NVIDIA Quadro FX 370M

The NVIDIA Quadro FX 370M is a mobile workstation GPU built on the Tesla architecture, fabricated at UMC on a 65 nm process. The chip, designated G98S, contains 210 million transistors on an 86 mm² die, yielding a transistor density of 2.4 million per square millimeter. Released on August 14, 2008, this end-of-life product belongs to the Quadro FX Mobile (x700M) generation and succeeds the Quadro FX Go, with the Quadro Fermi-M as its successor. It holds a 50th percentile rank among all GPUs in the database, with an average benchmark score of 0. The GPU features 8 shading units, 4 texture mapping units, and 4 render output units, with a 64-bit memory bus and 256 MB of GDDR3 memory.

Benchmark Performance

The average benchmark score for the Quadro FX 370M is 0, which indicates that no standardized performance measurements have been recorded in this database. Consequently, the 50th percentile rank is the only relative performance indicator available. This percentile places the GPU exactly at the median of the historical GPU performance distribution, meaning it is neither particularly strong nor weak when compared to the entire range of GPUs. However, without direct rival scores, the percentile alone does not convey how it stacks against contemporaneous parts. The absence of benchmark data also means that the theoretical peak figures are the sole quantitative basis for performance analysis.

The FP32 compute throughput is 23.20 GFLOPS, a figure that reflects the modest 8 shading units. This level of compute is suitable for basic vertex and pixel shader workloads, but it will be quickly overwhelmed by complex geometry or physics calculations. The pixel fill rate is 2.320 GPixel/s, and the texture fill rate is 2.320 GTexel/s, both limited by the 4 ROPs and 4 TMUs respectively. These numbers suggest a part designed for 2D desktop rendering and light 3D tasks at low resolutions. The DirectX 11.1 (10_0) feature level indicates that while the driver supports DirectX 11.1, the hardware only implements a DirectX 10-class feature set. This restricts the available shader model and rendering features, such as tessellation and compute shaders, which are not fully supported. The OpenGL 3.3 support is similarly limited to older GLSL versions.

The 50th percentile rank, when combined with the zero average score, suggests that the GPU is not an extreme outlier, but its practical performance is unverified. The theoretical fill rates and compute throughput are consistent with a low-end mobile part from the Tesla generation. The 64-bit memory bus and 6.4 GB/s bandwidth further constrain performance, particularly in memory-intensive scenes. The transistor density of 2.4 million per square millimeter indicates a relatively simple design, and the 65 nm process is an older node. All these factors contribute to a GPU that is unlikely to deliver high frame rates in any demanding application.

Ray Tracing and Feature Set

The Quadro FX 370M does not include dedicated ray tracing cores or tensor cores; both fields are null in the specification. This is consistent with the Tesla architecture, which predates hardware-accelerated ray tracing. The GPU relies entirely on traditional rasterization for graphics processing. The API support includes DirectX 11.1 with a feature level of 10_0, which means the hardware is limited to DirectX 10-class shader capabilities. OpenGL 3.3 is the maximum supported version. Vulkan is not listed, indicating no Vulkan driver support. The absence of tensor cores also means that any AI-accelerated features are unavailable. For professional workstation applications that rely on OpenGL or DirectX 10-era APIs, this GPU can function, but it will not support modern ray tracing or machine learning workloads. The feature set is strictly defined by the Tesla architecture's capabilities. The lack of hardware ray tracing means that any ray-traced effects must be software-emulated, which is impractical given the low compute throughput. The DirectX 11.1 (10_0) feature level also limits the use of advanced shading techniques such as deferred rendering with multiple render targets, though some are possible at reduced efficiency.

Memory Subsystem

The memory subsystem consists of 256 MB of GDDR3 memory connected via a 64-bit bus. The memory clock is 400 MHz, with an effective data rate of 800 Mbps per pin, yielding a total bandwidth of 6.4 GB/s. This is a small frame buffer by any standard, and the narrow bus width severely limits memory throughput. For high-resolution rendering, the 256 MB capacity is insufficient for large textures or complex geometry, forcing the GPU to rely on system memory over a slow interface. The 6.4 GB/s bandwidth is a fraction of what contemporary GPUs offer, and it will become a bottleneck in scenes with high texture detail or anti-aliasing. The pixel rate and texture rate are both 2.320, which means that the memory bandwidth is sufficient for those fill rates at low resolutions, but any increase in resolution will quickly saturate the bus. The memory type, GDDR3, is an older standard, and the effective speed of 800 Mbps is low even for that generation. The 400 MHz memory clock is also modest. The combination of low capacity and low bandwidth means that the GPU will struggle with modern texture-heavy workloads, and even 2D applications with large desktops or multiple monitors may exceed the available memory bandwidth.

Who Should Consider It

The Quadro FX 370M is suited for users who require a compact, low-power mobile GPU for basic workstation tasks such as 2D CAD, spreadsheet applications, or light 3D modeling at low resolutions. The 256 MB frame buffer and 6.4 GB/s bandwidth are adequate for older applications that do not demand large textures. The DirectX 11.1 (10_0) support allows running DirectX 10-era software, but the hardware will struggle with any game or professional application that uses advanced shaders. The 50th percentile rank indicates that it is not the weakest GPU ever produced, but the zero average benchmark score means there is no empirical evidence of its real-world performance. Users who need certified drivers for legacy professional software may find this GPU acceptable, provided the software's requirements are modest. The 20 W TDP makes it suitable for thin-and-light laptops where power efficiency is a priority, but it will not handle high-resolution displays or multi-monitor setups well. Given the end-of-life status, this GPU is only relevant for legacy systems or as a replacement part for existing hardware. It is not recommended for any modern 3D application, as the memory and compute resources are far below the minimum requirements for contemporary software.

Power and Cooling

The Quadro FX 370M has a TDP of 20 W, which is low for a GPU, even in its era. The module uses an MXM-II form factor and requires no auxiliary power connectors, as indicated by the "None" entry for power connectors. The bus interface is MXM-II, which is a standard for mobile graphics modules. The low power draw means that cooling solutions can be passive or use a small fan, but the exact cooling implementation is not specified. The absence of a suggested PSU is consistent with a mobile design where the host laptop's power delivery is used. The 20 W TDP allows for efficient thermal management in compact chassis. However, the performance is correspondingly limited, as the power budget restricts clock speeds and memory bandwidth. The slot width is listed as "MXM Module", and the display outputs are described as "Portable Device Dependent", meaning that the actual ports depend on the laptop's design. For users who prioritize battery life and low heat output, this GPU is a reasonable choice, but it will not deliver high frame rates or complex rendering performance. The lack of a dedicated power connector simplifies installation, but the low TDP also means that the GPU cannot be overclocked or pushed beyond its specified limits.