NVIDIA GeForce 940A

The NVIDIA GeForce 940A is a Maxwell-generation GPU built around the GM108S chip and manufactured by TSMC on a 28nm process. The database entry places it in the GeForce 900A generation and lists 1,020 million transistors on a 77 mm² die, for a transistor density of 13.2M per mm². It is a low-power part with a 33W TDP, an IGP slot width, and no power connectors. Its production status is end-of-life, its release date is March 12, 2015, and its predecessor is GeForce 800A. The entry also contains no nearest rivals, no benchmark scores, and no suggested PSU.

Who Should Consider It

The 940A is aimed at scenarios where the 64-bit memory bus and 2GB DDR3 frame buffer are not a bottleneck. With 384 shading units, 16 TMUs, and 8 ROPs, the GPU offers 903.2 GFLOPS of FP32 throughput, but that shader capacity is paired with only 16.02 GB/s of memory bandwidth. For lower resolutions and modest graphics settings, the supported API set — DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.4 — enables a range of software. The PCIe 3.0 x8 bus interface places another bandwidth limit on data transfers. Because display outputs are listed as "Portable Device Dependent", the physical display connections are determined by the host device rather than the GPU alone.

This profile suits portable or integrated systems that need a 33W, no-auxiliary-power-connector graphics solution and operate within a small memory footprint. High-resolution, texture-heavy workloads would quickly exceed the 16.02 GB/s bandwidth and 2GB capacity. The absence of stored benchmark scores means this recommendation is based on the listed hardware characteristics, not measured workload performance.

Memory Subsystem

The memory system is a 2GB DDR3 configuration on a 64-bit bus. The memory clock is 1001 MHz, with a 2 Gbps effective data rate. Across the 64-bit interface, that produces 16.02 GB/s of bandwidth. That bandwidth is the key constraint in this subsystem: the GPU can generate up to 9.408 GPixel/s and 18.82 GTexel/s, but the memory path can only move so much data per second. The 64-bit bus width limits how many bytes can be fetched per clock, and the DDR3 type limits the effective speed.

In high-resolution scenarios, the 2GB capacity can be insufficient, and the bandwidth ceiling reduces texture throughput and frame-buffer operations. The memory clocks are modest, and there is no separate game clock listed. As a result, the memory subsystem defines the overall performance ceiling more than the 384 shading units do.

How It Compares

The nearestRivals array for this entry is empty. There are no rival names, scores, or deltaPct values in the fact pack, so no direct comparison to a named alternative is possible. The only relational data points are the predecessor GeForce 800A, the GeForce 900A generation, and the end-of-life production status. The successor field is also empty.

The database's percentileVsAllGpus value is 50, which suggests a middle position among all GPUs, but this is not tied to any measured rival delta. The avgBenchmarkScore is 0 and the benchmarks array is empty, so no empirical comparison can be drawn from similar products. Exact percentage deltas are not present in this record. Thus the comparison section is limited to stating what the database does and does not contain.

FAQ

Q: What architecture and process does the 940A use?

A: It uses the Maxwell architecture, with the GM108S chip manufactured by TSMC on 28nm. The die contains 1,020 million transistors and measures 77 mm², with a transistor density of 13.2M per mm².

Q: How much memory does the 940A have and what is its bandwidth?

A: It has 2 GB of DDR3 memory on a 64-bit bus. The memory clock is 1001 MHz, with a 2 Gbps effective data rate, which yields 16.02 GB/s of bandwidth.

Q: What APIs are supported?

A: The database lists DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.4.

Q: Does it require a PCIe power connector?

A: No. The power connectors field is "None", the TDP is 33 W, and the slot width is IGP. The database does not list a suggested PSU.

Q: Is there ray tracing or AI acceleration hardware?

A: The fact pack lists no RT cores and no tensor cores, so no dedicated ray tracing or tensor-accelerated features are quantified in the database.

Q: What is the production status and predecessor?

A: It is marked end-of-life and lists GeForce 800A as its predecessor. The release date is March 12, 2015.

Ray Tracing and Feature Set

The fact pack lists no RT cores and no tensor cores for the 940A. Dedicated ray tracing and tensor-acceleration hardware are therefore not quantified in the entry. The feature set instead centers on Maxwell architecture support for the listed APIs: DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.4. DirectX 12 support is at the 11_0 feature level, per the database.

The GPU contains 384 shading units, 16 TMUs, and 8 ROPs, with a 1072 MHz base clock and a 1176 MHz boost clock. These resources produce 903.2 GFLOPS of FP32 compute, 18.82 GTexel/s texture fill, and 9.408 GPixel/s pixel fill. The display output specification is "Portable Device Dependent", meaning the host determines monitor connections. The bus interface is PCIe 3.0 x8. No Tensor Core or RT Core feature set beyond the absence of listed counts can be inferred from the data.

Power and Cooling

With a 33W TDP, the 940A is a low-power part in the database. The slot width is listed as IGP, and the power connectors field is None, meaning no auxiliary power connectors are listed. The fact pack does not include a suggested PSU; therefore, a specific power-supply recommendation cannot be made from this data. No physical dimensions are listed, consistent with a device-dependent mounting approach.

The low TDP and the absence of power connectors suggest that power delivery is handled through the host platform. Cooling requirements are modest, though the database does not describe a cooler. The end-of-life status implies that new platform integration may be limited.

Benchmark Performance

The benchmark records for the 940A are empty. The avgBenchmarkScore is 0, which is consistent with no stored benchmark samples rather than a measured result. The percentileVsAllGpus field places the GPU at 50, the midpoint of the database's GPU distribution. However, because no nearestRivals or benchmark entries exist, this percentile cannot be cross-checked against specific competitors.

The remaining performance data is theoretical. The 1072 MHz base and 1176 MHz boost clocks drive 384 shading units, 16 TMUs, and 8 ROPs. This produces 903.2 GFLOPS of FP32 throughput, 18.82 GTexel/s texture fill, and 9.408 GPixel/s pixel fill. The memory bandwidth is 16.02 GB/s. Those numbers define a small GPU whose memory path is the tightest constraint. The percentile value of 50 suggests a middle rank across all GPUs, but the lack of measured scores and rival deltaPct values means no exact percentage comparison to any other GPU can be sourced from this entry.