AMD Radeon R5 M335 vs NVIDIA GeForce GT 645M Comparison

The AMD Radeon R5 M335 and NVIDIA GeForce GT 645M are both end-of-life mobile graphics solutions from different generations, and the benchmark data reveals a split decision. The Radeon R5 M335 dominates in OpenCL compute workloads, while the GeForce GT 645M counters with a narrow win in Vulkan performance. Across the two head-to-head tests, each card claims one victory, with average benchmark scores of 4752 for the AMD part and 4411 for the NVIDIA part, placing them at the 28th and 26th percentiles of all GPUs, respectively.

Head-to-Head Benchmarks

The most striking disparity appears in the Geekbench OpenCL test. The AMD Radeon R5 M335 scores 4745, while the NVIDIA GeForce GT 645M manages only 2680. This translates to a massive 77.1% advantage for the AMD part, making it the clear winner for any OpenCL-accelerated application. This is not a marginal difference; it is a near-doubling of performance in this specific workload, suggesting that the R5 M335's architecture is far better optimized for the compute patterns exercised by OpenCL.

In contrast, the Vulkan benchmark tells a different story. The NVIDIA GeForce GT 645M edges out the AMD Radeon R5 M335 with a score of 4875 versus 4758, a delta of -2.4% from the perspective of the AMD card. While this is a win for NVIDIA, the margin is comparatively slender. The GT 645M's victory here demonstrates that its Kepler architecture holds up reasonably well in modern graphics API workloads, even though its overall compute throughput in OpenCL lags dramatically.

Looking at the broader context, the AMD Radeon R5 M335's average score of 4752 places it in a tight cluster of rivals. It sits just 0.5% ahead of the AMD Radeon R8 M445DX (4727), 0.7% behind the AMD Radeon R5 M255 (4788), and 0.8% behind the NVIDIA GeForce RTX 3080 12 GB (4791). This near-parity with a high-end desktop card like the RTX 3080 12 GB in this specific aggregate metric is an artifact of the benchmark suite's weighting, but it underscores that the R5 M335's compute performance is competitive within its immediate class. The NVIDIA GeForce GT 645M, with an average of 4411, sits 1.9% behind the AMD Radeon R7 M260 (4499) and 0.5% ahead of the NVIDIA GeForce 930M (4388), placing it in a slightly lower performance bracket overall.

Where Each One Wins

The AMD Radeon R5 M335 is the definitive choice for compute-heavy tasks that leverage OpenCL. Its 77.1% lead in that benchmark is the single most significant performance gap between the two cards and suggests that applications like video encoding, certain physics simulations, or data-parallel processing would run substantially faster on the AMD solution. The R5 M335's architecture appears to excel in raw parallel throughput for these workloads, as evidenced by its higher FP32 rating of 659.2 GFLOPS compared to the GT 645M's 599.0 GFLOPS.

The NVIDIA GeForce GT 645M wins in the Vulkan graphics API, albeit by a small margin. For gaming or applications that utilize Vulkan for rendering, the GT 645M holds a slight edge. This suggests that NVIDIA's driver maturity or hardware design for modern graphics pipelines provides a benefit, even if the underlying compute capability is weaker. The GT 645M also offers a higher texture rate of 24.96 GTexel/s versus 20.60 GTexel/s for the AMD card, which could contribute to better performance in texture-bound scenarios. However, the R5 M335 counters with a higher pixel rate of 8.240 GPixel/s compared to 6.240 GPixel/s, indicating a potential advantage in fill-rate-limited situations.

For general-purpose compute, the R5 M335's lead is overwhelming. For modern graphics API rendering, the GT 645M takes a narrow victory. The data does not suggest that either card is a universal winner; rather, they have complementary strengths.

Architecture Differences

The two GPUs are built on fundamentally different architectures from different eras. The AMD Radeon R5 M335 uses the GCN 1.0 architecture with a chip codenamed "Exo," part of the Gem System (R5 M300) generation. The NVIDIA GeForce GT 645M uses the Kepler architecture with the GK107 chip, from the GeForce 600M generation. Both are fabricated on a 28 nm process by TSMC, but the similarities end there.

The transistor counts and die sizes differ significantly. The AMD chip contains 690 million transistors on a 56 mm² die, resulting in a transistor density of 12.3M per mm². The NVIDIA chip is larger, with 1,270 million transistors on a 118 mm² die, yielding a lower density of 10.8M per mm². This suggests that NVIDIA's design is more complex in absolute terms, but AMD's is more efficiently packed.

Memory configurations are notably different. Both cards have 2 GB of DDR3 memory, but the bus widths diverge. The R5 M335 uses a 64-bit bus, providing a bandwidth of 14.40 GB/s. The GT 645M uses a 128-bit bus, doubling the bandwidth to 28.80 GB/s. This is a significant advantage for the NVIDIA card in memory-intensive tasks, potentially explaining its Vulkan performance edge. The core configurations also differ: the R5 M335 has 320 shading units, 20 texture mapping units (TMUs), and 8 ROPs, while the GT 645M has 384 shading units, 32 TMUs, and 16 ROPs.

Clock speeds vary as well. The AMD card's core clock is not specified in the data, but its memory runs at 900 MHz with 1800 Mbps effective. The NVIDIA card has a base clock of 709 MHz and a boost clock of 780 MHz, with the same memory speed of 900 MHz and 1800 Mbps effective. In terms of API support, the R5 M335 supports DirectX 12 (11_1), OpenGL 4.6, and Vulkan 1.2.170. The GT 645M supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175, giving it a slightly newer Vulkan version. The AMD card also lists a TDP, but the GT 645M is rated at 32 W and uses an IGP slot width with no power connectors.

The Verdict

The data presents a clear, albeit bifurcated, verdict. For users whose primary workload involves OpenCL compute, the AMD Radeon R5 M335 is the superior choice. Its 77.1% performance advantage in the Geekbench OpenCL test is decisive and should be the deciding factor for any application that relies on this API. The R5 M335's higher FP32 throughput of 659.2 GFLOPS reinforces this conclusion.

For users focused on Vulkan-based gaming or rendering, the NVIDIA GeForce GT 645M is the better option. Its 2.4% lead in the Vulkan benchmark, combined with its doubled memory bandwidth of 28.80 GB/s and higher texture rate, points to a stronger experience in modern graphics APIs. The GT 645M also has a higher average benchmark score relative to its closest rivals, landing 1.8% ahead of the NVIDIA GeForce RTX 4070 GDDR6 (4335), whereas the R5 M335 sits 0.8% behind the RTX 3080 12 GB.

Ultimately, the choice depends entirely on the target application. The R5 M335 is a compute specialist, while the GT 645M is a graphics generalist with a slight edge in the Vulkan API. Given that the GT 645M also has double the memory bandwidth, it may be the more balanced pick for a broader range of tasks, but the sheer magnitude of the R5 M335's OpenCL win cannot be ignored for compute-focused users.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The AMD Radeon R5 M335 has an average benchmark score of 4752, which is higher than the NVIDIA GeForce GT 645M's average of 4411.

Q: How large is the performance gap in OpenCL?

A: The AMD Radeon R5 M335 leads the NVIDIA GeForce GT 645M by 77.1% in the Geekbench OpenCL test, with scores of 4745 and 2680, respectively.

Q: Does the NVIDIA card win any benchmark?

A: Yes, the NVIDIA GeForce GT 645M wins the Geekbench Vulkan test with a score of 4875, compared to the AMD Radeon R5 M335's 4758, a margin of 2.4%.

Q: What is the memory bandwidth difference?

A: The NVIDIA GeForce GT 645M has a 128-bit memory bus providing 28.80 GB/s bandwidth, while the AMD Radeon R5 M335 has a 64-bit bus providing 14.40 GB/s.

Q: Which GPU has more shading units?

A: The NVIDIA GeForce GT 645M has 384 shading units, while the AMD Radeon R5 M335 has 320 shading units.

Q: What is the transistor count for each chip?

A: The AMD Radeon R5 M335's "Exo" chip contains 690 million transistors, while the NVIDIA GeForce GT 645M's GK107 chip contains 1,270 million transistors.