NVIDIA GeForce GT 645M vs NVIDIA GeForce GTX 460M Comparison

# NVIDIA GeForce GT 645M vs NVIDIA GeForce GTX 460M

The GeForce GTX 460M is the clear winner in the only directly comparable benchmark, delivering a 59.8% higher OpenCL score (4282 vs 2680) despite being an older architecture. However, the GeForce GT 645M counters with superior API support, a more modern feature set, and significantly lower power draw, making it the better choice for users prioritizing efficiency and forward-looking software compatibility over raw compute performance.

Where Each One Wins

The GTX 460M wins outright in raw compute performance. In the sole head-to-head benchmark available, Geekbench OpenCL, it scores 4282 against the GT 645M's 2680, a decisive 37.4% advantage. This is the only benchmark where both GPUs were tested under identical conditions, and the result is unambiguous — the Fermi-based GTX 460M delivers substantially higher throughput in OpenCL workloads. Its 60.00 GB/s memory bandwidth, enabled by GDDR5 memory on a 192-bit bus, gives it a massive data throughput edge that directly contributes to this compute victory.

The GT 645M wins in every other measurable category that matters for modern usage. It has a broader benchmark portfolio, with scores in Geekbench Metal (5679), Geekbench Vulkan (4875), and Geekbench OpenCL (2680), demonstrating its ability to run contemporary graphics APIs. The GTX 460M only has an OpenCL score and no Vulkan support whatsoever, meaning it cannot run Vulkan-based applications at all. The GT 645M also wins decisively on efficiency, with a 32 W TDP versus 50 W for the GTX 460M — a 36% reduction in power consumption. For mobile platforms where thermal headroom and battery life are critical, this is a substantial advantage.

The GT 645M also edges ahead in average benchmark score: 4411 across its three tests versus 4282 for the GTX 460M's single test. While this comparison is imperfect because it averages different test sets, it suggests that the GT 645M's overall capability profile is competitive, even if its peak OpenCL performance lags. The percentile rankings reinforce this: the GT 645M sits at the 26th percentile of all GPUs, just ahead of the GTX 460M's 25th percentile, meaning the newer card is marginally better positioned relative to the broader GPU landscape.

Architecture Differences

The two GPUs represent fundamentally different design philosophies from NVIDIA. The GT 645M uses the GK107 chip built on Kepler architecture, fabricated on TSMC's 28 nm process. The GTX 460M uses the GF106 chip on the older Fermi architecture, manufactured on a 40 nm process. This process difference is stark: 28 nm versus 40 nm, which explains much of the efficiency gap. The GK107 packs 1,270 million transistors into a 118 mm² die, achieving a density of 10.8M transistors per mm². The GF106, despite having fewer transistors at 1,170 million, requires a much larger 238 mm² die due to the older process, yielding only 4.9M transistors per mm². The GT 645M is more than twice as dense, which directly translates to its lower power draw.

Shader configuration differs dramatically. The GT 645M has 384 shading units, 32 texture mapping units, and 16 raster operation units. The GTX 460M has only 192 shading units but 32 TMUs and 24 ROPs. The GT 645M's shader count is double, yet its FP32 throughput is only 599.0 GFLOPS versus 518.4 GFLOPS for the GTX 460M — a modest 15.5% difference. This indicates the Kepler architecture's shaders are more efficient per unit, but the GTX 460M's higher clock rates and different scheduling compensate. The GTX 460M's additional ROPs (24 vs 16) suggest it may handle pixel-heavy workloads better relative to its compute capability.

Memory subsystems are fundamentally different. The GT 645M uses 2 GB of DDR3 on a 128-bit bus, yielding 28.80 GB/s bandwidth. The GTX 460M uses 1536 MB of GDDR5 on a 192-bit bus, delivering 60.00 GB/s — more than double the bandwidth. The GTX 460M's memory clock is 625 MHz with an effective 2.5 Gbps data rate, while the GT 645M runs at 900 MHz with 1800 Mbps effective. The GTX 460M compensates for its lower clock with a wider bus and faster memory type. The GT 645M's larger capacity (2 GB vs 1536 MB) matters for texture-heavy modern games, but the bandwidth deficit is severe.

API support is where the GT 645M pulls decisively ahead. Both support DirectX 12 (11_0) and OpenGL 4.6, but the GT 645M adds Vulkan 1.2.175 support while the GTX 460M has no Vulkan support listed. This means the GTX 460M cannot run Vulkan-based games or applications, which is increasingly relevant for modern titles and emulators. The GT 645M also uses PCIe 3.0 x16, while the GTX 460M is limited to PCIe 2.0 x16 — a bandwidth advantage for the newer card, though less impactful for mobile platforms. The GT 645M is an IGP (integrated graphics processor) form factor, while the GTX 460M uses an MXM module, reflecting their different design targets.

Head-to-Head Benchmarks

The only direct comparison available is Geekbench OpenCL, and the results are lopsided. The GTX 460M scores 4282, while the GT 645M manages only 2680, a delta of -37.4% for the newer card. This is a substantial defeat for the Kepler part. The GTX 460M's advantage in this test likely stems from its memory configuration — 60.00 GB/s bandwidth versus 28.80 GB/s provides a 108% advantage in data throughput, which OpenCL workloads heavily depend on. The GTX 460M's 192-bit bus and GDDR5 memory simply feed data to its compute units far faster than the GT 645M's 128-bit DDR3 setup.

However, the GT 645M's other benchmark results paint a more nuanced picture. Its Geekbench Metal score of 5679 and Vulkan score of 4875 are both well above its OpenCL score, suggesting that the card is better optimized for modern graphics APIs than for raw compute. The Metal score is 112% higher than the OpenCL score, and Vulkan is 82% higher. This indicates that the GT 645M's hardware is capable, but its OpenCL drivers or architecture implementation may be suboptimal. The GTX 460M has no Metal or Vulkan scores, so no comparison is possible on those fronts, but the absence of Vulkan support alone is disqualifying for many modern workloads.

Looking at the GT 645M's nearest rivals, its average score of 4411 places it just 0.5% above the GeForce 930M (4388) and 1.2% above the Intel Iris Pro Graphics 5200 (4360). It sits 1.8% above the GeForce RTX 4070 GDDR6, an anomaly that reflects the RTX 4070's low score in the same benchmark database. The GT 645M trails the AMD Radeon R7 M260 by 1.9% (4499). For the GTX 460M, its 4282 average is 0.3% below the AMD FirePro W2100 (4295) and 0.3% above the AMD Radeon Vega 3 (4268), while sitting 1% above the Quadro K3000M (4241) and 1.2% below the GeForce RTX 4070 GDDR6 (4335). Both cards cluster tightly with their peers, indicating they are mid-pack performers in their respective eras.

The Verdict

The data supports a clear split decision. For users who prioritize raw compute performance in OpenCL-heavy workloads, the GTX 460M is the superior choice. Its 37.4% lead in the head-to-head OpenCL benchmark is decisive, and the 60.00 GB/s memory bandwidth is a genuine advantage that cannot be overcome by the GT 645M's architecture improvements. If your applications are OpenCL-bound and you don't need Vulkan, the GTX 460M delivers more throughput.

For everyone else, the GT 645M is the better recommendation. It offers Vulkan support, which the GTX 460M completely lacks, making it compatible with a broader range of modern software. It consumes 36% less power (32 W vs 50 W), which is critical for mobile platforms. It has a higher average benchmark score (4411 vs 4282) and a marginally better percentile ranking (26th vs 25th). Its 2 GB memory capacity is 33% larger than the GTX 460M's 1536 MB, which helps with modern texture-heavy workloads even if bandwidth is lower. The GT 645M also supports PCIe 3.0, doubling the bus bandwidth available to the GTX 460M's PCIe 2.0 interface.

The GTX 460M is a legacy part from 2010, while the GT 645M is from 2012. The two-year gap shows in every architectural metric except raw compute throughput. The GT 645M's Kepler design is fundamentally more efficient and capable for modern use cases. The GTX 460M wins the benchmark it was tested in, but loses the broader war of compatibility and efficiency.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The GeForce GT 645M averages 4411 across its three benchmark tests (Metal, OpenCL, Vulkan), while the GeForce GTX 460M averages 4282 from its single OpenCL test.

Q: Does the GeForce GTX 460M support Vulkan?

A: No. The GTX 460M has no Vulkan support listed, while the GT 645M supports Vulkan 1.2.175.

Q: How much more power does the GTX 460M consume?

A: The GTX 460M has a TDP of 50 W, which is 18 W (or 56%) higher than the GT 645M's 32 W TDP.

Q: What is the memory bandwidth difference between the two?

A: The GTX 460M delivers 60.00 GB/s via GDDR5 on a 192-bit bus, while the GT 645M provides 28.80 GB/s via DDR3 on a 128-bit bus. The GTX 460M has 108% more bandwidth.

Q: Which GPU has more shading units?

A: The GT 645M has 384 shading units, exactly double the GTX 460M's 192. Despite this, the GTX 460M achieves higher OpenCL scores.

Q: What is the process node difference?

A: The GT 645M uses a 28 nm process, while the GTX 460M uses 40 nm. Both are fabricated by TSMC.

Specification Differences

| Specification | GeForce GT 645M | GeForce GTX 460M |

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

| Chip | GK107 | GF106 |

| Architecture | Kepler | Fermi |

| Process Node | 28 nm | 40 nm |

| Transistors | 1,270 million | 1,170 million |

| Die Size | 118 mm² | 238 mm² |

| Base Clock | 709 MHz | — |

| Boost Clock | 780 MHz | — |

| Memory Clock | 900 MHz / 1800 Mbps effective | 625 MHz / 2.5 Gbps effective |

| Memory Size | 2 GB | 1536 MB |

| Memory Type | DDR3 | GDDR5 |

| Memory Bus | 128 bit | 192 bit |

| Memory Bandwidth | 28.80 GB/s | 60.00 GB/s |

| Shading Units | 384 | 192 |

| ROPs | 16 | 24 |

| Pixel Rate | 6.240 GPixel/s | 5.400 GPixel/s |

| Texture Rate | 24.96 GTexel/s | 21.60 GTexel/s |

| FP32 | 599.0 GFLOPS | 518.4 GFLOPS |

| TDP | 32 W | 50 W |

| Slot Width | IGP | MXM Module |

| Bus Interface | PCIe 3.0 x16 | PCIe 2.0 x16 |

| Vulkan | 1.2.175 | — |

| Release Date | 2012-09-30 | 2010-09-02 |

| Predecessor | GeForce 500M | GeForce 300M |

| Successor | GeForce 700M | GeForce 500M |