on that version of Fedora. So, I just updated it and then Fedora. So, I just updated it and then Fedora. So, I just updated it and then it worked. In fact, Framework recommends it worked. In fact, Framework recommends it worked. In fact, Framework recommends you use Fedora 44, not 43 or 42. Now, you use Fedora 44, not 43 or 42. Now, you use Fedora 44, not 43 or 42. Now, this is the M5 MacBook Pro and I just this is the M5 MacBook Pro and I just this is the M5 MacBook Pro and I just wanted to see them side by side and run wanted to see them side by side and run wanted to see them side by side and run a few tests because that's what you all a few tests because that's what you all a few tests because that's what you all want to see probably. It is nice to see want to see probably. It is nice to see want to see probably. It is nice to see a non-mac machine besides a Surface Pro a non-mac machine besides a Surface Pro a non-mac machine besides a Surface Pro have a haptic trackpad like this where have a haptic trackpad like this where have a haptic trackpad like this where you can press it anywhere and the tap you can press it anywhere and the tap you can press it anywhere and the tap will be registered. The keyboard is will be registered. The keyboard is will be registered. The keyboard is something I'm not used to on the something I'm not used to on the something I'm not used to on the framework. However, it is a slightly framework. However, it is a slightly framework. However, it is a slightly longer travel and that might actually be longer travel and that might actually be longer travel and that might actually be easier for typing because when I'm in easier for typing because when I'm in easier for typing because when I'm in the office and I'm on a desktop or the office and I'm on a desktop or the office and I'm on a desktop or remoting in, I use a mechanical keyboard remoting in, I use a mechanical keyboard remoting in, I use a mechanical keyboard sometimes where the travel is a lot sometimes where the travel is a lot sometimes where the travel is a lot more. So, this feels a little bit closer more. So, this feels a little bit closer more. So, this feels a little bit closer to that than the membrane on the to that than the membrane on the to that than the membrane on the MacBook, but the space bar is pretty MacBook, but the space bar is pretty MacBook, but the space bar is pretty much the same. They even sound very much the same. They even sound very much the same. They even sound very similar. Let's see. How do I Oh, yeah. I similar. Let's see. How do I Oh, yeah. I similar. Let's see. How do I Oh, yeah. I want to get HDMI output. I like that.

want to get HDMI output. I like that. want to get HDMI output. I like that. Now, it is an Intelbased chip, so I will Now, it is an Intelbased chip, so I will Now, it is an Intelbased chip, so I will plug it in in order to do some of these plug it in in order to do some of these plug it in in order to do some of these tests. But just before I do, I'm going tests. But just before I do, I'm going tests. But just before I do, I'm going to run speedometer 3.1. Basically shows to run speedometer 3.1. Basically shows to run speedometer 3.1. Basically shows you how responsive web applications are you how responsive web applications are you how responsive web applications are going to be on the machine, whether going to be on the machine, whether going to be on the machine, whether you're developing them or you're the end you're developing them or you're the end you're developing them or you're the end user. And since these are all JavaScript user. And since these are all JavaScript user. And since these are all JavaScript based applications, there's uh to-do based applications, there's uh to-do based applications, there's uh to-do applications. There's charting applications. There's charting applications. There's charting libraries, Angular, React, Vue, things libraries, Angular, React, Vue, things libraries, Angular, React, Vue, things like that. Everything is basically a like that. Everything is basically a like that. Everything is basically a single threaded operation here because single threaded operation here because single threaded operation here because JavaScript is single threaded. So, we're JavaScript is single threaded. So, we're JavaScript is single threaded. So, we're getting 12.5, which is not the most getting 12.5, which is not the most getting 12.5, which is not the most impressive thing I've ever seen. That's impressive thing I've ever seen. That's impressive thing I've ever seen. That's better. 30.8. So, yeah, I'm going to better. 30.8. So, yeah, I'm going to better. 30.8. So, yeah, I'm going to keep this thing plugged in, but that's keep this thing plugged in, but that's keep this thing plugged in, but that's nothing new. That's just the way Intel nothing new. That's just the way Intel nothing new. That's just the way Intel chips have been for a long time now. I'm chips have been for a long time now. I'm chips have been for a long time now. I'm going to set it to performance mode and going to set it to performance mode and going to set it to performance mode and see if there's any fan noise that pops see if there's any fan noise that pops see if there's any fan noise that pops up. Nope. This time, let's raise these up. Nope. This time, let's raise these up. Nope. This time, let's raise these two machines. Now, it's very difficult two machines. Now, it's very difficult two machines. Now, it's very difficult to beat the M5 chip in single core to beat the M5 chip in single core to beat the M5 chip in single core performance. In fact, I got 60.6 over performance. In fact, I got 60.6 over performance. In fact, I got 60.6 over here. And here on performance mode, I here. And here on performance mode, I here. And here on performance mode, I got 37.3, which is actually very nice. got 37.3, which is actually very nice. got 37.3, which is actually very nice. Let's see if HDMI works. There it is. It Let's see if HDMI works. There it is. It Let's see if HDMI works. There it is. It comes up as a second screen, and I can comes up as a second screen, and I can comes up as a second screen, and I can drag stuff to it. Now, you might be drag stuff to it. Now, you might be drag stuff to it. Now, you might be wondering why I'm comparing the MacBook wondering why I'm comparing the MacBook wondering why I'm comparing the MacBook Pro M5 to this laptop to the Framework Pro M5 to this laptop to the Framework Pro M5 to this laptop to the Framework 13 Pro. And that's because they both 13 Pro. And that's because they both 13 Pro. And that's because they both have Pro in the name, obviously.

have Pro in the name, obviously. have Pro in the name, obviously. Well, also it has to do with price Well, also it has to do with price Well, also it has to do with price because these two, the starting price is because these two, the starting price is because these two, the starting price is about the same, very close. MacBook Pro about the same, very close. MacBook Pro about the same, very close. MacBook Pro starts out at $2,000. This is for the starts out at $2,000. This is for the starts out at $2,000. This is for the 14-in. So, it's just an inch bigger. 14-in. So, it's just an inch bigger. 14-in. So, it's just an inch bigger. Yeah, just a tiny bit bigger screen than Yeah, just a tiny bit bigger screen than Yeah, just a tiny bit bigger screen than the framework. I'll pick black. I'll the framework. I'll pick black. I'll the framework. I'll pick black. I'll pick standard M5 chip because if we go pick standard M5 chip because if we go pick standard M5 chip because if we go to the M5 Pro, we're at $2500. It's 500 to the M5 Pro, we're at $2500. It's 500 to the M5 Pro, we're at $2500. It's 500 bucks more than the Framework laptop bucks more than the Framework laptop bucks more than the Framework laptop start. So, I think that's kind of a start. So, I think that's kind of a start. So, I think that's kind of a different kind of comparison. Unified different kind of comparison. Unified different kind of comparison. Unified memory starts at 16 GB on the MacBook. memory starts at 16 GB on the MacBook. memory starts at 16 GB on the MacBook. We're going to configure the Framework We're going to configure the Framework We're going to configure the Framework Pro DIY edition. And we're going to go Pro DIY edition. And we're going to go Pro DIY edition. And we're going to go with Intel because that's about $200 with Intel because that's about $200 with Intel because that's about $200 cheaper than the AMD 300 series. And if cheaper than the AMD 300 series. And if cheaper than the AMD 300 series. And if you go with the Ultra 5 Intel, then you go with the Ultra 5 Intel, then you go with the Ultra 5 Intel, then you're in the MacBook Air territory. So, you're in the MacBook Air territory. So, you're in the MacBook Air territory. So, you're $800 less than the MacBook Pro you're $800 less than the MacBook Pro you're $800 less than the MacBook Pro starts, which is actually pretty good. starts, which is actually pretty good. starts, which is actually pretty good. That's just the starting price, though. That's just the starting price, though. That's just the starting price, though. We're going to configure it a little bit We're going to configure it a little bit We're going to configure it a little bit more, and it's going to get up there. more, and it's going to get up there. more, and it's going to get up there. This version is the Ultra X7, which This version is the Ultra X7, which This version is the Ultra X7, which brings our price to $1699. We're still brings our price to $1699. We're still brings our price to $1699. We're still below the starting price of the MacBook, below the starting price of the MacBook, below the starting price of the MacBook, but then we got to add memory. So, LP but then we got to add memory. So, LP but then we got to add memory. So, LP Cam 2 is going to bring this up to about Cam 2 is going to bring this up to about Cam 2 is going to bring this up to about the same price as the MacBook, just a the same price as the MacBook, just a the same price as the MacBook, just a little bit under, but we still need little bit under, but we still need little bit under, but we still need storage. MacBook gives us 1 TBTE. So, storage. MacBook gives us 1 TBTE. So, storage. MacBook gives us 1 TBTE. So, let's select that. And now we're above let's select that. And now we're above let's select that. And now we're above the MacBook price. We're at 2153, but the MacBook price. We're at 2153, but the MacBook price. We're at 2153, but I'd say that's considered pretty much I'd say that's considered pretty much I'd say that's considered pretty much within the same area. Plus or minus within the same area. Plus or minus within the same area. Plus or minus couple hundred bucks is not a huge deal couple hundred bucks is not a huge deal couple hundred bucks is not a huge deal considering all that you get with this considering all that you get with this considering all that you get with this one. You get options, bezel, keyboard, one. You get options, bezel, keyboard, one. You get options, bezel, keyboard, power adapter. They're going to charge power adapter. They're going to charge power adapter. They're going to charge you for a power adapter, but it's a 100

you for a power adapter, but it's a 100 you for a power adapter, but it's a 100 watt power adapter for 60 bucks. So, watt power adapter for 60 bucks. So, watt power adapter for 60 bucks. So, that brings the price up a little more. that brings the price up a little more. that brings the price up a little more. The MacBook includes a 70watt power The MacBook includes a 70watt power The MacBook includes a 70watt power adapter and you can modify that to a 96 adapter and you can modify that to a 96 adapter and you can modify that to a 96 one for $20 extra and I will do that one for $20 extra and I will do that one for $20 extra and I will do that just to get it a little bit closer to just to get it a little bit closer to just to get it a little bit closer to the power of the framework one. That's the power of the framework one. That's the power of the framework one. That's all the configuration you get with the all the configuration you get with the all the configuration you get with the MacBook. So, we're at $219. MacBook. So, we're at $219. MacBook. So, we're at $219. But with the framework, you get But with the framework, you get But with the framework, you get expansion cards, which you can get or expansion cards, which you can get or expansion cards, which you can get or you don't have to. But you should you don't have to. But you should you don't have to. But you should probably have some if this is your first probably have some if this is your first probably have some if this is your first one. If this is your second one, then one. If this is your second one, then one. If this is your second one, then you might just use the old ones. USBC. you might just use the old ones. USBC. you might just use the old ones. USBC. I'll grab one of those. Actually, let's I'll grab one of those. Actually, let's I'll grab one of those. Actually, let's grab two of those because well, the grab two of those because well, the grab two of those because well, the MacBook has three actually. So, let's go MacBook has three actually. So, let's go MacBook has three actually. So, let's go with three. That's 30 bucks. Graphite, with three. That's 30 bucks. Graphite, with three. That's 30 bucks. Graphite, orange, and red. I don't know. Don't orange, and red. I don't know. Don't orange, and red. I don't know. Don't judge me. I want a USBA one just in case judge me. I want a USBA one just in case judge me. I want a USBA one just in case I'm carrying one of these mice around I'm carrying one of these mice around I'm carrying one of these mice around and I don't feel like connecting with and I don't feel like connecting with and I don't feel like connecting with Bluetooth. I want to have that handy Bluetooth. I want to have that handy Bluetooth. I want to have that handy because that's good backup option. I do because that's good backup option. I do because that's good backup option. I do want an HDMI. That's 22 bucks. And want an HDMI. That's 22 bucks. And want an HDMI. That's 22 bucks. And Ethernet is nice to have. So, I normally Ethernet is nice to have. So, I normally Ethernet is nice to have. So, I normally ordered that, but I won't do it just for ordered that, but I won't do it just for ordered that, but I won't do it just for the price comparison sake. Finally, the the price comparison sake. Finally, the the price comparison sake. Finally, the MacBook does have an SD card, so let's MacBook does have an SD card, so let's MacBook does have an SD card, so let's add one of those. That's 25 bucks. We add one of those. That's 25 bucks. We add one of those. That's 25 bucks. We are at $2,277 are at $2,277 are at $2,277 compared to the MacBook's 2019. So, the compared to the MacBook's 2019. So, the compared to the MacBook's 2019. So, the MacBook is actually cheaper now. So, MacBook is actually cheaper now. So, MacBook is actually cheaper now. So, that's pricing. But, did I mention that that's pricing. But, did I mention that that's pricing. But, did I mention that framework also gives you more options framework also gives you more options framework also gives you more options like eventually I hope to be able to like eventually I hope to be able to like eventually I hope to be able to select this Ultra X9. It's out of stock, select this Ultra X9. It's out of stock, select this Ultra X9. It's out of stock, but maybe eventually it will be but maybe eventually it will be but maybe eventually it will be available. It's in the same machine.

available. It's in the same machine. available. It's in the same machine. Memory, you can go up to 64 here. And Memory, you can go up to 64 here. And Memory, you can go up to 64 here. And these LP CAM modules actually go up to these LP CAM modules actually go up to these LP CAM modules actually go up to 96 if you can find one here. You can 96 if you can find one here. You can 96 if you can find one here. You can only configure it with 64, but you don't only configure it with 64, but you don't only configure it with 64, but you don't have to buy it at all. You can just have to buy it at all. You can just have to buy it at all. You can just order it somewhere else if you want to order it somewhere else if you want to order it somewhere else if you want to save the $1,600. Like if you have a side save the $1,600. Like if you have a side save the $1,600. Like if you have a side deal going on somewhere or eBay, grab deal going on somewhere or eBay, grab deal going on somewhere or eBay, grab yourself a 96 GB chip and you're set. yourself a 96 GB chip and you're set. yourself a 96 GB chip and you're set. But if I raise this up to 64, we're over But if I raise this up to 64, we're over But if I raise this up to 64, we're over $3,000 now. 3,600. And the MacBook $3,000 now. 3,600. And the MacBook $3,000 now. 3,600. And the MacBook doesn't even give you that option doesn't even give you that option doesn't even give you that option because the M5 chip only supports up to because the M5 chip only supports up to because the M5 chip only supports up to 32 GB in memory. If you want to go with 32 GB in memory. If you want to go with 32 GB in memory. If you want to go with a 64, you would need to upgrade to an M5 a 64, you would need to upgrade to an M5 a 64, you would need to upgrade to an M5 Pro chip. So, let's do that. Now, we got Pro chip. So, let's do that. Now, we got Pro chip. So, let's do that. Now, we got a different chip in here. And our price a different chip in here. And our price a different chip in here. And our price is 3,699. is 3,699. is 3,699. Pretty close. Now, the framework is Pretty close. Now, the framework is Pretty close. Now, the framework is cheaper. Every time you clone a cheaper. Every time you clone a cheaper. Every time you clone a repository or install dependencies or repository or install dependencies or repository or install dependencies or open a project, you're hitting the SSD. open a project, you're hitting the SSD. open a project, you're hitting the SSD. Now, the SSD with a MacBook is the way Now, the SSD with a MacBook is the way Now, the SSD with a MacBook is the way it is, but with the framework, we can it is, but with the framework, we can it is, but with the framework, we can pick different kinds of SSDs we want to pick different kinds of SSDs we want to pick different kinds of SSDs we want to use. With the framework, I got the use. With the framework, I got the use. With the framework, I got the SanDisk SN7100, and that's gen 4. 7100 SanDisk SN7100, and that's gen 4. 7100 SanDisk SN7100, and that's gen 4. 7100 and 7200 on the M5 for read and write and 7200 on the M5 for read and write and 7200 on the M5 for read and write and 5800, 5600 for read and write on the and 5800, 5600 for read and write on the and 5800, 5600 for read and write on the framework. The read and write speeds are framework. The read and write speeds are framework. The read and write speeds are pretty good. They're not as high as on pretty good. They're not as high as on pretty good. They're not as high as on the M5, but those are sequential read the M5, but those are sequential read the M5, but those are sequential read and writes. The random read and writes and writes. The random read and writes and writes. The random read and writes is what matters a lot for doing is what matters a lot for doing is what matters a lot for doing development. you're reading and writing development. you're reading and writing development. you're reading and writing a lot of little files here and there a lot of little files here and there a lot of little files here and there when you're doing compilations. And that when you're doing compilations. And that when you're doing compilations. And that number is really impressive on the number is really impressive on the number is really impressive on the framework, especially on the right framework, especially on the right framework, especially on the right speeds. Even an M5 Max model doesn't

speeds. Even an M5 Max model doesn't speeds. Even an M5 Max model doesn't come that close. This is my M5 Max on come that close. This is my M5 Max on come that close. This is my M5 Max on the side here. And these are the the side here. And these are the the side here. And these are the numbers. Now, since we're talking about numbers. Now, since we're talking about numbers. Now, since we're talking about JavaScript, this is the web tooling JavaScript, this is the web tooling JavaScript, this is the web tooling benchmark. I sometimes run this. It's benchmark. I sometimes run this. It's benchmark. I sometimes run this. It's basically an oldie but a goodie. Look at basically an oldie but a goodie. Look at basically an oldie but a goodie. Look at that. 8 years ago is when it was that. 8 years ago is when it was that. 8 years ago is when it was updated, but it still works fine. Clone updated, but it still works fine. Clone updated, but it still works fine. Clone the repo, install the mpm packages, the repo, install the mpm packages, the repo, install the mpm packages, build it, and let's race. Boom. Got both build it, and let's race. Boom. Got both build it, and let's race. Boom. Got both of these machines running now. And I'm of these machines running now. And I'm of these machines running now. And I'm expecting since this is JavaScript expecting since this is JavaScript expecting since this is JavaScript again, I'm expecting the MacBook to win. again, I'm expecting the MacBook to win. again, I'm expecting the MacBook to win. But by how much? We've got Babel uh 25.8 But by how much? We've got Babel uh 25.8 But by how much? We've got Babel uh 25.8 runs per second on the framework here. runs per second on the framework here. runs per second on the framework here. 34.18 runs per second. So quite a bit 34.18 runs per second. So quite a bit 34.18 runs per second. So quite a bit faster on the MacBook there. I guess faster on the MacBook there. I guess faster on the MacBook there. I guess this will surface the single core this will surface the single core this will surface the single core performance of that M5 chip quite performance of that M5 chip quite performance of that M5 chip quite nicely. But we also got to do some real nicely. But we also got to do some real nicely. But we also got to do some real builds and multi-core tests. So stay builds and multi-core tests. So stay builds and multi-core tests. So stay tuned for that. Typescript 31.6 6 on the tuned for that. Typescript 31.6 6 on the tuned for that. Typescript 31.6 6 on the framework, TypeScript 46.65. framework, TypeScript 46.65. framework, TypeScript 46.65. Those are the ones I really care about. Those are the ones I really care about. Those are the ones I really care about. And the geometric mean is much higher on And the geometric mean is much higher on And the geometric mean is much higher on the MacBook. 37.18 and 26.36 the MacBook. 37.18 and 26.36 the MacBook. 37.18 and 26.36 on the framework. Now, how will these on the framework. Now, how will these on the framework. Now, how will these machines handle multicore using all machines handle multicore using all machines handle multicore using all their cores to the max in a Python their cores to the max in a Python their cores to the max in a Python interpreted test? Let's check that out.

interpreted test? Let's check that out. interpreted test? Let's check that out. For that, I got the manner broad test, For that, I got the manner broad test, For that, I got the manner broad test, which I commonly use on the channel which I commonly use on the channel which I commonly use on the channel here. It's written all in Python and it here. It's written all in Python and it here. It's written all in Python and it pegs the course to the max. And this is pegs the course to the max. And this is pegs the course to the max. And this is a race. Let's go check out that CPU load a race. Let's go check out that CPU load a race. Let's go check out that CPU load on the MacBook. Those are the green on the MacBook. Those are the green on the MacBook. Those are the green lines marching up and down here. Filling lines marching up and down here. Filling lines marching up and down here. Filling up all the cores. There's 10 cores here. up all the cores. There's 10 cores here. up all the cores. There's 10 cores here. And this is what it looks like on the 16 And this is what it looks like on the 16 And this is what it looks like on the 16 cores of the Framework laptop. cores of the Framework laptop. cores of the Framework laptop. Definitely filling up those cores. But Definitely filling up those cores. But Definitely filling up those cores. But you can hear the fans. Oh yeah, definitely hear those fans. Oh yeah, definitely hear those fans. Hopefully pays off. And the framework Hopefully pays off. And the framework Hopefully pays off. And the framework actually beat the MacBook here. I ran actually beat the MacBook here. I ran actually beat the MacBook here. I ran that a couple times. 30 seconds is the that a couple times. 30 seconds is the that a couple times. 30 seconds is the MacBook average. Don't quote me on that MacBook average. Don't quote me on that MacBook average. Don't quote me on that average part. I know somebody will say average part. I know somebody will say average part. I know somebody will say something. And we got about 23 to 25 something. And we got about 23 to 25 something. And we got about 23 to 25 seconds here on the framework. And it is seconds here on the framework. And it is seconds here on the framework. And it is that core difference. We have 10 cores that core difference. We have 10 cores that core difference. We have 10 cores on the MacBook, 16 on the framework. And on the MacBook, 16 on the framework. And on the MacBook, 16 on the framework. And since this is a multi-core program that since this is a multi-core program that since this is a multi-core program that utilizes all those cores, the framework utilizes all those cores, the framework utilizes all those cores, the framework is going to do better, which means that is going to do better, which means that is going to do better, which means that anytime you're going to have a anytime you're going to have a anytime you're going to have a compilation of code, that's going to use compilation of code, that's going to use compilation of code, that's going to use all the cores. The framework is probably all the cores. The framework is probably all the cores. The framework is probably going to be faster. Now, this test did going to be faster. Now, this test did going to be faster. Now, this test did peg every single core synthetically, but peg every single core synthetically, but peg every single core synthetically, but what happens when we take a look at a what happens when we take a look at a what happens when we take a look at a real compilation? This is Damian Edwards real compilation? This is Damian Edwards real compilation? This is Damian Edwards who is a software architect on the .NET who is a software architect on the .NET who is a software architect on the .NET team at Microsoft and he had this thing team at Microsoft and he had this thing team at Microsoft and he had this thing called Devbench. I have my own little called Devbench. I have my own little called Devbench. I have my own little benchmarks that I run for development benchmarks that I run for development benchmarks that I run for development purposes that arenet and Creat related.

purposes that arenet and Creat related. purposes that arenet and Creat related. And I took a look at this one and it's And I took a look at this one and it's And I took a look at this one and it's good because it's got all these uh good because it's got all these uh good because it's got all these uh different options like warm builds and different options like warm builds and different options like warm builds and cold builds and incremental builds. So I cold builds and incremental builds. So I cold builds and incremental builds. So I thought, hey, I'll check it out and use thought, hey, I'll check it out and use thought, hey, I'll check it out and use it. And I take a look at benchmarks. And it. And I take a look at benchmarks. And it. And I take a look at benchmarks. And one of the benchmarks, the big the big one of the benchmarks, the big the big one of the benchmarks, the big the big large net compile benchmark that I large net compile benchmark that I large net compile benchmark that I always show on my channel is here. It's always show on my channel is here. It's always show on my channel is here. It's mine in his repository. So I feel mine in his repository. So I feel mine in his repository. So I feel honored here to be part of this honored here to be part of this honored here to be part of this repository. And now I'm going to use repository. And now I'm going to use repository. And now I'm going to use Devbench to do the tests here. This has Devbench to do the tests here. This has Devbench to do the tests here. This has slightly more variation. So I'm going to slightly more variation. So I'm going to slightly more variation. So I'm going to run this. And what this let me do is run this. And what this let me do is run this. And what this let me do is select which test I want to do. So let's select which test I want to do. So let's select which test I want to do. So let's say I want to do hello world. It's a say I want to do hello world. It's a say I want to do hello world. It's a very short one. Small.net application. very short one. Small.net application. very short one. Small.net application. Boom. Let's go. So test the cold build, Boom. Let's go. So test the cold build, Boom. Let's go. So test the cold build, the warm build, and the incremental the warm build, and the incremental the warm build, and the incremental build. We have our results. Cold build build. We have our results. Cold build build. We have our results. Cold build on the MacBook is 836 milliseconds. Warm on the MacBook is 836 milliseconds. Warm on the MacBook is 836 milliseconds. Warm is 294, and incremental is 335. is 294, and incremental is 335. is 294, and incremental is 335. Definitely faster than on the framework. Definitely faster than on the framework. Definitely faster than on the framework. Cold build 2,172 Cold build 2,172 Cold build 2,172 milliseconds, warm 446, and incremental milliseconds, warm 446, and incremental milliseconds, warm 446, and incremental 491. What about that huge build, the one 491. What about that huge build, the one 491. What about that huge build, the one with 100,000 name spaces and classes with 100,000 name spaces and classes with 100,000 name spaces and classes that I created before? That's now part that I created before? That's now part that I created before? That's now part of this one. Boom. This one takes a of this one. Boom. This one takes a of this one. Boom. This one takes a little bit longer. Oh, in case you're little bit longer. Oh, in case you're little bit longer. Oh, in case you're curious, I do have this on my GitHub.

curious, I do have this on my GitHub. curious, I do have this on my GitHub. It's this uh project right here. And It's this uh project right here. And It's this uh project right here. And this is how it generates the code. I this is how it generates the code. I this is how it generates the code. I actually generate the code using Python, actually generate the code using Python, actually generate the code using Python, but I make sure that the C compiler but I make sure that the C compiler but I make sure that the C compiler can't optimize away and it has to build can't optimize away and it has to build can't optimize away and it has to build every single thing. Oh, I'm hearing the every single thing. Oh, I'm hearing the every single thing. Oh, I'm hearing the fans. I mean, it's not loud, but I can fans. I mean, it's not loud, but I can fans. I mean, it's not loud, but I can still hear them. still hear them. still hear them. Things are happening. Let's see what's Things are happening. Let's see what's Things are happening. Let's see what's happening with the thermals here. We got happening with the thermals here. We got happening with the thermals here. We got about 45 on the MacBook up to 47. And about 45 on the MacBook up to 47. And about 45 on the MacBook up to 47. And what's going on here? Where's the hot what's going on here? Where's the hot what's going on here? Where's the hot spot? The hot spot's around the same spot? The hot spot's around the same spot? The hot spot's around the same spot here on the framework. And we're a spot here on the framework. And we're a spot here on the framework. And we're a little cooler here. 43 44. Just a couple little cooler here. 43 44. Just a couple little cooler here. 43 44. Just a couple of degrees. And we have our results. of degrees. And we have our results. of degrees. And we have our results. Cold build, 85 seconds. Let's just call Cold build, 85 seconds. Let's just call Cold build, 85 seconds. Let's just call it seconds. It's in milliseconds, but it seconds. It's in milliseconds, but it seconds. It's in milliseconds, but whatever. 54 on the framework. Much whatever. 54 on the framework. Much whatever. 54 on the framework. Much faster on the framework this time. And faster on the framework this time. And faster on the framework this time. And the warm build, 10 1/2 seconds on the the warm build, 10 1/2 seconds on the the warm build, 10 1/2 seconds on the M5, 7.9 seconds on the framework. This M5, 7.9 seconds on the framework. This M5, 7.9 seconds on the framework. This is where more cores pays off when you is where more cores pays off when you is where more cores pays off when you have slightly larger projects like this. have slightly larger projects like this. have slightly larger projects like this. And there's one more in here, which is a And there's one more in here, which is a And there's one more in here, which is a build of an actual CMS called Orchard build of an actual CMS called Orchard build of an actual CMS called Orchard Core. This is as real as they get as far Core. This is as real as they get as far Core. This is as real as they get as far as projects go. Let's try that one out as projects go. Let's try that one out as projects go. Let's try that one out cuz that's the one that's not faked.

cuz that's the one that's not faked. cuz that's the one that's not faked. It's real. I mean, I didn't fake the It's real. I mean, I didn't fake the It's real. I mean, I didn't fake the other ones. It's just that they're more other ones. It's just that they're more other ones. It's just that they're more of a synthetic kind of thing. Hello of a synthetic kind of thing. Hello of a synthetic kind of thing. Hello world. Who's going to write that? This world. Who's going to write that? This world. Who's going to write that? This one, they're very close. Cold build is one, they're very close. Cold build is one, they're very close. Cold build is faster on the framework just by a couple faster on the framework just by a couple faster on the framework just by a couple of seconds. And the warm build and the of seconds. And the warm build and the of seconds. And the warm build and the incremental build are very close. incremental build are very close. incremental build are very close. They're within half a second. It's time They're within half a second. It's time They're within half a second. It's time to move on and talk about AI. I know, I to move on and talk about AI. I know, I to move on and talk about AI. I know, I know, your favorite topic, right? Well, know, your favorite topic, right? Well, know, your favorite topic, right? Well, for some of you it might be, but this for some of you it might be, but this for some of you it might be, but this machine has that LP cam memory in it. machine has that LP cam memory in it. machine has that LP cam memory in it. LP, by the way, that's the same kind of LP, by the way, that's the same kind of LP, by the way, that's the same kind of LP that's in this laptop. LP DDDR5X. And LP that's in this laptop. LP DDDR5X. And LP that's in this laptop. LP DDDR5X. And this one has LP DDR5X also, but it's in this one has LP DDR5X also, but it's in this one has LP DDR5X also, but it's in the LP cam format, which means you can the LP cam format, which means you can the LP cam format, which means you can swap it out. Apple solders the same swap it out. Apple solders the same swap it out. Apple solders the same exact memory onto the chip, so you can't exact memory onto the chip, so you can't exact memory onto the chip, so you can't swap it out and upgrade it later. this swap it out and upgrade it later. this swap it out and upgrade it later. this one you can but other than that it's the one you can but other than that it's the one you can but other than that it's the same memory and memory is very important same memory and memory is very important same memory and memory is very important when it comes to AI workflows LLM image when it comes to AI workflows LLM image when it comes to AI workflows LLM image generation because memory bandwidth or generation because memory bandwidth or generation because memory bandwidth or how much throughput memory can do that's how much throughput memory can do that's how much throughput memory can do that's what determines how fast it can generate what determines how fast it can generate what determines how fast it can generate tokens so let's check memory bandwidth tokens so let's check memory bandwidth tokens so let's check memory bandwidth with this good old stream benchmark and with this good old stream benchmark and with this good old stream benchmark and I say good old because it is pretty old I say good old because it is pretty old I say good old because it is pretty old you can find the code on GitHub but it's you can find the code on GitHub but it's you can find the code on GitHub but it's been around for a long time look at this been around for a long time look at this been around for a long time look at this website yeah it's all HTML and the website yeah it's all HTML and the website yeah it's all HTML and the stream benchmark Boom. Measures the stream benchmark Boom. Measures the stream benchmark Boom. Measures the memory bandwidth. It's got a triad of memory bandwidth. It's got a triad of memory bandwidth. It's got a triad of things that it measures. One particular things that it measures. One particular things that it measures. One particular thing is the copy command. This is very thing is the copy command. This is very thing is the copy command. This is very important when it comes to LLMs. 136,000 important when it comes to LLMs. 136,000 important when it comes to LLMs. 136,000 megabytes per second is what we have on megabytes per second is what we have on megabytes per second is what we have on the M5. Now, Apple quotes 150, but this

the M5. Now, Apple quotes 150, but this the M5. Now, Apple quotes 150, but this is the more realistic number here. And is the more realistic number here. And is the more realistic number here. And we pay a slight penalty for not having we pay a slight penalty for not having we pay a slight penalty for not having those chips soldered on. Now, on the those chips soldered on. Now, on the those chips soldered on. Now, on the framework laptop, we get 97,000 framework laptop, we get 97,000 framework laptop, we get 97,000 megabytes per second. So yeah, it is megabytes per second. So yeah, it is megabytes per second. So yeah, it is actually slower, but you get to go up to actually slower, but you get to go up to actually slower, but you get to go up to 64 GB. You get to go up to 96 if you 64 GB. You get to go up to 96 if you 64 GB. You get to go up to 96 if you want to, and you get to be able to swap want to, and you get to be able to swap want to, and you get to be able to swap it out. That's the push and pull of the it out. That's the push and pull of the it out. That's the push and pull of the whole thing, right? But let's see how whole thing, right? But let's see how whole thing, right? But let's see how this actually affects LLMs. I'm going to this actually affects LLMs. I'm going to this actually affects LLMs. I'm going to fire up LM Studio and it's not going to fire up LM Studio and it's not going to fire up LM Studio and it's not going to work very well. I'm going to warn you work very well. I'm going to warn you work very well. I'm going to warn you that ahead of time and I'm going to that ahead of time and I'm going to that ahead of time and I'm going to explain why. I'm going to load up Gemma explain why. I'm going to load up Gemma explain why. I'm going to load up Gemma 312B by Unslo. It's the 4bit 312B by Unslo. It's the 4bit 312B by Unslo. It's the 4bit quantization. And same thing over here. quantization. And same thing over here. quantization. And same thing over here. Now, we're not going to go too deeply Now, we're not going to go too deeply Now, we're not going to go too deeply into prompts here and different kinds of into prompts here and different kinds of into prompts here and different kinds of prompts, different lengths. I'm just prompts, different lengths. I'm just prompts, different lengths. I'm just going to issue something simple. So, we going to issue something simple. So, we going to issue something simple. So, we can see that even if I say hi here and can see that even if I say hi here and can see that even if I say hi here and hi here, we have a significantly hi here, we have a significantly hi here, we have a significantly different speed here. And this doesn't different speed here. And this doesn't different speed here. And this doesn't even make sense. We're less than half even make sense. We're less than half even make sense. We're less than half the speed on the framework laptop than the speed on the framework laptop than the speed on the framework laptop than we are over here on the Mac. 15.6 tokens we are over here on the Mac. 15.6 tokens we are over here on the Mac. 15.6 tokens per second on the Mac, but the same per second on the Mac, but the same per second on the Mac, but the same model, same prompt, and same model, same prompt, and same model, same prompt, and same quantization, and 6.2 tokens a second quantization, and 6.2 tokens a second quantization, and 6.2 tokens a second here on the framework. What is going on?

here on the framework. What is going on? here on the framework. What is going on? And you'd think, "Oh, maybe it's just And you'd think, "Oh, maybe it's just And you'd think, "Oh, maybe it's just not running on the GPU." Well, actually, not running on the GPU." Well, actually, not running on the GPU." Well, actually, it is. Here, let me run it again and it is. Here, let me run it again and it is. Here, let me run it again and tell it to write a story so we can tell it to write a story so we can tell it to write a story so we can capture it. And there it is. Look at capture it. And there it is. Look at capture it. And there it is. Look at that. It's running on the GPU with 93 that. It's running on the GPU with 93 that. It's running on the GPU with 93 94% utilization. Notice that the GPU is 94% utilization. Notice that the GPU is 94% utilization. Notice that the GPU is sharing the memory, that LP cam with uh sharing the memory, that LP cam with uh sharing the memory, that LP cam with uh the CPU. 62 GB available for the GPU to the CPU. 62 GB available for the GPU to the CPU. 62 GB available for the GPU to do its work. We can run larger models, do its work. We can run larger models, do its work. We can run larger models, way larger than on this MacBook with 16 way larger than on this MacBook with 16 way larger than on this MacBook with 16 gigs of memory. But why are we getting gigs of memory. But why are we getting gigs of memory. But why are we getting the slowdown? Well, this has to do with the slowdown? Well, this has to do with the slowdown? Well, this has to do with that Intel chip inside there. And it's that Intel chip inside there. And it's that Intel chip inside there. And it's not that the Intel chip is bad. It's not that the Intel chip is bad. It's not that the Intel chip is bad. It's just that it's too new. LM Studio just that it's too new. LM Studio just that it's too new. LM Studio doesn't yet support that architecture. doesn't yet support that architecture. doesn't yet support that architecture. So, for now, we have to use a different So, for now, we have to use a different So, for now, we have to use a different build of Llama CPP that's not yet build of Llama CPP that's not yet build of Llama CPP that's not yet supported in LM Studio. And you can do supported in LM Studio. And you can do supported in LM Studio. And you can do that. You can go to Llama CPP on GitHub, that. You can go to Llama CPP on GitHub, that. You can go to Llama CPP on GitHub, build it. It's pretty easy actually, but build it. It's pretty easy actually, but build it. It's pretty easy actually, but you have to include Cickle support. CL you have to include Cickle support. CL you have to include Cickle support. CL is the underlying API that's available is the underlying API that's available is the underlying API that's available for Intel that's going to allow it to for Intel that's going to allow it to for Intel that's going to allow it to run faster. Otherwise, it's going to run run faster. Otherwise, it's going to run run faster. Otherwise, it's going to run Vulcan. Vulcan is another kind of API Vulcan. Vulcan is another kind of API Vulcan. Vulcan is another kind of API that runs across all the platforms. So, that runs across all the platforms. So, that runs across all the platforms. So, Vulcan can run on AMD. It can run on Vulcan can run on AMD. It can run on Vulcan can run on AMD. It can run on Intel. But remember, when you're first Intel. But remember, when you're first Intel. But remember, when you're first customizing your framework laptop, you customizing your framework laptop, you customizing your framework laptop, you have the option to pick Intel, the brand have the option to pick Intel, the brand have the option to pick Intel, the brand new Ultra Series 3, which is Panther new Ultra Series 3, which is Panther new Ultra Series 3, which is Panther Lake. Support is probably coming soon.

Lake. Support is probably coming soon. Lake. Support is probably coming soon. Or you can pay $200 more and get the Or you can pay $200 more and get the Or you can pay $200 more and get the Ryzen AI300. Not that it's a better Ryzen AI300. Not that it's a better Ryzen AI300. Not that it's a better chip. It's an older chip, but it has chip. It's an older chip, but it has chip. It's an older chip, but it has more support. and Vulcan will work more support. and Vulcan will work more support. and Vulcan will work pretty well on that as well as the AMD pretty well on that as well as the AMD pretty well on that as well as the AMD libraries. Here I'm going to run Llama libraries. Here I'm going to run Llama libraries. Here I'm going to run Llama CPP and this is the sickle build. I'm CPP and this is the sickle build. I'm CPP and this is the sickle build. I'm going to point at the exact same model, going to point at the exact same model, going to point at the exact same model, same quantization. Boom. Let's go. There same quantization. Boom. Let's go. There same quantization. Boom. Let's go. There it is. And it's generating. And this is it is. And it's generating. And this is it is. And it's generating. And this is using the GPU again. If we take a look using the GPU again. If we take a look using the GPU again. If we take a look at that, there it is. It's up to 97% at that, there it is. It's up to 97% at that, there it is. It's up to 97% now. 99%. So, the utilization of the GPU now. 99%. So, the utilization of the GPU now. 99%. So, the utilization of the GPU is better here because of that optimized is better here because of that optimized is better here because of that optimized software. Still, it probably could be software. Still, it probably could be software. Still, it probably could be better. We got 9.9 tokens per second better. We got 9.9 tokens per second better. We got 9.9 tokens per second here, which is not quite two times here, which is not quite two times here, which is not quite two times faster than it was with Vulcan. So even faster than it was with Vulcan. So even faster than it was with Vulcan. So even though we saw 40% difference in stream though we saw 40% difference in stream though we saw 40% difference in stream benchmark, the difference in generation benchmark, the difference in generation benchmark, the difference in generation here is more like 50 to 55%. So what here is more like 50 to 55%. So what here is more like 50 to 55%. So what happens when we want to run a larger happens when we want to run a larger happens when we want to run a larger model? Well, on the Mac, this is a 16 model? Well, on the Mac, this is a 16 model? Well, on the Mac, this is a 16 gig machine, so I can't even run this uh gig machine, so I can't even run this uh gig machine, so I can't even run this uh Quen Coder 30B. It's likely too large. Quen Coder 30B. It's likely too large. Quen Coder 30B. It's likely too large. And that's true. Even the 3bit And that's true. Even the 3bit And that's true. Even the 3bit quantization is 15 GB. But of course, quantization is 15 GB. But of course, quantization is 15 GB. But of course, you'd need to buy a larger machine if you'd need to buy a larger machine if you'd need to buy a larger machine if you want to run the larger models. The you want to run the larger models. The you want to run the larger models. The 16 in machine, not the 14. I'm just 16 in machine, not the 14. I'm just 16 in machine, not the 14. I'm just kidding. I mean larger memory. But this kidding. I mean larger memory. But this kidding. I mean larger memory. But this one has 64 GB, so I'm able to run the one has 64 GB, so I'm able to run the one has 64 GB, so I'm able to run the larger models. Here's token generation larger models. Here's token generation larger models. Here's token generation using Vulcan compared to Sickle. And as using Vulcan compared to Sickle. And as using Vulcan compared to Sickle. And as you can see, Sickle is much faster. Quen you can see, Sickle is much faster. Quen you can see, Sickle is much faster. Quen coder 30B is down here. 22 tokens per coder 30B is down here. 22 tokens per coder 30B is down here. 22 tokens per second with Vulcan and 34.9 tokens per second with Vulcan and 34.9 tokens per second with Vulcan and 34.9 tokens per second with Sickle. That's pretty nice.

second with Sickle. That's pretty nice. second with Sickle. That's pretty nice. What I didn't expect is the prompt What I didn't expect is the prompt What I didn't expect is the prompt processing. And this one is a little bit processing. And this one is a little bit processing. And this one is a little bit confounding because I haven't found this confounding because I haven't found this confounding because I haven't found this pattern before in my testing where the pattern before in my testing where the pattern before in my testing where the prompt processing is actually faster in prompt processing is actually faster in prompt processing is actually faster in Vulcan than it is with Sickle except for Vulcan than it is with Sickle except for Vulcan than it is with Sickle except for Quen Coder. But this looks like an Quen Coder. But this looks like an Quen Coder. But this looks like an anomaly. It doesn't follow the pattern anomaly. It doesn't follow the pattern anomaly. It doesn't follow the pattern of the other models. Unfortunately, I of the other models. Unfortunately, I of the other models. Unfortunately, I don't know of a way that you'll be able don't know of a way that you'll be able don't know of a way that you'll be able to use prompt processing using Vulcan to use prompt processing using Vulcan to use prompt processing using Vulcan and then token generation, the second and then token generation, the second and then token generation, the second part of inference using Sickle. Once part of inference using Sickle. Once part of inference using Sickle. Once you've loaded that model up using one you've loaded that model up using one you've loaded that model up using one stack, you kind of have to use the same stack, you kind of have to use the same stack, you kind of have to use the same stack unless you want to use twice the stack unless you want to use twice the stack unless you want to use twice the memory amount and that's just not memory amount and that's just not memory amount and that's just not feasible. So, the framework, it's an feasible. So, the framework, it's an feasible. So, the framework, it's an interesting machine and it's solidly interesting machine and it's solidly interesting machine and it's solidly built. It feels premium. It's feels like built. It feels premium. It's feels like built. It feels premium. It's feels like it's quality. As far as performance it's quality. As far as performance it's quality. As far as performance goes, it's somewhat there. Is it going goes, it's somewhat there. Is it going goes, it's somewhat there. Is it going to be able to do all your development to be able to do all your development to be able to do all your development related tasks? Probably. Most likely. Am related tasks? Probably. Most likely. Am related tasks? Probably. Most likely. Am I going to be trading in my M5 Max, my I going to be trading in my M5 Max, my I going to be trading in my M5 Max, my daily driver for the framework? No, I daily driver for the framework? No, I daily driver for the framework? No, I won't be. But that's also an $8,000 won't be. But that's also an $8,000 won't be. But that's also an $8,000 laptop. Way more powerful. If I had an laptop. Way more powerful. If I had an laptop. Way more powerful. If I had an M5 MacBook Pro and I needed a Linux M5 MacBook Pro and I needed a Linux M5 MacBook Pro and I needed a Linux machine, this is an easy trade. Even machine, this is an easy trade. Even machine, this is an easy trade. Even though the M5 is an integrated system though the M5 is an integrated system though the M5 is an integrated system and it can do a lot of the tasks that we and it can do a lot of the tasks that we and it can do a lot of the tasks that we did slightly faster except for the did slightly faster except for the did slightly faster except for the multi-core stuff which is actually the multi-core stuff which is actually the multi-core stuff which is actually the more realistic workflows. This one does more realistic workflows. This one does more realistic workflows. This one does all that. I can see this framework being all that. I can see this framework being all that. I can see this framework being my dev machine and that's why I'm going my dev machine and that's why I'm going my dev machine and that's why I'm going to keep a close eye on this. I think to keep a close eye on this. I think to keep a close eye on this. I think they've come a long way. This is an they've come a long way. This is an they've come a long way. This is an impressive device and this is my first impressive device and this is my first impressive device and this is my first look at it. So if you have any

look at it. So if you have any look at it. So if you have any questions, write it down below. I'll questions, write it down below. I'll questions, write it down below. I'll probably be back doing more reviews on probably be back doing more reviews on probably be back doing more reviews on this. And if you want to see Windows on this. And if you want to see Windows on this. And if you want to see Windows on this thing, I didn't really want to, but this thing, I didn't really want to, but this thing, I didn't really want to, but maybe some of you do. Let me know in the maybe some of you do. Let me know in the maybe some of you do. Let me know in the comments down below. Now, this is just comments down below. Now, this is just comments down below. Now, this is just the base M5 MacBook Pro. If you want to the base M5 MacBook Pro. If you want to the base M5 MacBook Pro. If you want to see the comparison of this one with the see the comparison of this one with the see the comparison of this one with the M5 Pro and the M5 Max, watch this video M5 Pro and the M5 Max, watch this video M5 Pro and the M5 Max, watch this video here. Thanks for watching, and I'll see here. Thanks for watching, and I'll see here. Thanks for watching, and I'll see you next time.