NVIDIA GeForce GTX 965M vs NVIDIA T400 Comparison

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA T400 leads with an average benchmark score of 16508, while the NVIDIA GeForce GTX 965M trails at 14404. This places the T400 in the 60th percentile of all GPUs, compared to the GTX 965M's 56th percentile.

Q: What is the performance gap between the two in OpenCL workloads?

A: The T400 scores 17039 in Geekbench OpenCL, which is 17.4% higher than the GTX 965M's 14509. The T400 wins this test decisively.

Q: How do the two compare in Vulkan performance?

A: The T400 again comes out ahead with a Geekbench Vulkan score of 15976 versus 14299 for the GTX 965M, a delta of 11.7%. The T400 wins both recorded benchmark tests.

Q: What are the architectural generations of these two cards?

A: The T400 is built on the Turing architecture (TU117 chip) and belongs to the Quadro Turing (Tx000) generation. The GTX 965M uses the Maxwell 2.0 architecture (GM204 chip) from the GeForce 900M generation.

Q: Which card has a more advanced manufacturing process?

A: The T400 is fabricated on a 12 nm process at TSMC, while the GTX 965M uses a 28 nm process, also at TSMC. The T400's process node is considerably newer, allowing for a much higher transistor density of 23.5M per mm² compared to 13.1M per mm² for the GTX 965M.

Q: Do both cards support the same DirectX and Vulkan versions?

A: Yes, both support DirectX 12 (12_1) and Vulkan 1.4. They also both support OpenGL 4.6. However, the underlying architectures differ significantly in how they achieve this support.

Architecture Differences

The NVIDIA T400 and the NVIDIA GeForce GTX 965M represent two distinct generations of NVIDIA GPU design, separated by roughly six years of architectural evolution. The T400 is built on the Turing architecture, while the GTX 965M is based on Maxwell 2.0. This fundamental difference drives nearly every other specification disparity between the two cards.

The manufacturing process is a key differentiator. The T400 uses a 12 nm process at TSMC, while the GTX 965M uses a 28 nm process, also at TSMC. This allows the T400 to pack 4,700 million transistors into a 200 mm² die, yielding a transistor density of 23.5M per mm². The GTX 965M, by contrast, contains 5,200 million transistors on a much larger 398 mm² die, resulting in a density of just 13.1M per mm². The T400 achieves more transistors per area despite having fewer total transistors, a direct consequence of the newer manufacturing node.

Clock behavior also differs substantially. The T400 has a base clock of 420 MHz and a boost clock of 1425 MHz. The GTX 965M runs at a higher base clock of 924 MHz but only boosts to 950 MHz, meaning its boost headroom is minimal compared to the T400's large boost range. This suggests the T400 can scale its performance dynamically much more aggressively under load.

The memory subsystems are architecturally different as well. The T400 uses 2 GB of GDDR6 memory on a 64-bit bus, with a bandwidth of 80.00 GB/s and an effective data rate of 10 Gbps. The GTX 965M uses 2 GB of GDDR5 on a 128-bit bus, achieving a slightly higher bandwidth of 80.19 GB/s at an effective rate of 5 Gbps. Both have the same capacity and nearly identical bandwidth, but the T400 achieves this with half the bus width and faster memory.

Compute unit counts diverge sharply. The T400 has 384 shading units, 24 texture mapping units (TMUs), and 16 raster operation units (ROPs). The GTX 965M has 1024 shading units, 64 TMUs, and 32 ROPs. Despite having fewer than half the shading units, the T400's higher boost clock and architectural efficiency allow it to outperform the GTX 965M in the recorded benchmarks. The GTX 965M does maintain advantages in pixel rate (30.40 GPixel/s versus 22.80 GPixel/s) and texture rate (60.80 GTexel/s versus 34.20 GTexel/s).

The FP32 compute figures reflect this inversion. The T400 delivers 1,094.4 GFLOPS, while the GTX 965M delivers 1.946 TFLOPS, nearly double. However, the T400 also supports FP16 compute at 2.189 TFLOPS with a 2:1 ratio, a feature the GTX 965M does not list. Neither card has ray tracing cores or tensor cores.

Physical and interface differences are notable. The T400 is a single-slot card with no power connectors and a 30 W TDP, using a PCIe 3.0 x16 interface. It features three mini-DisplayPort 1.4a outputs. The GTX 965M is an MXM module with an MXM-B (3.0) interface and no power connectors; its display outputs are portable-device dependent. The GTX 965M has no listed TDP or suggested PSU in the database.

Where Each One Wins

The benchmark data gives a clear picture: the NVIDIA T400 wins both recorded tests. In Geekbench OpenCL, the T400 scores 17039 against 14509 for the GTX 965M, a 17.4% advantage. In Geekbench Vulkan, the T400 scores 15976 versus 14299, an 11.7% lead. The T400 therefore wins in every workload category captured by the database.

The T400's wins are likely driven by its newer architecture and more efficient design. The Turing architecture brings architectural improvements that allow a smaller GPU with fewer shading units to outperform a larger, older chip. The T400's much higher boost clock, combined with faster GDDR6 memory, helps it overcome the GTX 965M's raw resource advantages in shading units, TMUs, and ROPs.

The GTX 965M, despite losing both benchmarks, retains theoretical advantages in pixel rate and texture rate. Its 30.40 GPixel/s pixel throughput and 60.80 GTexel/s texture throughput are both higher than the T400's 22.80 GPixel/s and 34.20 GTexel/s. These figures suggest the GTX 965M could have an edge in purely rasterization-bound workloads with minimal compute requirements, though the database does not include such tests.

The T400 also has the benefit of being a desktop-oriented card with dedicated display outputs (three mini-DisplayPort 1.4a connectors), while the GTX 965M is a mobile module whose outputs depend on the host laptop. This makes the T400 more suitable for fixed workstation deployments where consistent display connectivity is required.

Specification Differences

The following fields differ between the NVIDIA T400 and the NVIDIA GeForce GTX 965M:

• Chip: TU117 (T400) versus GM204 (GTX 965M)
• Architecture: Turing versus Maxwell 2.0
• Generation: Quadro Turing (Tx000) versus GeForce 900M
• Process node: 12 nm versus 28 nm
• Transistors: 4,700 million versus 5,200 million
• Die size: 200 mm² versus 398 mm²
• Transistor density: 23.5M / mm² versus 13.1M / mm²
• Base clock: 420 MHz versus 924 MHz
• Boost clock: 1425 MHz versus 950 MHz
• Memory clock: 1250 MHz (10 Gbps effective) versus 1253 MHz (5 Gbps effective)
• Memory type: GDDR6 versus GDDR5
• Memory bus width: 64 bit versus 128 bit
• Memory bandwidth: 80.00 GB/s versus 80.19 GB/s
• Shading units: 384 versus 1024
• TMUs: 24 versus 64
• ROPs: 16 versus 32
• Pixel rate: 22.80 GPixel/s versus 30.40 GPixel/s
• Texture rate: 34.20 GTexel/s versus 60.80 GTexel/s
• FP32 compute: 1,094.4 GFLOPS versus 1.946 TFLOPS
• FP16 compute: 2.189 TFLOPS (2:1) versus not listed
• TDP: 30 W versus not listed
• Slot width: Single-slot versus MXM Module
• Suggested PSU: 200 W versus not listed
• Bus interface: PCIe 3.0 x16 versus MXM-B (3.0)
• Display outputs: 3x mini-DisplayPort 1.4a versus Portable Device Dependent
• Release date: 2021-05-05 versus 2015-01-08
• Predecessor: Quadro Volta versus GeForce 800M
• Successor: Workstation Ampere versus GeForce 10 Mobile

Fields that are the same include the 2 GB memory size, the lack of ray tracing and tensor cores, DirectX 12 (12_1) support, OpenGL 4.6 support, Vulkan 1.4 support, and the end-of-life production status.

Head-to-Head Benchmarks

The recorded benchmark data consists of two tests, and the NVIDIA T400 wins both. The first test, Geekbench OpenCL, shows the T400 at 17039 points against the GTX 965M's 14509 points. This is a delta of 17.4%, the largest margin between the two cards in any recorded test. The T400's advantage here is substantial and indicates a meaningful compute performance lead in OpenCL workloads.

The second test, Geekbench Vulkan, narrows the gap somewhat. The T400 scores 15976 while the GTX 965M scores 14299, a delta of 11.7%. The T400 still wins comfortably, but the smaller margin in Vulkan suggests the GTX 965M's architecture handles this API relatively better than it handles OpenCL, possibly due to its larger count of shading units and TMUs.

Looking at the broader context, the T400's average benchmark score of 16508 places it near the NVIDIA GeForce RTX 5090 D V2, which scores 16504 with a delta of 0%, and the AMD Radeon PRO W7500 at 16415 with a delta of 0.6%. The T400 also sits just ahead of the NVIDIA RTX PRO 6000 Blackwell (16408, delta 0.6%) and the AMD Radeon RX 5700 XT (16361, delta 0.9%). This places the T400 in a competitive mid-range position despite its modest specifications.

The GTX 965M's average score of 14404 puts it in the company of the AMD Radeon RX Vega 11 (14385, delta 0.1%), the NVIDIA GeForce GTX TITAN (14373, delta 0.2%), the AMD Radeon Vega 11 (14352, delta 0.4%), and the Intel Iris Xe MAX Graphics (14315, delta 0.6%). The GTX 965M sits at the top of this group, slightly ahead of all its nearest rivals.

The overall win tally is straightforward: the T400 records 2 wins and the GTX 965M records 0 wins. The data does not include any test where the GTX 965M outperforms the T400. This is consistent with the architectural analysis, which shows the T400's newer process node, higher boost clock, and faster memory more than compensate for its smaller compute resource pool.

The T400's superiority in both OpenCL and Vulkan indicates that its Turing architecture provides better compute efficiency per shading unit than the Maxwell 2.0 architecture in the GTX 965M. The GTX 965M's theoretical advantages in pixel rate and texture rate do not translate into benchmark wins in the recorded tests. Based on the available data, the T400 is the stronger performer across all measured workloads.