NVIDIA GeForce GTX 660 Ti vs NVIDIA GeForce GTX 690 Comparison

Head-to-Head Benchmarks

The recorded benchmark data shows a clear and consistent advantage for the NVIDIA GeForce GTX 690 across both tested workloads. In the Geekbench OpenCL test, the GTX 690 scores 17,399 points against 15,113 for the GTX 660 Ti, a lead of 15.1%. That is a substantial margin, roughly the difference you would expect between a flagship dual-GPU card and a mid-range single-GPU offering from the same generation.

The Vulkan test tells a similar story, though the gap narrows slightly. The GTX 690 posts 16,675 points, while the GTX 660 Ti manages 15,012. The delta here is 11.1%, still a comfortable win for the larger card, but the closer margin suggests that the GTX 660 Ti holds up relatively better in API-level workloads that may not scale perfectly with additional compute resources.

Looking at the average benchmark score across both tests, the GTX 690 sits at 17,037, while the GTX 660 Ti trails at 15,063. That works out to a roughly 13% overall advantage for the GTX 690, which aligns with the individual test deltas. The GTX 690 wins both recorded head-to-head matchups, giving it a 2-0 record in this comparison.

Context from the nearest rivals helps frame these numbers. The GTX 690's average score of 17,037 places it within 0.1% of the AMD Radeon HD 7970M (17,019) and 0.6% ahead of the NVIDIA Tesla M4 (16,932). It is essentially tied with those parts. Against the AMD Radeon RX 7600 XT (17,083), the GTX 690 trails by just 0.3%, and versus the NVIDIA GeForce RTX 3070 (17,208), it is 1% behind. These are modern cards, yet the old flagship still hangs close in raw compute benchmarks.

For the GTX 660 Ti, the nearest rival data shows it within 0.1% of the AMD Radeon Pro 560X (15,082), 0.7% behind the AMD Radeon RX 7600 (15,171), 0.9% behind the NVIDIA GeForce RTX 3050 OEM (15,199), and 1.4% behind the AMD Radeon 680M (15,270). The GTX 660 Ti is essentially competitive with these parts, but it does not beat any of them in the recorded averages.

The percentile rankings reinforce the hierarchy. The GTX 690 sits at the 60th percentile among all GPUs in the database, while the GTX 660 Ti sits at the 57th percentile. That is a modest gap in overall standing, but the head-to-head deltas show the GTX 690 is clearly the stronger performer in direct comparison.

Architecture Differences

Both cards share the same fundamental architecture: Kepler on the GK104 chip, built at TSMC's 28 nm process node. Both use the same transistor count of 3,540 million and the same die size of 294 mm², resulting in an identical transistor density of 12.0M per mm². The base clock is also identical at 915 MHz. This is not a case of a newer or older design; it is the same silicon, configured differently.

The critical differences lie in how the silicon is partitioned. The GTX 690 has 1,536 shading units, 128 texture mapping units, and 32 raster operation units. The GTX 660 Ti has fewer of each: 1,344 shading units, 112 TMUs, and 24 ROPs. That is 192 fewer shaders, 16 fewer TMUs, and 8 fewer ROPs on the smaller card. These reductions directly explain the lower pixel rate (27.44 GPixel/s versus 32.61 GPixel/s) and texture rate (109.8 GTexel/s versus 130.4 GTexel/s) for the GTX 660 Ti.

The boost clocks differ as well. The GTX 690 boosts to 1,019 MHz, while the GTX 660 Ti boosts to 980 MHz. That 39 MHz advantage on the flagship contributes to its higher FP32 throughput: 3.130 TFLOPS versus 2.634 TFLOPS. That is an 18.8% advantage in raw compute, which is larger than the shader count difference alone would suggest, because the GTX 690 also runs at a higher boost frequency.

Memory configuration is another major split. Both cards have 2 GB of GDDR5 memory running at 1,502 MHz (6 Gbps effective), but the bus width differs: 256 bit on the GTX 690 versus 192 bit on the GTX 660 Ti. That gives the GTX 690 a memory bandwidth of 192.3 GB/s, while the GTX 660 Ti is limited to 144.2 GB/s. The bandwidth gap of roughly 33% is significant for texture-heavy workloads and higher resolutions.

Power delivery and physical design follow the performance split. The GTX 690 draws 300 W TDP and requires two 8-pin power connectors with a suggested 700 W PSU. The GTX 660 Ti draws half that, 150 W, and uses two 6-pin connectors with a suggested 450 W PSU. The GTX 690 is also longer at 279 mm (11 inches) versus 241 mm (9.5 inches) for the GTX 660 Ti, and it is a dual-slot card in both cases. The GTX 690 measures 111 mm in height and 38 mm in width; the GTX 660 Ti has no recorded height or width dimensions.

Display outputs differ as well. The GTX 690 offers 3x DVI and 1x mini-DisplayPort 1.2, while the GTX 660 Ti offers 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2. The API support is identical: DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175 for both cards.

Both cards are end-of-life products, released in 2012. The GTX 690 launched on May 2, 2012, and the GTX 660 Ti followed on August 15, 2012. Both belong to the GeForce 600 generation, and both succeed the GeForce 500 series with the GeForce 700 series as their successor.

FAQ

Q: Which card is faster in compute benchmarks?

A: The GTX 690 wins both recorded tests. It scores 17,399 in OpenCL and 16,675 in Vulkan, versus 15,113 and 15,012 for the GTX 660 Ti. The deltas are 15.1% and 11.1%, respectively.

Q: Do these cards use the same GPU chip?

A: Yes, both use the GK104 chip with the Kepler architecture, built on TSMC's 28 nm process. They share the same transistor count of 3,540 million and the same die size of 294 mm².

Q: Why is the GTX 690 so much more power-hungry?

A: The GTX 690 has a 300 W TDP and needs two 8-pin connectors with a 700 W suggested PSU. The GTX 660 Ti has a 150 W TDP, two 6-pin connectors, and a 450 W suggested PSU. The higher power draw comes from more shading units, more TMUs, more ROPs, a wider memory bus, and a higher boost clock.

Q: What is the memory bandwidth difference?

A: The GTX 690 has a 256 bit bus and 192.3 GB/s bandwidth. The GTX 660 Ti has a 192 bit bus and 144.2 GB/s bandwidth. Both cards have 2 GB of GDDR5 memory at 6 Gbps effective.

Q: How does the GTX 690 compare to modern cards in the database?

A: Its average score of 17,037 is within 0.3% of the AMD Radeon RX 7600 XT and 1% behind the NVIDIA GeForce RTX 3070. The GTX 660 Ti, at 15,063, is 0.9% behind the NVIDIA GeForce RTX 3050 OEM.

Q: Are these cards still relevant for current games?

A: Both support DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The GTX 690 sits at the 60th percentile among all GPUs, while the GTX 660 Ti sits at the 57th percentile. They are end-of-life products but remain functional for older titles.

The Verdict

The data makes the choice straightforward for anyone comparing these two directly: the GTX 690 is the stronger card in every recorded benchmark. It wins OpenCL by 15.1% and Vulkan by 11.1%. It has more shading units, more texture units, more ROPs, a wider memory bus, higher memory bandwidth, and a higher boost clock. It also has a higher average benchmark score (17,037 versus 15,063) and a higher percentile rank (60th versus 57th).

The GTX 660 Ti is not without merit. It offers the same architecture, the same process node, and the same base clock. It delivers respectable performance that lands within 1.4% of the AMD Radeon 680M and 0.9% of the NVIDIA GeForce RTX 3050 OEM in the database. For a 150 W card with a 450 W suggested PSU, that is efficient performance. But in a direct head-to-head, it cannot match the GTX 690's compute resources.

The real question is whether the GTX 690's advantages matter for your use case. The GTX 690 is a power-hungry dual-slot card that demands a 700 W PSU and two 8-pin connectors. The GTX 660 Ti is much easier to accommodate in a typical system. If your priority is raw performance and you have the power delivery to support it, the GTX 690 is the clear pick. If you need a drop-in card for an older system with a 450 W PSU, the GTX 660 Ti is the practical choice.

Specification Differences

The two cards differ in the following recorded specifications:

• Boost Clock: 1,019 MHz (GTX 690) versus 980 MHz (GTX 660 Ti)
• Shading Units: 1,536 versus 1,344
• TMUs: 128 versus 112
• ROPs: 32 versus 24
• Pixel Rate: 32.61 GPixel/s versus 27.44 GPixel/s
• Texture Rate: 130.4 GTexel/s versus 109.8 GTexel/s
• FP32 Performance: 3.130 TFLOPS versus 2.634 TFLOPS
• Memory Bus Width: 256 bit versus 192 bit
• Memory Bandwidth: 192.3 GB/s versus 144.2 GB/s
• TDP: 300 W versus 150 W
• Power Connectors: 2x 8-pin versus 2x 6-pin
• Suggested PSU: 700 W versus 450 W
• Length: 279 mm (11 inches) versus 241 mm (9.5 inches)
• Height: 111 mm (4.4 inches) versus not recorded
• Width: 38 mm (1.5 inches) versus not recorded
• Display Outputs: 3x DVI, 1x mini-DisplayPort 1.2 versus 2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2
• Launch MSRP: 999 USD versus 299 USD

Identical specifications include the base clock (915 MHz), memory size (2 GB), memory type (GDDR5), memory clock (1,502 MHz / 6 Gbps effective), process node (28 nm), foundry (TSMC), transistor count (3,540 million), die size (294 mm²), transistor density (12.0M / mm²), bus interface (PCIe 3.0 x16), and all API versions.

Where Each One Wins

The GTX 690 wins in every recorded benchmark, so the use-case split is not about which card performs better, it is about which card fits the system and workload constraints.

Choose the GTX 690 if: you need maximum compute throughput from a Kepler-era card. It delivers 3.130 TFLOPS of FP32 performance, 192.3 GB/s of memory bandwidth, and a 15.1% OpenCL lead over the GTX 660 Ti. It is also the better option for multi-display setups with its three DVI outputs. The 60th percentile ranking places it above most GPUs in the database, so it can handle a wider range of tasks without bottlenecking.

Choose the GTX 660 Ti if: your system cannot support a 300 W card. The GTX 660 Ti draws 150 W, requires only a 450 W PSU, and uses two 6-pin connectors instead of two 8-pin. It is also significantly shorter at 241 mm versus 279 mm, which matters for smaller cases. Its Vulkan score of 15,012 is only 11.1% behind the GTX 690, so in Vulkan-based workloads the gap is more manageable. The GTX 660 Ti also offers HDMI output directly, which the GTX 690 lacks.

For gaming at lower resolutions: the GTX 660 Ti's 24 ROPs and 144.2 GB/s bandwidth are sufficient for many titles, and its lower power draw means less heat and noise. The GTX 690's extra ROPs and bandwidth matter more at higher resolutions and with texture-heavy settings.

For compute or content creation: the GTX 690 is the better choice. The 192 additional shading units, 16 extra TMUs, and 48.1 GB/s extra bandwidth translate to measurable gains in OpenCL workloads, as shown by the 15.1% delta.

For a period-correct build: the GTX 660 Ti is the more practical card for most systems from 2012, given its lower power requirements and shorter length. The GTX 690 was a flagship product with a launch MSRP of 999 USD, and it demands flagship-level power delivery.