NVIDIA GeForce GTX 1660 Ti vs NVIDIA GeForce GTX 650 Ti Boost Comparison

NVIDIA
GEFORCE

NVIDIA GeForce GTX 1660 Ti

CORE STATE TU116
VRAM 6 GB
CLOCK SPEED 1770 MHz
TDP 120 W
BUS WIDTH 192 bit
ARCHITECTURE Turing
nm
PROCESS 12 nm
LAUNCH DATE 2019
VS
NVIDIA
GEFORCE

GeForce GTX 650 Ti Boost

CORE STATE GK106
VRAM 2 GB
CLOCK SPEED 1032 MHz
TDP 134 W
BUS WIDTH 192 bit
ARCHITECTURE Kepler
nm
PROCESS 28 nm
LAUNCH DATE 2013

PERFORMANCE BENCHMARKS

3dmark_3dmark_steel_nomad_dx12
1,061
N/A
geekbench_opencl
51,029
9,302
geekbench_vulkan
5,827
10,041
passmark_directx_10
66
N/A
passmark_directx_11
103
N/A
passmark_directx_12
51
N/A
passmark_directx_9
190
N/A
passmark_g2d
804
N/A
passmark_g3d
12,871
N/A
passmark_gpu_compute
5,226
N/A
geekbench_metal
N/A
4,858

Analysis: NVIDIA GeForce GTX 1660 Ti vs NVIDIA GeForce GTX 650 Ti Boost

# NVIDIA GeForce GTX 650 Ti Boost vs NVIDIA GeForce GTX 1660 Ti

The head-to-head benchmark data presents a fascinating split between two NVIDIA graphics cards separated by six years of architecture development. The GTX 650 Ti Boost, built on the Kepler architecture with a 28 nm process, holds its own in one compute workload, while the GTX 1660 Ti, based on the Turing architecture with a 12 nm process, dominates the other. The average benchmark scores tell a nuanced story: the GTX 650 Ti Boost averages 8067 across its tested workloads, while the GTX 1660 Ti averages 7723, yet the individual test results reveal dramatic swings in performance depending on the API and workload type. The percentile rankings are nearly identical, with the GTX 650 Ti Boost at the 42nd percentile and the GTX 1660 Ti at the 41st percentile among all GPUs, suggesting that raw architectural superiority does not always translate directly to every benchmark scenario.

Head-to-Head Benchmarks

The two cards share only one common benchmark in the head-to-head dataset, yet that single comparison is remarkably lopsided. In the Geekbench OpenCL test, the GTX 1660 Ti scores 51,029 points, which is a staggering 81.8% higher than the GTX 650 Ti Boost's 9,302 points. This five-fold advantage reflects the massive compute capability difference between the two generations, driven by the GTX 1660 Ti's 5.437 TFLOPS of FP32 throughput compared to the GTX 650 Ti Boost's 1.585 TFLOPS. The OpenCL result is the clearest expression of raw compute power in this comparison, and the gap is so large that it dwarfs any other statistical difference between the cards.

However, the Vulkan benchmark tells an entirely different story. In the Geekbench Vulkan test, the GTX 650 Ti Boost scores 10,041 points, while the GTX 1660 Ti manages only 5,827 points. This gives the older card a 72.3% advantage, a reversal that is almost as dramatic as the OpenCL result. The Vulkan performance gap suggests that the GTX 650 Ti Boost's driver optimization or architectural characteristics under the Vulkan API are notably superior for this specific workload, despite its older design. The GTX 650 Ti Boost supports Vulkan 1.2.175, while the GTX 1660 Ti supports Vulkan 1.4, yet the newer card still falls behind in this test.

The average benchmark scores place the GTX 650 Ti Boost slightly ahead overall, with an 8067 average versus 7723 for the GTX 1660 Ti, a difference of approximately 4.5% in favor of the older card. This is primarily driven by the inclusion of the Geekbench Vulkan score in the GTX 650 Ti Boost's average, which skews its overall result upward. The GTX 1660 Ti has a much broader benchmark suite, including 3DMark Steel Nomad DX12 at 1,061 points, PassMark DX9 at 190 points, PassMark DX11 at 103 points, PassMark G3D at 12,871 points, and PassMark GPU Compute at 5,226 points, none of which have direct equivalents in the GTX 650 Ti Boost's dataset. The nearest rivals for each card reinforce this unusual positioning: the GTX 650 Ti Boost sits within 0.3% of the NVIDIA GRID K2, GTX 650 Ti, GTX 880M, and Quadro P5000, while the GTX 1660 Ti is within 2.3% of the AMD Radeon 540, Radeon Pro WX 3100, Radeon R7 250, and Intel Arc A310.

Architecture Differences

The architectural divide between these two cards is substantial, beginning with the manufacturing process. The GTX 650 Ti Boost uses TSMC's 28 nm node, while the GTX 1660 Ti is fabricated on a 12 nm process, also by TSMC. This process shrink allows the GTX 1660 Ti to pack 6,600 million transistors into a 284 mm² die, compared to the GTX 650 Ti Boost's 2,540 million transistors on a 221 mm² die. The transistor density more than doubles, from 11.5 million transistors per square millimeter on the Kepler card to 23.2 million per square millimeter on the Turing card. The GTX 650 Ti Boost's chip is the GK106, while the GTX 1660 Ti uses the TU116, representing the Kepler and Turing architectures respectively.

The compute resources differ by a factor of two across the board. The GTX 650 Ti Boost has 768 shading units, 64 texture mapping units, and 24 ROPs, while the GTX 1660 Ti doubles each of these to 1,536 shading units, 96 TMUs, and 48 ROPs. The pixel rate scales accordingly, from 16.51 GPixel/s to 84.96 GPixel/s, and the texture rate jumps from 66.05 GTexel/s to 169.9 GTexel/s. The FP32 throughput expands from 1.585 TFLOPS on the GTX 650 Ti Boost to 5.437 TFLOPS on the GTX 1660 Ti, and the newer card also adds FP16 capability at 10.87 TFLOPS with a 2:1 ratio, a feature entirely absent from the Kepler card. Clock speeds also differ significantly, with the GTX 650 Ti Boost running at a 980 MHz base and 1032 MHz boost, while the GTX 1660 Ti operates at 1,500 MHz base and 1,770 MHz boost.

Memory configuration is another major divergence. Both cards use a 192-bit memory bus, but the GTX 650 Ti Boost has 2 GB of GDDR5 memory running at 6 Gbps effective, yielding 144.2 GB/s of bandwidth. The GTX 1660 Ti has 6 GB of GDDR6 memory at 12 Gbps effective, doubling the bandwidth to 288.0 GB/s. The power characteristics are surprisingly close given the performance gap: the GTX 650 Ti Boost has a TDP of 134 W with a single 6-pin power connector, while the GTX 1660 Ti has a lower TDP of 120 W but requires a single 8-pin connector. Both are dual-slot cards, and both recommend a 300 W power supply. The GTX 1660 Ti is slightly shorter at 229 mm compared to 241 mm, and it has a defined height of 111 mm and width of 35 mm, while the GTX 650 Ti Boost's height and width are unspecified.

Display outputs differ as well: the GTX 650 Ti Boost offers 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2, while the GTX 1660 Ti has 1x DVI, 1x HDMI 2.0, and 1x DisplayPort 1.4a. The DirectX support also advances, with the GTX 650 Ti Boost supporting DirectX 12 (11_0) and the GTX 1660 Ti supporting DirectX 12 (12_1). OpenGL support is identical at 4.6, and Vulkan support differs from 1.2.175 on the older card to 1.4 on the newer one. The GTX 650 Ti Boost launched on March 25, 2013, with a launch MSRP of 169 USD, while the GTX 1660 Ti launched on February 21, 2019, with a launch MSRP of 279 USD. Both are now end-of-life products, with the GTX 650 Ti Boost's successor being the GeForce 700 series and the GTX 1660 Ti's successor being the GeForce 20 series.

The Verdict

The benchmark data indicates that the GTX 1660 Ti is the superior compute performer in OpenCL workloads, with an 81.8% advantage that aligns with its doubled shading units, TMUs, ROPs, higher clocks, and greater memory bandwidth. The GTX 650 Ti Boost, however, retains a decisive edge in Vulkan performance, coming out 72.3% ahead despite its older architecture and far lower raw specifications. The average benchmark scores slightly favor the GTX 650 Ti Boost at 8067 versus 7723, but this average includes only three benchmarks for the older card versus ten for the newer one, making direct comparison of averages misleading. The GTX 1660 Ti's additional benchmark results—including PassMark G3D at 12,871 and 3DMark Steel Nomad DX12 at 1,061—cover a broader range of workloads that the GTX 650 Ti Boost was never tested on.

For users prioritizing OpenCL compute, the GTX 1660 Ti is the clear choice based on its 51,029 score, which represents a five-fold improvement over the GTX 650 Ti Boost's 9,302. The GTX 1660 Ti also offers 6 GB of GDDR6 memory versus 2 GB of GDDR5, doubling bandwidth to 288.0 GB/s, which matters for memory-intensive tasks. The GTX 650 Ti Boost, however, should not be dismissed for Vulkan-based applications, where its 10,041 score exceeds the GTX 1660 Ti's 5,827 by a wide margin. The percentile rankings suggest both cards perform within the same tier overall, at the 42nd and 41st percentiles respectively, which is consistent with their close average scores when accounting for different test suites. The data does not support a universal verdict; instead, the choice depends entirely on the workload API.

FAQ

Q: Which card has the higher average benchmark score?

A: The GTX 650 Ti Boost has an average benchmark score of 8067, while the GTX 1660 Ti averages 7723, giving the older card a slight edge of approximately 4.5%.

Q: What is the memory configuration difference between the two cards?

A: The GTX 650 Ti Boost has 2 GB of GDDR5 memory with a 192-bit bus and 144.2 GB/s bandwidth, while the GTX 1660 Ti has 6 GB of GDDR6 memory with the same 192-bit bus but 288.0 GB/s bandwidth.

Q: How do the FP32 compute performances compare?

A: The GTX 650 Ti Boost delivers 1.585 TFLOPS of FP32 performance, while the GTX 1660 Ti delivers 5.437 TFLOPS, a difference of 3.852 TFLOPS in favor of the newer card.

Q: Which card performs better in the Geekbench Vulkan test?

A: The GTX 650 Ti Boost scores 10,041 points in Geekbench Vulkan, which is 72.3% higher than the GTX 1660 Ti's 5,827 points, making the older card the clear winner in this specific workload.

Q: What are the TDP ratings and power connector requirements?

A: The GTX 650 Ti Boost has a TDP of 134 W and requires a single 6-pin power connector, while the GTX 1660 Ti has a lower TDP of 120 W but requires a single 8-pin connector; both recommend a 300 W power supply.

Q: What DirectX versions do the cards support?

A: The GTX 650 Ti Boost supports DirectX 12 (11_0), while the GTX 1660 Ti supports DirectX 12 (12_1), reflecting the newer card's more advanced feature set.

Where Each One Wins

The GTX 1660 Ti dominates in OpenCL compute workloads, posting a score of 51,029 against the GTX 650 Ti Boost's 9,302, an 81.8% advantage. This makes it the appropriate choice for applications that leverage OpenCL for general-purpose GPU computing, such as scientific simulations, video encoding, or any compute-heavy task that uses this API. The card's doubled shading units (1,536 versus 768), higher clock speeds (1,770 MHz boost versus 1,032 MHz), and greater memory bandwidth (288.0 GB/s versus 144.2 GB/s) all contribute to this outcome. The GTX 1660 Ti also has the advantage of 6 GB of GDDR6 memory, which allows it to handle larger datasets than the GTX 650 Ti Boost's 2 GB GDDR5 frame buffer.

The GTX 650 Ti Boost wins decisively in the Geekbench Vulkan test, scoring 10,041 points compared to the GTX 1660 Ti's 5,827, a 72.3% margin. This suggests that for Vulkan-based games or applications, the older Kepler card may deliver better performance despite its older architecture. The GTX 650 Ti Boost's nearest rivals in average score include the NVIDIA GRID K2, GTX 650 Ti, GTX 880M, and Quadro P5000, all within 0.3% of its 8067 average. The GTX 1660 Ti's closest competitors are AMD Radeon 540, Radeon Pro WX 3100, Radeon R7 250, and Intel Arc A310, within 2.3% of its 7723 average. For users with workloads that rely on Vulkan for rendering or compute, the GTX 650 Ti Boost presents a compelling case based purely on the benchmark data, though its end-of-life status and older feature set should be considered. The GTX 1660 Ti, meanwhile, offers a broader benchmark footprint with additional tests like 3DMark Steel Nomad DX12, PassMark DX9 through DX12, and PassMark G2D and G3D, indicating it has been validated across a wider range of scenarios.

DETAILED SPECIFICATIONS

SPECIFICATION
GTX 1660 Ti
GTX 650 Ti Boost
Core Specs
Shading Units
1,536
768 -50.0%
Shaders
1,536
768 -50.0%
TMUs
96
64 -33.3%
ROPs
48
24 -50.0%
SM Count
24
—
Clocks
Base Clock
1500 MHz
980 MHz
Boost Clock
1770 MHz
1032 MHz
Memory Clock
1500 MHz 12 Gbps effective
1502 MHz 6 Gbps effective
Memory
Memory Size
6 GB
2 GB
VRAM (MB)
6,144
2,048 -66.7%
Memory Type
GDDR6
GDDR5
Memory Bus
192 bit
192 bit
Bandwidth
288.0 GB/s
144.2 GB/s
Cache
L1 Cache
64 KB (per SM)
16 KB (per SMX)
L2 Cache
1536 KB
384 KB
Performance
Pixel Rate
84.96 GPixel/s
16.51 GPixel/s
Texture Rate
169.9 GTexel/s
66.05 GTexel/s
FP32 (TFLOPS)
5.437 TFLOPS
1.585 TFLOPS
FP64 (TFLOPS)
169.9 GFLOPS (1:32)
66.05 GFLOPS (1:24)
FP16 (TFLOPS)
10.87 TFLOPS (2:1)
—
Power
TDP
120 W
134 W
TDP (W)
120
134 +11.7%
Suggested PSU
300 W
300 W
Power Connectors
1x 8-pin
1x 6-pin
Architecture
Architecture
Turing
Kepler
GPU Name
TU116
GK106
Generation
GeForce 16
GeForce 600
Process Size
12 nm
28 nm
Transistors
6,600 million
2,540 million
Die Size
284 mm²
221 mm²
Foundry
TSMC
TSMC
Density
23.2M / mm²
11.5M / mm²
API Support
DirectX
12 (12_1)
12 (11_0)
OpenGL
4.6
4.6
Vulkan
1.4
1.2.175
OpenCL
3.0
3.0
CUDA
7.5
3.0
Shader Model
6.8
6.5 (5.1)
Physical
Slot Width
Dual-slot
Dual-slot
Length
229 mm 9 inches
241 mm 9.5 inches
Height
111 mm 4.4 inches
—
Outputs
1x DVI1x HDMI 2.01x DisplayPort 1.4a
2x DVI1x HDMI 1.4a1x DisplayPort 1.2
Bus Interface
PCIe 3.0 x16
PCIe 3.0 x16
Other
Launch Price
279 USD
169 USD
Production
End-of-life
End-of-life
Predecessor
GeForce 10
GeForce 500
Successor
GeForce 20
GeForce 700
View GeForce GTX 1660 Ti Details View GeForce GTX 650 Ti Boost Details