NVIDIA GeForce GTX 690 vs NVIDIA Tesla K40m Comparison

FAQ

Q: Which GPU has a higher OpenCL benchmark score?

A: The NVIDIA Tesla K40m scores 19,885 in Geekbench OpenCL, while the GeForce GTX 690 scores 17,399. The Tesla K40m leads by 14.3%.

Q: How much memory does each card have?

A: The Tesla K40m has 12 GB of GDDR5 memory on a 384-bit bus, while the GTX 690 has 2 GB of GDDR5 on a 256-bit bus. The memory bandwidth figures are 288.4 GB/s for the Tesla and 192.3 GB/s for the GTX 690.

Q: What is the transistor count difference between the two chips?

A: The Tesla K40m's GK110B chip contains 7,080 million transistors on a 561 mm² die. The GTX 690's GK104 chip has 3,540 million transistors on a 294 mm² die. Both are built on a 28 nm process at TSMC.

Q: Which card has a higher boost clock?

A: The GeForce GTX 690 runs a base clock of 915 MHz and a boost clock of 1,019 MHz. The Tesla K40m runs at 745 MHz base and 876 MHz boost. The GTX 690's base clock is 170 MHz higher.

Q: What is the power draw of each card?

A: The Tesla K40m has a TDP of 245 W with a suggested PSU of 550 W. The GTX 690 has a TDP of 300 W and a suggested PSU of 700 W.

Q: Which card has a higher percentile ranking in the database?

A: The Tesla K40m sits at the 65th percentile against all GPUs, while the GTX 690 sits at the 60th percentile. The Tesla also has a higher average benchmark score at 19,885 versus 17,037 for the GTX 690.

Architecture Differences

Both cards share the Kepler architecture and a 28 nm TSMC process, but they are fundamentally different chips with different design goals. The Tesla K40m uses the GK110B chip, which is the larger, compute-oriented silicon in the Kepler lineup. The GTX 690 uses the GK104 chip, which was designed for gaming performance per watt and per dollar.

The sheer scale difference is visible in the silicon itself. The GK110B packs 7,080 million transistors into a 561 mm² die, while the GK104 uses 3,540 million transistors on a 294 mm² die. That is exactly double the transistor count on nearly double the die area. Transistor density is also slightly higher on the Tesla, at 12.6M per mm² versus 12.0M per mm².

The execution resources follow the same pattern. The Tesla K40m has 2,880 shading units, 240 texture mapping units, and 48 ROPs. The GTX 690 has 1,536 shading units, 128 TMUs, and 32 ROPs. For every major compute block, the Tesla has roughly twice the hardware.

Memory configuration differs sharply. The Tesla K40m carries 12 GB of GDDR5 across a 384-bit bus, yielding 288.4 GB/s of bandwidth. The GTX 690 carries 2 GB of GDDR5 across a 256-bit bus, yielding 192.3 GB/s. The Tesla's memory subsystem is built for large datasets and compute workloads, not just frame buffers.

The GTX 690 does have one notable architectural distinction: it is a dual-GPU card. The GK104 chip appears twice on the board, which is why the card consumes 300 W versus the Tesla's 245 W. However, the database records the GTX 690's shading units, TMUs, and ROPs as the per-GPU totals listed above, and its benchmark scores reflect the combined result of both GPUs.

Clock speeds favor the GTX 690. Its base clock is 915 MHz and boost reaches 1,019 MHz, while the Tesla K40m runs at 745 MHz base and 876 MHz boost. The GTX 690's clocks are higher, but its per-chip resources are smaller.

The feature sets differ in ways that matter for deployment. The Tesla K40m has no display outputs; it is a compute-only accelerator. The GTX 690 has 3x DVI and 1x mini-DisplayPort 1.2 outputs. The Tesla supports DirectX 12 (11_1) while the GTX supports DirectX 12 (11_0). Both support OpenGL 4.6 and Vulkan 1.2.175.

The Tesla K40m belongs to the Tesla Kepler (Kxx) generation and was released later, on 2013-11-21. The GTX 690 is part of the GeForce 600 generation and released on 2012-05-02. The Tesla's predecessor is Tesla Fermi and its successor is Tesla Maxwell. The GTX 690's predecessor is GeForce 500 and its successor is GeForce 700.

Head-to-Head Benchmarks

The database contains one head-to-head benchmark between these two cards: Geekbench OpenCL. The Tesla K40m scores 19,885 against the GTX 690's 17,399, a 14.3% advantage for the Tesla. That is the only direct comparison recorded, and the Tesla wins it.

Context from the nearest rivals helps interpret the margin. The Tesla K40m's closest competitor is the AMD FirePro W7000 at 19,905, which is only 0.1% ahead of the Tesla. The AMD Radeon RX 6650 XT sits 0.6% behind at 19,765, and the AMD FirePro D300 trails by 1.3% at 19,637. The NVIDIA Quadro K5200 is 1.4% behind at 19,602. The Tesla's score places it in a tight cluster of professional and consumer cards, and its 14.3% lead over the GTX 690 is far larger than any of those rival deltas.

The GTX 690's nearest rivals show a different competitive picture. Its average score across two benchmarks is 17,037, with OpenCL at 17,399 and Vulkan at 16,675. The AMD Radeon HD 7970M is 0.1% ahead at 17,019, the AMD Radeon RX 7600 XT is 0.3% ahead at 17,083, and the NVIDIA Tesla M4 is 0.6% behind at 16,932. The NVIDIA GeForce RTX 3070 is 1% ahead at 17,208. The GTX 690 sits in a dense pack where the nearest rivals are all within a single percentage point.

The 14.3% delta between the two cards is meaningful. It is more than ten times larger than any delta in the Tesla's nearest rival list, and more than fourteen times larger than any delta in the GTX 690's nearest rival list. In other words, the performance gap between the Tesla K40m and GTX 690 is not a rounding error; it is a decisive separation in this benchmark.

Pixel and texture rates support the same conclusion. The Tesla K40m delivers 52.56 GPixel/s and 210.2 GTexel/s, while the GTX 690 delivers 32.61 GPixel/s and 130.4 GTexel/s. The Tesla's pixel rate is roughly 61% higher and its texture rate roughly 61% higher. FP32 compute follows the same pattern: 5.046 TFLOPS for the Tesla versus 3.130 TFLOPS for the GTX 690.

Specification Differences

| Specification | NVIDIA Tesla K40m | NVIDIA GeForce GTX 690 |

| --- | --- | --- |

| Chip | GK110B | GK104 |

| Generation | Tesla Kepler (Kxx) | GeForce 600 |

| Transistors | 7,080 million | 3,540 million |

| Die Size | 561 mm² | 294 mm² |

| Transistor Density | 12.6M / mm² | 12.0M / mm² |

| Base Clock | 745 MHz | 915 MHz |

| Boost Clock | 876 MHz | 1,019 MHz |

| Memory Size | 12 GB | 2 GB |

| Memory Bus Width | 384 bit | 256 bit |

| Memory Bandwidth | 288.4 GB/s | 192.3 GB/s |

| Shading Units | 2,880 | 1,536 |

| TMUs | 240 | 128 |

| ROPs | 48 | 32 |

| Pixel Rate | 52.56 GPixel/s | 32.61 GPixel/s |

| Texture Rate | 210.2 GTexel/s | 130.4 GTexel/s |

| FP32 | 5.046 TFLOPS | 3.130 TFLOPS |

| TDP | 245 W | 300 W |

| Power Connectors | Not specified | 2x 8-pin |

| Suggested PSU | 550 W | 700 W |

| Display Outputs | No outputs | 3x DVI, 1x mini-DisplayPort 1.2 |

| DirectX | 12 (11_1) | 12 (11_0) |

| Length | 267 mm (10.5 inches) | 279 mm (11 inches) |

| Height | Not specified | 111 mm (4.4 inches) |

| Width | Not specified | 38 mm (1.5 inches) |

| Release Date | 2013-11-21 | 2012-05-02 |

| Predecessor | Tesla Fermi | GeForce 500 |

| Successor | Tesla Maxwell | GeForce 700 |

| Launch MSRP | 7,699 USD | 999 USD |

| OpenCL Score | 19,885 | 17,399 |

| Vulkan Score | Not recorded | 16,675 |

| Average Benchmark Score | 19,885 | 17,037 |

| Percentile vs All GPUs | 65 | 60 |

The two cards share the same process node, foundry, memory type, memory clock, bus interface, slot width, OpenGL version, Vulkan version, and production status. Those are the commonalities. Everything else in the table above is a point of difference.

The Verdict

The data points to a clear separation between these two cards, but the right choice depends entirely on the workload.

For compute work, the Tesla K40m is the stronger part. It wins the only head-to-head benchmark by 14.3%, holds a 65th percentile ranking versus the GTX 690's 60th, and delivers 5.046 TFLOPS of FP32 compute against 3.130 TFLOPS. Its 12 GB memory capacity and 384-bit bus are built for datasets that would overwhelm the GTX 690's 2 GB frame buffer. The Tesla also consumes less power at 245 W versus 300 W, and requires a smaller suggested PSU at 550 W versus 700 W. Its lack of display outputs is irrelevant for a compute accelerator.

For gaming or general desktop use, the GTX 690 is the more sensible device. It has display outputs, a dual-GPU design, and higher clock speeds at 915 MHz base and 1,019 MHz boost. Its OpenCL score of 17,399 is lower than the Tesla's 19,885, but it also carries a Vulkan score of 16,675, which the Tesla does not record. The GTX 690's nearest rival list includes the GeForce RTX 3070 at just 1% ahead, which places it in a competitive gaming performance tier despite its age.

The launch MSRP figures reflect the positioning: the Tesla K40m launched at 7,699 USD, while the GTX 690 launched at 999 USD. The Tesla demands a premium for its compute resources, and the benchmark data justifies that premium only for workloads that can use 12 GB of memory and 2,880 shading units.

The GTX 690 wins in clock speed, power connectors, display outputs, and physical compactness relative to its height and width. The Tesla wins in memory capacity, memory bandwidth, shading units, TMUs, ROPs, pixel rate, texture rate, FP32 compute, power efficiency, and every recorded benchmark comparison.

The verdict is straightforward. The Tesla K40m is the pick for compute, data-heavy workloads, and any task that scales with shading units and memory bandwidth. The GTX 690 is the pick for anyone who needs display outputs, higher clocks, and a lower launch price. Performance data alone favors the Tesla, but the GTX 690's feature set serves a different use case entirely.