AMD Radeon RX 5500M vs NVIDIA GeForce GTX 660 Ti Comparison

FAQ

Q: Which GPU is older, the NVIDIA GeForce GTX 660 Ti or the AMD Radeon RX 5500M?

A: The NVIDIA GeForce GTX 660 Ti was released on 2012-08-15, while the AMD Radeon RX 5500M launched on 2019-10-06. The GTX 660 Ti is from the GeForce 600 generation, whereas the RX 5500M belongs to the Navi Mobile (RX 5000M) generation.

Q: How do the two cards compare in average benchmark scores?

A: The GTX 660 Ti has an average benchmark score of 15063, placing it in the 57th percentile of all GPUs. The RX 5500M has a lower average score of 13356, putting it in the 54th percentile. Despite the RX 5500M winning the direct head-to-head tests, its overall average is dragged down by additional benchmark results.

Q: Which GPU wins in the OpenCL benchmark?

A: The AMD Radeon RX 5500M wins decisively, scoring 38725 compared to the GTX 660 Ti's 15113. This represents a 61% margin in favor of the AMD card.

Q: What are the memory specifications of each card?

A: The GTX 660 Ti has 2 GB of GDDR5 memory on a 192-bit bus, delivering 144.2 GB/s bandwidth. The RX 5500M has 4 GB of GDDR6 memory on a 128-bit bus, achieving 224.0 GB/s bandwidth.

Q: What is the process node difference between the two?

A: The GTX 660 Ti uses a 28 nm process from TSMC, while the RX 5500M uses a 7 nm process from the same foundry. The RX 5500M's node is significantly smaller, allowing for higher transistor density.

Q: Which card has a higher pixel rate?

A: The RX 5500M has a pixel rate of 52.64 GPixel/s, nearly double the GTX 660 Ti's 27.44 GPixel/s. This is a direct consequence of the AMD card's higher boost clock and larger ROP count.

Architecture Differences

The NVIDIA GeForce GTX 660 Ti is built on the Kepler architecture using the GK104 chip, fabricated on TSMC's 28 nm process. It packs 3,540 million transistors into a 294 mm² die, resulting in a transistor density of 12.0M per mm². The architecture supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175.

The AMD Radeon RX 5500M uses the RDNA 1.0 architecture with the Navi 14 chip, built on TSMC's 7 nm node. This newer process allows 6,400 million transistors in a much smaller 158 mm² die, achieving a transistor density of 40.5M per mm². The RDNA architecture supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4, giving it a more modern API feature set.

The shading unit counts are close: the GTX 660 Ti has 1344 shading units while the RX 5500M has 1408. However, the texture and ROP configurations differ notably. The NVIDIA card has 112 texture mapping units and 24 ROPs, while the AMD card has 88 TMUs and 32 ROPs. The RX 5500M's higher ROP count contributes to its superior pixel throughput.

Clock speeds are substantially different. The GTX 660 Ti operates at a 915 MHz base and 980 MHz boost, while the RX 5500M runs at 1375 MHz base, 1645 MHz boost, with a game clock of 1448 MHz. The AMD card's higher clocks, combined with its RDNA architecture, deliver significantly better raw compute performance.

The RX 5500M also supports FP16 computation at 9.265 TFLOPS (2:1 ratio), a feature absent from the GTX 660 Ti, which only lists FP32 at 2.634 TFLOPS. The AMD card's FP32 throughput of 4.632 TFLOPS is nearly double that of the NVIDIA card.

Where Each One Wins

The AMD Radeon RX 5500M wins in every direct benchmark comparison against the GTX 660 Ti. In OpenCL, it achieves 38725 versus 15113, and in Vulkan it scores 35693 versus 15012. These results indicate the RX 5500M is the clear performance leader in modern compute and graphics workloads.

The RX 5500M also excels in scenarios requiring high memory bandwidth. Its 224.0 GB/s bandwidth, enabled by GDDR6 memory, is substantially higher than the GTX 660 Ti's 144.2 GB/s. This advantages the AMD card in bandwidth-sensitive applications like high-resolution texturing and compute tasks.

However, the GTX 660 Ti holds an advantage in average benchmark score across all recorded tests. Its 15063 average is higher than the RX 5500M's 13356, despite losing the head-to-head tests. This suggests the NVIDIA card performs more consistently across a broader range of workloads, including legacy DirectX titles where its Kepler architecture remains competitive.

The GTX 660 Ti also benefits from its desktop form factor. It features a dual-slot design with 2x 6-pin power connectors and a suggested PSU of 450 W, making it suitable for traditional desktop builds. The RX 5500M is a mobile part with no discrete power connectors, designed for laptop integration.

Specification Differences

The most striking difference is the process node: 28 nm for the GTX 660 Ti versus 7 nm for the RX 5500M. This leads to dramatically different transistor densities, with the AMD card packing 40.5M transistors per mm² compared to 12.0M for the NVIDIA card.

Memory configurations differ in every aspect. The GTX 660 Ti uses 2 GB GDDR5 on a 192-bit bus at 1502 MHz (6 Gbps effective), while the RX 5500M uses 4 GB GDDR6 on a 128-bit bus at 1750 MHz (14 Gbps effective). The RX 5500M's bandwidth advantage is clear at 224.0 GB/s versus 144.2 GB/s.

Power requirements are vastly different. The GTX 660 Ti has a TDP of 150 W and requires 2x 6-pin power connectors, while the RX 5500M has an 85 W TDP with no power connectors, reflecting its mobile nature. The NVIDIA card is dual-slot, while the RX 5500M's slot width is not specified.

The bus interface also differs: the GTX 660 Ti uses PCIe 3.0 x16, while the RX 5500M uses PCIe 4.0 x8. The newer PCIe 4.0 standard offers higher bandwidth per lane, partially compensating for the reduced lane count.

Display outputs reflect their intended markets. The GTX 660 Ti provides 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2, while the RX 5500M's outputs are listed as "Portable Device Dependent," meaning they vary by laptop implementation.

The GTX 660 Ti has a launch MSRP of 299 USD. The RX 5500M has no recorded launch MSRP.

Head-to-Head Benchmarks

The two recorded head-to-head tests show a dominant performance from the AMD Radeon RX 5500M. In the Geekbench OpenCL test, the RX 5500M scores 38725 against the GTX 660 Ti's 15113, a delta of 61% in favor of AMD. This massive margin indicates the RX 5500M's RDNA architecture is far more efficient at compute workloads.

In the Geekbench Vulkan test, the RX 5500M again wins with 35693 versus 15012, a 57.9% advantage. The consistent performance gap across both API tests suggests the RX 5500M's architectural advantages, including its higher clock speeds, larger memory bandwidth, and more modern instruction set, translate directly into benchmark superiority.

However, the broader benchmark picture is more nuanced. The GTX 660 Ti's average score of 15063 is higher than the RX 5500M's 13356, despite losing both direct comparisons. This discrepancy arises because the RX 5500M has additional benchmark results, including PassMark tests where it scores 5848 in G3D and 2316 in GPU compute, alongside several low DirectX scores (39 in DirectX 10, 35 in DirectX 11, 28 in DirectX 12, 100 in DirectX 9). These additional data points lower its average, while the GTX 660 Ti's average is based solely on its two strong Geekbench scores.

The nearest rivals for each card provide context. The GTX 660 Ti sits near the AMD Radeon Pro 560X (15082, delta -0.1%), AMD Radeon RX 7600 (15171, delta -0.7%), NVIDIA GeForce RTX 3050 OEM (15199, delta -0.9%), and AMD Radeon 680M (15270, delta -1.4%). The RX 5500M's nearest rivals are the AMD FirePro M6100 (13354, delta 0%), AMD Radeon HD 8950M (13376, delta -0.1%), AMD Radeon Pro 555X (13321, delta 0.3%), and AMD Radeon Pro 555 (13407, delta -0.4%).

The Verdict

Based on the recorded data, the AMD Radeon RX 5500M is the clear performance winner in direct comparisons. Its 61% advantage in OpenCL and 57.9% advantage in Vulkan demonstrate that RDNA 1.0 with 4 GB GDDR6 memory and a 7 nm process provides substantially more compute and graphics capability than the aging Kepler architecture.

The GTX 660 Ti, despite being seven years older, maintains a higher average benchmark score of 15063 versus 13356. This suggests it remains competitive in legacy workloads and DirectX 9/10/11 scenarios, where its 1344 shading units and 192-bit memory bus still deliver respectable performance.

For users prioritizing modern API performance, compute workloads, and memory bandwidth, the RX 5500M is the superior choice. Its 224.0 GB/s bandwidth, 4 GB VRAM, and 4.632 TFLOPS FP32 throughput provide a substantial foundation for contemporary gaming and productivity tasks.

For those working with older software or seeking consistent performance across a wide range of benchmarks, the GTX 660 Ti's higher average score and desktop-oriented design (dual-slot, 2x 6-pin, 450 W PSU requirement) make it a viable option in legacy systems. Its launch MSRP of 299 USD reflects its original mid-range positioning.

The data ultimately shows that the RX 5500M wins where it matters most for modern use: raw compute performance, memory efficiency, and power efficiency (85 W TDP versus 150 W). The GTX 660 Ti's advantages are confined to average score and desktop form factor flexibility. For any modern workload, the RX 5500M is the recommended choice.