ASRock Industrial has been one of the most prolific vendors in the ultra-compact form-factor (UCFF) PC space over the last few years. They have been releasing 4"x4" systems based on the latest AMD and Intel platforms within a few weeks of their announcements by the processor vendors. On the AMD front, the company released the Rembrandt Refresh-based 4X4 BOX-7735U earlier this year, and followed that up a couple of quarters later with the Phoenix-based 4X4 BOX-7040 series.

The 4X4 BOX-7040 series builds upon the I/O and iGPU improvements in the 4X4 BOX-7735U with the incorporation of AMD's latest Zen 4 CPU cores and RDNA3-based iGPU. The move to a 4nm TSMC process (from the 6nm one used for Rembrandt Refresh) and the CPU / iGPU improvements should theoretically delivery much better performance and power efficiency for a range of workloads.

The Ryzen 7 7840U is a 28W TDP part that bears striking similarity to the Ryzen 7 7840HS that was already evaluated in a 65W mode in Beelink's GTR7. As a result of the lower power limit, the base clock is lowered. The iGPU configuration, I/Os, and other operating parameters remain the same. The U-series is primarily meant for notebook platforms, but ASRock Industrial (and other mini-PC vendors) have been deploying them in actively-cooled UCFF cases. The presence of a fan gives vendors a bit of leeway in configuring the power limits - and, as an extension, performance.

ASRock Industrial ships the 4X4 BOX-7840U in the 'Normal Mode' with AMD's suggested TDP of 28W. However, a toggle in the BIOS can push the system into a 'Performance Mode' with a souped-up TDP of around 40W. This review explores the performance profile of the PC in both modes, and provides detailed insights into the differentiating aspects of AMD's Phoenix in a UCFF mini-PC platform.

Introduction and Product Impressions

UCFF systems have been steadily replacing bulky tower desktops for many use-cases over the last decade. This category has been experiencing growth in both home consumer and industrial settings, with the latter prompting the B2B / industrial computing arms of many motherboard vendors to provide more attention to such systems. ASRock Industrial (spun out of ASRock's business unit in 2018) has been creating UCFF systems based on both AMD and Intel platforms since 2019.

This review delves in detail into the company's AMD Phoenix-based flagship UCFF offering - the 4X4 BOX-7840U. Based on AMD's high-end Phoenix 28W offering (Ryzen 7 7840U), the new system is meant to be a follow-up to the Rembrandt-R-based 4X4 BOX-7735U released earlier this year. The Ryzen 7 7840U continues with the same 8C/16T configuration of the Ryzen 7 7735U. However, the fabrication process has moved from TSMC's 6nm FinFET to 4nm FinFET, and the CPU cores are now based on Zen 4. The integrated GPU has also been re-architected, moving from the RDNA2-based Radeon 680M in Rembrandt-R to the RDNA3-based 780M in Phoenix.

ASRock Industrial's UCFF systems are non-descript machines that do not opt for a fancy case design. The functional chassis used in previous 4X4 BOX systems is retained for the 4X4 BOX-7840U also. This does mean an extension of the lifespan of the glossy polycarbonate fingerprint magnet casing, but that is set to change in 2024. The I/O port locations are exactly the same as in the previous generation. The internal board is also largely the same, as the USB4 ports and DDR5 support were already pencilled in during the design of the 4X4 BOX-7735U.

The company's 4X4 BOX-7040 series has only two members - one based on the Ryzen 5 7640U and the other based on the Ryzen 7 7840U. Some of the key relevant aspects are brought out in AMD's introductory slide to the product family back at the 2023 CES.

ASRock Industrial's choice of I/O ports for the Intel and AMD systems in the same generation is worth a note. While the rear USB ports are only USB 2.0 in the AMD ones, they are USB 3.2 Gen 2 in the Intel versions. However, the two front panel Type-C ports support the full USB4 feature set including PCIe tunneling in the AMD systems. The Intel boards are skimped on, with only one of the Type-C ports capable of full USB4 / Thunderbolt 4 support and the other falling back to USB 3.2 Gen 2 with Alt DP support.

The company offers both barebones version of the system as well as the motherboard alone. The former is typically sold in retail, while the motherboard is meant for the B2B channel. The barebones version package comes with a 120W DC power adapter (19V @ 6.32A), VESA mount (and associated screws), a geo-specific power cord, the main unit, and a product overview / user guide.

The barebones version needs two DDR5 SODIMMs and a M.2 2280 SSD to complete the build. Crucial's DDR5-5600 SODIMM kit (2x16GB) was complemented by a Samsung 990 PRO 2TB Gen 4 NVMe SSD for this purpose.

Access to the SODIMM and M.2 slots is via the underside, similar to the previous generation systems. Removal of the four screws at the bottom allows the panel to be popped off.

Despite the support for a 2.5" SATA drive in the system, ASRock Industrial strongly recommends not using it in order to aid with proper airflow. The installation process is otherwise similar to the older 4X4 BOX systems, and we were up and running with a freshly installed OS in no time (after installing the chipset drivers from the product support page). The Ryzen 7 7840U does include the XDNA engine (now rebranded as Ryzen AI in Phoenix Refresh), but the drivers for that (recognized as an IPU in Device Manager) are not yet available through Windows Update.

The full specifications of the review sample (as tested) are provided in the table below. As we will note in the next section, the BIOS allows the system to be configured in either of two modes with different TDPs, as specified in the Processor entry.

Systems Specifications (as tested)
	ASRock 4X4 BOX-7840U (Performance)	ASRock 4X4 BOX-7840U (Normal)
Processor	AMD Ryzen 7 7840U Zen 4 (Phoenix) 8C/16T, 3.3 - 5.1 GHz TSMC 4nm, 16MB L3, 28W Target TDP : 40W	AMD Ryzen 7 7840U Zen 4 (Phoenix) 8C/16T, 3.3 - 5.1 GHz TSMC 4nm, 16MB L3, 28W Target TDP : 28W
Memory	Crucial CT16G56C46S5.M8G1 DDR5-5600 SODIMM 46-45-45-90 @ 5600 MHz 2x16 GB	Crucial CT16G56C46S5.M8G1 DDR5-5600 SODIMM 46-45-45-90 @ 5600 MHz 2x16 GB
Graphics	AMD Radeon 780M (RDNA3 / Phoenix) - Integrated (12 CUs @ 2.7 GHz)	AMD Radeon 780M (RDNA3 / Phoenix) - Integrated (12 CUs @ 2.7 GHz)
Disk Drive(s)	Samsung SSD 990 PRO MZ-VP92T0B (2 TB; M.2 2280 PCIe 4.0 x4 NVMe;) (Samsung 7th Gen. V-NAND 176L (136T) 3D TLC; Samsung Pascal S4LV008 Controller)	Samsung SSD 990 PRO MZ-VP92T0B (2 TB; M.2 2280 PCIe 4.0 x4 NVMe;) (Samsung 7th Gen. V-NAND 176L (136T) 3D TLC; Samsung Pascal S4LV008 Controller)
Networking	1x 2.5 GbE RJ-45 (Realtek RTL8125BG) 1x GbE RJ-45 (Realtek RTL8111EPV) Mediatek MT7922 (RZ616) Wi-Fi 6E (2x2 802.11ax - 1.9 Gbps)	1x 2.5 GbE RJ-45 (Realtek RTL8125BG) 1x GbE RJ-45 (Realtek RTL8111EPV) Mediatek MT7922 (RZ616) Wi-Fi 6E (2x2 802.11ax - 1.9 Gbps)
Audio	Realtek ALC256 (3.5mm Audio Jack in Front) Digital Audio with Bitstreaming Support over HDMI and Display Port (Type-C)	Realtek ALC256 (3.5mm Audio Jack in Front) Digital Audio with Bitstreaming Support over HDMI and Display Port (Type-C)
Video	2x HDMI 2.0 2x Display Port 1.4a over USB4 Type-C	2x HDMI 2.0 2x Display Port 1.4a over USB4 Type-C
Miscellaneous I/O Ports	2x USB 2.0 (Rear) 2x USB4 Type-C (Front) 1x USB 3.2 Gen 2 Type-A (Front)	2x USB 2.0 (Rear) 2x USB4 Type-C (Front) 1x USB 3.2 Gen 2 Type-A (Front)
Operating System	Windows 11 Enterprise (22621.2861)	Windows 11 Enterprise (22621.2861)
Pricing	(Street Pricing on December 27th, 2023) US $570 (barebones) US $837 (as configured, no OS)	(Street Pricing on December 27th, 2023) US $570 (barebones) US $837 (as configured, no OS)
Full Specifications	ASRock Industrial 4X4 BOX-7840U Specifications	ASRock Industrial 4X4 BOX-7840U Specifications

In the next section, we take a look at the system setup and follow it up with a detailed platform analysis.

Setup Notes and Platform Analysis

The video below presents the entire gamut of available options in the BIOS of the 4X4 BOX-7840U. Of particular interest is the 'CPU Operating Mode' under 'Advanced > CPU Configuration'. It is set to 'Normal' by default, corresponding to a TDP of 28W. Altering it to 'Performance' sets the fan speed to maximum irrespective of the actual load, but ekes out extra performance by pushing up the TDP to around 40W.

The system is equipped with dual LAN ports (1x 2.5GbE + 1x 1GbE) backed up by Realtek controllers. Similar to previous 4X4 BOX systems, the 1GbE link comes with DASH support to make it easy for IT departments to deploy and manage the system with an out-of-band management interface. This support is disabled by default.

The block diagram below presents the overall high-speed I/O distribution.

Compared to the Rembrandt Refresh board, some functionalities seem to have been absorbed into the Phoenix package. For example, the two HDMI ports are driven directly from the package instead of using a protocol converter to switch Display Port to HDMI. The two power delivery controllers between the package and the USB4 ports seem to be absent too, but the Kandou Technologies KB8002 retimer remains (which also means that the USB4 ports still don't support the USB 3.2 Gen 2x2 20Gbps mode). The other board components are largely similar to the ones in the 4X4 BOX-7735U.

In today's review, we compare the 4X4 BOX-7840U and a host of other systems based on processors with TDPs ranging from 15W to 65W. The systems do not target the same market segments, but a few key aspects lie in common, making the comparisons relevant.

Comparative PC Configurations
Aspect	ASRock 4X4 BOX-7840U (Performance)	ASRock 4X4 BOX-7840U (Performance)ASRock 4X4 BOX-7840U (Normal)ASRock 4X4 BOX-5800U (Performance)Beelink GTR7ASRock NUC BOX-1360P-D5 (Performance)ASRock 4X4 BOX-7735U (Performance)Intel NUC13ANKi7 (Arena Canyon)ASRock NUCS BOX-1360P-D4ASRock 4X4 BOX-4800UGEEKOM A5Intel NUC12WSKi7 (Wall Street Canyon)ASRock NUC BOX-1260PGEEKOM AS 6 (ASUS PN53)GEEKOM Mini IT13
CPU	AMD Ryzen 7 7840U Zen 4 (Phoenix) 8C/16T, 3.3 - 5.1 GHz TSMC 4nm, 16MB L3, 28W Target TDP : 40W	AMD Ryzen 7 7840U Zen 4 (Phoenix) 8C/16T, 3.3 - 5.1 GHz TSMC 4nm, 16MB L3, 28W Target TDP : 40W
GPU	AMD Radeon 780M (RDNA3 / Phoenix) - Integrated (12 CUs @ 2.7 GHz)	AMD Radeon 780M (RDNA3 / Phoenix) - Integrated (12 CUs @ 2.7 GHz)
RAM	Crucial CT16G56C46S5.M8G1 DDR5-5600 SODIMM 46-45-45-90 @ 5600 MHz 2x16 GB	Crucial CT16G56C46S5.M8G1 DDR5-5600 SODIMM 46-45-45-90 @ 5600 MHz 2x16 GB
Storage	Samsung SSD 990 PRO MZ-VP92T0B (2 TB; M.2 2280 PCIe 4.0 x4 NVMe;) (Samsung 7th Gen. V-NAND 176L (136T) 3D TLC; Samsung Pascal S4LV008 Controller)	Samsung SSD 990 PRO MZ-VP92T0B (2 TB; M.2 2280 PCIe 4.0 x4 NVMe;) (Samsung 7th Gen. V-NAND 176L (136T) 3D TLC; Samsung Pascal S4LV008 Controller)
Wi-Fi	1x 2.5 GbE RJ-45 (Realtek RTL8125BG) 1x GbE RJ-45 (Realtek RTL8111EPV) Mediatek MT7922 (RZ616) Wi-Fi 6E (2x2 802.11ax - 1.9 Gbps)	1x 2.5 GbE RJ-45 (Realtek RTL8125BG) 1x GbE RJ-45 (Realtek RTL8111EPV) Mediatek MT7922 (RZ616) Wi-Fi 6E (2x2 802.11ax - 1.9 Gbps)
Price (in USD, when built)	(Street Pricing on December 27th, 2023) US $570 (barebones) US $837 (as configured, no OS)	(Street Pricing on December 27th, 2023) US $570 (barebones) US $837 (as configured, no OS)

The next few sections will deal with comparative benchmarks for the above systems.

System Performance: UL and BAPCo Benchmarks

Our 2022 Q4 update to the test suite for Windows 11-based 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. New additions include BAPCo's CrossMark multi-platform benchmarking tool, as well as UL's Procyon benchmark suite.

UL PCMark 10

UL's PCMark 10 evaluates computing systems for various usage scenarios (generic / essential tasks such as web browsing and starting up applications, productivity tasks such as editing spreadsheets and documents, gaming, and digital content creation). We benchmarked select PCs with the PCMark 10 Extended profile and recorded the scores for various scenarios. These scores are heavily influenced by the CPU and GPU in the system, though the RAM and storage device also play a part. The power plan was set to Balanced for all the PCs while processing the PCMark 10 benchmark. The scores for each contributing component / use-case environment are also graphed below.

UL PCMark 10 - Performance Scores

The Phoenix-based systems are on top across every segment, except for the 'Essentials' workload. On account of their multi-core / multi-threaded performance advantage and the iGPU capabilities, that is no surprise. The single-threaded performance is a little bit behind Intel's Raptor Lake, even when the power limit is bumped up to 65W, and that is evident in the Essentials score.

UL Procyon v2.1.544

PCMark 10 utilizes open-source software such as Libre Office and GIMP to evaluate system performance. However, many of their professional benchmark customers have been requesting evaluation with commonly-used commercial software such as Microsoft Office and Adobe applications. In order to serve their needs, UL introduced the Procyon benchmark in late 2020. There are five benchmark categories currently - Office Productivity, AI Inference, Battery Life, Photo Editing, and Video Editing. AI Inference benchmarks are available only for Android devices, while the battery life benchmark is applicable to Windows devices such as notebooks and tablets. We presents results from our processing of the other three benchmarks.

UL Procyon - Office Productivity Scores

The workloads involved in the 'Office Productivity' segment all benefit from single-threaded performance, and that is evident as the Raptor Lake-based systems all have a good lead over the Phoenix-based ones. At 65W, the GTR7 manages to inch closer to the 35W Core i9-13900H in the GEEKOM Mini IT13, but the Phoenix systems are a step behind irrespective of the power limits.

From an energy consumption viewpoint, the 4nm process for Phoenix seems to be working wonders, as the 4X4 BOX-7840U in its 28W avatar manages to complete the workload with minimal energy consumption. It might not be the fastest machine for the workload, but it is quite power efficient in getting the job done.

Moving on to the evaluation of Adobe Photoshop and Adobe Lightroom, we find that the multi-core performance and iGPU advantage helps the Phoenix systems lock up the top spots. Interestingly, there is not much difference between the 40W and 65W versions, and this points to the overdrive to 65W being of not much benefit.

In fact, the energy consumption for the GTR7 is discernably more than the one for the 40W version of the 4X4 BOX-7840U, pointing to a sweet spot of around 35W for the processor to deliver a good performance / power efficiency balance.

UL Procyon evaluates performance for video editing using Adobe Premier Pro.

This workload appears to benefit from the extra power budget, and the 40W+ configurations of both Raptor Lake-P and Phoenix provide similar performance numbers.

However, the 4X4 BOX-7840U configurations are the most energy efficient of the lot, with the 40W mode presenting a good balance between performance and energy consumption with the second place in both graphs.

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 - Sub-Category Scores

The use of idle time compression favors Intel-based systems, but that is not indicative of real-world performance. Based on these results, it would appear that scripting and automating workloads to keep the system busy all the time would probably be better on the Intel-based PCs. Since CrossMark attempts to consolidate different workloads together without idle time intervals and play it back in a non-real-time environment, it is not entirely representative of real-world performance like PCMark 10 and UL Procyon. Therefore, tasks requiring frequent user interaction are better represented by those other benchmarks.

System Performance: Application-Specific Workloads

Standardized benchmarks such as UL's PCMark 10 and BAPCo's SYSmark take a holistic view of the system and process a wide range of workloads to arrive at a single score. Some systems are required to excel at specific tasks - so it is often helpful to see how a computer performs in specific scenarios such as rendering, transcoding, JavaScript execution (web browsing), etc. This section presents focused benchmark numbers for specific application scenarios.

3D Rendering - CINEBENCH R23

We use CINEBENCH R23 for 3D rendering evaluation. R23 provides two benchmark modes - single threaded and multi-threaded. Evaluation of different PC configurations in both supported modes provided us the following results. Interestingly, moving the Phoenix package beyond its suggested TDP of 28W actually seems to have a slight detrimental effect on single-threaded performance.

Amplifying the available power budget gives the eight cores more headroom to work with, and that reflects in the massive jump in the multi-threaded score for the 40W and 65W Phoenix versions.

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.

Software transcoding performance is a function of the number of available cores and available power budget (assuming that the instruction sets are similar). Therefore, it is no surprise to see the souped-up Phoenix versions in the top two spots. Interestingly, there doesn't seem to be much benefit in pushing Phoenix beyond 40W for transcoding, though 3D rendering was able to take advantage of the headroom in a much better manner.

Moving on to the evaluation of the hardware transcoding performance, it is not a surprise to see the 4X4 BOX-7840U configurations with its latest iGPU in the top two spots, differentiated only by their power budgets. The vce_h265 performance of the Radeon 780M has probably improved due to driver updates, as that could be the only possible explanation for the 65W GTR7 falling below the 40W 4X4 BOX-7840U by as much as 10 fps.

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.

Similar to 3D rendering and software transcoding, archival operations benefit immensely from multiple cores and higher power budgets. This translates to the 40W+ Phoenix configurations taking the top two spots for decompression. Compression is a different story, with the GTR7 falling behind significantly to the middle of the pack. It is possible that BIOS updates for performance tuning over the last few months have made Phoenix perform better in the 4X4 BOX-7840U series for such workloads even with lower power limits

Web Browsing: JetStream, Speedometer, and Principled Technologies WebXPRT4

Web browser-based workloads have emerged as a major component of the typical home and business PC usage scenarios. For headless systems, many applications based on JavaScript are becoming relevant too. In order to evaluate systems for their JavaScript execution efficiency, 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.

From a real-life workload perspective, we also process WebXPRT4 from Principled Technologies. WebXPRT4 benchmarks the performance of some popular JavaScript libraries that are widely used in websites.

JavaScript performance seems to be primarily impacted by single-threaded performance. As a result, we see the Intel-based systems comfortably surpassing the Phoenix-based ones despite the lower power budget.

Application Startup: GIMP 2.10.30

A new addition to our systems 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. The AMD Phoenix systems are bunched together in the middle, with the recent Intel-based systems proving to be faster at application loading.

Cryptography Benchmarks

Cryptography has become an indispensable part of our interaction with computing systems. Almost all modern systems have some sort of hardware-acceleration for making cryptographic operations faster and more power efficient. In the case of IoT servers, many applications - including web server functionality and VPN - need cryptography acceleration.

BitLocker is a Windows features that encrypts entire disk volumes. While drives that offer encryption capabilities are dealt with using that feature, most legacy systems and external drives have to use the host system implementation. Windows has no direct benchmark for BitLocker. However, we cooked up a BitLocker operation sequence to determine the adeptness of the system at handling BitLocker operations. We start off with a 4.5GB RAM drive in which a 4GB VHD (virtual hard disk) is created. This VHD is then mounted, and BitLocker is enabled on the volume. Once the BitLocker encryption process gets done, BitLocker is disabled. This triggers a decryption process. The times taken to complete the encryption and decryption are recorded. This process is repeated 25 times, and the average of the last 20 iterations is graphed below.

Hardware acceleration is available for the operations in all the considered systems. The time taken for processing is directly dependent on the available power budget. That said, the Phoenix system at 28W manages to eke out almost as good a performance as the 40W and 65W versions. Cryptography acceleration on AMD systems seems to be better utilized compared to Intel ones based on the benchmark results across a wide range of PCs.

GPU Performance: Synthetic Benchmarks

AMD's Phoenix SoCs include an integrated GPU with a microarchitectural update over the one in the Rembrandt-R SoCs. The new RDNA3 microarchitecture is present in the Ryzen 7 7840U in the form of the Radeon 780M. With 12 CUs and 768 shader units clocked at 2.7 GHz, AMD claims that the GPU should be capable of playing virtually any modern game at Full HD resolutions. The slide below references the HS series, but the iGPU clock rates are the same for the U-series as well as the HS-series as per AMD's specifications. Obviously, the power budget dictates sustained performance numbers.

For full-blown desktop systems or mini-PCs targeting the gaming market, we look at gaming workloads. However, for general-purpose mini-PC models like the ASRock Industrial 4X4 BOX-7840U, we restrict ourselves to a series of canned benchmarks from Kishonti and Futuremark / UL. Prior to that, a look at the capabilities of the GPU via AIDA64 and GPU-Z is warranted.

Rembrandt-R's Radeon 680M had already achieved an industry-first by integrating hardware-accelerated ray tracing, and the Radeon 780M in Phoenix builds upon that. There are some improvements in the media engine too (such as AV1 encode support), but those are not revealed in the GPU-Z screenshot. The remaining subsections below look into the performance aspects.

GFXBench

The DirectX 12-based GFXBench tests from Kishonti are cross-platform, and available all the way down to smartphones. As such, they are not very taxing for discrete GPUs and modern integrated GPUs. We processed the offscreen versions of the 'Aztec Ruins' benchmark.

At 1080p, the Core i7-1360P with a 40W power budget slightly outscores the Phoenix-based systems (including the 65W GTR7). However, at 1440p, the AMD RDNA2 and RDNA3 systems have a clean break-out and move to the top of the pack.

UL 3DMark

Four different workload sets were processed in 3DMark - Fire Strike, Time Spy, Night Raid, and Wild Life.

3DMark Fire Strike

The Fire Strike benchmark has three workloads. The base version is meant for high-performance gaming PCs. It uses DirectX 11 (feature level 11) to render frames at 1920 x 1080. The Extreme version targets 1440p gaming requirements, while the Ultra version targets 4K gaming system, and renders at 3840 x 2160. The graph below presents the overall score for the Fire Strike Extreme and Fire Strike Ultra benchmark across all the systems that are being compared.

UL 3DMark - Fire Strike Workloads

The lead for the RDNA2 iGPU over RPL-P's Iris Xe was already evident in previous reviews, and we see that Phoenix cements this position further - with a bigger lead as the power budget is increased.

3DMark Time Spy

The Time Spy workload has two levels with different complexities. Both use DirectX 12 (feature level 11). However, the plain version targets high-performance gaming PCs with a 2560 x 1440 render resolution, while the Extreme version renders at 3840 x 2160 resolution. The graphs below present both numbers for all the systems that are being compared in this review.

UL 3DMark - Time Spy Workloads

The observations made in the Fire Strike workloads hold true here also. However, we can note that the benefit of increasing the available power budget from 40W to 65W is not as significant as what is obtained in moving from 28W to 40W. This indicates that the 65W configuration is beyond its operating sweet spot.

3DMark Wild Life

The Wild Life workload was initially introduced as a cross-platform GPU benchmark in 2020. It renders at a 2560 x 1440 resolution using Vulkan 1.1 APIs on Windows. It is a relatively short-running test, reflective of mobile GPU usage. In mid-2021, UL released the Wild Life Extreme workload that was a more demanding version that renders at 3840 x 2160 and runs for a much longer duration reflective of typical desktop gaming usage.

UL 3DMark - Wild Life Workloads

At 1440p, the Iris Xe iGPU in RPL-P puts up a creditable performance and almost matches the Radeon 780M performance, but 4K leaves that well behind, as the Phoenix configurations take up the top three spots with a significant lead.

3DMark Night Raid

The Night Raid workload is a DirectX 12 benchmark test. It is less demanding than Time Spy, and is optimized for integrated graphics. The graph below presents the overall score in this workload for different system configurations.

The Radeon 680M and 780M configurations have a clear lead over the Iris Xe iGPU in RPL-P and the lead only grows with the available power budget for the AMD processor package.

3DMark Port Royal

UL introduced the Port Royal ray-tracing benchmark as a DLC for 3DMark in early 2019. The scores serve as an indicator of how the system handles ray-tracing effects in real-time.

Ray tracing performance has improved in moving from Radeon 680M to 780M, as expected from a new generation. Interestingly, there doesn't seem to be much benefit to power budgets beyond 40W for this workload.

Workstation Performance - SPECworkstation 3.1

SFF PCs traditionally do not lend themselves to workstation duties. However, a recent trend towards miniaturized workstations has been observed. While the 4X4 BOX-7840U is primarily marketed towards office workloads and home users, its capabilities encouraged us to benchmark the system for both content creation workloads as well as professional applications. Towards this, we processed the SPEC benchmark geared towards workstations without discrete GPUs - SPECworkstation 3.10.

SPECworkstation 3.1

The SPECworkstation 3.1 benchmark measures workstation performance based on a number of professional applications. It includes more than 140 tests based on 30 different workloads that exercise the CPU, graphics, I/O and memory hierarchy. These workloads fall into different categories.

• Media and Entertainment (3D animation, rendering)
• Product Development (CAD/CAM/CAE)
• Life Sciences (medical, molecular)
• Financial Services
• Energy (oil and gas)
• General Operations
• GPU Compute

Individual scores are generated for each test and a composite score for each category is calculated based on a reference machine (HP Z240 tower workstation using an Intel E3-1240 v5 CPU, an AMD Radeon Pro WX3100 GPU, 16GB of DDR4-2133, and a SanDisk 512GB SSD). Official benchmark results generated automatically by the benchmark itself are linked in the table below for the systems being compared.

SPECworkstation 3.1 Official Results (2K)
ASRock 4X4 BOX-7840U (Performance)	Run Summary
ASRock 4X4 BOX-7840U (Normal)	Run Summary
ASRock 4X4 BOX-5800U (Performance)	Run Summary
Beelink GTR7	Run Summary
ASRock NUC BOX-1360P-D5 (Performance)	Run Summary
ASRock 4X4 BOX-7735U (Performance)	Run Summary
Intel NUC13ANKi7 (Arena Canyon)	Run Summary
ASRock NUCS BOX-1360P-D4	Run Summary
ASRock 4X4 BOX-4800U	Run Summary
GEEKOM A5	Run Summary
Intel NUC12WSKi7 (Wall Street Canyon)	Run Summary
ASRock NUC BOX-1260P	Run Summary
GEEKOM AS 6 (ASUS PN53)	Run Summary
GEEKOM Mini IT13	Run Summary

Details of the tests in each category, as well as an overall comparison of the systems on a per-category basis are presented below.

Media and Entertainment

The Media and Entertainment category comprises of workloads from five distinct applications:

• The Blender workload measures system performance for content creation using the open-source Blender application. Tests include rendering of scenes of varying complexity using the OpenGL and ray-tracing renderers.
• The Handbrake workload uses the open-source Handbrake application to transcode a 4K H.264 file into a H.265 file at 4K and 2K resolutions using the CPU capabilities alone.
• The LuxRender workload benchmarks the LuxCore physically based renderer using LuxMark.
• The Maya workload uses the SPECviewperf 13 maya-05 viewset to replay traces generated using the Autodesk Maya 2017 application for 3D animation.
• The 3ds Max workload uses the SPECviewperf 13 3dsmax-06 viewset to replay traces generated by Autodesk's 3ds Max 2016 using the default Nitrous DX11 driver. The workload represents system usage for 3D modeling tasks.

Product Development

The Product Development category comprises of eight distinct workloads:

• The Rodinia (CFD) workload benchmarks a computational fluid dynamics (CFD) algorithm.
• The WPCcfd workload benchmarks another CFD algorithm involving combustion and turbulence modeling.
• The CalculiX workload uses the Calculix finite-element analysis program to model a jet engine turbine's internal temperature.
• The Catia workload uses the catia-05 viewset from SPECviewperf 13 to replay traces generated by Dassault Systemes' CATIA V6 R2012 3D CAD application.
• The Creo workload uses the creo-02 viewset from SPECviewperf 13 to replay traces generated by PTC's Creo, a 3D CAD application.
• The NX workload uses the snx-03 viewset from SPECviewperf 13 to replay traces generated by the Siemens PLM NX 8.0 CAD/CAM/CAE application.
• The Solidworks workload uses the sw-04 viewset from SPECviewperf 13 to replay traces generated by Dassault Systemes' SolidWorks 2013 SP1 CAD/CAE application.
• The Showcase workload uses the showcase-02 viewset from SPECviewperf 13 to replay traces from Autodesk's Showcase 2013 3D visualization and presentation application

Life Sciences

The Life Sciences category comprises of four distinct test sets:

• The LAMMPS set comprises of five tests simulating different molecular properties using the LAMMPS molecular dynamics simulator.
• The NAMD set comprises of three tests simulating different molecular interactions.
• The Rodinia (Life Sciences) set comprises of four tests - the Heartwall medical imaging algorithm, the Lavamd algorithm for calculation of particle potential and relocation in a 3D space due to mutual forces, the Hotspot algorithm to estimate processor temperature with thermal simulations, and the SRAD anisotropic diffusion algorithm for denoising.
• The Medical workload uses the medical-02 viewset from SPECviewperf 13 to determine system performance for the Tuvok rendering core in the ImageVis3D volume visualization program.

Financial Services

The Financial Services workload set benchmarks the system for three popular algorithms used in the financial services industry - the Monte Carlo probability simulation for risk assessment and forecast modeling, the Black-Scholes pricing model, and the Binomial Options pricing model.

Energy

The Energy category comprises of workloads simulating various algorithms used in the oil and gas industry:

• The FFTW workload computes discrete Fourier transforms of large matrices.
• The Convolution workload computes the convolution of a random 100x100 filter on a 400 megapixel image.
• The SRMP workload processes the Surface-Related Multiples Prediction algorithm used in seismic data processing.
• The Kirchhoff Migration workload processes an algorithm to calculate the back propogation of a seismic wavefield.
• The Poisson workload takes advantage of the OpenMP multi-processing framework to solve the Poisson's equation.
• The Energy workload uses the energy-02 viewset from SPECviewperf 13 to determine system performance for the open-source OPendTec seismic visualization application.

General Operations

In the General Options category, the focus is on workloads from widely used applications in the workstation market:

• The 7zip workload represents compression and decompression operations using the open-source 7zip file archiver program.
• The Python workload benchmarks math operations using the numpy and scipy libraries along with other Python features.
• The Octave workload performs math operations using the Octave programming language used in scientific computing.
• The Storage workload evaluates the performance of the underlying storage device using transaction traces from multiple workstation applications.

GPU Compute

In the GPU Compute category, the focus is on workloads taking advantage of the GPU compute capabilities using either OpenCL or CUDA, as applicable:

• The LuxRender benchmark is the same as the one seen in the media and entertainment category.
• The Caffe benchmark measures the performance of the Caffe deep-learning framework.
• The Folding@Home benchmark measures the performance of the system for distributed computing workloads focused on tasks such as protein folding and drug design.

We only process the OpenCL variants of the benchmark, as the CUDA version doesn't process correctly with default driver installs.

Most of the workloads in the SPECworkstation 3.1 suite benefit immensely from multiple fast cores. The AMD Phoenix systems have an advantage in that aspect, and it shows across all of the workloads. Interestingly, the 65W GTR7 doesn't always come out on top, and we suspect this is due to the availability of a more mature BIOS in the 4X4 BOX-7840U.

System Performance: Multi-Tasking

One of the key drivers of advancements in computing systems is multi-tasking. On mobile devices, this is quite lightweight - cases such as background email checks while the user is playing a mobile game are quite common. Towards optimizing user experience in those types of scenarios, mobile SoC manufacturers started integrating heterogenous CPU cores - some with high performance for demanding workloads, while others were frugal in terms of both power consumption / die area and performance. This trend is now slowly making its way into the desktop PC space.

Multi-tasking in typical PC usage is much more demanding compared to phones and tablets. Desktop OSes allow users to launch and utilize a large number of demanding programs simultaneously. Responsiveness is dictated largely by the OS scheduler allowing different tasks to move to the background. The processor is required to work closely with the OS thread scheduler to optimize performance in these cases. Keeping these aspects in mind, the evaluation of multi-tasking performance is an interesting subject to tackle.

We have augmented our systems benchmarking suite to quantitatively analyze the multi-tasking performance of various platforms. The evaluation involves triggering a ffmpeg transcoding task to transform 1716 3840x1714 frames encoded as a 24fps AVC video (Blender Project's 'Tears of Steel' 4K version) into a 1080p HEVC version in a loop. The transcoding rate is monitored continuously. One complete transcoding pass is allowed to complete before starting the first multi-tasking workload - the PCMark 10 Extended bench suite. A comparative view of the PCMark 10 scores for various scenarios is presented in the graphs below. Also available for concurrent viewing are scores in the normal case where the benchmark was processed without any concurrent load, and a graph presenting the loss in performance.

UL PCMark 10 Load Testing - Digital Content Creation Scores

UL PCMark 10 Load Testing - Productivity Scores

UL PCMark 10 Load Testing - Essentials Scores

UL PCMark 10 Load Testing - Gaming Scores

UL PCMark 10 Load Testing - Overall Scores

Even at 28W, the Phoenix SoC has plenty of headroom to tackle multiple concurrent demanding workloads triggered by mainstream users (of the type benchmarked by PCMark 10). The 4X4 BOX-7840U configurations already had leading performance in the absence of concurrent loading in most segments. Activating the ffmpeg load reduces the raw score, but the performance loss is less than the competition and the Phoenix systems continue to stay on top.

Following the completion of the PCMark 10 benchmark, a short delay is introduced prior to the processing of Principled Technologies WebXPRT4 on MS Edge. Similar to the PCMark 10 results presentation, the graph below show the scores recorded with the transcoding load active. Available for comparison are the dedicated CPU power scores and a measure of the performance loss.

Principled Technologies WebXPRT4 Load Testing Scores (MS Edge)

The observations made for the overall PCMark 10 scores continue to hold for WebXPRT 4 also. Interestingly, the handling by Phoenix is good enough to even prop up the systems to the top spot in the presence of the transcoding load despite being in the middle of the pack without the extra load.

The final workload tested as part of the multitasking evaluation routine is CINEBENCH R23.

3D Rendering - CINEBENCH R23 Load Testing - Single Thread Score

3D Rendering - CINEBENCH R23 Load Testing - Multiple Thread Score

Similar to the web browsing workload, the Intel systems lose their ST performance advantage in the presence of concurrent loading. As a result, the Phoenix systems rise to the top. In the MT version, they retain their top billing despite the drop in the raw scores.

After the completion of all the workloads, we let the transcoding routine run to completion. The monitored transcoding rate throughout the above evaluation routine (in terms of frames per second) is graphed below.

In the above scenario, the transcoding task is supposed to happen in the background. The user should ideally not care about the transcoding performance in that case, but it is still interesting to look into how that varies based on the concurrent load being applied.

ASRock Industrial 4X4 BOX-7840U (Ryzen 7 7840U @ 28W) ffmpeg Transcoding Rate (Multi-Tasking Test)
Task Segment	Transcoding Rate (FPS)
	Minimum	Average	Maximum
Transcode Start Pass	4.5	14.5	44.5
PCMark 10	0	12.79	44.5
WebXPRT 4	4.5	12.66	22.5
Cinebench R23	0	13.55	42
Transcode End Pass	3	14.65	45

ASRock Industrial 4X4 BOX-7840U (Ryzen 7 7840U @ 40W) ffmpeg Transcoding Rate (Multi-Tasking Test)
Task Segment	Transcoding Rate (FPS)
	Minimum	Average	Maximum
Transcode Start Pass	5.5	17.12	51.5
PCMark 10	0	14.81	49
WebXPRT 4	6	14.72	30.5
Cinebench R23	2.5	15.16	59
Transcode End Pass	6	17.03	50.5

Despite the addition of heavy concurrent loading, we see that the drop in transcoding performance is not too worrisome outside of full saturation regions from the foreground tasks.

Digital Signage and HTPC Credentials

The 2022 Q4 update to our system reviews brings an updated HTPC evaluation suite for systems. After doing away with the evaluation of display refresh rate stability and Netflix streaming evaluation, the local media playback configurations have also seen a revamp. This section details each of the workloads processed on the ASRock 4X4 BOX-7840U (in both modes) as part of the HTPC suite.

YouTube Streaming Efficiency

YouTube continues to remain one of the top OTT platforms, primarily due to its free ad-supported tier. Our HTPC test suite update retains YouTube streaming efficiency evaluation as a metric of OTT support in different systems. Mystery Box's Peru 8K HDR 60FPS video is the chosen test sample. On PCs running Windows, it is recommended that HDR streaming videos be viewed using the Microsoft Edge browser after putting the desktop in HDR mode.

YouTube Streaming Statistics - Normal Mode

YouTube Streaming Statistics - Performance Mode

The iGPU in ASRock 4X4 BOX-7840U supports hardware decoding of VP9 Profile 2, and we see the stream encoded with that codec being played back. The streaming is perfect, thanks to the powerful GPU and hardware decoding support - the few dropped frames observed in the statistics below are due to mouse clicks involved in bringing up the overlay.

The streaming efficiency-related aspects such as GPU usage and at-wall power consumption are also graphed below.

The 4X4 BOX-7840U in the Normal Mode (28W) ties with the NUC BOX-1360P/D5 for the best energy efficiency in network streaming.

Hardware-Accelerated Encoding and Decoding

The transcoding benchmarks in the systems performance section presented results from evaluating the VCE encoder within Handbrake's framework. The capabilities of the decoder engine are brought out by DXVAChecker.

Video Decoding Hardware Acceleration in ASRock 4X4 BOX-7840U

On paper, this codec list is quite comprehensive and should cover most home consumer and digital signage requirements.

Local Media Playback

Evaluation of local media playback and video processing is done by playing back files encompassing a range of relevant codecs, containers, resolutions, and frame rates. A note of the efficiency is also made by tracking GPU usage and power consumption of the system at the wall. Users have their own preference for the playback software / decoder / renderer, and our aim is to have numbers representative of commonly encountered scenarios. Our Q4 2022 test suite update replaces MPC-HC (in LAV filters / madVR modes) with mpv. In addition to being cross-platform and open-source, the player allows easy control via the command-line to enable different shader-based post-processing algorithms. From a benchmarking perspective, the more attractive aspect is the real-time reporting of dropped frames in an easily parseable manner. The players / configurations considered in this subsection include:

• VLC 3.0.20
• Kodi 20.2
• mpv 0.37 (hwdec auto, vo=gpu-next)
• mpv 0.37 (hwdec auto, vo=gpu-next, profile=gpu-hq)

Fourteen test streams (each of 90s duration) were played back from the local disk with an interval of 30 seconds in-between. Various metrics including GPU usage, at-wall power consumption, and total energy consumption were recorded during the course of this playback.

All our playback tests were done with the desktop HDR setting turned on. It is possible for certain system configurations to automatically turn on/off the HDR capabilities prior to the playback of a HDR video, but, we didn't take advantage of that in our testing.

The 8Kp60 AV1 playback is at half-rate even when hardware acceleration is in the picture, and this could point to driver issues or even software compatibility. This is a retrogression from the progress we saw when evaluating the Beelink GTR7 where the playback was perfect without any dropped frames for this case. In terms of energy efficiency, the Phoenix systems are at the top, except when mpv is used.

Power Consumption and Thermal Characteristics

The power consumption at the wall was measured with a 4K display being driven through the HDMI port of the system. In the graph below, we compare the idle and load power of the ASRock 4X4 BOX-7840U with other systems evaluated before. For load power consumption, we ran the AIDA64 System Stability Test with various stress components, as well as our custom stress test with Prime95 / Furmark, and noted the peak as well as idling power consumption at the wall.

The numbers are consistent with the TDP and suggested PL1 / PL2 values for the processors in the systems, and do not come as any surprise. Idle power consumption of Intel-based systems is very impressive, with the Arena Canyon NUC able to hit a sub-5W number. The 4X4 BOX-7840U in its 28W mode idles at 8.17W. While a part of this number could be attributed to the Gen 4 x4 NVMe SSD in the system, we believe ASRock Industrial can tweak BIOS settings further to optimize this number.

Stress Testing

Our thermal stress routine is a combination of Prime95, Furmark, and Finalwire's AIDA64 System Stability Test. The following 9-step sequence is followed, starting with the system at idle:

• Start with the Prime95 stress test configured for maximum power consumption
• After 30 minutes, add Furmark GPU stress workload
• After 30 minutes, terminate the Prime95 workload
• After 30 minutes, terminate the Furmark workload and let the system idle
• After 30 minutes of idling, start the AIDA64 System Stress Test (SST) with CPU, caches, and RAM activated
• After 30 minutes, terminate the previous AIDA64 SST and start a new one with the GPU, CPU, caches, and RAM activated
• After 30 minutes, terminate the previous AIDA64 SST and start a new one with only the GPU activated
• After 30 minutes, terminate the previous AIDA64 SST and start a new one with the CPU, GPU, caches, RAM, and SSD activated
• After 30 minutes, terminate the AIDA64 SST and let the system idle for 30 minutes

Traditionally, this test used to record the clock frequencies - however, with the increasing number of cores in modern processors and fine-grained clock control, frequency information makes the graphs cluttered and doesn't contribute much to understanding the thermal performance of the system. The focus is now on the power consumption and temperature profiles to determine if throttling is in play.

There were some unforeseen issues with tracking STAPM numbers and other power-related quantities in the system. Therefore, only the at-wall numbers are presented. They track our observations in the 4X4 BOX-7735U review, which helped in inferring the package power limits in both modes.

The thermal solution is able to easily handle the 28W configuration, but we suspect it is not enough to handle the 40W version even with the fan running at full tilt. The comparison of the temperature numbers show the die temperature reaching 95C+ in the performance mode, while it is well below 85C even under extreme stress in the normal mode. This was unlike our experience with ASRock Industrial's Rembrandt-R system, where both modes came out with flying colors in the stress test. Additionally, the lack of support for an effective thermal solution for the NVMe SSD may limit storage performance. We see temperatures in excess of 75C (and possibly kept in check by the SSD controller's throttling) for the drive.

Miscellaneous Aspects and Concluding Remarks

Networking and storage are aspects that may be of vital importance in specific PC use-cases. The ASRock 4X4 BOX-7840U comes with dual LAN ports. From a management perspective, the system is DASH-enabled. This out-of-band management path for deployment and maintenance is a huge plus for IT departments. The WLAN component is with the AMD-branded Mediatek-based Wi-Fi 6E solution that was also used in the previous generation 4X4 BOX-7735U.

On the storage side, we configured the barebones system with a high-end Gen 4 NVMe SSD - the Samsung SSD 990 PRO. From a benchmarking perspective, we provide results from the WPCstorage test of SPECworkstation 3.1. This benchmark replays access traces from various programs used in different verticals and compares the score against the one obtained with a 2017 SanDisk 512GB SATA SSD in the SPECworkstation 3.1 reference system.

SPECworkstation 3.1.0 - WPCstorage SPEC Ratio Scores

The graphs above present results for different verticals, as grouped by SPECworkstation 3.1. The storage workload consists of 60 subtests. Access traces from CFD solvers and programs such as Catia, Creo, and Soidworks come under 'Product Development'. Storage access traces from the NAMD and LAMMPS molecular dynamics simulator are under the 'Life Sciences' category. 'General Operations' includes access traces from 7-Zip and Mozilla programs. The 'Energy' category replays traces from the energy-02 SPECviewperf workload. The 'Media and Entertainment' vertical includes Handbrake, Maya, and 3dsmax. As expected, the Samsung 990 PRO comes out with a huge lead in most workloads. A couple of segments see the SSD slipping to the middle. It is not clear how much of this can be attributed to the SSD itself, as against throttling due to the lack of an effective thermal solution. In general, it is advisable to use low-power DRAM-less NVMe SSDs in such systems, as the performance benefits of high-end SSDs can't be realized in this form factor.

Closing Thoughts

The ASRock 4X4 BOX-7840U provided us with the opportunity to evaluate an AMD Phoenix-based UCFF PC with premium features. The Phoenix platform had already impressed us in its 65W avatar in the Beelink GTR7. That product had its share of early-adopter troubles, which we thankfully didn't encounter in the ASRock Industrial 4X4 BOX-7840U. There are still a few worrisome iGPU driver issues related to media playback that keep reappearing in different releases. Hopefully that is something AMD can address, now that they are gaining significant market share in this segment.

In terms of I/O capabilities, the two USB4 40 Gbps ports bring these systems on par with the mainstream Intel NUCs. The iGPU performance is better for most workloads. Unless single-threaded performance is of vital importance, these Phoenix systems offer a much better value proposition compared to the Raptor Lake-P ones.

ASRock Industrial has delivered two different AMD-based UCFF system sets this year. The Rembrandt-R ones came to market only in Q1, and the Phoenix-based ones have come in early Q4. Due to the relatively short time between the introductions, many of the concerns we voiced in the 4X4 BOX-7735U review remain unaddressed. In the performance mode, the fan is set to run at 100% duty cycle - even when the system is at idle. The company could have delivered support for similar TDP values with a better noise profile by allowing for fine-grained fan speed control based on the SoC temperature. A thermal solution for the M.2 SSD is needed, and at least one of the two USB 2.0 ports in the rear panel needs to be replaced with a USB 3.2 Gen 2 port.

At $570 for the barebones version, there is a big pricing advantage over the Arena Canyon NUC and even ASRock Industrial's own NUC BOX-1300 series. Unless the dual USB 3.2 Gen 2 rear ports and slight edge in ST performance are needed, it is difficult to recommend the Arena Canyon NUC or any other RPL-P platform. The Phoenix platform wins on the energy efficiency metric as well.

Looking forward to 2024, ASRock Industrial has already introduced the Phoenix-R-based 4X4 BOX-8040 series and Meteor Lake-based NUC 100 Ultra Box series of UCFF systems. With Intel going in for architectural changes in both the CPU and GPU (compared to the refresh approach from AMD), we might finally see Intel reclaim some of the lost ground. That said, newer generations always come at a price premium, and the software / BIOS also needs some time to mature. Given that context, the value proposition of AMD Phoenix-based systems like the ASRock Industrial 4X4 BOX-7840U will continue to be very attractive for the next few months.