AMD Radeon Vega 8 vs NVIDIA GeForce GTX 1660 Ti Comparison

Where Each One Wins

The recorded benchmark data splits cleanly between these two GPUs, with each taking one decisive victory in the two available head-to-head comparisons. The AMD Radeon Vega 8 wins the Vulkan compute test, while the NVIDIA GeForce GTX 1660 Ti dominates the OpenCL workload. This is not a case of one card being universally faster; rather, the two accelerators show complementary strengths depending on the API in use.

The Vega 8’s Vulkan score of 8,134 against the GTX 1660 Ti’s 5,827 represents a 39.6% advantage. This is a substantial margin, indicating that in Vulkan-based compute workloads, the integrated part from AMD is the stronger option. The GTX 1660 Ti, however, posts an OpenCL score of 51,029 versus the Vega 8’s 8,822, an 82.7% deficit for the AMD part. That OpenCL result is more than five times higher, showing NVIDIA’s discrete card to be in a completely different performance class when the workload runs through that API.

Looking at the broader database, the Vega 8 sits at the 45th percentile of all GPUs, while the GTX 1660 Ti is at the 41st percentile. These percentile figures are close, but they do not tell the whole story. The average benchmark score for the Vega 8 is 9,221, while the GTX 1660 Ti averages 7,723. This is an uncommon result: the lower-percentile card has the higher average score, which indicates the GTX 1660 Ti’s benchmark suite includes more demanding tests that pull its average down, even though its peak results in OpenCL are far higher.

The GTX 1660 Ti’s PassMark results show its strengths in legacy APIs. It scores 190 in DirectX 9, 103 in DirectX 11, 66 in DirectX 10, and 51 in DirectX 12. The G3D score of 12,871 is the highest among its PassMark numbers, while its compute score is 5,226 and its G2D is 804. These figures indicate that the NVIDIA part excels at traditional rasterization and compute tasks, even if the newer API performance is not as dominant as the OpenCL result would suggest.

Architecture Differences

The two GPUs come from fundamentally different design philosophies. The AMD Radeon Vega 8 uses the GCN 5.0 architecture on a chip codenamed Raven, built on a 14 nm process at GlobalFoundries. The die size is 210 mm² with 4,940 million transistors, giving a transistor density of 23.5 million per square millimeter. The NVIDIA GeForce GTX 1660 Ti uses the Turing architecture on the TU116 chip, manufactured by TSMC on a 12 nm node. Its die is larger at 284 mm² with 6,600 million transistors, but the density is slightly lower at 23.2 million per square millimeter.

The Vega 8 is an integrated graphics processor (IGP), sharing its memory with the system. It has 512 shading units, 32 texture mapping units, and 8 raster output pipelines. The GTX 1660 Ti, by contrast, is a discrete dual-slot card with 1,536 shading units, 96 TMUs, and 48 ROPs. Those figures are all three times higher on the NVIDIA part, which explains its higher fill rates: 84.96 GPixel/s pixel rate and 169.9 GTexel/s texture rate versus the Vega 8’s 8.8 GPixel/s and 35.2 GTexel/s.

Neither GPU has any ray tracing cores or tensor cores, so both rely on traditional shader-based rendering. The FP32 throughput tells the story of raw compute: the GTX 1660 Ti delivers 5.437 TFLOPS, while the Vega 8 delivers 1,126.4 GFLOPS. The FP16 figures are 10.87 TFLOPS for the NVIDIA part and 2.253 TFLOPS for the AMD part. Both have a 2:1 ratio for FP16, meaning they halve the work for half precision.

The memory subsystems are completely different. The Vega 8 uses system-shared memory with a bus width and bandwidth that are system dependent. The GTX 1660 Ti has 6 GB of GDDR6 on a 192-bit bus, providing 288.0 GB/s of bandwidth. The NVIDIA part also has a 1,775 MHz boost clock and a 1,500 MHz base clock, while the Vega 8 runs at 300 MHz base and 1,100 MHz boost. The memory clock on the GTX 1660 Ti is 1,500 MHz with 12 Gbps effective transfer rate.

Head-to-Head Benchmarks

The two recorded head-to-head tests are the only direct comparisons in the database. Starting with the Vulkan test, the AMD Radeon Vega 8 scores 8,134, which is 39.6% higher than the GTX 1660 Ti’s 5,827. That delta is significant and places the Vega 8 clearly ahead in Vulkan compute. The GTX 1660 Ti, despite being a discrete card with far more shaders, cannot match the AMD part in this specific API test.

The OpenCL test flips the result entirely. The GTX 1660 Ti scores 51,029 against the Vega 8’s 8,822, a delta of 82.7% in NVIDIA’s favor. This is a massive margin, and it is the kind of difference that separates a dedicated accelerator from an integrated solution. The Vega 8’s OpenCL score is roughly 17% of the GTX 1660 Ti’s, which suggests that the OpenCL workload scales heavily with raw shader count and memory bandwidth, both of which favor the NVIDIA part.

The dynamic between these two tests is instructive. The Vulkan result shows that API efficiency and driver maturity can make a modest IGP competitive, while the OpenCL result shows raw compute throughput dominating. In practice, users would see a very different experience depending on which API the software uses.

The GTX 1660 Ti also has a set of PassMark scores that do not have a direct counterpart on the Vega 8. These include a G3D score of 12,871, a compute score of 5,226, a 2D score of 804, and API-specific results: 190 for DirectX 9, 103 for DirectX 11, 66 for DirectX 10, and 51 for DirectX 12. The DirectX 12 score is notably low relative to the others, which may reflect the specific workload in that test.

Specification Differences

The two parts differ in almost every measurable specification. The process node is 14 nm for AMD versus 12 nm for NVIDIA. The foundry is GlobalFoundries for the Vega 8, while TSMC makes the GTX 1660 Ti. Transistor count is 4,940 million versus 6,600 million, which leads to the die size difference of 210 mm² versus 284 mm².

The memory is a major split: the Vega 8 has system shared memory, while the GTX 1660 Ti has 6 GB of GDDR6 on a 192-bit bus with 288.0 GB/s bandwidth. The base clock is 300 MHz versus 1,500 MHz, and the boost clock is 1,100 MHz versus 1,770 MHz. The memory clock is system dependent on the AMD part, while the NVIDIA has a fixed 1,500 MHz with 12 Gbps effective.

The shading units are 512 versus 1,536, TMUs are 32 versus 96, and ROPs are 8 versus 48. The pixel rate is 8.8 GPixel/s versus 84.96 GPixel/s, and the texture rate is 35.2 GTexel/s versus 169.9 GTexel/s. FP32 compute is 1,126.4 GFLOPS versus 5.437 TFLOPS, and FP16 is 2.253 TFLOPS versus 10.87 TFLOPS.

The TDP is 25 W for the Vega 8 and 120 W for the GTX 1660 Ti, which is a 95 W difference. The Vega 8 is an IGP with no power connectors, while the GTX 1660 Ti is a dual-slot card requiring a 1x 8-pin connector and a recommended 300 W PSU. The bus interface is IGP versus PCIe 3.0 x16. Display outputs are motherboard dependent on the Vega 8, while the GTX 1660 Ti has 1x DVI, 1x HDMI 2.0, and 1x DisplayPort 1.4a.

The physical dimensions differ as well: the GTX 1660 Ti is 229 mm long, 111 mm tall, and 35 mm wide, while the Vega 8 has no physical dimensions as it is an integrated part. The release dates are about a year apart, with the Vega 8 released in February 2018 and the GTX 1660 Ti in February 2019. The GTX 1660 Ti has a launch MSRP of 279 USD, but the Vega 8 has no such figure because it is an IGP sold as part of a processor.

Both support DirectX 12 (12_1) and OpenGL 4.6, but the Vulkan version differs: the Vega 8 supports 1.3, while the GTX 1660 Ti supports 1.4.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon Vega 8 has an average benchmark score of 9,221, while the NVIDIA GeForce GTX 1660 Ti has an average of 7,723.

Q: In which API does the AMD Radeon Vega 8 beat the GTX 1660 Ti?

A: The Vega 8 wins the Vulkan test, scoring 8,134 against the GTX 1660 Ti’s 5,827, a 39.6% difference.

Q: What is the largest benchmark margin between the two?

A: The OpenCL test shows the largest margin, with the GTX 1660 Ti scoring 51,029 versus the Vega 8’s 8,822, an 82.7% difference in NVIDIA’s favor.

Q: How do the power requirements compare?

A: The Vega 8 has a TDP of 25 W and uses no power connectors, while the GTX 1660 Ti has a TDP of 120 W and requires one 8-pin power connector with a recommended 300 W power supply.

Q: What is the memory configuration difference?

A: The Vega 8 uses system shared memory, while the GTX 1660 Ti has 6 GB of GDDR6 on a 192-bit bus with 288.0 GB/s bandwidth.

Q: Which GPU supports a newer Vulkan version?

A: The GTX 1660 Ti supports Vulkan 1.4, while the Vega 8 supports Vulkan 1.3.