System Performance

Our updated test suite for Windows 11-based SFF systems carries over some of the standard benchmarks we have been using over the last several years, including UL's PCMark and BAPCo's SYSmark. Starting this year, we are also including BAPCo's CrossMark multi-platform benchmarking tool.

UL Benchmark: PCMark 10

PCMark is a full-system benchmark which highlights CPU performance, but also includes the GPU, memory, and storage as factors in the outcome depending on the test being performed.

UL PCMark 10 Benchmarks
(Performance Scores - Higher is Better)
  Intel NUC12DCMi9
(Dragon Canyon)
Intel NUC11BTMi9
(Beast Canyon)
Digital Content Creation 6478 5734
Gaming 2189 2029
Productivity 7103 6722
Essentials 11056 10182
Overall 4505 4143

All workloads see a significant upgrade in scores as we move from Tiger Lake to Alder Lake in two machines with essentially identical thermal designs.

BAPCo SYSmark 25

BAPCo's SYSmark 25 replays traditional business workloads in a tightly controlled environment for consistent reproducibility. Energy measurement is also done to determine efficiency of the PC under test.

BAPCo SYSmark 25 Benchmarks
(Performance Scores - Higher is Better)
(Energy Consumption in Wh - Lower is Better)
  Intel NUC12DCMi9
(Dragon Canyon)
Intel NUC11BTMi9
(Beast Canyon)
Productivity 1927 (32.76 Wh) 1681 (26.22 Wh)
Creativity 2222 (31.44 Wh) 2006 (25.54 Wh)
Responsiveness 1476 1273
Overall 1972 (64.20 Wh) 1743 (51.76 Wh)

The Dragon Canyon NUC delivers significantly better performance at the cost of increased energy consumption - a 13% increase in scores for a 24% increase in energy consumption.

BAPCo CrossMark 1.0.1.86

BAPCo's CrossMark aims to simplify benchmark processing while still delivering scores that roughly tally with SYSmark. The main advantage is the cross-platform nature of the tool - allowing it to be run on smartphones and tablets as well.

BAPCo CrossMark 1.0.1.86 Benchmark
(Performance Scores - Higher is Better)
  Intel NUC12DCMi9
(Dragon Canyon)
Intel NUC11BTMi9
(Beast Canyon)
Productivity 1890 1733
Creativity 2164 1718
Responsiveness 1908 1702
Overall 2006 1722

The improvements seen in SYSmark 25 translate to CrossMark also, as expected.

Rendering: Cinebench R23

The Cinebench rendering benchmark can operate in two modes - single-threaded and multi-threaded.

Cinebench R23 Benchmarks
(Performance Scores - Higher is Better)
  Intel NUC12DCMi9
(Dragon Canyon)
Intel NUC11BTMi9
(Beast Canyon)
Single Thread 1833.69 1456.62
Multiple Threads 17951.68 11583.86

The upgrade in the single-threaded performance is quite welcome, given that both processors carry similar TDPs and clocks. On the multi-threaded front, the presence of 24 threads gives the Dragon Canyon a distinct advantage over the 16 threads in the Beast Canyon NUC.

Transcoding: Handbrake 1.5.1

Handbrake is one of the most user-friendly open source transcoding front-ends in the market. It allows users to opt for either software-based higher quality processing or hardware-based fast processing in their transcoding jobs. Our new test suite uses the 'Tears of Steel' 4K AVC video as input and transcodes it with a quality setting of 19 to create a 720p AVC stream and a 1080p HEVC stream.

Handbrake 1.5.1 Transcoding Benchmarks
(Transcoding FPS - Higher is Better)
  Intel NUC12DCMi9
(Dragon Canyon)
Intel NUC11BTMi9
(Beast Canyon)
Software (x264)
(4K AVC to 720p AVC)
101.29 76.70
Hardware (qsv_h264)
(4K AVC to 720p AVC)
121.17 100.30
Software (x265_10bit)
(4K AVC to 1080p HEVC)
28.79 22.14
Hardware (qsv_h265_10bit)
(4K AVC to 1080p HEVC)
107.77 93.98

Hardware transcoding for HEVC is a lot faster than software (when compared to AVC in the same modes). Between Tiger Lake and Alder Lake, we see improvements in the transcoding rates across the board in the latter.

Archiving: 7-Zip 21.7

The 7-Zip benchmark is carried over from our previous test suite with an update to the latest version of the open source compression / decompression software.

7-Zip 21.7 Benchmarks
(LZMA:x5:MT2 Algorithm Processing Rate in MBps - Higher is Better)
  Intel NUC12DCMi9
(Dragon Canyon)
Intel NUC11BTMi9
(Beast Canyon)
Compression 71.41 55.86
Decompression 1177.77 854.12

The higher number of threads in Alder Lake provides it with a distinct advantage in this multi-threading friendly benchmark.

Web Browsing: JetStream and Speedometer

Web browser-based workloads have emerged as a major component of the typical home and business PC usage scenarios. We are carrying over the browser-focused benchmarks from the WebKit developers used in our notebook reviews. Hosted at BrowserBench, JetStream 2.0 benchmarks JavaScript and WebAssembly performance, while Speedometer measures web application responsiveness. We also process MotionMark, but the confidence level of the results vary as much as ±35%.

Intel NUC12DCMi9 (Dragon Canyon) Browser Bench
  Speedometer 2.0 JetStream 2.0 MotionMark 1.2
Microsoft Edge
(92.0.902.55)
279 ± 4 267.32 1221.43 ± 4.18%
Google Chrome
(92.0.4515.107)
289 ± 3.3 264.5 1101.27 ± 1.55%
Mozilla Firefox
(90.0.2.7872)
183 ± 3.2 153.96 989.06 ± 7.85%
Intel NUC11BTMi9 (Beast Canyon) Browser Bench
  Speedometer 2.0 JetStream 2.0 MotionMark 1.2
Microsoft Edge
(92.0.902.55)
181 ± 2.2 201.313 900.46 ± 14.16%
Google Chrome
(92.0.4515.107)
212.1 ± 2.1 201.097 1115.74 ± 5.54%
Mozilla Firefox
(90.0.2.7872)
164 ± 2.2 123.612 1072.12 ± 3.44%

Other than MotionMark, we see a significant uptick in performance for all browsers across all benchmarks when moving from Tiger Lake to Alder Lake.

Application Startup: GIMP 2.10.30

A new addition to our SFF test suite is AppTimer - a benchmark that loads up a program and determines how long it takes for it to accept user inputs. We use GIMP 2.10.30 with a 50MB multi-layered xcf file as input. What we test here is the first run as well as the cached run - normally on the first time a user loads the GIMP package from a fresh install, the system has to configure a few dozen files that remain optimized on subsequent opening. For our test we delete those configured optimized files in order to force a ‘fresh load’ every second time the software is run. As it turns out, GIMP does optimizations for every CPU thread in the system, which requires that higher thread-count processors take a lot longer to run. So the test runs quick on systems with fewer threads, however fast cores are also needed.

AppTimer: GIMP 2.10.30 Startup
(Time in Seconds - Lower is Better)
  Intel NUC12DCMi9
(Dragon Canyon)
Intel NUC11BTMi9
(Beast Canyon)
First Run 7.43 7.76
Decompression 4.01 3.87

Despite more cores in Dragon Canyon, their fast nature helps in achieving parity for first load times. However, the Beast Canyon enjoys a slight edge in the cache start.

Setup Notes and Platform Analysis GPU Performance
Comments Locked

38 Comments

View All Comments

  • thestryker - Thursday, February 24, 2022 - link

    So you basically want a bigger case? That's the only way you'd get what you're looking for cooling wise because you have to account for the baseboard and power supply. I believe the DAN Case A4 is the only ITX case on the market smaller than the Intel NUC 11/12 Extreme (I'd assume there are probably knockoffs of this case which might be as well). As for the cooling there's a shroud which goes over the Element to draw air in from the back of the case for CPU cooling and the three fans at the top pull air out. There's nothing "random" about how they designed the cooling nor is it "off the shelf".

    For CPU heavy workloads I definitely wish they'd have a better cooling solution as they seem limited by their choice to utilize the compute element. I think the only way to solve this would be by doing a completely custom board setup, but I also highly doubt this would be financially viable.
  • Operandi - Thursday, February 24, 2022 - link

    A larger case isn't need, just use the space more intelligently. The compute board is using a super compact heatsink with a blower fan (notebook cooling components), the case itself just has fans exhausting the hot air like any regular case. Its a lazy repackaging of components that already exist into a new form fact that offers zero benefits over what you can already accomplish with ITX.

    Repackage all of it so you are using all that active cooling to push the air through a passive heatsink, easily accomplished given how little power this uses. The rest of the case size and cooling would be dictated by how much space, cooling, and power is required for the GPU.
  • thestryker - Thursday, February 24, 2022 - link

    You can't use top to bottom while using the compute element, and the PSU is located at the front of the case so you can't use front to back either. So you'd have to increase the width or height to be able to move the PSU out of the way for a front to back airflow which would make the case larger than it is now.
  • Operandi - Thursday, February 24, 2022 - link

    Only if you look at it in the narrow confines of how its setup now. Even with the compute element there are enormous ways you could achieve a top to bottom layout. PSU can go anywhere and be any form factor and completely external for lower power variants. Frankly I don't get why there are power cables at all side from for the GPU, power for the compute should be carried via the baseboard.
  • thestryker - Thursday, February 24, 2022 - link

    Name *one* way you could still utilize the compute element with the baseboard that also has an externally accessible M.2 slot with a top/bottom airflow setup. Now you want them to have a proprietary 650W gold PSU made for
    just the enthusiast class NUC? There are *no* low power variants of the *enthusiast* NUC as they have a ton of regular NUCs to fit that bill. The compute element adheres to PCIe so they should redesign that just for this unit to fit the extra power consumption?

    It doesn't seem that you actually understand how any of the components for these units work and just don't like the setup.
  • QChronoD - Friday, February 25, 2022 - link

    If they are going to go through the trouble of making a custom baseboard for the compute module to plug into, it wouldn't have been that much more difficult to have the connectors positioned for the ATX and GPU power outputs on the PSU. Also likely could have flipped the compute board so that it was back to back with the GPU and then have a larger fan directly drawing air in from the outside of the case.
  • thestryker - Friday, February 25, 2022 - link

    Flipping the compute element has been the most suggested thing ever since Ghost Canyon, and Intel has never addressed this as far as I know. In the case of the enthusiast NUCs this would be the easiest and most obvious change, but it hasn't happened. I can't think of a logical reason not to do it, but it must have something to do with the other applications they're used in at the enterprise level.
  • Operandi - Monday, February 28, 2022 - link

    Name one? You want me to verbally draw you a mental image? Look at what systems like the NZXT H1 are able to do with top/down layout using ITX. The limiting factor is going to be where the ports exit the system because of the compute element but thats just another problem with how this is designed.

    Is the PSU supplying anything other than 12v? I doubt it, and if it is it shouldn't be. And yeah design a custom one its not hard and Intel is a huge company. Its all proprietary anyway and we already know this is going to be insanely expensive anyway so they have no excuse.

    What exactly is the point of them using PCI-e for the compute element again? Just so they don't have to design and manufacture a purpose built connector for their entirely proprietary form factor? It certainly adds zero functionality to the system.
  • Dug - Friday, March 11, 2022 - link

    nzxt h1 is one of the worst cooling cases around, that is what you call garbage. The aio would not be able to handle this chip. Air in from sides and out one 92mm fan is not a good design.

    The cooling in nuc is actually very good for what it is. The shroud separates the hot air from the gpu from the cpu, which is usually the problem in itx cases. Even those that use pci-e riser to put graphics card on the other side, no one actually isolates the compartments, which ends up over heating some other component like an ssd on the back of the motherboard. But the extension itself is troublesome and has shown to be more hassle than it's worth.
  • Dug - Friday, March 11, 2022 - link

    I agree with a custom power supply. Look at what Apple did in the iMac pro. 8 to 18-core xeon, 5k display, Radeon Pro Vega 56, speakers, TB, etc, all from a super slim power supply built into the monitor. All one cable, no power brick. The archaic model of giant atx power supplies with thick cables going everywhere needs to go away.

Log in

Don't have an account? Sign up now