GPU Comparison

The NVIDIA GeForce GT 645M and NVIDIA GeForce 930M are two end-of-life mobile graphics processors from different generations, yet benchmark results show they are remarkably close in overall performance. The GT 645M, a Kepler-based chip from the GeForce 600M series, posts an average benchmark score of 4411. The 930M, built on Maxwell architecture from the GeForce 900M series, averages 4388. That difference is a mere 0.5%, placing both cards at the 26th percentile of all GPUs. For a builder looking at used laptops, the choice between these two is not about raw speed, but about specific workload strengths and platform details.

Head-to-Head Benchmarks

The two GPUs split their head-to-head benchmark wins exactly evenly, with one victory each. This parity is the defining characteristic of this comparison.

The most decisive difference appears in the Geekbench OpenCL test. The 930M scores 5046, while the GT 645M manages only 2680. That is a 46.9% advantage for the 930M, a massive margin that indicates a clear dominance in compute-heavy, general-purpose GPU workloads. This result aligns with the newer Maxwell architecture’s efficiency in parallel compute tasks.

However, the GT 645M strikes back in the Geekbench Vulkan test. Here, the older card scores 4875 against the 930M’s 3729. The GT 645M leads by 30.7%, a substantial reversal that suggests its Kepler design handles the low-level graphics API more effectively. This is a notable win for a card from 2012, showing that API-level optimization can favor older hardware.

Looking at the broader rival landscape, both cards sit within a tight cluster. The Intel Iris Pro Graphics 5200 scores 4360, which is 1.2% behind the GT 645M and 0.7% ahead of the 930M. The NVIDIA GeForce RTX 4070 GDDR6, despite its modern high-end positioning, scores 4335 in this particular benchmark aggregation, placing it 1.8% below the GT 645M. On the other side, the AMD Radeon R7 M260 scores 4499, which is 1.9% higher than the GT 645M. The 930M’s rival list shows the AMD FirePro W2100 at 4295, sitting 2.2% below it. These deltas are all within a few percentage points, confirming that neither mobile GPU has a commanding lead over its immediate peers.

Where Each One Wins

The benchmark data creates a clear use-case split based on which API or compute framework you prioritize.

The 930M is the choice for OpenCL-based applications. Its 5046 OpenCL score is not just a win over the GT 645M; it is the highest score recorded for either card in any test. This makes it the stronger candidate for tasks like video encoding acceleration, physics simulations, or any productivity software that leverages OpenCL for GPU compute. If you are running applications that offload parallel processing to the GPU, the 930M delivers nearly half again the performance of the GT 645M.

Conversely, the GT 645M wins decisively in Vulkan-based workloads. Its 4875 Vulkan score demonstrates that it can handle modern graphics APIs with greater efficiency than the 930M. This translates to a practical advantage in games or applications that use the Vulkan renderer, where the GT 645M will provide a smoother frame rate experience. For a gamer playing titles with Vulkan support, this older card is the better performer.

In terms of overall average score, the GT 645M edges out the 930M by 23 points (4411 vs 4388). This is a negligible difference, but it does mean the GT 645M is the nominal leader in aggregate performance. However, this lead is so small that it should not be the deciding factor. The choice depends entirely on whether your primary software leans on OpenCL or Vulkan.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce GT 645M has a slightly higher average benchmark score of 4411, compared to the 930M’s 4388. This is a 0.5% difference, which is statistically insignificant.

Q: Is the newer GeForce 930M faster than the older GT 645M in all tests?

A: No. The 930M wins the Geekbench OpenCL test by a large margin (5046 vs 2680), but the GT 645M wins the Geekbench Vulkan test decisively (4875 vs 3729). Each card wins one benchmark.

Q: What is the performance gap in the Vulkan benchmark?

A: The GT 645M leads the 930M by 30.7% in the Geekbench Vulkan test, with scores of 4875 and 3729 respectively.

Q: How large is the OpenCL performance difference?

A: The 930M is 46.9% faster than the GT 645M in Geekbench OpenCL, scoring 5046 against the GT 645M’s 2680.

Q: Do these GPUs have similar overall performance to other rivals?

A: Yes. The GT 645M is within 1.9% of the AMD Radeon R7 M260 and 1.8% of the NVIDIA GeForce RTX 4070 GDDR6. The 930M is within 2.2% of the AMD FirePro W2100 and 0.7% of the Intel Iris Pro Graphics 5200.

Q: Which card has a higher transistor density?

A: The GeForce 930M has a higher transistor density of 13.2M per mm², compared to the GT 645M’s 10.8M per mm². This is due to the 930M’s smaller die size despite having fewer total transistors.

Specification Differences

The two cards differ significantly in their core specifications, which explains their divergent benchmark behavior.

Clock speeds are a primary difference. The GT 645M runs at a base clock of 709 MHz with a boost clock of 780 MHz. The 930M operates at a static 549 MHz for both base and boost, meaning it does not have a dynamic overclocking headroom. The GT 645M also has faster memory, running at 900 MHz (1800 Mbps effective) compared to the 930M’s 800 MHz (1600 Mbps effective).

Memory bandwidth is a major point of separation. The GT 645M uses a 128-bit memory bus, delivering 28.80 GB/s of bandwidth. The 930M is constrained to a 64-bit bus, resulting in only 12.80 GB/s. This is a 16 GB/s deficit for the 930M and a likely factor in its Vulkan performance shortfall. Both cards have 2 GB of DDR3 memory, but the GT 645M’s wider bus gives it a clear theoretical advantage in memory-intensive scenes.

The pixel and texture rates also favor the GT 645M. It achieves a pixel rate of 6.240 GPixel/s and a texture rate of 24.96 GTexel/s. The 930M is lower on both counts, with 4.392 GPixel/s and 13.18 GTexel/s. The FP32 compute performance follows the same trend: the GT 645M delivers 599.0 GFLOPS, while the 930M manages 421.6 GFLOPS. Despite this, the 930M wins in OpenCL, which shows that raw theoretical throughput does not always predict real-world compute results.

The bus interface also differs. The GT 645M uses PCIe 3.0 x16, while the 930M uses PCIe 3.0 x8. This halved interface bandwidth could affect data transfer rates in some scenarios, though for most mobile workloads the impact is minimal. Power consumption is nearly identical, with the GT 645M rated at 32 W TDP and the 930M at 33 W TDP.

Architecture Differences

Architecturally, these GPUs represent two distinct NVIDIA design philosophies, despite both being built on a 28 nm process at TSMC.

The GT 645M uses the GK107 chip, based on the Kepler architecture, and has a die size of 118 mm². It packs 1,270 million transistors, resulting in a transistor density of 10.8M per mm². The 930M uses the GM108S chip, based on the Maxwell architecture, and is a smaller die at 77 mm². It contains 1,020 million transistors, which yields a higher density of 13.2M per mm². The Maxwell chip is more efficiently packed, a sign of the architectural refinements made between generations.

Shading unit counts are identical at 384, but other execution resources differ. The GT 645M has 32 texture mapping units (TMUs) and 16 render output units (ROPs). The 930M has fewer of both, with 24 TMUs and 8 ROPs. This reduction in ROPs is substantial and directly impacts the pixel fill rate, which is a likely contributor to the GT 645M’s Vulkan advantage.

The API support shows a key difference in Vulkan version. Both cards support DirectX 12 (11_0) and OpenGL 4.6. However, the 930M supports Vulkan 1.4, while the GT 645M only lists Vulkan 1.2.175. Despite the newer Vulkan version on the 930M, the GT 645M scores higher on the Vulkan benchmark, indicating that hardware design matters more than API revision level. The release dates also differ, with the GT 645M launching on September 30, 2012, and the 930M on March 12, 2015. The 930M’s predecessor is the GeForce 800M, and its successor is the GeForce 10 Mobile, while the GT 645M sits between the GeForce 500M and 700M.

The Verdict

The data presents a clear, albeit narrow, verdict: choose the GT 645M for Vulkan-based gaming and choose the 930M for OpenCL compute workloads.

For a gamer or user who frequently runs Vulkan applications, the GT 645M is the superior pick. Its 30.7% lead in the Vulkan benchmark is a tangible performance gap that will be noticeable in frame rates. The wider 128-bit memory bus and higher fill rates support this advantage, making it the more capable graphics card for modern low-level APIs. The GT 645M also holds a slight edge in average overall score, which reinforces its position as the more balanced performer.

For a user whose software stack relies on OpenCL, the 930M is the clear winner. Its 46.9% advantage in the OpenCL test is the single largest performance gap in this comparison. This makes it the better choice for GPU-accelerated productivity tasks, even if its gaming performance in Vulkan is weaker. The 930M’s higher transistor density and newer Maxwell architecture suggest it is more efficient at general compute, which the OpenCL results confirm.

Ultimately, neither card is a dominant winner. They are evenly matched in overall average score, with a 0.5% difference. The deciding factor must be the specific workload. A builder prioritizing Vulkan gaming should select the GT 645M. A builder prioritizing OpenCL compute should select the 930M. Both are end-of-life products, so availability will dictate the final purchase, but the performance data gives a clear directive for each use case.