NVIDIA GeForce GTX 1660 Ti vs NVIDIA GRID K2 Comparison

NVIDIA’s GRID K2 and GeForce GTX 1660 Ti are both end-of-life products, but they target entirely different workloads. The GRID K2 is a 2013-era Kepler board built for virtualized cloud graphics, while the GTX 1660 Ti is a 2019 Turing consumer card for local rendering. Benchmark data shows a single head-to-head test, with the GTX 1660 Ti dominating in raw compute. However, the GRID K2’s legacy lies in its server-oriented design, not in gaming or desktop performance.

The Verdict

Pick the NVIDIA GeForce GTX 1660 Ti if you need a modern, self-contained graphics card for a desktop PC. The data is unambiguous: in the only shared benchmark, Geekbench OpenCL, the GTX 1660 Ti scores 51,029 versus the GRID K2’s 10,602, a 79.2% advantage. That gap reflects a newer architecture, faster memory, and far higher clock speeds. The GTX 1660 Ti also has a functional display output suite (1x DVI, 1x HDMI 2.0, 1x DisplayPort 1.4a), while the GRID K2 has no outputs — it is meant to be accessed remotely, not plugged into a monitor.

Pick the NVIDIA GRID K2 only if your use case is a virtualized environment where physical display outputs are irrelevant. Its 4 GB GDDR5 memory and 160.0 GB/s bandwidth, combined with a Kepler architecture, made it a capable cloud GPU in its day. But its 42nd percentile ranking among all GPUs and its nearest rivals — the GTX 650 Ti Boost (0.2% ahead), GeForce 945M (0.2% behind), GTX 650 Ti (0.3% ahead), and GTX 880M (0.5% ahead) — show it is a low-end performer by modern standards. The GTX 1660 Ti sits at the 41st percentile, but its nearest rivals are low-end AMD and Intel parts (Radeon 540, Radeon Pro WX 3100, Radeon R7 250, Arc A310), meaning its average score is dragged down by a broader benchmark suite that includes older API tests.

For a practical builder, the GTX 1660 Ti is the only one that makes sense for a physical machine. The GRID K2 is a niche artifact for legacy virtualization servers.

Where Each One Wins

The GTX 1660 Ti wins the only direct comparison, but the GRID K2 has structural advantages for specific deployments. In Geekbench OpenCL, the GTX 1660 Ti’s 51,029 score crushes the GRID K2’s 10,602. That is a 5x difference in raw compute throughput, driven by the Turing chip’s 5.437 TFLOPS FP32 versus the Kepler chip’s 2.289 TFLOPS. The GTX 1660 Ti also has a higher pixel rate (84.96 GPixel/s vs 23.84 GPixel/s) and texture rate (169.9 GTexel/s vs 95.36 GTexel/s), so it wins in any rasterization-heavy workload.

The GRID K2’s wins are not in performance but in form factor and power delivery. It uses a 1x 6-pin + 1x 8-pin power connector setup, suggesting it was designed for server chassis with robust power delivery. Its 225 W TDP is higher than the GTX 1660 Ti’s 120 W, meaning the K2 draws more power but may fit into pre-existing server power budgets. The K2 also has a longer board (267 mm vs 229 mm), which is unusual for a card without outputs — again, it is a server card.

In API support, the GTX 1660 Ti leads with DirectX 12 (12_1) and Vulkan 1.4, while the GRID K2 is limited to DirectX 12 (11_0) and Vulkan 1.2.175. For modern games or compute frameworks, the GTX 1660 Ti is the only viable option.

Architecture Differences

The two cards are generations apart. The GRID K2 uses the GK104 chip on the Kepler architecture, built on TSMC’s 28 nm process. It packs 3,540 million transistors into a 294 mm² die, giving a density of 12.0M transistors per mm². The GTX 1660 Ti uses the TU116 chip on the Turing architecture, also from TSMC but on a 12 nm process. It has 6,600 million transistors in a 284 mm² die, a density of 23.2M / mm².

The GTX 1660 Ti’s memory subsystem is newer and faster: 6 GB GDDR6 on a 192-bit bus with 288.0 GB/s bandwidth, versus 4 GB GDDR5 on a 256-bit bus with 160.0 GB/s. Effective memory speed is 12 Gbps versus 5 Gbps. The K2’s memory clock is listed as 1250 MHz (5 Gbps effective), while the GTX 1660 Ti runs at 1500 MHz (12 Gbps effective).

Core counts differ in texture and raster units. Both have 1536 shading units, but the K2 has 128 TMUs and 32 ROPs, while the GTX 1660 Ti has 96 TMUs and 48 ROPs. The higher ROP count on the GTX 1660 Ti explains its superior pixel rate. The K2’s higher TMU count is offset by its lower clock speeds — the GTX 1660 Ti has a base clock of 1500 MHz and boost of 1770 MHz, while the K2 has no listed base or boost clocks.

Neither card has ray tracing or tensor cores. The GTX 1660 Ti does have FP16 support at 10.87 TFLOPS (2:1 ratio), while the K2 has no FP16 listing.

FAQ

Q: Which card is faster in the only shared benchmark?

A: The NVIDIA GeForce GTX 1660 Ti scores 51,029 on Geekbench OpenCL, versus 10,602 for the NVIDIA GRID K2, a 79.2% lead.

Q: Can I connect a monitor to the GRID K2?

A: No. The GRID K2 has no display outputs. The GTX 1660 Ti has 1x DVI, 1x HDMI 2.0, and 1x DisplayPort 1.4a.

Q: Why is the GRID K2’s average benchmark score higher than the GTX 1660 Ti’s?

A: The GRID K2’s average benchmark score is 8,080, while the GTX 1660 Ti’s is 7,723. This is because the K2 has only two benchmark entries (Geekbench Metal and OpenCL), while the GTX 1660 Ti has ten entries, including older DirectX 9 and 10 tests that pull its average down.

Q: Which card has better API support for modern games?

A: The GTX 1660 Ti supports DirectX 12 (12_1) and Vulkan 1.4, while the GRID K2 supports DirectX 12 (11_0) and Vulkan 1.2.175.

Q: What is the power draw difference?

A: The GRID K2 has a 225 W TDP and suggests a 550 W PSU, while the GTX 1660 Ti has a 120 W TDP and suggests a 300 W PSU.

Q: Are these cards still in production?

A: Both are end-of-life. The GRID K2 was released in 2013, and the GTX 1660 Ti in 2019.

Head-to-Head Benchmarks

The sole head-to-head result is Geekbench OpenCL. The GTX 1660 Ti scores 51,029, the GRID K2 scores 10,602, and the delta is -79.2% for the K2. That is not a marginal win; it is a landslide. To put it in context, the GRID K2’s nearest rivals in average score are the GTX 650 Ti Boost (8,067), GeForce 945M (8,099), GTX 650 Ti (8,053), and GTX 880M (8,040) — all within 0.5% of its 8,080 average. The GTX 1660 Ti’s nearest rivals are the AMD Radeon 540 (7,673), Radeon Pro WX 3100 (7,580), Radeon R7 250 (7,557), and Intel Arc A310 (7,550) — all within 2.3% of its 7,723 average.

The 79.2% deficit in OpenCL is consistent with the architectural gap. The GTX 1660 Ti has 5.437 TFLOPS FP32, more than double the K2’s 2.289 TFLOPS. Memory bandwidth is 288.0 GB/s versus 160.0 GB/s. Even the texture rate, where the K2 has more TMUs (128 vs 96), ends up favoring the GTX 1660 Ti (169.9 GTexel/s vs 95.36 GTexel/s) because of clock speed.

There is no head-to-head test for gaming or DirectX workloads, but the API support alone tells the story: the GTX 1660 Ti supports DirectX 12 (12_1), while the K2 is stuck at DirectX 12 (11_0). The GTX 1660 Ti also supports Vulkan 1.4, a newer revision than the K2’s 1.2.175.

Specification Differences

The following fields differ between the two cards, based on the data:

• Chip: GK104 vs TU116
• Architecture: Kepler vs Turing
• Process node: 28 nm vs 12 nm
• Transistors: 3,540 million vs 6,600 million
• Die size: 294 mm² vs 284 mm²
• Transistor density: 12.0M / mm² vs 23.2M / mm²
• Base clock: Not listed vs 1500 MHz
• Boost clock: Not listed vs 1770 MHz
• Memory clock: 1250 MHz (5 Gbps effective) vs 1500 MHz (12 Gbps effective)
• Memory size: 4 GB vs 6 GB
• Memory type: GDDR5 vs GDDR6
• Memory bus width: 256 bit vs 192 bit
• Memory bandwidth: 160.0 GB/s vs 288.0 GB/s
• TMUs: 128 vs 96
• ROPs: 32 vs 48
• Pixel rate: 23.84 GPixel/s vs 84.96 GPixel/s
• Texture rate: 95.36 GTexel/s vs 169.9 GTexel/s
• FP32: 2.289 TFLOPS vs 5.437 TFLOPS
• FP16: Not listed vs 10.87 TFLOPS (2:1)
• TDP: 225 W vs 120 W
• Power connectors: 1x 6-pin + 1x 8-pin vs 1x 8-pin
• Suggested PSU: 550 W vs 300 W
• Display outputs: No outputs vs 1x DVI, 1x HDMI 2.0, 1x DisplayPort 1.4a
• DirectX support: 12 (11_0) vs 12 (12_1)
• Vulkan support: 1.2.175 vs 1.4
• Dimensions: 267 mm (10.5 inches) vs 229 mm (9 inches) length; height and width only listed for the GTX 1660 Ti (111 mm / 4.4 inches, 35 mm / 1.4 inches)
• Release date: 2013-05-10 vs 2019-02-21
• Launch MSRP: 5,199 USD vs 279 USD