The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
Verdict: The ASUS RS520QA-E13-RS8U is a specialized memory-density platform, not a universal high-performance server. Its CXL Type-3 memory expansion lets each compact, single-socket AMD EPYC node add capacity without sacrificing four-node-per-2U density or moving the local memory subsystem to 2DPC. That makes it especially interesting for capacity-bound virtualization. The trade-offs are higher CXL memory latency, more platform-specific hardware and firmware, and limited storage and PCIe expansion.
This assessment is based on ServeTheHome’s June 9, 2025 review, which disclosed sponsorship and special access from ASUS. It should be read as a review of the tested configuration rather than independent testing of every possible production configuration.
What the ASUS RS520QA-E13-RS8U is
The RS520QA-E13-RS8U is a 2U server containing four front-accessible, single-socket AMD EPYC nodes. It supports AMD EPYC 9005 “Turin” processors, with the reviewed node using the 128-core EPYC 9755. ASUS rates the platform for processors with up to 400W configurable thermal design capability, although actual support depends on the CPU, firmware, cooling configuration and vendor qualification.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
The chassis is approximately 900mm (35.4 inches) deep. Nodes are serviceable from the cold aisle, while the redundant Delta power supplies sit at the rear in the hot aisle. This arrangement prioritizes compute and memory density: four independent servers occupy only 2U, but each node has less room for storage and expansion than a conventional full-width server.
#1 Best Overall
- Server Memory RAM compatible for the ASUS Rack models: RS500-E8-RS4, RS520-E8-RS8 V2, RS700-E8-RS4, RS700-E8-RS8 V2, RS720Q-E8-RS12
- 128 GB Kit ( 8 x 16 GB Modules ) | DDR4 2400 MHz ( PC4-19200 / PC4-2400 ) | DDR4 DIMM ( 288-Pin ) | 1Rx4 | ECC Registered | 1.2V - DDR4 Standard Voltage
- ECC Registered; JEDEC DDR4 standard 1.2V
- Improves performance by expanding your system's memory (RAM) capacity, allowing your system to take on more while maintaining a fast and smooth experience
- Quick and easy to install at home, no expertise required (Please refer to your system's manual for seating and channel guidelines)
Why CXL is needed here
A typical EPYC memory configuration can use two DIMMs per memory channel, but a half-width four-node chassis cannot easily provide the same physical DIMM layout as a full-width single-node motherboard. The alternatives each involve a compromise:
| Approach | Advantage | Trade-off |
|---|---|---|
| Larger DIMMs | Keeps memory directly attached to the CPU | Higher cost, availability constraints and platform capacity limits |
| 2DPC local memory | Adds directly attached capacity | Can reduce supported memory speed and requires more motherboard space |
| Second CPU socket | Adds memory channels and capacity | More power, heat, cost and NUMA complexity |
| CXL Type-3 memory | Adds capacity outside the cramped node motherboard while retaining single-socket density | Higher latency, additional components and platform-specific support |
The RS520QA-E13-RS8U uses CXL to place additional DDR5 memory outside the node’s conventional CPU-attached DIMM slots. The result is a way to expand capacity without abandoning the four-node-per-2U design or forcing the local memory channels into a 2DPC configuration.
Physical design and node hardware
Each node includes two 2.5-inch NVMe bays, a low-profile PCIe Gen5 x16 slot connected through a riser, an OCP NIC 3.0 slot, a dedicated management port, two USB 3 ports, VGA and a POST-code display. Cooling is handled by a large heatsink with eight heatpipes.
Recommended Free Tools
The external CXL hardware is located toward the rear of the chassis. The design uses redundant power supplies, node-specific connection boards, PCIe/CXL retimers and high-speed cables routed around the fan region. This is a clever way to fit CXL connectivity into a half-width node, but it also means more signal-path components than a conventional motherboard DIMM layout. The review does not establish long-term failure rates or field-service data for the retimers, cables or external boards.
How the CXL memory is connected
The simplified path is:
EPYC CPU → PCIe/CXL connection → retimer → cable → Montage CXL controller → DDR5 DIMMs
Each node uses two ASUS CXL-R2H-Q boards. Each board contains two Montage CXL memory controllers, four DDR5 DIMM slots and power inputs. Together, the two boards add eight CXL-connected DDR5 DIMM positions per node.
In other words, CXL does not mean that eight extra DIMMs are inserted into the front node itself. The memory is installed on external CXL expansion boards and reached through the node’s CXL/PCIe signal path.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →The tested 1.28TB memory configuration
ServeTheHome tested one node with twelve directly attached 64GB DDR5 DIMMs and eight additional 64GB DDR5 DIMMs connected through CXL:
Rank #2
- Server Memory RAM compatible for the ASUS Rack models: RS500-E8, RS520-E8, RS520-E8-RS8, RS700-E9-RS12, RS700-E9-RS4, RS720Q-E8-RS12
- 64 GB Kit ( 2 x 32 GB Modules ) | DDR4 2666 MHz ( PC4-21300 / PC4-2666 ) | DDR4 DIMM ( 288-Pin ) | 2Rx4 | ECC Registered | 1.2V - DDR4 Standard Voltage
- ECC Registered; JEDEC DDR4 standard 1.2V
- Improves performance by expanding your system's memory (RAM) capacity, allowing your system to take on more while maintaining a fast and smooth experience
- Quick and easy to install at home, no expertise required (Please refer to your system's manual for seating and channel guidelines)
| Memory area | Population | Capacity |
|---|---|---|
| CPU-attached DDR5 | 12 × 64GB | 768GB |
| CXL-attached DDR5 | 8 × 64GB | 512GB |
| Total per tested node | 20 × 64GB | 1.28TB |
The 1.28TB figure is per node, not automatically the total for the entire four-node chassis. A four-node chassis populated identically would mathematically amount to four times that capacity, but the detailed review configuration is presented at the individual-node level and should not be treated as a verified full-chassis bill of materials.
The four Montage controllers each exposed 128GB through two 64GB DIMMs. The CXL memory was reported operating at DDR5-4400.
How the operating system sees CXL memory
The tested system exposed local CPU-attached memory as NUMA node 0. The CXL memory appeared as NUMA node 1, with memory but no CPU cores associated with it.
This is the most important operational fact about the platform. CXL memory is not indistinguishable from local DDR5. It is better understood as a separate, slower memory tier. Latency-sensitive working sets should generally remain in local memory, while CXL capacity can absorb less latency-sensitive data or prevent a capacity-constrained workload from paging or reducing its virtual-machine density.
Workload behavior will depend on the operating system, hypervisor and allocation policy. Before deployment, administrators should inspect:
- NUMA topology and distance information;
- virtual-machine placement rules;
- first-touch and memory-allocation policies;
- page migration or memory-tiering behavior;
- whether hot pages can accidentally land in CXL memory; and
- whether the hypervisor explicitly understands the CXL-backed NUMA node.
A system that fits comfortably in local memory may gain little from CXL and could lose performance if hot pages are placed remotely. A system that is constrained by total capacity may benefit substantially even though CXL access is slower.
Performance: what the review demonstrates
CPU cooling
ServeTheHome first examined whether the compact four-node chassis could cool the EPYC 9755. Its result was approximately comparable to a 1U single-node EPYC 9005 server, within normal run-to-run variation. That supports the basic feasibility of putting a high-power EPYC processor into this dense chassis, but it is not evidence that every CPU, ambient temperature or firmware profile will behave identically.
Local memory versus CXL memory
The meaningful comparison was not “fast server versus slow server.” It was a comparison between different ways of reaching a desired capacity:
Rank #3
- Next gen Wi-Fi standard 802.11Ax Wi-Fi standard for better efficiency and throughput; ultrafast Wi-Fi speed 6000 Mbps Wi-Fi speed to handle even the busiest network with ease.Power Supply : AC Input : 110V-240V(50-60Hz), DC Output : 19 V with max. 2.37 A current.
- Certified for Humans – Smart home made easy for non-experts. Setup with Alexa is simple.
- Wider usage and more convenience 4 antennas plus 8 LAN ports to support more clients at the same time
- Commercial-grade security – AiProtection powered by Trend Micro blocks internet security threats for all your connected smart devices
- Better partner with mesh system; compatible with ASUS AiMesh Wi-Fi system for seamless whole home coverage’s support: Windows 10, Windows 8, Windows 7, Mac OS X 10.6, Mac OS X 10.7, Mac OS X 10.8
- a 12-DIMM local-memory configuration;
- a 24-DIMM 2DPC configuration; and
- a configuration combining 12 local DIMMs with eight CXL DIMMs.
2DPC can add local capacity, but memory speed may fall. CXL adds capacity while allowing the direct CPU-attached memory channels to remain in the preferred 1DPC arrangement. CXL memory also provides a separate path rather than simply adding more DIMMs to the CPU’s direct channels, although that benefit comes with higher latency.
The virtualization testing supports a specific conclusion: CXL is most valuable when memory capacity is the limiting resource. It can allow more virtual machines or larger VM allocations without immediately resorting to paging, overcommitment or a larger server. The advantage narrows, and may disappear, when the workload is primarily limited by memory latency or bandwidth.
The available review material supports this qualitative conclusion more clearly than a universal performance percentage. It does not establish one latency or bandwidth figure that can be applied to every EPYC model, DIMM population, firmware version or workload.
Management and serviceability
Each node uses an ASPEED AST2600 baseboard management controller with ASUS ASMB12-iKVM based on MegaRAC SP-X. The review observed HTML5 iKVM, normal out-of-band management, power and fan telemetry, system-management information and visibility into the Montage controllers.
That visibility is valuable because a CXL deployment has more components to monitor than a conventional DIMM-only system. However, firmware interfaces can change. The review demonstrates what was visible on the tested system; it does not guarantee that every production firmware revision exposes identical CXL controls or telemetry fields.
Cold-aisle node replacement is another practical advantage. Operators can service the compute nodes from the front while the power supplies and rear CXL infrastructure remain at the back. Buyers should nevertheless confirm the exact replacement procedure for CXL boards, retimers and cables, including whether a board replacement requires a full node shutdown.
What CXL improves—and what it does not
CXL improves
- Memory capacity per single-socket CPU.
- Memory density per rack unit.
- The ability to retain four nodes in 2U.
- The ability to keep local DDR5 at a 1DPC configuration.
- Virtualization density when total RAM, rather than memory latency, is the bottleneck.
- Flexibility in placing additional memory outside a cramped node motherboard.
CXL does not automatically improve
- Minimum memory latency.
- Local DDR5 bandwidth.
- Every application’s performance.
- Storage capacity or PCIe expansion.
- GPU density.
- Software simplicity or operational maturity.
Calling CXL “faster RAM” would be misleading. Its main value here is capacity and topology: it creates another memory tier while preserving the compact node design.
Free tools Windows power users keep installed
One-click scans. No signup required.
Who should consider this server?
The strongest use cases are:
- Virtualization-heavy infrastructure: especially when RAM capacity limits VM consolidation before CPU capacity does.
- Memory-constrained cloud platforms: where rack-level memory density matters more than maximizing local-memory performance for every workload.
- CXL evaluation and development: for organizations building NUMA-aware provisioning, monitoring and memory-tiering workflows.
- Dense compute clusters: where four independent nodes per 2U are more valuable than large local storage or extensive PCIe expansion.
The platform is a poorer fit for latency-critical databases, tightly optimized HPC codes, workloads that require all memory to be local, storage-heavy nodes and organizations that cannot validate CXL firmware and hypervisor behavior.
Rank #4
- Server Memory RAM compatible for the ASUS Rack models: RS100-X7, RS300-H8-PS12, RS320Q-E7/RS12, S1016P
- Single 8 GB Module | DDR3 1333 MHz ( PC3-10600 / PC3-1333 ) | DDR3 DIMM ( 240-Pin ) | 2Rx8 | ECC Unbuffered | 1.5V - DDR3 Standard Voltage
- ECC Unbuffered; JEDEC DDR3 standard 1.5V
- Improves performance by expanding your system's memory (RAM) capacity, allowing your system to take on more while maintaining a fast and smooth experience
- Quick and easy to install at home, no expertise required (Please refer to your system's manual for seating and channel guidelines)
How it compares with the alternatives
Larger local DIMMs
Larger DIMMs avoid the additional CXL latency and are operationally simpler. They may be preferable when the platform supports the required capacity and the cost and supply situation are acceptable. The available review does not provide a current price comparison, so CXL should not be described as universally cheaper.
A 2DPC local-memory configuration
2DPC offers more directly attached memory, but it can reduce memory speed and may require a larger full-width server. In this chassis, a 24-DIMM arrangement cannot physically fit in the half-width node, so choosing 2DPC may mean giving up four-node-per-2U density.
A dual-socket EPYC server
A second socket can provide additional memory channels and capacity, but it increases power consumption, cooling requirements, cost and NUMA complexity. CXL offers a different compromise: retain a single socket and add a slower memory tier. Which is better depends on whether the workload values local latency, total capacity, node density or power efficiency most.
A conventional high-I/O server
The RS520QA-E13-RS8U is not a replacement for a storage-dense or expansion-heavy 2U system. Each node has two NVMe bays, one low-profile PCIe Gen5 x16 slot and an OCP NIC 3.0 slot. Buyers needing many NVMe drives, multiple accelerators or numerous PCIe cards should compare against a lower-density full-width server.
Deployment risks and validation checklist
CXL support is a platform integration problem involving the EPYC processor, motherboard, retimers, CXL controllers, DIMMs, firmware, operating system, hypervisor and management software. A generic statement that a component is “CXL-capable” is not enough.
Before placing an order, ask ASUS or an authorized reseller to confirm:
- Supported EPYC 9005 processor SKUs and power profiles.
- Validated CXL-R2H-Q board and DIMM combinations.
- Maximum local and CXL memory per node.
- Maximum supported memory for the complete four-node chassis.
- BIOS, CXL-device firmware and operating-system support.
- Hypervisor certification and NUMA behavior.
- RAS features, error reporting and replacement procedures.
- CXL hot-plug or hot-replacement behavior, if required.
- Power draw under the intended CPU and memory population.
- Noise limits and permitted inlet temperature.
- Warranty coverage for retimers, cables, controllers and external CXL boards.
- Actual delivery date and availability of replacement parts.
Bottom line
The ASUS RS520QA-E13-RS8U is a compelling demonstration of where CXL can matter in real server design. In the tested setup, it delivered 1.28TB per node by combining 768GB of local DDR5 with 512GB of CXL-attached memory, while retaining a compact single-socket node and four-node-per-2U density.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallThat is a strong answer to a specific problem: adding memory capacity when motherboard space, rack density and local 1DPC operation all matter. It is not a general solution for workloads that demand uniformly low latency, maximum memory bandwidth, extensive storage or broad commodity support. Treat CXL as a slower NUMA memory tier, validate placement behavior with the actual hypervisor and workload, and compare the complete enterprise quote against larger local-DIMM, 2DPC and dual-socket alternatives.
For further platform details, see the ServeTheHome review, its pages on the chassis and CXL hardware, memory topology and testing, and the final assessment. ASUS’s server review index also lists the coverage.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
