Microsoft continues to design and ship new form factors for its Surface range of devices. As part of the start of availability of the Surface Laptop Studio in Australia I was asked to try out one and provide feedback on how it would work for developers. The trial Surface Laptop Studio I was shipped came with an i7 Intel CPU, 32 GB RAM and an NVidia A2000 GPU.
Disclaimer: I work for Microsoft and was sent a Surface Laptop Studio to trial and provide feedback on for developers. What you’ll read here is my honest feedback, but I’d encourage you to read other sources before you make a decision!
Now, with that out of the way let’s get into the meaty stuff!
Since joining Microsoft I’ve been working on Surface Laptop devices and have really enjoyed the experience. Initially I worked on a Laptop 2 which came with an Alcantara cover on the wrist area, and 2021 I moved to a Laptop 3 which has a nice aluminium finish instead which doesn’t attract the dirt! Given what I’m doing in my day-to-day right now the Laptop 3 really suits me, so for the rest of this post it’s really going to be the benchmark!
The usability of the keyboard is a massive part of choosing any device for a developer. If you’re stuck with a bad layout, it can really impact your productivity, and sometimes even physical health.
Previously I spent a lot of time working on a Lenovo X1 Carbon, partly because of it’s form factor, but also because of its keyboard. Lenovo devices inherited the IBM keyboard design which is amongst the best, even if on the Carbon X1 they have the Function and Control keys in an odd configuration!
I mention the keyboard, because I’ve found the keyboard on the Surface Laptop to be excellent to type on, with one main issue: the power button is right next to the Delete key on the keyboard! The number of times I sleep my Laptop screen isn’t funny, and it took me a while to realise you don’t hit the power button twice as a toggle – the second time actually sleeps machine and not just the screen.
On the Surface Laptop Studio I was hoping to find a different layout, but it retains the power button in the same position. Maybe in a future release we’ll see that pesky button move elsewhere!
If you intend to use the Laptop Studio in a docked configuration with an external mouse and keyboard then the keyboard may not be a determining factor. In my previous job, where I would be moving between locations a lot, I never used external keyboards or mice, because you break muscle memory for typing – for me it’s better to use the laptop keyboard instead. Regardless, I find the Surface keyboard great to work with.
I carry around a Microsoft Bluetooth 3600 mouse and I’ve been doing that for ages as using in-built pointing devices on any computer for long periods feels odd to me. Even with my Surface Laptop I still use an external mouse, only periodically using the touchpad.
The Surface Laptop Studio has a nice large touchpad that at first glance appears similar to the Surface Laptop, but which in fact is larger and much more tactile due to in-built haptic feedback. If you’re a trackpad-fist person then I think it will be hard to beat!
Some people will praise the simplicity of the Surface Laptop Studio having just two Thunderbolt 4 ports for connectivity, but my expectation is it will be some time before people have easy access to devices that can connect via those ports.
It really feels like the Surface Laptop Studio continues the trend of needing an external dock to provide all the legacy device connectivity you’ll need. This isn’t a new trend, and I’ve been using a dock for external monitors and networking while working from home and it does make life a lot easier, but it’s definitely one to look out for depending on your setup.
As an aside, I have to say I found the Laptop Studio having a 3.5mm audio jack a little odd, but it’s there if you still want to plug in headphones and/or a microphone!
Pen or not pen?
I don’t understand why there isn’t a pen bundled with the Surface Laptop Studio because I think without it you miss the full experience. If you’re getting one of these computers, definitely spend the extra money on a Pen. The new Surface Slim Pen 2 is also super-tactile and responsive to write with. To be clear – you’re not going to be churning out code using the pen as it’s more suitable for note-taking and diagramming. The handwriting recognition is pretty slick and when the Laptop Studio is in tablet or “cinema” mode the Pen comes into it’s own.
Usually I don’t RTFM until I really have to, and to my mind, if I don’t need to then we’re on a winner with most people. I have to admit to being frustrated with the out-of-the-box experience with the Pen and the Surface Laptop Studio. I’ve used Pens and Surface devices before, but this was the first time I couldn’t actually get the pair working together without having to lookup online how to!
I attached the Pen to the side of the screen (like I have with other Surface devices) and had no luck getting it to pair with the computer. It turns out that the Pen charging location on the Surface Laptop Studio is underneath the lip of the computer’s base – not exactly intuitive. You can see it on the Surface website, not that I’d expect you’d have to go and look! Once you place the Pen there for the first time the magic happens and your two devices are now besties.
Yes, the Surface Laptop Studio has three modes – Laptop, Stage and Studio. You’ve already seen two of those above, so here’s what Stage looks like.
So where would you use this setting? As one of my colleagues pointed out, this is a great setting to have when you are docked at a workstation. On a standard laptop you will often close the screen because the angle is weird and the keyboard takes too much space, but the Laptop Studio in this setting hides the keyboard and gives you a screen position that works well underneath larger screens.
It’s also a super-crisp screen, though I’ve been spoilt at 4K screens for a few years, so I am really used to it and expect no less. One item that I did notice is the rounded edge corners on the screen really suit Windows 11’s new design aesthetic which ships with rounded window corners. On Windows 10 it hit my OCD that the window corners were cropped by the screen!
Developer and Data Scientist Productivity
I thought rather than dig out benchmarking tools to see what the computer can do that I would simply run some realistic developer scenarios to see what the computer could handle.
Secure all the things
While not strictly a performance or productivity boost, I have to say that I’ve gotten very used to using Windows Hello to sign into computers using either face identify or other biometrics. The Surface Laptop Studio ships with a good camera that supports Windows Hello, so it’s always a pleasure to not have to remember my password. 🙂
Compiling software while multi-tasking
I started off by installing Visual Studio 2022 (now with 32 extra bits!), followed by cloning the .NET SDK GitHub repository (at time of writing it contains .NET 7 Alpha).
As it turns out, you don’t need to install Visual Studio to compile the .NET SDK – you simply run the
build.cmd file in the repository root. This will download anything required to build the SDK from scratch, so it does take a while to run. At the end though, the actual msbuild task reports a time. One the Laptop Studio it reported build time of 4 minutes, 14 seconds.
I might add this time was also while the machine was running a long Anaconda resolve request for a TensorFlow GAN sample in the background, as well as having Visual Studio open.
In order to understand the performance of the Laptop Studio I also ran the same compilation process on my Surface Laptop 3. The result on that computer was over 5 minutes. Not a huge performance difference, but then the Laptop 3 is also using an i7 processor from the last 12 months so I would expect CPU-bound workloads to not differ massively in their behaviour.
Containers and Kubernetes
The shift to cloud native architectures has seen the rise of Containers for both building and hosting applications, along with their use for novel approaches to development such as Dev Containers. Often times developers want to work with a realistic set of services locally (this hasn’t changed from the old N-tier architectures!) and with the Surface Laptop Studio you have the performance to easily run minikube or similar solutions so you can develop productively.
Local Machine Learning PoCs
While the NVidia GPUs for both editions of the Surface Laptop Studios are no slouches (for example, the NVidia A2000 has 3,328 CUDA cores), where you will notice these are mobile GPUs is with their memory. Both cards ship with 4GB which means you’ll be in a position to build and test models locally, but once they get large enough you will be looking for either an external GPU (linked via one of the Thunderbolt 4 connectors) or pushing the model baseline to a service like Azure Machine Learning Service to perform larger-scale training.
I downloaded and ran the WSL-based version of DeOldify (find the setup instructions on GitHub) which ran fine with the inbuilt Intel i7 GPU, taking about a minute to colourise the sample photo, but ran into ‘out of memory’ issues when I kicked it across to use the NVidia GPU with the default quality settings for DeOldify, though I will note the DeOldify authors say they have GPUs running with 11 GB of memory. I will note that NVidia provides a specific WSL CUDA toolkit which gives you full access to the power of these GPUs with official drivers.
The hardware in the Surface Laptop Studio is at the same amazing level as the more recent Surface devices – particularly the Surface Laptop. Given the GPU in the Laptop Studio it is a pretty chunky device, but still portable (how I don’t miss those old 17″ workstation machines from ~2008). In order to maximise performance you will find warm air being expelled from both sides of the base by the cooling fans and while this might sound like a complaint, it is not. If you are buying this device you are buying it for performance – something that is adversely impacted by overheating.
Right now I’m sticking with my Surface Laptop 3 (it’s also Sandstone coloured which is nice 😉 ) as the portability and performance suits my needs. If I was in a technical lead, agile lead, BA or data scientist role, and I was looking for a new device, then I think this would be a strong contender!