NVIDIA GeForce GTX 650 Ti vs NVIDIA GeForce GTX 960M Comparison

FAQ

Q: Which GPU is faster in OpenCL compute workloads?

A: The NVIDIA GeForce GTX 960M wins decisively, scoring 11045 in Geekbench OpenCL versus 7877 for the GTX 650 Ti, a 40.2% advantage.

Q: How do the two cards compare in Vulkan performance?

A: The GTX 960M edges out the GTX 650 Ti by a narrow margin, scoring 8245 versus 8229, a 0.2% difference that is effectively negligible for real-world purposes.

Q: What is the average benchmark score for each GPU?

A: The GTX 960M has an average benchmark score of 9645, while the GTX 650 Ti averages 8053. This places the 960M at the 46th percentile of all GPUs, compared to the 42nd percentile for the 650 Ti.

Q: Which card has more memory and a larger memory capacity?

A: The GTX 960M offers 4 GB of GDDR5 memory, while the GTX 650 Ti comes with 1024 MB (1 GB) of GDDR5 memory. Both use a 128-bit memory bus.

Q: What is the difference in power consumption between the two?

A: The GTX 960M has a TDP of 75 W, whereas the GTX 650 Ti draws 110 W. The 960M is the more power-efficient design, despite delivering higher compute performance.

Q: Which GPU is newer and what is the release timeline?

A: The GTX 960M was released on 2015-03-12, while the GTX 650 Ti launched on 2012-10-08. The 960M is a later-generation mobile part from the GeForce 900M series, whereas the 650 Ti belongs to the GeForce 600 desktop family.

Architecture Differences

The two GPUs come from different NVIDIA architectures. The GTX 960M is built on the Maxwell architecture using the GM107 chip, while the GTX 650 Ti uses the Kepler architecture with the GK106S chip. Both are manufactured on a 28 nm process at TSMC, but the underlying designs diverge significantly.

The GTX 960M packs 1,870 million transistors on a die size of 148 mm², resulting in a transistor density of 12.6M per mm². The GTX 650 Ti, by contrast, integrates 2,540 million transistors on a larger 221 mm² die, yielding a lower density of 11.5M per mm². This means the Maxwell chip achieves more logic per unit area, a hallmark of architectural refinement.

Shading unit counts differ: the GTX 960M has 640 shading units, while the GTX 650 Ti has 768. However, the Maxwell architecture's efficiency allows the 960M to outperform the Kepler card despite fewer shaders. Texture mapping units also favor the 650 Ti, with 64 TMUs versus 40 on the 960M, while both share 16 ROPs.

The GTX 960M has a base clock of 1097 MHz and a boost clock of 1176 MHz. The GTX 650 Ti has no recorded base or boost clocks in the database. Memory clock rates are 1253 MHz (5 Gbps effective) for the 960M and 1350 MHz (5.4 Gbps effective) for the 650 Ti. The 650 Ti's higher memory clock gives it a slight bandwidth edge at 86.40 GB/s versus 80.19 GB/s, despite the same 128-bit bus width.

API support is nearly identical for DirectX (both 12, 11_0) and OpenGL (both 4.6). Vulkan support differs: the 960M supports Vulkan 1.4, while the 650 Ti supports Vulkan 1.2.175. The 960M is an MXM module with a bus interface of MXM-B (3.0), whereas the 650 Ti uses a standard PCIe 3.0 x16 slot. The 650 Ti requires a 1x 6-pin power connector and a 300 W suggested PSU; the 960M has no power connectors and a lower 75 W TDP.

Where Each One Wins

The GTX 960M dominates in compute-heavy scenarios. Its Geekbench OpenCL score of 11045 versus 7877 for the 650 Ti represents a 40.2% lead. This makes the 960M the clear choice for OpenCL-accelerated applications, including video encoding, physics simulation, and general-purpose GPU compute tasks. The architecture's higher efficiency and newer design translate directly into faster raw computation.

The GTX 960M also wins in Vulkan, but by a tiny margin: 8245 versus 8229, a 0.2% difference. In practice, this means the two GPUs deliver essentially identical Vulkan performance for gaming or other Vulkan-based workloads. Neither card has a meaningful edge here, so the winner is purely nominal.

The GTX 650 Ti does hold some specification-level advantages. Its 86.40 GB/s memory bandwidth exceeds the 960M's 80.19 GB/s, and its higher texture rate of 59.39 GTexel/s beats the 960M's 47.04 GTexel/s. In texture-heavy workloads that are not compute-bound, the 650 Ti could theoretically perform competitively, though the recorded benchmark data does not show this translating into a win. The 650 Ti also has twice the TMUs (64 versus 40), which could benefit certain fill-rate-limited scenarios.

For power-constrained environments, the GTX 960M wins decisively. Its 75 W TDP versus 110 W for the 650 Ti means it delivers superior compute performance while drawing 35 W less power. This makes the 960M the better option for laptops or compact systems where thermal and power budgets are tight.

Specification Differences

The two cards differ across multiple specification categories. The GTX 960M uses the GM107 chip (Maxwell architecture) from the GeForce 900M generation, while the GTX 650 Ti uses the GK106S chip (Kepler architecture) from the GeForce 600 generation. The 960M's die size is 148 mm² versus 221 mm² for the 650 Ti, and its transistor count is 1,870 million versus 2,540 million. Transistor density is 12.6M per mm² for the 960M and 11.5M per mm² for the 650 Ti.

Clock specifications show the 960M with a base clock of 1097 MHz and boost of 1176 MHz; the 650 Ti has no recorded base or boost clocks. Memory clocks are 1253 MHz (5 Gbps effective) for the 960M and 1350 MHz (5.4 Gbps effective) for the 650 Ti. Memory capacity is 4 GB for the 960M and 1024 MB for the 650 Ti. Both use GDDR5 and a 128-bit bus, but bandwidth differs: 80.19 GB/s for the 960M versus 86.40 GB/s for the 650 Ti.

Compute resources differ: the 960M has 640 shading units and 40 TMUs, while the 650 Ti has 768 shading units and 64 TMUs. Both have 16 ROPs. Pixel rate is 18.82 GPixel/s for the 960M versus 14.85 GPixel/s for the 650 Ti. Texture rate is 47.04 GTexel/s for the 960M versus 59.39 GTexel/s for the 650 Ti. FP32 performance is 1.505 TFLOPS for the 960M and 1,425.4 GFLOPS for the 650 Ti.

Form factor and power differ substantially: the 960M is an MXM Module with no power connectors and a 75 W TDP, while the 650 Ti is Single-slot, requires 1x 6-pin power, and has a 110 W TDP with a 300 W suggested PSU. The 650 Ti has a length of 145 mm (5.7 inches), while the 960M has no recorded dimensions. Display outputs are Portable Device Dependent for the 960M and 2x DVI plus 1x mini-HDMI 1.4a for the 650 Ti. Vulkan support is 1.4 for the 960M and 1.2.175 for the 650 Ti. The 960M has no launch MSRP; the 650 Ti launched at 149 USD. Release dates are 2015-03-12 for the 960M and 2012-10-08 for the 650 Ti.

Head-to-Head Benchmarks

The database records two head-to-head benchmark comparisons between these GPUs. The first is Geekbench OpenCL, where the GTX 960M scores 11045 against the GTX 650 Ti's 7877. The delta is 40.2%, a massive margin that places the 960M far ahead in compute performance. This is the single largest difference between the two cards in any measured test. The 960M's average benchmark score of 9645 versus 8053 for the 650 Ti reflects this substantial gap, as does the percentile ranking: 46th versus 42nd.

The second benchmark is Geekbench Vulkan, where the 960M scores 8245 and the 650 Ti scores 8229. The delta is a mere 0.2%, making this effectively a tie. Both cards sit within a hair of each other in Vulkan workloads, so users should not expect any meaningful difference in Vulkan-based applications or games.

Looking at the nearest rivals provides additional context. The GTX 960M's closest competitor is the NVIDIA Quadro K5000 with an average score of 9637, just 0.1% lower. The AMD Radeon Pro WX 2100 scores 9653, 0.1% higher. The 960M also sits near the NVIDIA Quadro P4000 (9665, 0.2% higher) and the NVIDIA Tesla C2070 (9716, 0.7% higher). This places the 960M in a tight performance cluster where small margins separate cards.

The GTX 650 Ti's nearest rivals include the NVIDIA GeForce GTX 880M at 8040 (0.2% lower), the NVIDIA GeForce GTX 650 Ti Boost at 8067 (0.2% higher), the NVIDIA Quadro P5000 at 8039 (0.2% lower), and the NVIDIA GRID K2 at 8080 (0.3% higher). The 650 Ti sits in a similar tight cluster, but at a lower absolute performance level than the 960M.

Across both recorded benchmarks, the GTX 960M wins 2 and the GTX 650 Ti wins 0. The Vulkan result is so close that it is statistically negligible, but the OpenCL result is decisive and drives the overall comparison.

The Verdict

The data points to a clear winner: the NVIDIA GeForce GTX 960M outperforms the GTX 650 Ti in every recorded benchmark. The OpenCL margin of 40.2% is substantial, and even the Vulkan test, though nearly identical, goes to the 960M. For anyone choosing between these two for compute workloads, the 960M is the obvious pick.

The GTX 960M also wins on efficiency. Its 75 W TDP versus 110 W for the 650 Ti means it delivers higher performance while consuming less power. This is particularly relevant for mobile or compact systems, where the 960M's MXM form factor and lack of power connectors make it a cleaner integration. The 650 Ti's requirement for a 6-pin connector and a 300 W suggested PSU adds installation complexity.

Memory capacity is another decisive factor. The 960M's 4 GB versus 1024 MB for the 650 Ti gives it four times the framebuffer, which matters for modern workloads with larger data sets or higher-resolution textures. The 650 Ti's slight bandwidth advantage (86.40 GB/s versus 80.19 GB/s) does not compensate for the capacity shortfall.

The GTX 650 Ti does retain some specification strengths: more shading units (768 versus 640), more TMUs (64 versus 40), and a higher texture rate (59.39 GTexel/s versus 47.04 GTexel/s). However, these advantages do not translate into benchmark wins. The Maxwell architecture's efficiency in the 960M overcomes the raw count disadvantage.

For Vulkan-specific applications, the choice is essentially moot: a 0.2% difference will not be perceptible. For OpenCL workloads, the 960M is the only rational choice. The percentile rankings reinforce this: the 960M sits at the 46th percentile of all GPUs, while the 650 Ti sits at the 42nd.

The verdict: choose the GTX 960M for superior compute performance, lower power draw, and four times the memory. The GTX 650 Ti is an older, less capable part that only makes sense if legacy PCIe slot compatibility or its specific display outputs (2x DVI, mini-HDMI) are required, and even then the performance gap is hard to ignore. The recorded data is unambiguous: the 960M wins 2 benchmarks, the 650 Ti wins 0.