Quick Answer
Low GPU usage means CPU bottleneck, active frame cap, or bad power settings. VSync and limiters pause GPU work intentionally. CPU-heavy games make low GPU load normal. Fix: inspect single-core CPU usage, turn off VSync, set OS power plan to maximum performance, and update graphics drivers.
Pros
- Supports the latest AMD AM5 CPU generations, including Ryzen 9000 series.
- Includes PCIe 5.0 and Gen5 M.2 support for future-focused upgrade headroom.
- Offers both wired and wireless networking features, including 2.5G LAN and WiFi 6E.
- Micro-ATX sizing makes it suitable for smaller PC cases without dropping core modern features.
Cons
- Only two DDR5 DIMM slots, so memory expansion is more limited than on four-slot boards.
- Micro-ATX format may not fit all pre-built systems, so compatibility should be checked carefully before buying.
- Realtek ALC897 audio is functional, but it is not a premium onboard audio solution.
Overview: The ASRock B850M-X R2.0 is a micro-ATX AM5 motherboard aimed at builders who want a compact platform for modern AMD Ryzen systems. It combines current-generation connectivity with a practical layout for smaller cases.
Performance and features: With support for Ryzen 9000, 8000, and 7000 series processors, plus a 6+1+1 phase power design with Dr.MOS, it is positioned for stable everyday use. The board also adds DDR5 support up to 8200+ OC, PCIe 5.0 x16, a Blazing M.2 Gen5x4 slot, a Hyper M.2 Gen4x4 slot, and four SATA3 ports for flexible storage setups.
Drawbacks and considerations: The two-DIMM design limits memory expansion compared with larger boards, and the micro-ATX format may not be compatible with every pre-built system. The Realtek ALC897 audio solution is adequate for general use, but it is not the strongest option for users who want higher-end onboard sound.
Verdict: This motherboard is a strong fit for budget-conscious or space-conscious AM5 builders who want modern platform features without moving to a larger board. It makes the most sense for gaming or general-purpose PC builds that value PCIe 5.0, fast storage, and broad Ryzen compatibility.
Pros
- Supports Intel Core 14th, 13th, and 12th Gen processors on the LGA 1700 socket.
- Includes PCIe 5.0 x16 and two M.2 slots for modern GPU and SSD upgrades.
- DDR4 support is cost-friendly for builders reusing compatible memory.
- Built-in Wi-Fi 6 and 2.5Gb Ethernet cover both wireless and wired networking needs.
Cons
- microATX format offers fewer expansion options than a full-size ATX motherboard.
- DDR4-only support means it is not compatible with DDR5 memory.
- Only two M.2 slots are listed, which may be limiting for storage-heavy builds.
Overview: The ASUS B760M-AYW WiFi D4 II is a microATX motherboard for Intel LGA 1700 processors, including 14th, 13th, and 12th Gen Core chips as well as Pentium Gold and Celeron models. Its compact layout makes it a practical fit for smaller desktop and gaming builds.
Performance and features: This board brings PCIe 5.0 x16 support, two M.2 slots, DDR4 memory compatibility, Realtek 2.5Gb Ethernet, and Wi-Fi 6. It also includes VRM and PCH heatsinks, hybrid fan headers, and Fan Xpert 2+ to help manage temperatures in everyday use and longer sessions.
Considerations: The microATX design naturally provides fewer expansion options than a larger ATX board, and DDR4 support means it does not work with DDR5 memory. The listed storage expansion is limited to two M.2 slots, so users with larger drive plans should check whether that is enough.
Verdict: This motherboard is a solid choice for builders who want a modern Intel platform without moving to DDR5 pricing. It is best suited to compact gaming PCs and mainstream upgrades where fast networking, reliable cooling features, and flexible storage matter.
Pros
- AM5 platform ensures broad CPU upgrade path across multiple generations
- PCIe 5.0 ready for next-generation graphics cards and NVMe drives
- Established B650 chipset with solid BIOS support and stability
- ATX form factor offers room for multiple GPU and expansion cards
Cons
- Missing WiFi and Bluetooth compared to higher-end B650 variants
- Simpler power delivery than premium boards, limiting extreme overclocking
- Minimal onboard features aimed at budget segment
The ASUS Prime B650-PLUS suits builders who want a stable, forward-compatible foundation for mid-tier Ryzen 7000 systems without paying for WiFi or heavy power delivery. It delivers PCIe 5.0 and DDR5 support at mainstream pricing. Builders seeking integrated WiFi or advanced overclocking should step up to premium B650 or X870 models.
Socket: AM5
Chipset: AMD B650
Form Factor: ATX
Memory Support: DDR5
Expansion: PCIe 5.0
What Low GPU Usage Means
GPU usage, as reported by Windows Task Manager, NVIDIA’s NVML, AMD’s ADLX, or utilities such as HWiNFO64 and MSI Afterburner, is a time-sliced utilization metric. It reports the percentage of a sampling interval (typically 500 ms to 1000 ms) during which at least one engine on the GPU was executing work. It does not report the fraction of shader cores, ALUs, or memory bandwidth in use. A GPU can therefore read 99% utilization while using only part of its power envelope, and it can read 55% when the render queue is repeatedly starved of work.
Low utilization in a game means the GPU is idle for a measurable share of each frame interval. At 60 FPS the frame budget is 16.67 ms. If the GPU finishes its work in 8 ms and then waits for the next command buffer, it reads about 48% utilization. The cause lies upstream (CPU submission, frame limiters, synchronization) or is intentional (power-saving behavior, a display-synchronized cap). Treating it as a fault of the GPU itself is rarely correct.
Reading the Correct Counters
Task Manager’s default GPU graph often displays the 3D engine only, and it may hide Compute or Copy engine load. Engineering-grade diagnosis should compare several counters at once:
- GPU Core Load / 3D Engine: time-based busy percentage. Values at or above 95-99% indicate a GPU-bound condition.
- Board Power: a high-end card at 99% load typically draws close to its rated TBP (for example 320 W on a 320 W-class card). A reading of 55% load at 150 W indicates the card is not saturated.
- Core Clock: modern GPUs commonly boost to 2400-2900 MHz under load. Clocks sitting near 1200-1800 MHz at low load show that power management has downclocked the card because the workload is light.
- Per-Core CPU Load: a single logical core at 95-100% while total CPU load reads 15-25% is the classic signature of a main-thread bottleneck.
- Frame Time (ms): flat, consistent frame times at a round value (16.67, 8.33, 6.94 ms) suggest a limiter. Irregular spikes suggest CPU stalls or shader compilation.
CPU Limits and Frame Caps
In the rendering pipeline, the CPU runs game logic, physics, animation, and draw-call generation, then submits command buffers through the graphics API. The GPU consumes them. If the CPU needs 12 ms to prepare a frame and the GPU needs 7 ms to render it, throughput is set by the CPU at about 83 FPS and the GPU sits idle roughly 42% of the time (5 ms of 12 ms).
Identifying a CPU Bottleneck
Diagnostic indicators of a CPU-limited condition include:
- GPU usage that drops when resolution is held constant but the scene becomes busier (crowded cities, large AI counts, simulation-heavy titles).
- Frame rate that does not rise when resolution or graphical presets are lowered. Moving from 1440p to 1080p raising FPS by under 5% points to a CPU limit, whereas a GPU-bound scene typically scales by 20-40% for that change.
- One or two logical cores pinned near 100% on an 8-, 12-, or 16-core processor. Many engines still rely on a primary render thread and a main game thread.
- CPU package temperature near its thermal throttling threshold (commonly 95-100 °C on current desktop parts) with clocks falling below rated boost, which reduces single-thread throughput.
Mitigation options include raising the render resolution or enabling heavier GPU-side features (higher-quality anti-aliasing, ray tracing, higher texture filtering) to shift the load balance, reducing CPU-heavy settings (crowd density, view distance, simulation quality), and ensuring the CPU maintains its rated boost. Memory configuration also matters: a platform running DDR5 at 4800 MT/s instead of its XMP/EXPO rated 6000 MT/s can lose a measurable percentage of frame rate in CPU-limited titles, and DDR4 in single-channel mode halves peak bandwidth (for example 25.6 GB/s versus 51.2 GB/s at 3200 MT/s).
Frame Caps and Synchronization
Frame limiters deliberately insert idle time. Common sources include:
- VSync: locks presentation to the display refresh. On a 60 Hz panel the GPU finishes each frame and then waits for the vertical blank, so usage falls in lighter scenes.
- In-game FPS limiters: engine-level caps (for example 60, 90, 120, or 144 FPS) release the render thread for the rest of the frame interval.
- Driver-level limiters: NVIDIA Control Panel’s Max Frame Rate and AMD Radeon Chill or Frame Rate Target Control cap output deliberately. Radeon Chill additionally reduces frame rate during low input activity to lower power draw.
- G-SYNC / FreeSync: with variable refresh enabled and VSync on, drivers commonly cap frame rate a few FPS below the panel’s maximum (for example about 141-142 FPS on a 144 Hz panel) to remain inside the VRR window.
- Third-party limiters: RTSS and similar tools can enforce a cap that persists across sessions and is easily forgotten.
