 And we're live. Hi. Hey, what's up? What's up? So, uh, so please introduce yourself. Hey, I'm Sahaj from, uh, Leonardo. I work as an application engineer with the 96 board steam, but I've been an arm of fanboy for a while. And, um, how long, how long have you been an arm? How long? Oh, well, since the Raspberry Pi came out, that's a long time. Seven, eight years, something like that. Yeah. Yeah, I think. So, so arm is, uh, how many buildings of chips per second? Per per per year? I don't know. Yeah, it's a massive. How many arm chips do you have in your office right there? Oh, I can't count them on my hand. I have so my SVC count. The last I checked, uh, about a year ago was around 2021. And then I've gotten a bunch more stuff from there. So I'd give it a good 30. Nice. And this is your YouTube channel right here, geek till it hurts. Yeah. And you have like, how many videos? Uh, I don't remember honestly. It's been a while. And you show awesome stuff like this, right? What is, what is this? That's, that's, that's the Pinebook Pro. It's, it's an arm powered laptop. It came out before the M, the M one Mac and it's powered by the Archi399. Pretty neat, pretty neat stuff. 3 3 9 9. It's like, uh, it's an awesome processor and it's been going for four or five years now. Yeah. I mean, it's, it's fun. Once you upstream a processor into Linux kernel and you would, you know, the support never stops. People just pick it up and use it. How's it going with the, with the, what you call it, the, the reverse engineering of the GPU and that, that, that's fair because this is Mali. Yeah. So, uh, yeah, that, that laptop actually runs that, uh, reverse engineered panfros driver and it's really good. Um, it's catching up pace. It's much easier to integrate. Actually, if you just install a Linux OS, it's integrated by default. Um, you don't have to do anything special for it. No binary is nothing. Just install. And it's really good. How good is it compared to like the official, uh, like Android GPU driver? Uh, I'm not entirely sure, but I think they're getting really close in terms of feature, future comparison. Uh, I hope they start working on Vulkan drivers soon because I see that sort of kind of the future that the GP world is heading in. So yeah, it's, it's good. It's very, very much usable. You can play some games as well. You probably have videos about, uh, panfrost, right? I do. Um, I found the, uh, panfrost a bunch. Like when I look at this, for example, there, geek till it hurts and then you jump in. That's a very old me. And what do you talk about in there? Like, uh, you're showing some, Oh, these are really early. Um, all right. So these are from Lima. Uh, Lima was the older 400 series for 50 series reverse engineer driver. And, uh, this was at a time when Lima was just good enough to show some spinning cubes. You can't, you couldn't really do a whole lot with it. And till now, like Lima has gained a lot of traction. It's working really well. Nice. And then you just talk a lot, right? Yeah, I do, I do talk, I do talk a lot. And, uh, there is showing that it can boot or So that's already booting. It's just, uh, running the GL mark two demo. All right. Cool. Uh, so it's gotten a lot of support over the last couple years. All this open source GPU driver and, but you can say that it's, uh, how much more can be done to get it like perfect? So to get it perfect, I think you need to be feature-complete on OpenGL3. You need to have Vulkan support and then have OpenCL support. All of them can come at a very fast pace, especially because, uh, Mesa, the folks that run the whole Linux, um, open source driver stack, they now have something called Clover and I've been testing it and it's really fun because Clover provides open CL over, uh, over, over just that Mesa stack by default. So AMD GPUs, you don't need to install even, um, their GPU AMD GPU Pro drivers for compute. Now you can just use Clover and that provides open source open CL driver as well. But that works on the AMD stuff and also on the mount, not, what do you mean? Not yet, not yet, but it's going to come to the Mali maybe. Yeah. Yeah. I see that sort of a path they should go on. Nice. And here you, are we screen sharing? This is your screen share. So, uh, what's some of the other kind of awesome stuff you've been talking about recently on your YouTube channel? Some of the most recent stuff has been with, um, ARM desktop because I was lucky enough for Solid Run to send me a Honeycomb LX2, which is based on, um, uh, LX2160A from NXP. It's a 16 core 72, um, sorry, 16 core codec 6872 CPU. But more importantly, it's, it's like, um, it's, um, uh, compliant. So systems ready ES compliant. So it just boots as a desktop. Um, and that's really fun because you can even install windows on it. So if I may, yeah, please, maybe you can, you can screen share. I'm there. I'm just going on their, uh, official video on their website. So that's the board right there. And you can run windows. So on, on my camp because I'm screen sharing, like we do captioning into a camera device. Um, this is the Honeycomb. And as you can see, it looks like it's in the bias and that's because it is like it has a full UEFI driver. Um, and I've just rebooted it so I can, you can like see the, uh, the, the boot up screen. And this has just been a really, really fun toy for the last, um, month. How's the performance? It's really good. Uh, you can add your own memory and I have like 64 gigs of memory on there. So I have enough memory. You can add your own GPU. I have an RX 5, uh, 560 on there right now. Uh, GP only works under, you know, Linux for now, uh, because there aren't any drivers on the windows side, but just to prove that it can do things differently. I need to show like proper, this is Windows 10 on ARM. This is what the stuff you get on ARM powered, uh, Microsoft Surface laptops. How do you get that image? Uh, do they just let people download it and stuff? Yeah, kind of. I have a video on it coming, uh, tomorrow. So yeah, it's, it's basically you have a virtual image and then you convert it into a disk image and you just install it. Oh, how smooth is the, is this Windows? So it's kind of okay. Um, but because there is no GPU driver, you can't really game. I've tried to like, you can see I have steam installed and that steam is x86. So it's running under, it's the under the emulation layer, but, uh, like you can't really game on it because like AMD doesn't provide ARM drivers. Um, so that's a limitation, but as for testing stuff out on ARM, uh, I think it's really good. Nice. Uh, so that's, uh, 16 core. Yeah. Let me actually ARM Cortex a 72. Yeah. Uh, so it's sold by NXP. It's part of the yeah, uh, layerscape, right? They call it. Yeah. Yeah. So 16 core Cortex a 72 and then I have 64 gigs of memory on there. So yeah, uh, running this machine on Linux is even more better because you just get to do so much fun stuff. So you, you, are you going to reboot into Linux now? Okay. Let's, let's do a quick reboot. Also doesn't take too much longer because all I have in there are SSDs. How do you share? How do you share that screen right now? Do you use some kind of switcher? I have a cam link for K USB stick. Um, and it just acts as a webcam device. So I just changed my webcam in the settings. That's all. Nice. It takes a second to reboot. Um, so it's, uh, um, is it kind of like server desktop? Yeah. What kind of market are you trying to do with this? So it's, um, it's network server ish. It has a bigger brother. Um, that's, uh, the clear frog. And oops, why does it boot in one dose? I don't want it booting on windows. Please don't just give me a second. Okay. So anyway, so it has a bigger brother that has a hundred gigabit ethernet port on it, not ethernet, but SFP port on it. And this one doesn't have that port. It has four, uh, four 40 gigabit ports on it. I think I said that right. Or a single 40 gig. All right. Four 10 gigabits port. And then you can act as a single 40 gigabit or something like that. But yeah, it's a very networking focused, um, so a network for network server, but it's also in this standardized desktop format and has this really nice EFI. So you can just install it. The installation process on it is, is basically you have an ISO. You can even write an ISO 20 to a DVD and it will work. Um, but you can have an ISO on a USB stick. You plug in that USB stick and it boots like a desktop. There's no board specific image. There's nothing like that. So, um, uh, are you playing around with this as part of the 96 boards ecosystem or it's going to be some kind of support in there? Or, uh, this is just my personal fraud, personal stuff right now. Um, 96 boards is separate. That's work. Um, and we'll talk about that in a little while. I have some interesting project on the 96 both sides of, uh, so again, that just standard Fedora, um, install and it works just fine. And yeah, the GPU works, uh, see, uh, AMD GPU. Okay. So for some reason that wasn't working. So, uh, it is exciting world we are living in 2021. No, we're, we're the, um, I mean, there's been armed, armed laptops and Chromebooks and some Windows 10 on the arm happening. There was, uh, um, what's it called? Uh, uh, there was this desktop that, uh, that was featured a few times at the, at the Lenaro connect. Um, with how many cores? The EMAG one, the EMAG one with 200 or something threads, right? Yeah. Uh, what's it called? Um, EMAG. I think it was, oh, ampere EMAG. Okay. There's that, that too. Yes. That's a, starts at like 4,300 USB. And, um, uh, what's it called that? Sorry. There was this desktop video I did. I'm just trying to find it right here. Uh, the developer box. Did you do that one as well? Yeah, there was the thunder thunder X station from social next. Yeah, social next to a 11. Yeah. So that's, that's the 96 votes one. So that's been happening. Yeah, that's, that's been happened. And it's, and developers hand and like it's, it's already upstreamed everywhere. Uh, it's a nice desktop. I don't have it sadly. Surprise. Lindi never got one, but, um, yeah, it was, it was really special beast because it was just one of the first desktop form form factor arm devices. And the CPU was really aimed towards, uh, low power, but heavily multi threaded applications. So it has like codex a 53 and, but it had like 24 of them. So you could compile a kernel, you could like fully, uh, force that CPU to go 100% and you could still put your hand on the CPU and it would be barely warm. But yeah, their, their main, I think goal was for a lot of AI and, and video encoding stuff. So there, uh, 2017 already. Yeah. Yeah. Hong Kong connect. If I remember correctly, but yeah, the idea with that was to just have the CPU push data on to the PCIe devices. Um, yeah. Uh, so it's been a dream of, uh, I guess all the guys at the narrow to, to develop on arm devices, right? Yeah. Yeah. I'm, uh, I've been using for last couple months, uh, M1 MacBook Air, but it's just, it's just in the Mac OS, which is a little bit weird thing to play with. But, uh, you've heard of people also trying to port Linux to it. Yeah. So, um, there have been work. Let me share my screen. There it is. I have, I have it on my screen. So there, uh, a Sahi Linux has been a work by Hector Martin to, uh, to get, uh, Linux ported and upstreamed properly, you know, in a proper way to, um, you know, to the M1 and they have been many updates, but it's not like super usable right now. And, and sometimes it's usable. I haven't, um, really, how are you going to make it work? Um, I mean, at the end of the day, it's just code compiled to run on an architecture. So they have to start really at the beginning, see what the, um, you know, how, how do you get? So luckily, um, Apple doesn't like super hardlock their bootloader. So like, officially you can still unlock its bootloader and run your own kernel on it. It's just, uh, like, you know, figuring out how to get the M1 specific stuff into Linux. So, yeah, that's sort of the work being done. There's also Corellium. They've also done some work on it. Um, I think Corellium's build, um, works already pretty okay, but, um, not sure if it's 100% like upstream friendly. So they have an entire video on it. So there's like Ubuntu running on, on, on the M1. So yeah, two projects, uh, hopefully both of their efforts can be combined into a proper running. Um, I remember seeing a QMU, uh, uh, Twitter posts that were, um, uh, talked to a bunch of people talking about that, but that's just, uh, that's virtualization or what do you call it? So Apple already supported virtualization out of the box and to, to run Linux clients on it, uh, Linux VMs. And then Alex Graf, um, just ported that, did a few changes into QMU, built it for Apple and was able to run, uh, over QEMU instead of using Apple's, um, Apple's, uh, the Apple's app for virtualization. I don't know what they called it. Do you think, uh, when is the next Linux is that in April? So the ones coming up in March is going to be virtual and then the September one is also going to be virtual. So yeah, but before March, uh, March, you say, we're in March. So it's like how soon? Uh, connect.linar.org. Let's talk about that as well. Um, yeah. March 23 to 25th. That's kind of like in a week. So it's too soon to get, uh, uh, a MacBook Air to run, uh, native Linux for all the, uh, in our engineers too soon. I don't know why, because I don't think that our engineers will really like the MacBook. It's limited in its memory. It only has like 16 gigs. Um, we, we really need a beefy system. So like, um, I've, all my systems are 32 gigs and up. I only have eight gigabyte RAM on my MacBook Air. I'm kind of like imagining that it would be good enough for, uh, 4k 60 video editing because I'm imagining that somehow they, uh, they don't really use the extra RAM for, I mean, maybe they do, but when you multitask, I guess, but, but otherwise they swap with that 3,000 megabyte per second built in SSD. That allows them to do a lot. It does, but it shortens the life as well of the SSD. Yeah. That's, that's what some people are saying, right? But, uh, I kind of think that it's, they, they, they, hopefully they have, I don't know, some kind of magical SSD that really has a long life or that's maybe, you know, like, uh, so shipping something that gets bad after a year or two. Um, because like for watch, watch zero, the Apple watch zero, the support was barely a year. So, uh, first gen devices, um, I think it's, uh, okay to assume that the support might not be too long. Uh, but like the M M one X Apple or M two Apple might be a better, um, thing. I'm ready to join the class action lawsuit. I'm good. I'm a, if the, if my MacBook Air M one stops working, I'll just suit them to get it with a 10 million other Apple fanboys. It'll be fine. I don't know. It's just some weird articles on the internet. People saying that, that the SSD is being used too much. I, I, yeah, because like the problem is the really good SSDs, the SLC or single level, SSDs are only with Google and Apple and some with Amazon because they are solely used for, for, for cloud providers for the storage. Um, it's, it's like you don't even get them at a regular market or like Apple can't get their hands on them. Samsung can't get their hands on them. We'll see. So I'll be very surprised if the, if the flash is a single level on there. Here's a barrow on my, my top search results. Oh yeah. Do you, do you know what he's doing? Uh, I don't remember his latest stuff. I know he left Leonardo about a year ago. Yeah. Yeah. It'd be cool to film an update with Barrow. Check out what he's up to. Yeah. What, what's his latest projects are because he liked to, it was a 2017 and he was building in the ArmDex. They say they were using the Macchiato bin, arm desktop. Yeah. And there's also videos with him doing arm laptops and stuff using call comes. Yeah. There's also the Thunder X, you know, like, yeah, I've seen this. They had lots of traction when it comes to uh, uh, like the arm server world and stuff. They had acquired by Marvell, but then they, they're not doing the custom CPU anymore or now they, they kill Thunder X3 and it's just like, I think a la carte. So if someone like a big player like Microsoft wants to go, Hey, we need a bunch of arm silicon for our, our main, our servers and our, our laptops, they'll put the Thunder X3 technology in that, but they're not selling Thunder X3 as a separate SOC right now. So they kind of kill that project already. Sadly. I was thinking maybe the, the reason they did that was, uh, uh, to, to just make, you know, like, uh, arm, arm cores and not make them custom, but they haven't announced what they were doing. No, they haven't. No, they haven't. As far as I know, the Thunder X3, okay, don't quote me on this, but I think the Thunder X3 team has been laid off already. There was a big layoff at Marvell, if I remember correctly. Yeah. But maybe it has to, had to do with the customizing. I'm not sure, but maybe this because there hasn't been any updates, but yeah, the projects were really good. Yeah. That's so many videos on the Thunder X over the time, but it doesn't mean that the, the ARM servers are not going gangbusters right now, like, uh, even ARM, ARM announced a bunch of cores that were optimized for mega servers, no? Yeah. N1, N2, V1 and E, E something as far as I remember correctly. That was another one. Um, and then, uh, we, again, we saw Graviton 2 with Amazon and then, um, and then with, with, with Altair, Altra, the Ampere Altra, that's been released recently as well. So here's the Ampere Altra, which is, uh, which are the companies, was it, uh, uh, Nutsisco or some kind of huge company is saying they have a huge deal going on with, with this chip. Uh, is it Dell? Is it HP? Gigabyte already has announced new servers with, um, with the Altra, uh, server motherboards. So Mount Jade is their older platform, not older, but like the initial platform, and then there was, uh, Mount Snow, which released recently, and that's, that motherboards made by Gigabyte. Um, I, I'm not sure who the OEM at the end of the day is. Maybe it's HP, maybe it's Dell, um, but I'm hoping to see deployment soon. Yeah. It'd be nice to, to see some, some big, big stuff. Uh, you muted yourself. Oh, sorry. Yeah. Can you hear me? Okay. So, so the Ampere's vision and, um, uh, Renee James was, uh, like a vice president or president, uh, over there at Intel before and there, so they're going big. Uh, there was this, uh, cool project. Uh, John Masters was part of the Nubia, right? Yeah. It's not like, uh, really intriguing to come and imagine what they were doing. It was very intriguing. Um, I, I don't know how to feel about them being bought up by, um, Qualcomm, but, you know, things happen. They, they, they need money to survive and, uh, Qualcomm's not that bad of a place. Um, if, if Qualcomm's really smart, they'll use Nubia tech in their laptops. And if they're really smart, they'll let Nubia also make servers and get back into the server market. Maybe, uh, maybe that chip that, uh, Nubia was working on is, uh, optimized, maybe not only for the data center, maybe it could also help Qualcomm challenge the M1 and the laptop and desktops, which makes really real sense because the folks at Nubia were the ones who made the M1 chip. Oh really? The, the, yeah, the X Apple, all of them, all of the top tier ones. Not John Masters. No, no, no. Not John Masters. Yeah. Um, so, so, uh, maybe, uh, maybe let's get back to your, your, your channel or, um, do you want to talk about some of the latest stuff that's happening in the 96 boards? Yeah. I can actually, if you, if you are on my screen, um, that's a drone. And so that's some of the newer things I've been working with at 96 boards to try and get a drone platform out. Uh, so 96 boards, drone platform. Um, we have a few prototypes in production. Um, and hopefully I get them soon enough. And, um, yeah. So I've been building drone, testing it out, just understanding what the PIXHawk ecosystem looks like. So, uh, and, and hopefully soon we have a PIXHawk based 96 boards drone board out there. So, so what is the chip there going on on this one? Oh, this is a PX4 or PX5. Um, this is from PIXHawk. It's a, it's a module that we're going to use later on. Uh, so what is, uh, what is a PIXHawk? Um, PIXHawk. I'll just speak. PIXHawk. Is that a ARM Cortex? Oh, PIXHawk is a entire stack of hardware, uh, built to run, um, and control drones, uh, autonomously or not. More often like an autopilot flight management system. Um, yeah. And, uh, they have their standard. So I have some of the HolyBrow stuff. Um, HolyBrow is like, um, one of their vendors. And, uh, they sent me a V5X, um, module to plug that in and try it out. And then we'll base our boards out of that. So, yeah. Yeah. Remember, I think I did this video right here. Uh, I think they were at the ARM booth. They were showing. Yeah. Kind of like they're doing an open ecosystem in, uh, around the drones. Yeah. Yeah. Because they, I think they use Cortex M7 now. They used to use M4s. Um, so that one is based on, PIXHawk 4 is based on an M4. And I think PIXHawk 5 and V5 forward is based on Cortex M7. So, yeah. And there, this was, uh, three years ago or something. There was a bunch of drones. So they, they had stuff running Intel, Qualcomm, uh, I don't know, a whole bunch of different, it's like an open ecosystem kind of thing, right? Yeah. It is. So they're, they're, um, the designs are open. The specs, specifications are open. The, um, uh, the firmware is open. So you can even go in and add your own. If you build your own controller, you can add that at PIXHawk firmware onto it. All right. Uh, how about some other examples of, uh, the latest stuff? What's, is there, uh, like, what, how would you define the 96 boards ecosystem of the last, uh, I mean, it's, uh, how it's been going on for four or five years now or even more? Yeah. I think, um, we are maturing a lot. Um, we are also coming very close to a new, a review, a spec revision. So, um, look at, look out for that in the coming months. We are, uh, revising our, um, CE spec or our consumer edition spec, like this board right here. Um, and adding some new pins, new buses and all the fun stuff. So yeah, uh, we are maturing. We are partnering up with a lot of big players. Um, and hopefully we see a lot of new boards. Um, so CE spec has been announced. Uh, what's going to happen to it? Uh, no, it's in, it's in the coming months. I can't say exactly when it's a surprise. And what you just showed was, uh, like a surprise or? Oh, no, this, this is a CE V1 spec. So, um, this is the Avenger 92. Whoops. It's okay, right? It's drop proof. Yeah. Um, this is the Avenger 96 and it's based on the SCM 32 MP1, uh, module by DH Core and this one's made by Arrow. Um, so yeah, a very IoT focused board has a dual core Cortex-A7 and then I think a Cortex-M4 in there for real time stuff. So it's basically like the size of 296 board CE. Yeah, this is this is the extended spec. Um, and this is the normal, the, the non extended spec. This is the RB3, uh, which has a Qualcomm, uh, Snapdragon, uh, 845. So a very, very, very powerful board. Uh, this thing can like clock up to 2.7 gigahertz, uh, on, on, on, on the top four cores and the little cores go like 1.8 or 1.7. But yeah, really nice, uh, really powerful board. This one. Um, so this is the CE spec. Uh, this is the CE extended spec. The form factor remains the same on, on the revision. Um, and it will be backwards compatible. Uh, but just new pins and new, new, um, new buses to be introduced much more high speed stuff. I remember some of the, um, uh, talk around the, the 96 boards when it was introduced originally was, uh, the stuff like having a standardized, the standard, uh, charging connector or what was the other things people are thinking about? The power connectors standardized, um, all the port, port layout is standardized. So we're not touching any of the stuff that's already standardized to standardized. We're just adding things on. Uh, people needed USB three broken out. People needed something with you are, uh, broken out on the board. So things like that are getting into it, uh, into the new spec. Um, a, a 13 tech says hi. Hey, what's up? Uh, a 13 is the opener. No, it's, it's, it's his favorite SOC. Uh, there's a question zooming in there. PCIE. Yes. Yes. Okay. Um, so, so what else could you talk about? Maybe you can show your, share your screen in terms of, uh, what do you like to talk about here? On your, yeah. So amongst the newer things, um, so last connect last physical connect September, 2019, we showed off the RB three robotic arms. So it was a robotic arm based on the RB three, but I just show, um, I just showed here and, uh, this board, uh, this robotic arm was, uh, basically a tech demo to show the, the power of the RB three, but we are not going to improve upon it and let's see if I can get a close up. So we are now planning on improving upon it. Uh, and to, to get like a better version of it, much more smoother running version of it. Why did it fade to black? I don't know, but yeah. Uh, so yeah, trying to do new stuff with PID controllers, uh, smoother movement, more AI stuff indicated onto it. So yeah, that's sort of something I'm looking forward to work on. Yeah. Apart from that, really excited for the new spec. Um, there are a few new boards that I can't really talk about. Do we get a preview of CE spec today? Nope. Nope. It's not in my hands. You don't, you don't get, get to see what the CE spec two holds. It, it'll be announced very soon. Like, uh, we just said that there was a, uh, Lenaro connect virtual event next week. So who knows? Who knows, right? Yeah. Yeah. Um, and, uh, if, if you go back to your share screen there, I mean, you've got so many videos. Did you, did you mention before how many it was? No, wait, I'll have to check. Not that many. Uh, I think when I search on YouTube, I can see you have 500 videos right now. Yeah. That's around, around what I'm saying as well. And like, what are some of the coolest videos you've been post having over the, let's say last year, what you, what could you say? I think one of the coolest ones was, uh, building my new workstation as well. It was really good. Um, good, uh, good feeling project because it was just the start of pandemic. You couldn't get anything. So there was a lot of stuff in, okay, source this. We're getting this now. The vendor says he has it. Just rush to the shop, go get it. Can you make it full screen? Yeah, sure. So what are we looking at here? So this is Ryzen 9, 3950X, 60, uh, so 32 cores, 16 physical cores, rest logical, around four gigahertz all core and, um, uh, 128 gigs of memory. Um, and, uh, and, and a lot of SSDs on it. I think there's like right now, two plus one, three plus three and a half terabytes of SSDs. Um, so yeah, it's, it's a beast of a system and all this is like 32 giga, 32 gig sticks each for 128 gig. Um, yeah. So all Ryzen platform, PCI for goodness. Um, I do have a weaker GPU, but just getting a GPU is impossible these days. So I'm just using something that I've got. Um, yeah. So this had been an interesting build and, uh, yeah, it's, it's just been so easy to work with these days. The AMD guys are doing a good job, uh, destroying Intel in the x86 arena. Uh, I'd say the Intel guys are doing a poor job to show what they're capable of. Oh yeah. You think they can do more than what they're showing? I think Intel can do a lot more. It's just like poor, uh, poor leadership, uh, past few years. Um, I think the new guy should really handle it better. That's a big, uh, it's a big heat sink, big heat sink. The heat sink is capable of dissipating like 250 watts. You could, uh, totally keep your ice cream cold with it, right? Or something. Yeah. All right. So does that mean you can build stuff faster? What can you do with this? Oh yeah. Uh, Android builds in an hour where it was taking me four hours earlier on my Xeon laptop. Um, what else? Linux kernel takes barely five minutes. Entire Yachter builds are done in an hour or less, actually much less 30 minutes or something like that. Um, also like Android builds these days takes huge amounts of memory. I've been hitting like 65 gigs already. Yes. I've not, I've, I've, I see the question from 813, but, uh, I have not done that test yet. Are you talking about this one? Yeah, no, there's a new one called how many kernels per hour you can build. Um, they just came out. Yeah. And, uh, that was a benchmark, uh, written by Marchion. And I haven't run this on, on, on the system, because honestly, who cares? Like it's so fast. Nice. I'm just going to open one of your videos. Uh, when I go on this one, uh, what's happening here? Uh, you remember this one? Do you have the title? Uh, x86 emulation. Oh, yeah. So, uh, so the guy, so Alex Graf, who wrote the queue, who did the QEMU stuff on Apple M1 many years ago. Um, when they were all the whole UEFI scene on arm 64 was really coming up. And around the time the developer box was released, uh, Alex Graf wrote a QEMU emulator within the UEFI bootloader. Um, so what it enables you to do is to run x86 EFI drivers for your GPU or your, or your network cards. And run that on an arm system during boot up. So that means if you have a network card that only works with x86 systems, uh, you can actually pixie boot using that network card because the EFI drivers loaded and emulated. But what does it run on? Radeon. Radeon. What, what are you looking at here? Uh, that's a USB. I think which, which one is it? Yeah, that's, uh, that's one of the USB, uh, hubs. I had some issue with them. I was just talking about something else for a moment. Right. Yeah. Yeah. Um, so we go around your channel. Sure. Um, is it fun to do this YouTube channel? It's, it's fun. It's kind of been, um, less rewarding recently because YouTube has been, you know, upload every day or you don't get paid as much or we'll cut your views and blah, blah, blah, blah, blah. So it's just been like, I only upload when I really want to upload something sadly enough. Like I'd be happy for a platform that promotes my, oh yes. Um, that's my, uh, folding at home, um, farm. So another interesting story if you can switch me over to, to my screen share. Yeah. During the start of the pandemic, uh, few friends and I in the armsy ecosystem did some work to port a Rosetta at home and eventually folding at home to arm. So, um, we worked with the Rosetta at home team and the holding at arm or home team to, uh, make sure that we can have our arm devices crunch, um, protein folding sequences to, you know, run, uh, on, on, on, on, uh, and, and, and contribute to the coronavirus vaccine related, um, research. So, uh, this is actually the first image that, um, verified like a proper for protein for being done on at the actual arm image. Um, this is like my arm cluster, which I run my folding at home stuff on. Um, and there's just, just the Seattle system, uh, that I used to develop this whole, um, used to put, uh, Rosetta at home and, um, so equinox and metal, we have friends there and they lent us like a few arm servers to run, like, um, run, run the protein folding sequences on their arm servers. So, there's like 64 core arm server running poor protein folding here and then I don't 32 core arm servers, couple of them running protein folding right there. Uh, a packet server still crunching. Yeah. Yeah. So packet servers are still crunching packet is now equinox metal. So yeah, those I think of my servers are gone. They had to be repurposed, but I think a couple of servers are still crunching. Uh, because when we're talking about, uh, 250 billion arm processors, I don't, I don't remember how many they are, but it'd be nice to, to have some kind of option to use them all and help make the world a better place, right? So like some kind of switch and then we have a supercomputer that's distributed. So one of the folks who helped us in this endeavor were folks from Neocortex and, uh, they are trying to do this thing. So you just download the app, uh, and you give them, uh, resources of your phone when your phone is not in use, uh, and you can actually get paid for it. Um, so yeah, you can actually run folding at home and Rosetta at home on your phone, uh, just download the app. Nice. Because how about all these IoT devices everywhere and, uh, like security camera or something should be helping, helping them out? Does that make sense? Eventually, but, um, sometimes like the return on investment, like if the process is too slow to even return the, uh, protein, uh, math in, in time, uh, then it's of no use. So there's obviously a low end, uh, low threshold to it, but do you have cell phone, uh, mobile phone style processors are just good enough? Because, uh, right now there's a big, big news. The, the Fugaku got turned on, uh, three, four days ago. Um, and this is, I mean, this could be, I mean, this is an arm, a supercomputer that's like five times faster than the second biggest, uh, which is a power PC. Uh, and they, they, they, they were presenting the part of the narrow, right? The, uh, yeah. So they've been working with it now to try to optimize everything of the, I mean, this is over there in the, uh, we call it the enterprise, you know, enterprise, right? Uh, no, no, they are part of the narrow HPC or high, high performance computing group. So that's separate from enterprise. And, um, yeah. Um, it's, uh, it's a different sort of beast because the way it uses its processor, it's, it, it has like, um, 48 cores, but only two cores run Linux and, uh, and I like assistant cores and like the rest of the cores are, are, are running their own, uh, actually, actually just there for the supercomputing math using, uh, neon and other SIMD stuff that they developed. Um, and then they have all of that connected to like an HBM to, um, Ram buffer, uh, very close to the CPU. So the memory transactions also like a terabyte a second. So yeah, a different beast, a very purpose built beast, but it's, it's a beast nonetheless. Yeah. Because it's, uh, it was this event they had just like, uh, just recently where they, they invited the Japanese media and they, they, they show this gigantic huge supercomputer and it switched on and maybe it's going to help, uh, you know, like predict the tsunamis and the earthquakes and the, the, uh, understand the climate change, the weather and, uh, I mean, there's so many, and maybe help in the, understanding the corona stuff and all the, uh, like medical research and everything. If we can, if, if we can predict pandemics, that would be great. Just, just let me know a year in advance that the world's going to shut down after a year for a year. I'll be fine. I'll prepare a year is enough, but don't just go, all right, we are shutting down today. I'm like, what? Yeah. Yeah. One would think that maybe those supercomputers could have helped a little bit more, uh, manage this weird corona time. Oh, I should have used the, the word because Google is going to downrank the video. Uh, the, this is like, it looks so big. Like it goes all the way down there and they all, the biggest arm CPU has ever built like so much performance in there, right? And the chip is so big you touched it, you hold it, you held it, right? At the, yeah. And I don't remember how many there are. There's like two on each board and I don't know how many boards there is in each rack. Uh, so that's, but when we, when we, when we look at this amazing supercomputer, maybe what you're talking about is maybe doing a supercomputer at home with all the different Raspberry Pi and that's, that's, yeah, that's, that's what Boeing is. That's what the distributive computing sector is all about. Yeah. All right. Uh, how about, uh, uh, what's the latest news in the arm world? There's been so much consolidation, uh, different companies buying other companies. Yeah. I mean, it's just we, things move forward. Um, we can hope for less, less fragmentation now. Hopefully the few companies that own most of the R my P, um, are, are, are going to, you know, just come together and make a nice pack to do things a particular way. And not everybody just doing their own things. So I'm hoping for less fragmentation, um, but in a fragmentation. So there's like constant pressure and constant competition in the market. Um, yeah, the whole arm on NVIDIA thing is funny. Uh, you know, no one wants NVIDIA to buy on, but like NVIDIA really wants to buy arms. So everyone's like, please don't buy arm. I think I, especially China is a little bit nervous, but I guess other people, China has problems with the US blocking them accessing a bunch of stuff. Uh, so if the Americans own arm, they might block even more. They might like disconnect, disconnect China. I think China's situation is a bit different because arm never directly dealt with China. Uh, they did it through the arm China organization, which is a separate ish entity. It's partially owned by arm and then partially owned by like Chinese companies. Um, but then I think UK is also having an issue because their entire arm IP goes into the US hands. Um, but, uh, and then all of the arm vendors have issues because then NVIDIA is also an arm vendor and they fear, uh, non, uh, non-competition competition friendly practices from them. So, you know, we'll see how it comes out. But it was, uh, it was, uh, it was kind of a big deal when SoftBank came and just bought arm to save them from Brexit kind of. That was my feeling of what was happening there. I was like trying to get out of the whole UK mess, uh, maybe in terms of ownership, but then this, how do you call it? I don't want to say the word Coro, uh, this thing happening right now that's happening in all the, oh, I guess the big invest 100 billion dollar investment they had got like challenged, right? So he needs to, to upload arm. Yeah. There's, there's some thing on, um, on, uh, SoftBank not making the right decisions later on, but, uh, yeah, arm was one of their money making, um, branches in a, in a way. And, um, yeah, they sadly had to let it go. Yeah. In a perfect world, I'd have arm become its own company and like be listed and all, but, um, and we, I'm not too, too concerned about NVIDIA buying it, but let's see where this all goes. Maybe they should just, uh, hand it over to Leonardo and you can, you can manage the future of arm over there. Uh, somebody, uh, if it has to be the EU or UK or somebody, uh, this also AMD, a chorus Linux, uh, Xilinx Xilinx. Yeah. This was, this is interesting and I'd like, and I'd very much like this to happen. They make some, uh, some nice, um, what do you call it? FPGAs. Yeah. So that's, uh, they, they paid more than arm for it. Mm hmm. Xilinx is worth a lot more, um, than arm, honestly, because they've been longer, you know, they were, they were the ones to invent FPGAs in a way and they just, they just have a lot of stuff going on for them. And, you know, it'll be good. They'll have some room to breathe and, you know, it should be fun. And when we look at, um, uh, the, the Graviton 2, how big of a, how, how big of a push has that done helping the arm take over the whole cloud server market? Quite a lot in the server, server ecosystem because there was like, there had to be one major player showing what, how much power benefit you get, you know, how, how much, uh, performance for what there is in an arm ecosystem. So they actually just showed the world that and people are saving money head over heels. Um, and just, uh, just, uh, just a good, good outlook. And I'm happy for, for that to happen. I hope Netflix, Netflix maybe is running on the arm now. It's just some examples of some companies that might be using it right now to save a bunch of, oh, arm is using arm. That's good. I wish, you know, it would be very weird if they don't. Yeah. Oh, they prefer to stay on the x86. Now that'd be weird. I wish I could run my arm devices that I, I should just click on something and run it over there. Um, on a little, what you call it instance, right? Yeah. And here's get 40% better, better price performance using arm. So it's been kind of like talked about over the last year. And I don't know how much that gets talked about over there in the narrow a bit, not a whole lot. You know, when it came out, everyone was talking about it a lot. Um, I'm sure there are folks working at Lenovo who are working much closer to this. But yeah. All right. I think, I think cloud's going to go to arm pretty soon. Hopefully just we just need more vendors pushing silicon out there. Do you want to talk about some other stuff on your YouTube's? Um, I think I'm good for the moment. But yeah, I just just been doing some random stuff as well. Um, you know, what else? I have some like old systems now that I sometimes take out and just upload a bunch of videos on them just trying stuff out. I have this, I have this very interesting one, which was the VRC3. It's one of the few non AMD or Intel vendors of x86 CPU. So they've licensed the course from Intel themselves. And then they just put that in there. It's a weird system. Not a whole lot works on it. But it's just a nice niche product to have. Just a random early express find. All right. Yeah, so a lot of fun over there. And also the the how's it going to go in the open hours stuff from the 96 boards channel. So openers was a bit of a different beast to handle for me. What ended up happening? So if you don't know, Robert actually left Lenovo and joined join joined arm. And he has been doing his own streams over there under innovation coffee on their own channel. And they have they have a momentum with calling people in for interviews. And it's always two of them. It's always Alessandro and Robert doing these things. Like Robert and I used to do for open hours. And it just ended up being just too much for me because I'd be sitting here. I'll be talking about something. And because the stream was live, no one would really talk. And open hours was always built by like having the community in the call and being a part of the conversation. And so that eventually stopped happening. People were shy enough to, you know, in talk during things were live. So I ended up making it to preserve the community into like just have that conversation going. I ended up making a private a non live streamed call. So open as what I do we meet four p.m. Thursday, UTC on zoom, there's a few of us six to 11 people that meet, we talk about arm a lot. And then so I had a guest this last week from Colcom. So I'll be I have that recorded, I'll edit that and I'll upload that and was a nice talk. So whenever guest comes, I usually have a same offline call, but recorded and then uploaded later on. So how can people join the zoom 96 boards.org slash open hours and you should have the name right there. All right. So right there, open out 96 boards slash open hours. Next one is in four days. Yep. Maybe five days. All right. Cool. And so maybe do you have some cool boards you want to show like in your online. Yeah. So one of the boards I have here is the 96 board some addition. And you're going to see more or some more projects out of this soon enough. This is one of our newer specs. So we have some spec now. And this is a some module itself. This is the rest is a test board to test everything out that's broken out on this particular song. I can take the song out and show it to you. So that's the 399 again, very nice. So I think this is the pro variant. And then on here is the main song connectors. And then the 96 boards connector and then a bunch of other stuff at the back. So yeah, this is one of our newer specs. And hopefully there's going to be a lot more content around this particular song module. So who's making it? I think this one is by hope run or no be be quick cloud be quick cloud. No, come on focus. Right. So that that's 3 3 9 9 is getting some play. Yeah, 3 9 is a very famous SoC. My my Samsung Chromebook Plus is still my my number one favorite. Yeah, for like just browsing and stuff. Yeah, I mean, it's hard to say that now that I have an M1, but I mean, it's totally a better value, I would say. Yeah, I have my ASUS one again, our category 99. And this one's this one is a flippy boy. So I can just like that. I would have it right here. So yeah, again, our category 99 works everywhere. Nice. So I'm just switching my battery for a second. Maybe show more stuff you have. Oh, I have some mezzanines on the board on the table. I have a dragon board. This is the LTE mezzanine again from like just plug this on into 96 boards and you get LTE connectivity. What else do I have on my board on my desk? A Hike 960. These are a relic now because of the trade war, they aren't really being built anymore. Do I have more 96 board stuff? Why they're not being built anymore, you say? The trade war. Ah, yeah, that's weird. Yeah, it's a little bit weird. So so they're blocking off high silicon. Not exactly because like this board uses a lot of components that's either shipped or made in the U.S. and the U.K. So like they can't get components to actually put on the board and then sell the board. Oh, yeah, they can't get the other stuff. Maybe in there. Do you have more? More cool stuff? More cool stuff. Something that's not secret. Secret. So this is another cool stuff on the face. It looks like a regular GPU. But then I made a video recently explaining how you can make this an arm only GPU. So now if you have x86 machine that has UEFI boot loader, this will not show you the boot menu. But if you have an arm machine that has an that has a UEFI boot loader, this will show you actually the boot stuff because I have flashed its internal internal EEPROM or SPI flash with UEFI drivers that AMD released, but they are ARM64 specific. So I think this is one of the more cool things. I really like the GPU because if I have an arm server, I can just plug this in and it just works as a display. So that means you turn all the headless stuff into you add a display to the headless stuff. You add a head to the headless. How about you showed this earlier? Yeah, I already showed it. Yeah. You want to show it again? No, I can switch over. Yeah, in the beginning of the video, you were doing stuff. This is live from your... This is live from my honeycomb. This has a 560 GPU. I don't know. Let's play some mind test maybe. Let's see if it works. Hello. So what does the mind test do? It's a Minecraft clone, but looks like it's running into an issue again. I'm running Fedora 34 beta and I'm just having issues with it. Sadly, no mind test today, I guess. You see there and it sort of crashed. That's just a thing. Let's try it again. Maybe it works this time. What did I click? Yeah, there you go. And the clouds are red. Not sure what happened. Again, this is like running on ARM. Also very dark because it's nighttime in the game. Let's make a new map. What does it do? The mind test? It's an open source Minecraft clone. And you can do all your Minecrafty things. I think I need to make an axe or something. Let's find some trees. So yeah, but yeah, because it's open source. I mean, Minecraft also works with the Java version, but because mind test is open source, I can just compile it for ARM and it will work just fine. Coder hex says I am here too. Hello. All right. So how do you enjoy your job working with all this Linux and ARM stuff? I love it. Have you been upstreaming a bunch of stuff? Not too much recently. I'm looking to upstream a few boards, but it's all been a lot of demos and stuff. Just building demos, trying things out. He says greetings from Turkey. Hi, over Turkey. I hope the weather is good. All right. So yeah, that's a lot of topics. We did an ARM video, but there's so much other stuff. Wait, what's that? Why are you magnified? I was out of focus. I refocused something like that just to get my smile bigger. Yeah. All right. For the narrow employees, it's a little bit like a normal time. This corona, you're just always working from home. So what's the difference? We can't have conferences anymore. I just miss meeting people. So yeah. Yeah. No real connects. No real connects. Yeah. You need to get one of these VR things where you can create in a VR space. It's not really fun. It's not that thing. I can't really take big projects and show them. It's so difficult to demo something. For the longest time, my projects have been a lot of physical stuff. LEDs, motors, servos, and it's how do you demo something like that? You just have a pre-made recording. But I want people to take my project in their hands and experience for themselves. Can't do that. There's no demo day Friday, nothing like that. Yeah, that's sort of been the tougher part of the last year or so. But let's do more videos like this. Sure, I mean. If you want. Yeah. As soon as you have some cool stuff to show, let's do a stream yard or the other platform live remote video chat. And then we can do it the best possible way. Maybe if you have a space in your desk that's like you want to show. Is that any two secrets? You can clear like a corner and we can do like a set a nice camera. Yeah, I actually have a desk come. Maybe it shows up. There you go. Oh yeah, perfect. Yeah, so there you can be showing stuff on my channel if you want also. Just to get some more views on the stuff. That'd be cool. All right. Kodur was also saying, I still prefer IMAX 8 board. Yeah, we have a few IMAX 8 boards. We have the IML and Tor 96. The IMAX 8 is here to stay. It's not going to stay. No, it's not the IMAX 7. Oh yeah. It's going to be like 10 years of support or something like that. Right. This is going to be like forever available. Yeah, I hope so. Yeah, that's what they usually say. Yeah. Why Tor 96 is so expensive? Ask Arrow. I mean, there was some mix up at Arrow and all the boards are so expensive now. We are working on it to get the price down. They'll be regularly priced board soon. They'll all be $35, right? No. That's unrealistic expectation made by like one company. And yeah, it's not easy to follow that. Let's say ARM 64 dollars. ARM 64 dollars, sure. Something like that. Okay. All right. Cool. So that's awesome. I mean, there was so many other topics, but sorry, I didn't really prepare the articles and everything because I'm sure you have, you have, you tweet a lot, right? You tweet about everything. I do. So this, this your Twitter right here, let me get on your Twitter. And people can follow you on the Twitter. And everything that has to do with anything, you have opinions. Yeah. Here's happening something. Vulcan compute. Rad V. Yeah, that's on the honeycomb. What's this guy? This is a perfect packaging for potatoes. Okay. Nuts. Ah, for nuts. Yeah. What is this guy doing here? Okay. But there's a lot of stuff on your Twitter and geek till it hurts on YouTube. Yeah. And, and it says here, do you have an FB account? What is that? No, I'm off. Fan, fan, fanboy, fanboy, this, what is it? Facebook. Ah, Facebook. Yeah. Okay. You're off the, I'm off Facebook. Yeah. Yeah. You don't want Zuckerberg to track you. No. Okay. It's not just that, like, FBs become bad. Like the content on there is bad. I don't like logging into FBNC and stuff. What is this guy doing? Okay. That wasn't real, right? It's, it's real. Yeah. It's a real footage from World War II. Yeah. That's a real footage from World War II. All right. And this guy is running in 96 boards on Mars, right? Maybe. There is a Snapdragon 800 on Mars now. So maybe 96 boards soon. Who knows? Do you think it's 800, 800, 801 or what do you think it is? Isn't it 835? No, no, no. It's older. It's the 32-bit one. It's either 801 or 802. I don't remember. They didn't bring LTE on board? Yeah. I can't get LTE here on Earth properly. And then you want to get that on Mars, sure. Who's this guy? Well, not mine. I mean, I just shared. Did you have another one, right? I heard earlier. Yeah. Yeah. All right. Cool. All right. So thanks a lot for giving an update. Yeah. Oh, thanks for having me. There's a couple comments here going on. The dual HDMI and quantity. What is that? For, for, for the Tor 96 and wired, I guess. And it is 801 on Mars. Yeah. Cool. So all the, all the work you've been doing is, is helping on Mars. Not quite yet. Not my work. Okay. Cool. That's, that's, that's way before I joined Linar or anything like that. Yeah. But I mean, they're all, all the Linux and ARM support is changing the world. Yeah. That's for sure. Yeah. All right. Cool. So thanks a lot. Thanks for showing some of the latest stuff. Yeah. And thanks for talking about it. Thanks for having me. It's been fun. It's nice to be back at the, doing a Linar or connect over YouTube. Yeah. Yeah. So it's a different world now. How about your, your friends that we were doing videos with last time? Money. Yeah. Working also on the 96 boards or? No. So money left 96 boards and joined the Linar or Qualcomm team. So what's he, what's he's doing over there? I think some, a lot of upstream for sure, but a lot of like bus related stuff the last time I talked to him about. The Qualcomm team is busy. Very busy. Very busy. Yeah. Nice. I hope I can do a video. Maybe you can help later and we can do Qualcomm updates. We'll see what, what we can get out of that, right? Because I'm sure there's a bunch of stuff happening. Cool. I just wanted to get to a feature film length. And then we can stop joking. What is the official time? It has to be one hour and 16 minutes. Then we can call it a feature film, but we don't have to wait until then. Okay. Yeah. Cool. Thanks a lot. Thanks for doing the video. Yeah. No worries. Thanks. I hope to see you soon. Yeah. See you in the next next video and thanks everybody for watching. Yeah.