AMD Radeon RX 6600S vs NVIDIA GeForce GTX 660 Comparison

FAQ

Q: How does the AMD Radeon RX 6600S compare to the NVIDIA GeForce GTX 660 in raw compute performance?

A: The AMD Radeon RX 6600S delivers 7.168 TFLOPS of FP32 compute, while the NVIDIA GeForce GTX 660 delivers 1.981 TFLOPS. In the Geekbench OpenCL test, the RX 6600S scores 66,435 against 11,347 for the GTX 660, a delta of 485.5%.

Q: Which GPU has a higher transistor density?

A: The AMD Radeon RX 6600S is built on a 7 nm process at TSMC and packs 11,060 million transistors into a 237 mm² die, yielding a density of 46.7M transistors per mm². The NVIDIA GeForce GTX 660 uses a 28 nm process with 2,540 million transistors on a 221 mm² die, for a density of 11.5M per mm².

Q: What memory configurations do these two cards ship with?

A: The AMD Radeon RX 6600S has 4 GB of GDDR6 memory on a 128-bit bus with 224.0 GB/s of bandwidth. The NVIDIA GeForce GTX 660 has 2 GB of GDDR5 memory on a 192-bit bus with 144.2 GB/s of bandwidth.

Q: Does the AMD Radeon RX 6600S support hardware ray tracing?

A: Yes, the AMD Radeon RX 6600S includes 28 ray tracing cores as part of its RDNA 2.0 architecture. The NVIDIA GeForce GTX 660, based on Kepler, has no ray tracing cores listed.

Q: What are the power requirements for each card?

A: The AMD Radeon RX 6600S has a TDP of 80 W and uses no external power connectors, since it is an IGP (integrated graphics processor) solution. The NVIDIA GeForce GTX 660 has a 140 W TDP, requires a single 6-pin power connector, and carries a suggested PSU rating of 300 W.

Q: Which API levels does each card support?

A: The AMD Radeon RX 6600S supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The NVIDIA GeForce GTX 660 supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175.

Architecture Differences

The AMD Radeon RX 6600S and NVIDIA GeForce GTX 660 come from different architectural eras and process generations. The RX 6600S is built on RDNA 2.0, using the Navi 23 chip, and is manufactured on TSMC's 7 nm process. The GTX 660 uses the Kepler architecture with the GK106 chip, fabricated on a 28 nm process. The process gap is substantial: the RX 6600S packs 11,060 million transistors into a 237 mm² die for a density of 46.7M per mm², while the GTX 660 fits 2,540 million transistors into a 221 mm² die at 11.5M per mm². That is a fourfold difference in transistor density, which directly explains the enormous performance gap.

The execution resources tell a similar story. The RX 6600S has 1,792 shading units, 112 texture mapping units, and 64 ROPs. The GTX 660 has 960 shading units, 80 TMUs, and only 24 ROPs. Beyond the raw counts, the RX 6600S adds 28 ray tracing cores, a feature the Kepler-based GTX 660 lacks entirely. The RX 6600S also supports DirectX 12 Ultimate (12_2) and Vulkan 1.4, whereas the GTX 660 is limited to DirectX 12 (11_0) and Vulkan 1.2.175.

Clock speeds also differ significantly. The RX 6600S has a base clock of 1700 MHz, a game clock of 1881 MHz, and a boost clock of 2000 MHz. The GTX 660 runs at 980 MHz base and 1032 MHz boost. Memory technology differs as well: the RX 6600S uses GDDR6 at 14 Gbps effective, while the GTX 660 uses GDDR5 at 6 Gbps effective. The RX 6600S has a narrower 128-bit memory bus but achieves 224.0 GB/s of bandwidth, versus 144.2 GB/s for the GTX 660's wider 192-bit bus.

Form factor and power delivery also separate the two. The RX 6600S is an IGP with no slot width, no power connectors, and an 80 W TDP. The GTX 660 is a dual-slot card, 241 mm long, with a 1x 6-pin power connector and a 140 W TDP. The RX 6600S connects over PCIe 4.0 x8, while the GTX 660 uses PCIe 3.0 x16. The RX 6600S relies on portable-device-dependent display outputs, whereas the GTX 660 provides 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2.

Where Each One Wins

The AMD Radeon RX 6600S wins decisively in every measured benchmark category. Its Geekbench OpenCL score of 66,435 dwarfs the GTX 660's 11,347. The RX 6600S also has the advantage in modern feature support: DirectX 12 Ultimate, Vulkan 1.4, and hardware ray tracing make it the more future-proof option for games and compute workloads that leverage these APIs. Its higher pixel rate of 128.0 GPixel/s and texture rate of 224.0 GTexel/s, compared to 20.64 GPixel/s and 82.56 GTexel/s for the GTX 660, mean it handles resolution-heavy and texture-heavy rendering far more comfortably.

The NVIDIA GeForce GTX 660 is the older part, and its strengths are confined to legacy scenarios. It has more conventional desktop connectivity with fixed display outputs, a dual-slot form factor, and a 300 W suggested PSU, which makes it easier to slot into an older desktop system. It also supports Geekbench Metal and Vulkan tests, scoring 4,305 in Metal and 11,415 in Vulkan. Those are the only benchmark entries recorded for the GTX 660 that have no RX 6600S equivalent in the database. For users running older DirectX 9 or DirectX 10 titles, the GTX 660's PassMark DirectX 9 score of 176 and DirectX 10 score of 81 are its best showings, though no comparable RX 6600S scores for those tests are recorded in the database.

In practical terms, the RX 6600S is the clear winner for modern gaming and general compute. The GTX 660 retains relevance only in very old software ecosystems or as a low-cost desktop filler, but the recorded data shows it cannot compete with the RX 6600S on any shared metric.

Specification Differences

The two cards differ in nearly every major specification field. The RX 6600S uses a 7 nm process, the GTX 660 uses 28 nm. Transistor counts are 11,060 million versus 2,540 million. Die size is similar, 237 mm² versus 221 mm², but transistor density is 46.7M per mm² versus 11.5M per mm².

Clock behavior differs: the RX 6600S has a 1700 MHz base and 2000 MHz boost, with a game clock of 1881 MHz. The GTX 660 has a 980 MHz base and 1032 MHz boost, with no game clock listed. Memory configurations are different in size, type, bus width, and bandwidth: 4 GB GDDR6 on 128-bit at 224.0 GB/s versus 2 GB GDDR5 on 192-bit at 144.2 GB/s. Memory clocks are 1750 MHz (14 Gbps effective) versus 1502 MHz (6 Gbps effective).

Compute resources differ: 1,792 shading units, 112 TMUs, 64 ROPs, and 28 ray tracing cores for the RX 6600S, versus 960 shading units, 80 TMUs, and 24 ROPs for the GTX 660. The RX 6600S has no tensor cores; neither does the GTX 660. Pixel rate is 128.0 GPixel/s versus 20.64 GPixel/s. Texture rate is 224.0 GTexel/s versus 82.56 GTexel/s. FP32 throughput is 7.168 TFLOPS versus 1.981 TFLOPS. The RX 6600S also lists FP16 at 14.34 TFLOPS (2:1), while the GTX 660 has no FP16 figure.

Power and physical specs diverge sharply: 80 W TDP versus 140 W, IGP versus dual-slot, no power connectors versus 1x 6-pin, no suggested PSU versus 300 W. Bus interface is PCIe 4.0 x8 versus PCIe 3.0 x16. Display outputs are portable-device-dependent versus 2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2. The GTX 660 is 241 mm (9.5 inches) long; the RX 6600S has no listed dimensions. API support differs: DirectX 12 Ultimate (12_2) versus DirectX 12 (11_0), Vulkan 1.4 versus Vulkan 1.2.175, and OpenGL 4.6 on both. Launch dates are far apart: 2022-01-03 for the RX 6600S and 2012-09-05 for the GTX 660. The GTX 660 has a launch MSRP of 229 USD; the RX 6600S has no launch MSRP recorded. Both are end-of-life products. The RX 6600S succeeded Polaris Mobile, while the GTX 660 succeeded GeForce 500 and was succeeded by GeForce 700.

Head-to-Head Benchmarks

The database records one shared benchmark between the AMD Radeon RX 6600S and the NVIDIA GeForce GTX 660: Geekbench OpenCL. The RX 6600S scores 66,435, and the GTX 660 scores 11,347. That is a 485.5% advantage for the RX 6600S, which is a massive single-test margin. The RX 6600S wins the head-to-head category with 1 win and 0 losses.

Interpreting that score: the RX 6600S is roughly 5.85 times faster in OpenCL compute. This aligns with the theoretical FP32 figures, where 7.168 TFLOPS is about 3.6 times the GTX 660's 1.981 TFLOPS, though the real-world OpenCL gap is even larger due to architectural efficiency and memory bandwidth advantages. The RX 6600S also has more than double the memory bandwidth and nearly three times the ROP count, which helps in memory-bound compute kernels.

The average benchmark scores tell a similar story. The RX 6600S has an average benchmark score of 10,629, and the GTX 660 has an average of 9,022. The RX 6600S sits at the 49th percentile among all GPUs in the database, while the GTX 660 sits at the 45th percentile. The RX 6600S's nearest rivals by average score are the AMD Radeon R9 M275X at 10,582 (0.4% behind), the NVIDIA GeForce GTX 560 Ti at 10,690 (0.6% ahead), the NVIDIA Quadro K2200 at 10,761 (1.2% ahead), and the AMD Radeon RX 550X at 10,481 (1.4% behind). The GTX 660's nearest rivals are the NVIDIA TITAN V CEO Edition at 9,037 (0.2% ahead), the NVIDIA GeForce GTX 560 at 9,058 (0.4% ahead), the AMD Radeon 550X at 8,918 (1.2% behind), and the AMD Radeon Pro WX 5100 at 8,863 (1.8% behind).

Notably, the RX 6600S's average score is dominated by its PassMark G3D result of 12,649 and its Geekbench OpenCL result of 66,435, while its PassMark GPU compute score is 4,879. The GTX 660's recorded benchmarks are limited to Geekbench Metal at 4,305, Geekbench OpenCL at 11,347, and Geekbench Vulkan at 11,415. The GTX 660 has no PassMark entries in the database, so its average score of 9,022 is derived solely from those three Geekbench results.

Given the massive OpenCL delta and the architectural advantages in every measured category, the RX 6600S is the clear performance winner. The GTX 660 cannot match it in any shared workload, and the only benchmarks where the GTX 660 appears without a rival score from the RX 6600S are Metal and Vulkan, which reflect its legacy API support rather than any competitive advantage.