NVIDIA GeForce GTX 1070 vs NVIDIA GeForce GTX 960 Comparison

The NVIDIA GeForce GTX 1070 and GTX 960 represent two distinct generations of NVIDIA’s mainstream and mid-range offerings, and the benchmark data shows a decisive performance gap between them. In the four head-to-head tests available, the GTX 1070 wins every single matchup, with margins ranging from roughly 114% to a staggering 568%. The most extreme difference appears in the 3DMark Steel Nomad DX12 test, where the GTX 1070 scores 1082 against the GTX 960’s 162, a 567.9% advantage. While that specific workload heavily favors newer architectures, the gap remains substantial across all other tests: 114.3% in Geekbench Metal, 136.2% in Geekbench OpenCL, and 139.6% in Geekbench Vulkan. This is not a close contest; it is a generational leap expressed in raw numbers.

Head-to-Head Benchmarks

The data paints a clear picture of dominance for the GTX 1070. Starting with the most demanding test, 3DMark Steel Nomad DX12, the GTX 1070 posts a score of 1082, while the GTX 960 manages only 162. This 567.9% delta is the largest of any comparison and highlights how much more capable the newer card is under modern DirectX 12 workloads. The GTX 960’s score here is low enough to suggest it struggles with the geometry and shading complexity of contemporary game engines, whereas the GTX 1070 handles it with relative ease.

In compute-oriented benchmarks, the GTX 1070 again leads by a wide margin. In Geekbench OpenCL, the GTX 1070 scores 44700, compared to the GTX 960’s 18925, resulting in a 136.2% advantage. This is a strong indicator of general-purpose compute performance, useful for tasks like video encoding, physics simulations, or GPU-accelerated rendering. The Geekbench Vulkan test shows a similar story: 22121 for the GTX 1070 versus 9231 for the GTX 960, a 139.6% delta. Vulkan is increasingly common in modern games, and the GTX 1070’s lead here suggests it will age better as more titles adopt this API.

The Geekbench Metal benchmark, which is specific to Apple’s graphics API, shows the GTX 1070 scoring 18801 against the GTX 960’s 8773, a 114.3% improvement. While Metal is primarily relevant on macOS systems, the result reinforces the overall performance hierarchy. Across all four tests, the GTX 1070 wins 4 out of 4, and the GTX 960 fails to secure a single victory. The smallest margin of victory is still over double the performance of the GTX 960, which is a decisive gap by any measure.

Where Each One Wins

Given the benchmark results, there is very little territory where the GTX 960 can claim a win. The GTX 1070 wins every single head-to-head test, so the practical question is where the GTX 960 remains usable rather than competitive. The data shows that the GTX 960’s best relative performance is in the Geekbench Metal test, where its score of 8773 is roughly 47% of the GTX 1070’s 18801. While that is still a large gap, it is the closest the GTX 960 gets to its rival. For users running older DirectX 11 titles or lightweight esports games, the GTX 960’s 2 GB of memory and 112.2 GB/s bandwidth might still suffice for 1080p gaming at medium settings, though the benchmark data does not directly confirm this.

The GTX 1070, on the other hand, wins decisively in every category. Its 8 GB of memory and 256.3 GB/s bandwidth make it far more suitable for high-resolution textures and modern game engines. The 3DMark Steel Nomad result, where the GTX 1070 scores 1082 versus 162, indicates that the GTX 960 is effectively obsolete for future DirectX 12 titles. The GTX 1070’s compute lead (44700 vs 18925 in OpenCL) also makes it the clear choice for any GPU-accelerated productivity work. In short, the GTX 960 wins only in the sense of being a lower-power (120 W vs 150 W) and physically shorter card (241 mm vs 267 mm), which might matter in small form factor builds, but in pure performance, there is no contest.

Architecture Differences

The two cards are built on fundamentally different architectures. The GTX 1070 uses the Pascal architecture on a 16 nm process, while the GTX 960 is based on Maxwell 2.0 on a 28 nm process. This node shrink is the primary driver of the performance gap, allowing the GTX 1070 to pack 7,200 million transistors into a 314 mm² die, compared to the GTX 960’s 2,940 million transistors on a 228 mm² die. The transistor density tells the story: 22.9M per mm² for the GTX 1070 versus 12.9M per mm² for the GTX 960. This means the GTX 1070 has more than twice the transistors in only 38% more die area, enabling significantly more compute units.

The core configurations differ dramatically. The GTX 1070 features 1920 shading units, 120 TMUs, and 64 ROPs, while the GTX 960 has 1024 shading units, 64 TMUs, and 32 ROPs. This means the GTX 1070 has 87.5% more shading units and 100% more ROPs, which directly translates to higher pixel and texture fill rates. The pixel rate is 107.7 GPixel/s for the GTX 1070 versus 37.70 GPixel/s for the GTX 960, and the texture rate is 202.0 GTexel/s versus 75.39 GTexel/s. The FP32 compute output is 6.463 TFLOPS for the GTX 1070 and 2.413 TFLOPS for the GTX 960, a 168% difference. The GTX 1070 also supports FP16 at 101.0 GFLOPS (1:64), while the GTX 960 has no FP16 capability listed. Both cards support DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4, so feature-level API support is identical, but the underlying hardware is vastly different in capability.

Specification Differences

The most impactful specification difference is memory. The GTX 1070 comes with 8 GB of GDDR5 on a 256-bit bus, delivering 256.3 GB/s of bandwidth. The GTX 960 has 2 GB of GDDR5 on a 128-bit bus, offering only 112.2 GB/s. This is a 128% bandwidth advantage for the GTX 1070, and the capacity difference means the GTX 1070 can hold far more texture data and assets in VRAM, which is critical for modern games at high resolutions. The memory clock also differs: the GTX 1070 runs at 2002 MHz (8 Gbps effective) versus 1753 MHz (7 Gbps effective) for the GTX 960.

Clock speeds also favor the GTX 1070. Its base clock is 1506 MHz with a boost of 1683 MHz, compared to the GTX 960’s base of 1127 MHz and boost of 1178 MHz. This is a 34% higher base clock and a 43% higher boost clock. The power draw is higher for the GTX 1070 at 150 W versus 120 W, and it requires a 1x 8-pin power connector versus the GTX 960’s 1x 6-pin. The suggested PSU is 450 W for the GTX 1070 and 300 W for the GTX 960. Both cards are dual-slot, but the GTX 1070 is longer at 267 mm (10.5 inches) versus 241 mm (9.5 inches) for the GTX 960. The GTX 1070 also supports DisplayPort 1.4a, while the GTX 960 is limited to DisplayPort 1.2. Both use a PCIe 3.0 x16 interface and share the same display outputs (1x DVI, 1x HDMI 2.0, 3x DisplayPort). The GTX 1070 has a launch MSRP of 379 USD, while the GTX 960 launched at 199 USD.

FAQ

Q: How much faster is the GTX 1070 in the 3DMark Steel Nomad DX12 test?

A: The GTX 1070 scores 1082, while the GTX 960 scores 162, making the GTX 1070 567.9% faster in that specific workload.

Q: Which card has more video memory, and why does that matter?

A: The GTX 1070 has 8 GB, while the GTX 960 has 2 GB. The larger capacity allows the GTX 1070 to store more textures and assets, which is critical for modern games at high resolutions and detail settings.

Q: Is the GTX 960 competitive in any benchmark?

A: No. In the four head-to-head tests (3DMark Steel Nomad, Geekbench Metal, OpenCL, and Vulkan), the GTX 1070 wins all of them. The GTX 960’s closest result is in Geekbench Metal, where it scores 8773 versus the GTX 1070’s 18801, a 114.3% delta.

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

A: The GTX 1070 has a 256-bit bus with 256.3 GB/s bandwidth, while the GTX 960 has a 128-bit bus with 112.2 GB/s. The GTX 1070 offers 128% more bandwidth.

Q: Do both cards support the same modern APIs?

A: Yes, both support DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. However, the GTX 1070’s superior hardware makes it far more effective in these APIs, as shown by the 3DMark Steel Nomad result.

Q: Which card requires a more powerful power supply?

A: The GTX 1070 has a suggested PSU of 450 W and a 150 W TDP, while the GTX 960 has a suggested PSU of 300 W and a 120 W TDP.

The Verdict

The benchmark data is unambiguous: the GTX 1070 is in a completely different performance class than the GTX 960. With a 100% win rate in head-to-head tests and minimum performance lead of 114.3%, there is no scenario where the GTX 960 outperforms the GTX 1070. The GTX 1070’s advantages in memory capacity (8 GB vs 2 GB), bandwidth (256.3 GB/s vs 112.2 GB/s), shading units (1920 vs 1024), and compute throughput (6.463 TFLOPS vs 2.413 TFLOPS) make it the superior choice for any modern gaming or productivity workload.

The GTX 960 remains viable only for users with strict power or size constraints, as it draws 30 W less power (120 W vs 150 W) and is 26 mm shorter (241 mm vs 267 mm). It also requires a less substantial PSU (300 W vs 450 W) and a smaller power connector (6-pin vs 8-pin). However, these physical advantages come at the cost of drastically reduced performance. For anyone building a system for current games or GPU-accelerated tasks, the GTX 1070 is the only sensible option from this pair. The GTX 960’s 2 GB VRAM and 112.2 GB/s bandwidth are simply insufficient for the demands of modern titles, as evidenced by its 162-point score in 3DMark Steel Nomad. The data recommends the GTX 1070 without reservation for performance-oriented users; the GTX 960 should only be considered for secondary or legacy systems where power and space are the absolute priorities.