In the video, the creator builds a powerful Linux workstation designed for software development, featuring a high-performance CPU and ample RAM. The process includes assembling parts and benchmarking the system against other machines to ensure optimal performance. The workstation showcases impressive speed and efficiency for development tasks, with future plans to share software setup details.
Introduction
I'm building a brand new workstation, and this time it's going to be Linux. The goal? To build a powerful machine built for software development. It needs to be fast, with a great CPU, plenty of RAM, blazing fast storage, and maybe even some AI capabilities. This rig has to be able to handle everything from compiling code to running multiple VS Code instances to, let's be honest, opening at least three Chrome tabs, or maybe even four if we're feeling bold.
You Might Like: Gaming Lptop ASUS Z13 Review
In thisYou Might Like article, I'll walk you through the entire process from choosing all of the parts, assembling the build, firing it up for the first time, and benchmarking it against some other machines to make sure it delivers. And a big thanks to Micro Center for sponsoring today's video. I even stopped by my local Micro Center to grab everything that I needed for this build. So let's get started.
Choosing the Right Components
CPU: Intel Core Ultra 7
The first one is my CPU, and it's the Intel Core Ultra 7. A powerful CPU with 20 cores. Now eight of those are performance cores and 12 of those are efficiency cores. And it turbo speeds up to 5.4 gigahertz. It's built for multitasking, compiling, and even AI workloads thanks to its onboard NPU. It also supports DDR5 and has 24 PCIe lanes, which is four more than its predecessors. This is a huge upgrade from my previous 8th gen model.
Motherboard: ASUS
Now that is going to be seated inside of this ASUS motherboard. It supports DDR5, up to 192 gigabytes of RAM, PCI Express 5.0 for next-gen GPUs, and four M.2 slots for plenty of fast storage. It's packed with features like Wi-Fi 7, Thunderbolt 4, 10 gigabit USB, and 2.5 gigabit NIC, which is plenty fast for me. I think this is the perfect motherboard for my build.
GPU: RTX 3060
For the GPU, I picked up an RTX 3060. Now I picked this one up specifically not for gaming. You might be wondering why a 3000 series and why this model specifically. Well, if you're looking into getting into large language models or ML, this is my "budget" pick because it has 12 gigabytes of RAM, which is great to store those LLMs. This makes the 3060 a good choice because it has 12 gigabytes, and the more RAM you have the less swapping that will happen, which means the faster they will run.
Cooling: Noctua CPU Cooler
To cool it, I'm going with air cooling and going with this Noctua CPU cooler. I typically go for larger CPU coolers. However, since this CPU runs much cooler than its predecessors, I want to keep this as quiet as possible. I have no doubt that this cooler will keep it cool and hopefully not at the expense of noise.
RAM: 64GB Corsair Vengeance DDR5
I mentioned DDR5, I'm going with 64 gigs of Corsair Vengeance RAM. This kit is 2x32 gigabytes, which lets me fill two slots and leaves another two for the future. Not much more to say about this other than it's going to be fast, and I'll be sure to set the XMP profile for it.
You Might Like: Lenovo ThinkPad T14s (Gen 6, Snapdragon)
Storage: Samsung 990 NVMe 2TB
Speaking of storage, I went with the Samsung 990 NVMe, and this is a 2 terabyte drive. Now I'm a huge fan of Samsung flash storage, and this one does not disappoint. It's blazing fast and has enough space to store the operating system and all of my apps, tools, and along with any projects that I'll be working on.
Power Supply: Corsair 850W Gold
So what's powering all this? This big old heavy power supply. So this is the 850 watt Corsair power supply. It has a gold rating and is fully modular, so I only need to plug in the cables that I'm using, making it easy to keep my case tidy. These power supplies are dead silent. They run so efficiently and so cool that I have yet to see the fan turn on. For all of these reasons, Corsair power supplies are always at the top of my list.
Case: PowerSpec 4U Server Chassis
And last but not least is the case for this build, and I decided to put this inside of a rack mount case. Now cases are usually a personal preference, but I went with one that has more utility than looks. It's not a traditional case that you would think of for a workstation, but I went with this PowerSpec 4U server chassis. It's a 4U rack mountable case that supports radiators and air cooling if you like. I chose to go with air cooling, but it still gives me plenty of room for drives, components, and even a full-length video card. There are quite a few configurations for this chassis, but since it's going to be a workstation that I want to be quiet, I'm not going to install any additional components.
This case will fit nicely inside of my studio rack where my other Mac and Windows workstation will be. If you're not into rack mounting gear, this build will easily fit into any other standard ATX case. Before diving into assembly, a quick shout out to MicroCenter where I got all of the parts for this build. They wanted me to let you know their holiday sale is still going on, with deals like the AMD Ryzen 5 5600X for just $99, the 12th gen Core i5 for $119, and even $120 off the CPU I'm using today in this build. Plus, they're opening a new store in Santa Clara in 2025, so sign up now for early access and get a free 128GB flash drive when it opens.
Assembling the Workstation
Assembling the workstation was straightforward, but not without its quirks. The biggest challenge was actually aligning the universal knock to a CPU cooler. Now it took a couple of tries to get it right, which allowed me to check how well the 5 dot thermal paste worked out, but the knock to a cooler fit perfectly in the end. A tip I can never quite remember is to pre-connect your power supply to the cables before installing it into the case. It saves you a lot of hassle later since it's hard to see once it's mounted. One small gripe with the case is that the fan cables are comically short, so I'll be swapping them out for knock to a replacements in the future. And while I'm not a fan of the brand name in the front, that's a problem for future me to solve with the Sharpie.
So now it was time to power it on, but will it boot? Well yeah, after I remembered to turn on the smart plug that it was connected to. After turning that on, it powered up right away. I noticed that it was using around 44 watts of power at idle, which I thought was a little bit high, and I will investigate this some more in the future.
Installing Ubuntu and Testing Performance
Now hardware is only half of the equation when it comes to building a new workstation, and software is the other. For the operating system, I'm going to go with the wrong distribution according to future comments, but I went with Ubuntu desktop. I've been a long time user of Ubuntu, and I can say that most of their software just works, which isn't always the case with Linux. It has a nice blend of features for the casual user, business user, and the developer, like me. I installed all of my usual tools, and most of them work without a hitch. There were a few gotchas though, and I'll be walking through all of the apps and services I use in a future Linux video.
You Might Like: Top Laptop Accessories to Boost Your Productivity | Essential Tools for Your Setup
With Ubuntu installed, it was now time to test. I wanted to test real-world scenarios like compiling, image processing, productivity apps, and even things like using a browser. I found Geekbench that can test all of that and more. And the good news is that it's cross-platform, so I can run these tests on Windows, Mac, and Linux, and the free version does all of this for free, although the results are automatically uploaded to their servers, which is fine for my use case and testing.
When running the CPU test on this new workstation build, I got a single core score of 3,303, and a multi-core score of 20,679. Now I didn't have anything to compare this to at the moment, so I decided to test two other machines that I have here. I ran the same test on a Mac Studio M2 Macs running Sequoia, the Beelink EQ13 with an Intel N200 running Windows 11, and the Minisforum MS-A1 with an AMD Ryzen 7 8700G and running Windows 11, and the results were pretty surprising. My new Intel Core Ultra 7 265K beat out all of these systems in the single core and multi-core CPU test. Now I knew this would beat the Beelink and the Minisforum, but I didn't think it would beat out my Mac Studio M2. It performed almost 20% faster in single core tests and almost 35% faster in multi-core tests over the Mac, which I think is pretty impressive. Now I'm sure this comes at the cost of power, but I wasn't able to measure power reliably at this time.
Next I decided to run the GPU test. Remember, this isn't testing the GPU to see how many FPS we can get, but testing common GPU tasks like blurring the background, face detection, applying a Gaussian blur, and more. While running the GPU test on my new build that has the Nvidia RTX 3060, it got a 96,933 score with OpenCL and a lower score using Vulkan. Again, not having anything to compare it to, I then ran it against my Mac Studio, Beelink and the MS-A1. You can see that my new workstation outperforms the Intel UHD graphics, the AMD Radeon 780M, and even slightly outperforms the Apple M2 Max, at least in the OpenCL test. If you look at the Metal test on the M2 Max, it blows away the competition with 154,471. Now I'm happy with the results for the 3060 GPU, but I'd love to run it on Windows just to see if there are any quirks with running it on Linux. And then on Windows, I can test it with Vulkan too. I also wanted to test AI tasks like image classification and object detection, but unfortunately, Geekbench doesn't currently work with Nvidia GPUs on Linux for AI tests. I'll be sure to revisit this in the future with Windows.
Final Thoughts
So all of those tests are great, but how does it feel overall when using it? Fantastic. Everything is so snappy and I'm amazed at how fast applications launch when clicking on them. Chrome and VS Code opened near instantly. Now I've never seen applications open this fast on anything that I've used, including my Mac Studio, which makes sense because after seeing the test, it did beat my Mac in almost every single one of them. And as a quick gut check, I cloned one of my repos, opened it with VS Code, installed all of my dependencies, and was able to get it up and running in about 30 seconds. This is going to make for a great development machine.
Comments
Post a Comment