Can a cell workstation actually maintain up in opposition to a equally specified desktop system in manufacturing? Jason Lewis places one in every of Asus’s new Nvidia Studio cell programs by means of an array of real-world checks to search out out.
On this evaluate, we’re going to be having a look at a cell workstation constructed upon the Nvidia Studio platform, the Asus ProArt Studiobook 16 OLED. It has been fairly a while since I’ve had the chance to check a cell system tailor-made to graphics work (the final time was again in 2015 with the HP ZBook 17 G2) so I used to be curious to see how seven years of technical advances have formed cell workstations.
This evaluate is damaged up into two elements, the primary specializing in the system, and the second on Nvidia Studio itself. However first, let’s deal with a query that artists have been asking since cell workstations first turned out there: is it higher to make use of a desktop or cell system for skilled graphics work?
I’ve touched on this in earlier evaluations, and the reply at all times will depend on your wants as an artist. Again in 2015, my conclusion was: if you might want to be moveable, a cell workstation could be a viable various to a desktop system, albeit at a barely increased value. For those who want absolutely the quickest system you may get, or if upgradeability is essential, you’re higher off with a desktop. However does the identical maintain true at the moment?
Bounce to a different a part of this evaluate
System specs
Testing process
Benchmark outcomes
Different issues
The Nvidia Studio platform
Verdict
System specs
For cell programs, Nvidia Studio laptops have fairly spectacular specs. After I began reviewing cell workstations within the early 2010s, four-core CPUs had been the most effective you possibly can hope for. As we speak, some programs go as excessive as 16 CPU cores. Pair that with the present Nvidia RTX GPUs (on the time this evaluate was written, the GeForce RTX 30 Collection, however since up to date to the 40 Collection), and you’ve got a recipe for efficiency.
The Asus ProArt Studiobook 16 OLED we’re right here isn’t Asus’s top-of-the-range cell system, but it surely’s nonetheless fairly highly effective, sporting an eight-core AMD Ryzen 9 5900HX CPU working at 3.3 GHz (with a most enhance pace of 4.6 GHz), and a Nvidia GeForce RTX 3070 cell GPU along with the Radeon Vega GPU on the 5900HX die. For reminiscence, the Studiobook 16 is provided with 32GB of three,200MHz DDR4 RAM in a 2 x 16GB configuration, and for storage, 2 x 1TB PCIe 4.0 M.2 SSD drives in a RAID 0 configuration.
For all the high-performance {hardware} packed into it, the Studiobook 16 is remarkably skinny: about half the thickness of the HP ZBook 17 I checked out in 2015. It’s heavier than it seems – the load caught me off guard the primary time I picked it up – but it surely’s nonetheless considerably lighter than the ZBook.
Flipping up the display reveals a fairly normal back-lit mobile-format keyboard with skinny, short-throw keys, a generously sized trackpad, and the Asus Dial, a mechanical wheel that controls varied capabilities in DCC purposes. It may be mapped to customized shortcuts by means of Asus’s ProArt Creator Hub app.
Connectivity is respectable: not the most effective I’ve seen in a cell system, however not the worst both, with two full measurement USB 3.2 Gen2 Kind-A ports, two USB 3.2 Gen2 Kind-C ports, a HDMI 2.1 port, a RJ45 Ethernet port, a 3.5mm audio jack, a SD Specific 7.0 card reader and the DC energy enter. These compete for house alongside the edges of the system with a pair of reasonably massive cooling exhaust vents.
The Studiobook 16 additionally features a standard-fare HD webcam constructed into the highest of the show bezel, however not like many cell programs, it has a built-in privateness shutter: a pleasant contact.
This explicit configuration has a 16-inch 4K OLED show with a decision of three,840 x 2,400px in a 16:10 facet ratio. It’s DisplayHDR 500 True Black-certified, with a 1,000,000:1 distinction ratio and a peak brightness of 550nits, runs at 60Hz with a 0.2ms response time, shows as much as 1.07 billion colors (a color gamut of 100% DCI-P3 or 133% sRGB), and is Pantone-validated.
Testing process
On paper, the Studiobook 16 ought to have the ability maintain as much as loads of desktop programs with out breaking a sweat. To check that idea, I benchmarked it in opposition to a desktop system with related specs: a Dell XPS workstation with a 10-core Intel Core i9-10900K working at 3.7GHz (with a most enhance pace of 5.3GHz), a Nvidia GeForce RTX 3070 GPU, 64GB of three,200MHz DDR4 RAM, and a 2TB PCIe 3.0 M.2 SSD drive.
The desktop system has a bonus in CPU clock pace and two further CPU cores, however the Studiobook has quicker storage. (The desktop system additionally has double the reminiscence, however so long as our benchmark checks don’t exceed 32GB, the additional RAM shouldn’t have a lot impact on efficiency.) Probably the most attention-grabbing comparability goes to be how Studiobook’s GeForce RTX 3070 cell GPU stacks as much as its desktop counterpart.
For testing, I used the next purposes:
Viewport efficiency
3ds Max 2023, Blender 3.2.1, Chaos Vantage 1.6.2, Maya 2023, Substance 3D Painter 8.1.1, Unreal Engine 5.0.3
Rendering
Blender 3.2.1 (utilizing Cycles, examined on CPU and GPU), Corona 8 for 3ds Max (CPU solely), Arnold (through MtoA 5.1.0, utilizing Arnold GPU), V-Ray 6 for 3ds Max (utilizing V-Ray GPU)
Different benchmarks
Axiom 2.1 for Houdini, Metashape 1.8.2, Premiere Professional 2022, World Machine 4027
Artificial benchmarks
3DMark, Cinebench R23, OctaneBench 2020
Within the viewport benchmarks, the body price scores symbolize the figures attained when manipulating the 3D property proven, averaged over 5 testing periods to remove inconsistencies. Testing was carried out at 60Hz on the native decision of the Studiobook 16’s OLED show (3,840 x 2,400px), and on desktop, on a 32-inch 4K show working at its native decision of three,840 x 2,160px.
Benchmark outcomes
Viewport efficiency
The viewport benchmarks embody key DCC purposes 3ds Max, Blender and Maya, plus 3D portray device Substance 3D Painter, scene exploration device Chaos Vantage, and sport engine Unreal Engine.
Within the viewport benchmarks, the desktop model of the GeForce RTX 3070 GPU within the Dell XPS beats its cell counterpart within the Studiobook 16 by a median of round 5-10% – though which may be much less concerning the efficiency of the {hardware} than the truth that the desktop card is driving a conventional 3,840 x 2,160px 4K show, whereas the cell card has to push roughly 11% extra pixels to drive the Studiobook’s 3,840 x 2,400px show. In any case, at these body charges, the distinction is barely noticeable.
Rendering
The rendering benchmarks embody each CPU and GPU renderers. Corona runs on the CPU; Arnold and V-Ray had been examined on the GPU; and Blender’s Cycles renderer on each.
The rendering benchmarks yield related outcomes to the viewport show checks, offering the GPU is getting used for rendering. For the 2 CPU rendering benchmarks – Blender’s Cycles renderer and Corona – the efficiency hole will increase, with the 2 further CPU cores of the desktop system’s Core i9-10900K offering further rendering horsepower in comparison with the Studiobook’s Ryzen 9 5900HX.
Different benchmarks
The subsequent set of benchmarks covers extra specialist apps: volumetric fluid solver Axiom, photogrammetry software program Metashape, video editor Premiere Professional and terrain generator World Machine.
The extra benchmarks comply with an analogous sample to the viewport and rendering checks, with the Studiobook 16 working neck-and-neck with the desktop system on the GPU-based Axiom solver, however the desktop system taking the lead within the different benchmarks, the place the CPU comes into play.
Artificial benchmarks
Lastly, we have now three artificial benchmark checks. Artificial benchmarks don’t precisely predict how a machine will carry out in manufacturing, however I’ve embody them right here as a technique to make approximate efficiency comparisons to different programs, since databases of scores can be found on-line.
Our artificial checks yield some attention-grabbing outcomes. 3DMark and OctaneBench carry out roughly as anticipated, with the desktop system taking the lead over the Studiobook, albeit not by a lot.
However with CPU benchmark Cinebench, the Studiobook truly beats the desktop system by a small margin, regardless of its drawback in each CPU core rely and clock pace. I’m not certain why that is, however my suspicion is that Cinebench simply performs nicer with AMD’s Zen 3 CPU structure, which is barely newer than the Intel Comet Lake CPU within the desktop system.
Different issues
Energy consumption
Given how related the efficiency of the 2 programs is, you would possibly count on them to make use of an analogous quantity of power. In truth, the Studiobook 16 makes use of considerably much less energy than the desktop system: in some instances, lower than half. This might clarify why it doesn’t get uncomfortably sizzling underneath regular use. A couple of spots across the keyboard get fairly heat with prolonged use, however I’ve used cell programs that run a lot hotter.
Show high quality
Lastly, we come to the Studiobook 16’s large, vivid, stunning show. This was my first expertise with an OLED PC show, and I beloved it. Colours are vivid and superbly saturated, the blacks are actually black, and the general distinction is superb.
My desk is cluttered with LCD and LED desktop shows, and for my part, it beats all of them simply – together with a pair which are dearer than your entire Studiobook 16. Even when the system itself had been full rubbish (it isn’t – fairly the reverse), the show alone could be a powerful argument in its favour.
The Nvidia Studio platform
Earlier than we get to the ultimate verdict, I wish to dive into the Nvidia Studio ecosystem that the ProArt Studiobook 16 is a part of. Nvidia touts its Studio line of desktop and cell programs as instruments to permit artists to ‘create on the pace of creativeness’. The Studio ecosystem has three parts: {hardware}, drivers and software program.
The {hardware} part is a set of requirements that system distributors want to satisfy to obtain the Nvidia Studio stamp of approval. There are particular necessities for the CPU, GPU, RAM, storage and show, however programs don’t want to incorporate GPUs from Nvidia’s RTX Axxxx sequence {of professional} GPUs: most of these at present out there through Nvidia’s e-store have shopper GeForce RTX GPUs. Skilled playing cards nonetheless have their place for duties that want very excessive graphics reminiscence capacities, however for almost all of artistic work, shopper GPUs are completely succesful, and evidently Nvidia is lastly acknowledging this.
The subsequent part of the Nvidia Studio ecosystem is GPU drivers. For a very long time, solely Nvidia’s skilled GPUs (the RTX Axxxx sequence, and earlier than it, the Quadro playing cards) received drivers particularly tuned for DCC purposes, however the Nvidia Studio drivers now give shopper GPUs the identical remedy.
The ultimate part of the Nvidia Studio ecosystem is a set of software program purposes out there to Nvidia Studio customers (you may obtain them even when you don’t have a Nvidia Studio system, however all of them require a Nvidia GPU): Nvidia Canvas, Nvidia Broadcast, Nvidia Omniverse and the Nvidia Texture Instruments Exporter. I haven’t had an opportunity to do a deep dive into them, however these are my preliminary reactions.
Nvidia Canvas is an AI-based image-creation device that lets you paint easy brushstrokes on a clean canvas and have Canvas generate a photorealistic panorama based mostly on them. It could possibly generate fairly spectacular outcomes underneath very particular circumstances, however more often than not, the output isn’t {of professional} high quality. For idea artists, it may very well be a fast technique to generate a base picture that may then be refined in Photoshop or different image-editing software program, but it surely appears extra like an AI showcase than a sensible DCC device, and is massively outclassed by AI-based picture mills like Midjourney.
Nvidia Broadcast is a set of AI-based audio and video filters that may be utilized to the output of your system’s microphone and digital camera to boost streaming and video conferencing. It may be used with commonest platforms, together with Discord, Zoom and Google Meet. That is the Studio utility I actually like. It doesn’t do something revolutionary: simply processes audio and video in the identical approach that almost all streaming platforms do natively, however a lot better! Because it’s free for anybody with a present Nvidia GPU, there’s no cause to not use it, though customers with sub-$100 webcams and mics will see essentially the most profit.
Subsequent is Nvidia Omniverse. Omniverse is an attention-grabbing beast. It’s marketed as a scene-assembly hub that may hook up with fashionable DCC purposes, enabling customers to construct, animate and render DCC initiatives, collaborating with different artists in actual time. Consider it as Nvidia’s reply to real-time graphics engines like Unreal Engine or Unity, with multi-user collaboration and stay hyperlinks to DCC purposes thrown in.
From a technical perspective, Omniverse is spectacular. Nvidia mainly wrote its personal real-time graphics engine, and added multi-user capabilities at a core degree. However I’ve but to come across knowledgeable sport improvement or content material creation studio utilizing it. Which will change in future, however DCC apps can already construct, animate and render scenes, and have a loyal and entrenched person base, whereas studios have their very own tried-and-tested strategies of collaboration. Leisure artists are a finicky bunch, and we concern change!
Nonetheless, I need to admit that I’ve not spent sufficient time with Omniverse to actually decide its usefulness. I’ll spend extra time with it sooner or later to get a greater understanding of its supposed function and advantages.
Lastly, we have now the Nvidia Texture Instruments Exporter. It’s a useful little app that generates a number of varieties of real-time pleasant texture maps from from a supply picture, and is accessible as a standalone utility or as a Photoshop plugin. I exploit it fairly a bit to generate regular maps from pictures in these few instances the place I’m not utilizing Substance 3D supplies or photoscanned textures, and it really works fairly effectively, particularly since Photoshop’s native regular map conversion device is, for my part, full crap.
Verdict
It’s time to see if we will reply the large query that was requested early on on this article: is is best to make use of a desktop or cell system for skilled graphics work?
Let’s have a look at efficiency first. For essentially the most half, the Asus ProArt Studiobook 16 OLED retains up with a Dell desktop system with related specs. The desktop system comes out prime in many of the benchmarks, however by a comparatively small margin, and except just a few CPU-heavy checks, you’d hardly discover the distinction in day-to-day work.
Subsequent, value. As of the time of testing, I used to be unable to search out any on-line outlets promoting Studiobook 16 programs with precisely the identical configuration as our check machine. Nonetheless, I did discover just a few with a GeForce RTX 3060 GPU as an alternative of a RTX 3070, and people had been promoting for round $2,300, so I’d estimate a price ticket of round $2,500. A Dell XPS system with related specs to the desktop machine, however a more recent 16-core Intel Core i9-12900 CPU, price somewhat over $2,400. Provided that with the desktop system, you would want to pay further to improve to an OLED show, I’d say the Studiobook 16 truly wins on price.
So what conclusions can we draw? First, that Nvidia Studio cell workstations can certainly be used for demanding content material creation duties. And second, in the case of price, there may be now little distinction between equally performing cell and desktop programs.
A few of it’s possible you’ll now be considering: ‘If this cell system is so nice, why would I ever wish to use a desktop system once more?’ Properly, regardless of the strengths of the ProArt Studiobook 16 OLED and different Nvidia Studio laptops, cell programs do have two actually large drawbacks.
The primary is {that a} desktop system is definitely upgradeable, whereas a cell system just isn’t. Typically, you may solely improve the reminiscence and storage, and even these prime out at a lot decrease values than in a comparable desktop system.
The second is that whereas a cell system might carry out on par with an equal desktop machine, you may construct desktop programs which are way more highly effective than out there cell programs. Presently, cell programs prime out at 8-12 CPU cores, 32-64GB of RAM and 2-4TB of storage. With desktop programs, you may stand up to 64 CPU cores, a number of terabytes of RAM, and storage arrays of over 40TB. Granted, these elements are very costly, however you’ve gotten the choice to make use of them, whereas with a cell system, you don’t.
For those who want your main system to be moveable, otherwise you simply want the shape issue of a laptop computer, a Nvidia Studio cell system will probably be match. (And within the case of the Asus ProArt Studiobook 16 OLED, that stunning OLED show is an actual promoting level.) If upgradeability is essential to you, otherwise you merely want extra uncooked processing energy – and may afford to pay for it – a desktop system will suit your wants higher.
Lastly, thanks for stopping by and taking the time to take a look at this evaluate. My purpose is to supply digital artists an in-depth have a look at trendy {hardware} and software program options, and the way they actually deal with DCC duties. I hope that this has been an pleasing and informative learn, and to see you once more on the subsequent one!
Hyperlinks
Learn extra concerning the ProArt Studiobook 16 OLED laptop computer on the Asus web site
Learn extra concerning the Nvidia Studio ecosystem
In regards to the reviewer
Jason Lewis is Atmosphere Artwork Lead at Staff Kaiju and CG Channel’s common reviewer. You’ll be able to see extra of his work in his ArtStation gallery. Contact him at jason [at] cgchannel [dot] com
Acknowledgements
I wish to give particular because of the next individuals for his or her help in bringing you this evaluate:
Shannon Baker of Zeno
Daria Bianchini of Zeno
Stephen G Wells
Adam Hernandez
Thad Clevenger
Tags: 3d, 3D modeling, 3D scanning, 3DMark, 3ds max, 4K, AMD, arch viz, architectural visualization, are Nvidia Studio laptops definitely worth the cash, Arnold, Asus, Asus Dial, Asus ProArt 16 OLED, Asus ProArt Studiobook 16, Axiom, base clock, benchmark, benchmark scores, benchmark checks, Blender, enhance clock, CAD, can a cell workstation compete with a desktop system, can the Asus Studiobook Professional compete with a desktop system, Chaos Vantage, clock pace, shade gamut, compositing, compute efficiency, connectivity, shopper GPU, content material creation, distinction ratio, cooling, Corona, CPU, CPU rendering, CUDA, Cycles, DCC, DCI-P3, DDR4, Dell, dimensions, show, show decision, DisplayHDR 500 TrueBlack, fluid solver, FP32, sport artwork, sport improvement, Recreation Engine, gaming GPU, GeForce RTX, GeForce RTX 3070, GPU, GPU driver, GPU rendering, HDMI 2.1, picture based mostly modeling, Intel Core i9-10900K, Jason Lewis, lighting, materials authoring, Maya, Metashape, mtoa, Nvida Broadcast, NVIDIA, Nvidia Canvas, NVIDIA Studio, Nvidia Studio driver, Nvidia Studio software program, Nvidia. Omniverse, Nvidia. Texture Instruments, OctaneBench, OctaneRender, OLED, OptiX, Pantone, efficiency, Photogrammetry, post-production, energy consumption, energy use, Premiere Professional, value, value comparability, manufacturing benchmark, skilled GPU, Quadro, RAID 0, RAM, ray tracing, real-world benchmark, real-world check, rendering, response time, Assessment, RTX Axxxx, Ryzen 9 5900HX, specs, sRGB, storage, Studio driver, Substance 3D Designer, TDP, Tensor core, terrain era, check, texturing, thickness, Kind-A, Kind-C, Unreal Engine, USB 3.2, V-Ray, V-Ray GPU, vfx, video modifying, viewport efficiency, visible results, VRAM, weight, workstation GPU, World Machine, XPS, Zen 3
