AMD Radeon 740M vs NVIDIA GeForce GTX 660 Ti Comparison

Head-to-Head Benchmarks

The recorded data splits this comparison cleanly down the middle. Across the two benchmark tests in the database, each GPU takes one win. The margin of victory, however, is far from symmetrical.

In the Geekbench OpenCL test, the NVIDIA GeForce GTX 660 Ti scores 15113 against the AMD Radeon 740M's 10172. That is a 48.6% advantage for the older NVIDIA card, a decisive gap in compute workloads that lean on OpenCL. The GTX 660 Ti's raw shading unit count and memory bandwidth appear to carry it here, despite its age.

The Vulkan test flips the result, but by a much narrower margin. The AMD Radeon 740M scores 15568, while the GTX 660 Ti trails at 15012. That is only a 3.6% difference in favor of the AMD part. For a modern API like Vulkan, the newer architecture and driver support of the Radeon 740M give it a slight edge, but it is nowhere near the scale of the OpenCL loss it suffers.

Looking at the broader database context, the GTX 660 Ti's average benchmark score is 15063, placing it at the 57th percentile of all GPUs. Its nearest rival in the database is the AMD Radeon Pro 560X at 15082, a delta of just -0.1%. It also sits close to the AMD Radeon RX 7600 (15171, -0.7%) and the NVIDIA GeForce RTX 3050 OEM (15199, -0.9%). The AMD Radeon 740M's average score is 12870, at the 53rd percentile. Its closest rival is the AMD FirePro W5100 at 12847, a delta of 0.2%, with the NVIDIA GeForce GTX 590 (12830, 0.3%) and AMD Radeon Pro 455 (12831, 0.3%) also nearby.

So while the two cards trade wins head-to-head, the GTX 660 Ti sits in a higher performance tier overall in the database's aggregate scoring. The 740M's Vulkan win is real but modest, while the GTX 660 Ti's OpenCL win is dominant.

Architecture Differences

These two GPUs come from completely different eras and design philosophies. The NVIDIA GeForce GTX 660 Ti uses the GK104 chip, built on Kepler architecture and manufactured by TSMC on a 28 nm process. The die measures 294 mm² and packs 3,540 million transistors, for a density of 12.0M per mm². It belongs to the GeForce 600 generation and was released on August 15, 2012, with a launch MSRP of 299 USD.

The AMD Radeon 740M is a modern integrated graphics processor based on the Phoenix2 chip, using RDNA 3.0 architecture. It is also made by TSMC, but on a 4 nm process. The die is much smaller at 137 mm², yet it contains 20,900 million transistors, giving a density of 152.6M per mm². It is part of the Navi III IGP (Phoenix) generation and released on January 30, 2024, nearly twelve years after the GTX 660 Ti.

The transistor density difference is enormous: the RDNA 3.0 part packs over twelve times more transistors per square millimeter. The GTX 660 Ti has 1344 shading units, 112 texture mapping units, and 24 ROPs. The Radeon 740M has 256 shading units, 16 TMUs, and 8 ROPs. Despite having far fewer of these classic compute units, the 740M also includes 4 ray tracing cores, a feature the Kepler card lacks entirely.

Clock speeds tell a similar story. The GTX 660 Ti runs at a base of 915 MHz with a boost of 980 MHz. The Radeon 740M has a much lower base clock of 800 MHz but boosts to 2800 MHz. The memory situation is entirely different: the GTX 660 Ti has 2 GB of dedicated GDDR5 on a 192-bit bus, delivering 144.2 GB/s of bandwidth. The 740M uses System Shared memory, with its bandwidth listed as System Dependent.

The FP32 throughput favors the newer card slightly: 2.634 TFLOPS for the GTX 660 Ti versus 2.867 TFLOPS for the Radeon 740M. The 740M also lists FP16 at 2.867 TFLOPS (1:1), while the GTX 660 Ti has no FP16 data recorded. Pixel rate favors NVIDIA at 27.44 GPixel/s versus 22.40 GPixel/s, but texture rate is a clear NVIDIA win at 109.8 GTexel/s versus 44.80 GTexel/s.

Power and physical requirements are night and day. The GTX 660 Ti is a dual-slot card with a 150 W TDP, requires two 6-pin power connectors, and a 450 W suggested PSU. It is 241 mm long (9.5 inches). The Radeon 740M is an IGP with a 45 W TDP, no power connectors, and no suggested PSU, with dimensions listed as motherboard dependent. The GTX 660 Ti uses PCIe 3.0 x16; the 740M uses PCIe 4.0 x8.

API support differs as well. The GTX 660 Ti supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The Radeon 740M supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The 740M's DirectX 12 Ultimate feature set is a significant modern advantage.

FAQ

Q: Which GPU is faster in OpenCL workloads?

A: The NVIDIA GeForce GTX 660 Ti wins decisively, scoring 15113 versus 10172 for the AMD Radeon 740M, a 48.6% advantage.

Q: Which GPU wins in Vulkan benchmarks?

A: The AMD Radeon 740M takes the Vulkan test at 15568 against 15012 for the GTX 660 Ti, a 3.6% margin in favor of AMD.

Q: How do their average benchmark scores compare?

A: The GTX 660 Ti averages 15063, placing it at the 57th percentile of all GPUs. The Radeon 740M averages 12870, at the 53rd percentile.

Q: What are the TDP requirements for each card?

A: The GTX 660 Ti has a 150 W TDP, requires two 6-pin power connectors, and a 450 W suggested PSU. The Radeon 740M has a 45 W TDP and requires no power connectors.

Q: Does the Radeon 740M support ray tracing?

A: Yes, the Radeon 740M includes 4 ray tracing cores. The GTX 660 Ti has no ray tracing cores at all.

Q: What memory configurations do the two GPUs use?

A: The GTX 660 Ti uses 2 GB of dedicated GDDR5 on a 192-bit bus with 144.2 GB/s bandwidth. The Radeon 740M uses System Shared memory with system dependent bandwidth.

The Verdict

The data points to a clear split based on workload type and use case. The NVIDIA GeForce GTX 660 Ti holds a significant overall performance edge in the database's aggregate scoring (15063 versus 12870). It is the stronger choice for compute tasks that leverage OpenCL, where its 48.6% lead is substantial. Its dedicated memory subsystem with 144.2 GB/s of bandwidth and 192-bit bus also gives it predictable performance independent of system RAM.

The AMD Radeon 740M is the more modern part in almost every architectural dimension: 4 nm process, RDNA 3.0, DirectX 12 Ultimate, Vulkan 1.4, and ray tracing support. Its Vulkan score is higher, and its FP32 throughput (2.867 TFLOPS) edges out the GTX 660 Ti (2.634 TFLOPS). It also consumes only 45 W versus 150 W, requires no external power, and fits into any system with a compatible motherboard as an IGP.

For a desktop user with an existing PCIe 3.0 x16 slot and power supply headroom, the GTX 660 Ti is the faster card overall, especially in compute-heavy OpenCL tasks. For a modern system builder looking for an IGP that can handle Vulkan APIs and newer DirectX features with minimal power draw, the Radeon 740M is the sensible choice. The GTX 660 Ti is end-of-life; the Radeon 740M is active production. Neither card is a universal winner, but the GTX 660 Ti has the better aggregate benchmark record.

Specification Differences

| Specification | NVIDIA GeForce GTX 660 Ti | AMD Radeon 740M |

|---|---|---|

| Architecture | Kepler | RDNA 3.0 |

| Process Node | 28 nm | 4 nm |

| Foundry | TSMC | TSMC |

| Transistors | 3,540 million | 20,900 million |

| Die Size | 294 mm² | 137 mm² |

| Transistor Density | 12.0M / mm² | 152.6M / mm² |

| Base Clock | 915 MHz | 800 MHz |

| Boost Clock | 980 MHz | 2800 MHz |

| Memory | 2 GB GDDR5 | System Shared |

| Memory Bus | 192 bit | System Shared |

| Memory Bandwidth | 144.2 GB/s | System Dependent |

| Shading Units | 1344 | 256 |

| TMUs | 112 | 16 |

| ROPs | 24 | 8 |

| Ray Tracing Cores | None | 4 |

| Pixel Rate | 27.44 GPixel/s | 22.40 GPixel/s |

| Texture Rate | 109.8 GTexel/s | 44.80 GTexel/s |

| FP32 | 2.634 TFLOPS | 2.867 TFLOPS |

| FP16 | Not listed | 2.867 TFLOPS (1:1) |

| TDP | 150 W | 45 W |

| Slot Width | Dual-slot | IGP |

| Power Connectors | 2x 6-pin | None |

| Suggested PSU | 450 W | Not listed |

| Bus Interface | PCIe 3.0 x16 | PCIe 4.0 x8 |

| DirectX | 12 (11_0) | 12 Ultimate (12_2) |

| Vulkan | 1.2.175 | 1.4 |

| Production Status | End-of-life | Active |

| Release Date | 2012-08-15 | 2024-01-30 |

Where Each One Wins

NVIDIA GeForce GTX 660 Ti:

• OpenCL compute: 48.6% ahead of the Radeon 740M.
• Aggregate benchmark score: 15063 versus 12870, a roughly 17% overall lead.
• Texture throughput: 109.8 GTexel/s versus 44.80 GTexel/s, more than double.
• Pixel fill rate: 27.44 GPixel/s versus 22.40 GPixel/s.
• Dedicated memory bandwidth: 144.2 GB/s versus system dependent, ensuring consistent performance.
• Higher percentile placement in the database: 57th versus 53rd.

AMD Radeon 740M:

• Vulkan performance: 15568 versus 15012, a 3.6% win.
• FP32 compute: 2.867 TFLOPS versus 2.634 TFLOPS.
• FP16 support: 2.867 TFLOPS (1:1), not available on the GTX 660 Ti.
• Ray tracing cores: 4, versus none on the GTX 660 Ti.
• Power efficiency: 45 W TDP versus 150 W, with no power connectors required.
• Modern API support: DirectX 12 Ultimate (12_2) and Vulkan 1.4.
• Process technology: 4 nm versus 28 nm, with much higher transistor density (152.6M / mm² versus 12.0M / mm²).
• Active production status versus end-of-life for the GTX 660 Ti.