NVIDIA GeForce GTX 960 OEM

The NVIDIA GeForce GTX 960 OEM is a GeForce 900-series part built around the GM204 chip on TSMC’s 28 nm process. It integrates 5,200 million transistors on a 398 mm² die, for a transistor density of 13.1M per mm². The GPU has 1280 shading units, 80 texture mapping units, and 48 ROPs. It is listed as end-of-life, with a release date of 2015-11-25. In the product sequence, it follows the GeForce 700 family and precedes the GeForce 10 family. Its bus interface is PCIe 3.0 x16, and its form factor is dual-slot.

How It Compares

This entry has no nearest-rival data. The nearestRivals list in the fact pack is empty, so there are no rival names, scores, or deltaPct values to examine. The only comparative signal is percentileVsAllGpus, which is 50. That places the card exactly at the median of the GPU database. Because the benchmarks array is empty and the avgBenchmarkScore is 0, the percentile cannot be tied to a particular workload result. It is a positional statement only: half of all GPUs in the database fall below this card, and half fall above it. Generationally, the card sits between the GeForce 700 series and the GeForce 10 series, but no measured delta against either is provided. In the absence of rival deltas, any attempt to rank it against specific products would go beyond the available data.

Power and Cooling

The TDP field is null, meaning no thermal design power is stated in the record. The suggested PSU is 200 W, and the power connector field is “None”, so no auxiliary PCIe power connectors are required. The slot width is dual-slot, and the interface is PCIe 3.0 x16. These details describe a board that can be installed in a standard dual-slot chassis without extra power cabling. The 200 W system PSU recommendation is the only power-supply constraint provided. No cooler specifications are included, but the dual-slot width is the physical cooling envelope. The absence of power connectors suggests the card is intended for systems that supply all power through the motherboard’s PCIe slot, although the data does not explicitly state that transfer path.

Benchmark Performance

The benchmark data is not populated. The benchmarks list is empty, and avgBenchmarkScore is 0. As a result, there are no measured workload scores to analyze and no exact percentage deltas to report. The performance-related numbers that do exist are peak specification values. The base clock is 924 MHz, and the boost clock is 1038 MHz. FP32 throughput is 2.657 TFLOPS. Texture fill rate is 83.04 GTexel/s, and pixel fill rate is 49.82 GPixel/s. These rates are derived from the 1280 shading units, 80 TMUs, and 48 ROPs at the listed clocks. The 50th percentile is the only relative performance marker, and it positions the card in the middle of the database, but without score data it cannot indicate whether that position derives from raster throughput, compute work, or memory-bound scenarios.

Who Should Consider It

Because the fact pack contains no resolution-specific benchmark scores, a concrete resolution and settings recommendation cannot be made from measured results. What the data does provide is a memory envelope: 3 GB of GDDR5 with 120.3 GB/s of bandwidth. Any workload that fits within that capacity and bandwidth is the relevant target for this card. The 200 W suggested PSU and the absence of power connectors make it a candidate for systems that do not have auxiliary GPU power cables. The dual-slot width and PCIe 3.0 x16 interface define the necessary physical and electrical support. A 50th-percentile rank means this card sits in the middle of the database rather than at either extreme. The 2.657 TFLOPS FP32 rate, 83.04 GTexel/s texture fill, and 49.82 GPixel/s pixel fill describe a rendering pipeline suited to workloads bounded by those peak rates.

Ray Tracing and Feature Set

The ray tracing and tensor core fields are both null, so no hardware ray tracing core count or tensor core count is available in the data. The feature set is instead defined by API support: DirectX 12 with feature level 12_1, OpenGL 4.6, and Vulkan 1.4. These are the graphics API capabilities listed for the card. The architecture is Maxwell 2.0, which is the generation-specific foundation for those APIs. The display output configuration includes 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.2. That output set defines how many monitors can be connected directly and which connector types are available. No RT core or tensor core acceleration can be inferred from the null values.

Memory Subsystem

The memory subsystem consists of 3 GB of GDDR5 on a 192-bit bus, with a total bandwidth of 120.3 GB/s. The memory clock is listed as 1253 MHz, with 5 Gbps effective data rate. For high-resolution work, the 3 GB capacity limits how much texture and geometry data can reside locally, while 120.3 GB/s limits how quickly that data can be moved to the rendering pipeline. The 49.82 GPixel/s pixel fill rate and 83.04 GTexel/s texture fill rate provide the raster-side context for that bandwidth. The memory clock expression of 1253 MHz and effective 5 Gbps are separate ways of looking at the same transfer speed, not two different memory rates. The combination of 192-bit width and 120.3 GB/s is the only memory bandwidth figure in the record.

FAQ

Q: What is the manufacturing process and die size?

A: The GPU is manufactured by TSMC on a 28 nm process. The die size is 398 mm², with 5,200 million transistors and a transistor density of 13.1M per mm².

Q: What memory configuration does the card have?

A: It has 3 GB of GDDR5 memory, a 192-bit bus, and 120.3 GB/s bandwidth. The memory clock is 1253 MHz, with 5 Gbps effective data rate.

Q: What power supply and connector requirements are listed?

A: The suggested PSU is 200 W. The power connector field is “None”, and the card uses a dual-slot form factor with a PCIe 3.0 x16 interface.

Q: Does it support hardware ray tracing?

A: The data lists rtCores as null and tensorCores as null, so no hardware ray tracing or tensor core information is provided. API support includes DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4.

Q: What are the peak compute and fill rates?

A: FP32 throughput is 2.657 TFLOPS. Pixel fill rate is 49.82 GPixel/s, and texture fill rate is 83.04 GTexel/s. Base clock is 924 MHz and boost clock is 1038 MHz.

Q: Which display outputs are available?

A: The card has 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.2 outputs.