NVIDIA GeForce GTX 860M OEM

The NVIDIA GeForce GTX 860M OEM is a mobile graphics module built on the Maxwell architecture, fabricated by TSMC on a 28 nm process. It integrates 1,870 million transistors on a 148 mm² die, yielding a transistor density of 12.6M per mm². The GPU sits at the 50th percentile of all GPUs in the database, indicating a median performance position, though its average benchmark score is recorded as 0, suggesting a lack of standardized test submissions.

Benchmark Performance

The GTX 860M OEM operates with a base clock of 1097 MHz and a boost clock of 1176 MHz. These clocks drive 640 shading units, 40 texture mapping units, and 16 ROPs. The resulting compute throughput is 1.505 TFLOPS for FP32 operations, a figure that defines its raw shader capability. Pixel fill rate reaches 18.82 GPixel/s, while texture fill rate is 47.04 GTexel/s. These rates are derived directly from the clock speeds and unit counts. Given the 50th percentile placement, the data indicates that this GPU delivers a middle-of-the-road experience in the broader landscape of graphics hardware. The FP32 throughput of 1.505 TFLOPS is consistent with a chip designed for mainstream mobile workloads, balancing resolution capabilities with power constraints. The pixel rate of 18.82 GPixel/s suggests that at standard 1080p resolutions, the GPU can maintain acceptable fill rates for older or less demanding titles, but the numbers do not point to high-refresh-rate or high-fidelity performance at higher resolutions. The texture rate of 47.04 GTexel/s, combined with the 40 TMUs, indicates a texture filtering capability that is adequate for moderate texture loads. Without specific benchmark scores in the database, the percentile ranking serves as the primary comparative metric. The 50th percentile means exactly half of all tracked GPUs perform better, and half perform worse, placing this part in a neutral, central position. The boost clock of 1176 MHz provides a modest uplift over the base 1097 MHz, which is a typical behavior for Maxwell parts, allowing for dynamic performance scaling under thermal and power headroom. The 1.505 TFLOPS FP32 figure is the key arithmetic throughput number, and it is the figure that defines its compute class. The 18.82 GPixel/s pixel rate is a product of the 16 ROPs and the clock speed, and the 47.04 GTexel/s texture rate is a product of the 40 TMUs and the clock speed. These metrics collectively paint a picture of a balanced, mid-range mobile GPU from its era. The boost behavior, while modest, provides a measurable performance uplift over sustained workloads, assuming the thermal solution can maintain the necessary headroom.

Power and Cooling

The thermal design power for this GPU is set at 75 W. This is a modest power envelope, typical for an MXM module designed for portable devices. The slot width is listed as "MXM Module", and the bus interface is "MXM-B (3.0)". This form factor means the card is intended for laptops or small-form-factor systems that use replaceable graphics modules. The power connector requirement is "None", indicating that the card draws all its operating power from the MXM slot itself, without needing auxiliary 6-pin or 8-pin power cables. The absence of a suggested PSU in the data reflects its mobile nature, where power delivery is handled by the system's own power supply and battery management. The 75 W TDP is a critical specification for system integrators, as it dictates the thermal solution required. A 75 W TDP typically requires a capable cooling solution, but it is within the range that can be managed by standard laptop cooling designs. The data shows that the GPU relies entirely on the MXM interface for power, simplifying installation in compatible chassis. Since no auxiliary power connectors are present, the system's power delivery design must accommodate the full 75 W through the slot. This is a significant consideration for mobile workstations or gaming laptops, as the overall system power budget must account for this draw. The MXM-B (3.0) interface ensures compatibility with a specific class of motherboards, and the 75 W envelope allows for a relatively thin cooling solution compared to higher-TDP desktop cards. The lack of a suggested PSU rating is consistent with its OEM and mobile orientation, where the host system's power supply is predefined and not user-upgradable. The 75 W figure places it in a power class that is manageable for thin-and-light gaming laptops, though sustained loads will still generate substantial heat that must be dissipated through the laptop's cooling system.

Memory Subsystem

The GTX 860M OEM is equipped with 2 GB of GDDR5 memory, connected via a 128-bit memory bus. The memory clock runs at 1253 MHz, which translates to an effective data rate of 5 Gbps. This configuration yields a total memory bandwidth of 80.19 GB/s. For high-resolution gaming, this bandwidth is a limiting factor. The 128-bit bus width is relatively narrow, which constrains the amount of data that can be transferred between the GPU and memory per clock cycle. The 80.19 GB/s bandwidth is sufficient for 1080p gaming with moderate texture settings, but at higher resolutions like 1440p or 4K, the bandwidth becomes a bottleneck, leading to potential frame rate drops when texture streaming demands exceed the available throughput. The 2 GB capacity is also a consideration; while adequate for older titles, modern games at high resolutions often require more than 2 GB of VRAM, which can result in texture pop-in or reduced detail settings. The memory type, GDDR5, provides a good balance of speed and power efficiency for its time. The effective data rate of 5 Gbps is a standard figure for that generation. The 128-bit bus width, combined with the 5 Gbps effective rate, mathematically produces the 80.19 GB/s bandwidth (128 bits / 8 = 16 bytes per clock; 16 bytes * 5 Gbps = 80 GB/s, matching the specified 80.19 GB/s). This figure is a hard ceiling on memory throughput. For users targeting high resolutions, the data suggests that the GPU will struggle to maintain smooth performance in memory-intensive scenarios. The 2 GB frame buffer limits the resolution and detail settings that can be used without exceeding capacity. Overall, the memory subsystem is adequate for its intended mainstream segment, but it is not designed for high-end, high-resolution workloads. The 80.19 GB/s bandwidth figure is the primary constraint when pushing beyond 1080p, and the 2 GB capacity further compounds this limitation in modern titles with large texture packs.

How It Compares

The nearestRivals array in the data is empty, meaning no direct comparative scores or delta percentages are available from the database. Therefore, the analysis relies on the percentile and generational context. The GPU is positioned as a successor to the GeForce 700M generation and a predecessor to the GeForce 900M generation. This places it in the Maxwell architecture era, following the Kepler-based 700M series and preceding the improved Maxwell 900M series. The 50th percentile ranking indicates that it sits exactly at the median of all GPUs tracked by the database. This is a neutral position, neither particularly strong nor weak. Without rival data, one can infer that its performance is typical of a mid-range mobile part from its generation. The 75 W TDP and MXM form factor are characteristic of mobile OEM parts, which often sacrifice peak performance for thermal and power efficiency. The lack of benchmark scores (average score of 0) further complicates direct comparison, but the percentile provides a global context. In the absence of named rivals, the discussion must center on its own specifications and its place in the product stack. The move from GeForce 700M to GeForce 800M (this part) and then to GeForce 900M represents a progression in architecture and efficiency. The 28 nm process node is shared with its immediate predecessor and successor generations, so the improvements come from architectural changes rather than process shrinks. The 1.505 TFLOPS FP32 performance and 80.19 GB/s bandwidth are the key metrics that would define its standing against any hypothetical competitor. Since no rivals are listed, the data cannot support claims of being ahead or behind any specific product. The percentile is the only comparative anchor, and it shows a perfectly average standing. This suggests that in the context of all GPUs, it offers a baseline level of performance that is neither exceptional nor deficient. Its position between two generations of mobile GPUs indicates that it inherits some architectural traits from the 700M series while laying groundwork for the 900M series, but without concrete rival data, any specific performance deltas remain unquantified.

Ray Tracing and Feature Set

The GTX 860M OEM does not include dedicated ray tracing cores, as the rtCores field is null. Similarly, it lacks tensor cores, with the tensorCores field also null. This is consistent with the Maxwell architecture, which predates the introduction of hardware-accelerated ray tracing and tensor-based AI processing in NVIDIA's consumer GPUs. Consequently, any ray tracing workloads would have to be handled by the standard shader units, which would result in significantly reduced performance compared to dedicated hardware. The feature set is defined by its API support. It supports DirectX 12, but at the feature level 11_0. This means it can run DirectX 12 applications, but it does not support the higher-tier features that require DirectX 12 Ultimate or feature level 12_0 and above. The OpenGL support is at version 4.6, which is a mature version that provides broad compatibility with professional and creative applications. Vulkan support is at version 1.4, offering a modern low-overhead API that can improve performance in supported games. The display outputs are listed as "Portable Device Dependent", meaning the actual ports depend on the specific laptop or system in which the MXM module is installed. This is typical for mobile GPUs, as the manufacturer integrates the necessary display connectors. The absence of RT and tensor cores means that features like DLSS (Deep Learning Super Sampling) and hardware-accelerated ray tracing are not available. The FP32 compute of 1.505 TFLOPS is the primary compute resource, and it is what would be used for any software-based ray tracing or AI tasks, albeit with limited efficiency. The API support, particularly Vulkan 1.4 and OpenGL 4.6, ensures that the GPU remains functional for a wide range of software, but the DirectX 12 (11_0) limitation is a notable constraint for the latest game titles that require higher feature levels. The data clearly indicates that this is a rasterization-focused GPU from an era before dedicated ray tracing hardware became standard. The lack of tensor cores also precludes any AI-accelerated features, leaving the 640 shading units as the sole computational engine for all rendering and compute tasks.