NVIDIA GeForce GTX 690 vs NVIDIA T400 4 GB Comparison

The NVIDIA GeForce GTX 690 and NVIDIA T400 4 GB represent two very different eras of GPU design, separated by nearly a decade of architectural evolution. The GTX 690 is a dual-GPU flagship from the Kepler generation, while the T400 is a low-power Turing-based workstation card. Benchmark data shows the GTX 690 edges out the T400 in both available tests, but the story is far more nuanced than raw scores alone, given the substantial differences in power, memory, and feature support.

Head-to-Head Benchmarks

The head-to-head benchmark results show a consistent, though narrow, victory for the older GTX 690. In the Geekbench OpenCL test, the GTX 690 scores 17,399 against the T400's 17,320, a margin of just 0.5%. This near-parity in OpenCL performance is remarkable given the architectural gulf between the two cards. The GTX 690’s dual-GPU configuration and massive 3,130 TFLOPS of FP32 compute power are offset by the T400’s newer Turing architecture, which brings efficiencies that Kepler simply cannot match.

The Vulkan results tell a similar story with a slightly larger gap. The GTX 690 posts 16,675 points, while the T400 manages 16,263, giving the GTX 690 a 2.5% advantage. This suggests that in modern API workloads, the Kepler architecture holds up better than one might expect, though the T400’s support for DirectX 12 (12_1) and Vulkan 1.4 indicates it is better positioned for future software compatibility.

When placed against their respective nearest rivals, both cards occupy a similar performance tier. The GTX 690’s average benchmark score of 17,037 places it 0.1% above the AMD Radeon HD 7970M and 0.6% above the NVIDIA Tesla M4, while sitting 1% below the NVIDIA GeForce RTX 3070 and 0.3% below the AMD Radeon RX 7600 XT. The T400’s average score of 16,792 is 0.6% above the AMD Radeon RX 7600S, 0.8% below the Tesla M4, 1.3% below the HD 7970M, and 1.4% below the GTX 690 itself. Both cards sit at the 60th percentile among all GPUs, indicating they are firmly mid-pack performers in the current landscape.

The data shows the GTX 690 wins both head-to-head tests, but the margins are thin. A 0.5% lead in OpenCL and a 2.5% lead in Vulkan are within the territory of driver-level variance and workload-specific optimizations. For practical purposes, these two cards deliver nearly interchangeable compute performance in the tested benchmarks, despite their wildly different specifications.

Architecture Differences

The architectural chasm between these two cards is vast. The GTX 690 is built on TSMC's 28 nm process node, packing 3,540 million transistors into a 294 mm² die, yielding a transistor density of 12.0M per mm². It uses the GK104 chip from the Kepler architecture, part of the GeForce 600 generation. In contrast, the T400 uses the TU117 chip from the Turing architecture, manufactured on TSMC's 12 nm node. This newer process allows 4,700 million transistors in a smaller 200 mm² die, achieving a much higher density of 23.5M per mm².

The compute resources differ enormously. The GTX 690 boasts 1,536 shading units, 128 texture mapping units, and 32 ROPs. The T400 is far more modest, with 384 shading units, 24 TMUs, and 16 ROPs. This gives the GTX 690 a massive theoretical advantage in raw throughput: 3,130 TFLOPS FP32 versus the T400's 1,094.4 GFLOPS. The T400 does offer FP16 compute at 2,189 TFLOPS (2:1 ratio), a capability the Kepler-based GTX 690 lacks entirely.

Memory configurations are equally divergent. The GTX 690 uses 2 GB of GDDR5 on a 256-bit bus, delivering 192.3 GB/s of bandwidth at 6 Gbps effective. The T400 comes with 4 GB of GDDR6 on a 64-bit bus, but its bandwidth is just 80.00 GB/s at 10 Gbps effective. The T400 has half the memory bus width but double the capacity and a faster memory type, resulting in less than half the bandwidth of the GTX 690.

Power consumption tells a stark tale of efficiency gains. The GTX 690 is rated at 300 W TDP, requires a dual-slot cooler with two 8-pin power connectors, and needs a 700 W power supply. It is a large card at 279 mm long, 111 mm high, and 38 mm wide. The T400, by contrast, sips just 30 W, uses no power connectors, fits in a single slot, and requires only a 200 W PSU. Its dimensions are not listed, but the single-slot design and lack of power connectors indicate a compact form factor.

Feature support also differs by generation. The GTX 690 supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175, with display outputs of 3x DVI and 1x mini-DisplayPort 1.2. The T400 supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4, with 3x mini-DisplayPort 1.4a outputs. The newer API support on the T400, particularly the higher Vulkan version and DirectX 12 feature level, indicates better compatibility with modern software stacks.

The Verdict

The benchmark data presents a clear but counterintuitive picture. The GTX 690, despite being a 2012-era dual-GPU monster, edges out the 2021 T400 in both OpenCL and Vulkan tests. The GTX 690's win margins are small—0.5% and 2.5%—but consistent across both benchmarks. Its average benchmark score of 17,037 is 1.4% higher than the T400's 16,792, a difference that is statistically minor but directionally clear.

However, the verdict is not simply "the GTX 690 is better." The T400 offers substantial advantages in efficiency and modern feature support. Its 30 W TDP versus the GTX 690's 300 W represents a tenfold reduction in power draw, making it viable in systems where the GTX 690 would be impractical or impossible to run. The T400 also has double the memory capacity (4 GB versus 2 GB), which matters for workloads that exceed the GTX 690's frame buffer.

The architecture differences explain the performance parity despite the GTX 690's overwhelming compute advantage. The T400's Turing architecture is more efficient per transistor and per watt, compensating for its smaller resource pool. The data shows both cards achieving a 60th percentile ranking among all GPUs, meaning they occupy the same performance tier in the broader market.

For users prioritizing raw compute density and are willing to accept a 300 W power draw, the GTX 690 is the stronger choice based on benchmark scores. For users needing low power consumption, modern API support, or larger memory capacity, the T400 is the more sensible selection, even if it loses the head-to-head tests.

FAQ

Q: Which card has a higher average benchmark score?

A: The NVIDIA GeForce GTX 690 has an average benchmark score of 17,037, which is 1.4% higher than the NVIDIA T400 4 GB's average of 16,792.

Q: What is the difference in FP32 compute performance?

A: The GTX 690 delivers 3,130 TFLOPS of FP32 compute, while the T400 provides 1,094.4 GFLOPS. The GTX 690 has approximately 2.9 times the FP32 throughput.

Q: How do their power requirements compare?

A: The GTX 690 has a TDP of 300 W and requires two 8-pin power connectors with a suggested 700 W power supply. The T400 has a TDP of 30 W, uses no power connectors, and requires only a 200 W power supply.

Q: Which card supports a newer version of Vulkan?

A: The T400 supports Vulkan 1.4, while the GTX 690 supports Vulkan 1.2.175. The T400 also supports DirectX 12 (12_1) compared to the GTX 690's DirectX 12 (11_0).

Q: What is the memory bandwidth difference?

A: The GTX 690 has a 256-bit bus with 192.3 GB/s bandwidth, while the T400 has a 64-bit bus with 80.00 GB/s bandwidth. The GTX 690 provides more than double the bandwidth.

Q: What are the transistor counts and die sizes?

A: The GTX 690 contains 3,540 million transistors on a 294 mm² die (28 nm), while the T400 contains 4,700 million transistors on a 200 mm² die (12 nm).

Where Each One Wins

The GTX 690 wins in raw compute throughput, memory bandwidth, and benchmark scores. Its 3,130 TFLOPS FP32 performance and 192.3 GB/s bandwidth are far superior to the T400's specifications, which explains its slight edge in both Geekbench tests. For workloads that are bandwidth-bound or require massive parallel compute, such as large-scale simulations or high-resolution texture processing, the GTX 690's architecture is better suited. Its 128 TMUs and 32 ROPs also give it advantages in texture-heavy and pixel-heavy rendering tasks.

The T400 wins decisively in power efficiency, memory capacity, and modern feature support. Its 30 W TDP is a fraction of the GTX 690's 300 W, making it suitable for small form factor systems, workstations with limited power budgets, or environments where heat dissipation is a concern. The 4 GB memory capacity is double the GTX 690's 2 GB, which is critical for workloads that require larger datasets or higher resolution textures that exceed 2 GB. The T400's support for Vulkan 1.4 and DirectX 12 (12_1) ensures compatibility with the latest graphics APIs, while the GTX 690 is limited to older API versions.

In practical terms, the T400 is the better choice for modern workstation tasks like CAD, video encoding, or AI inference that benefit from FP16 compute and newer API features. The GTX 690, despite its age, remains viable for compute-intensive tasks that leverage its raw FP32 power and high memory bandwidth. The benchmark data shows these cards deliver nearly identical overall performance in the tested workloads, so the deciding factors are power, memory capacity, and software compatibility rather than raw speed.