GPU Comparison

The GeForce GTX 960 and Intel Arc A380 are separated by seven years of GPU architecture, yet they land in a similar performance class. The data shows a clear generational shift: the Arc A380 wins every head-to-head benchmark in this comparison, but the GTX 960 remains competitive in its own right, with both cards sitting within a single percentile point of each other globally. The Arc A380 is not a marginal victor; it delivers decisive wins in modern API workloads, while the GTX 960’s legacy lies in its compatibility with older software ecosystems.

Where Each One Wins

The Intel Arc A380 is the dominant performer across every benchmark where both cards were tested. In 3DMark Steel Nomad DX12, the Arc A380 scores 808 versus the GTX 960’s 162, a delta of -80% from the NVIDIA card’s perspective, meaning the Arc A380 is roughly five times faster in this specific DirectX 12 test. The gap is similarly stark in Geekbench Vulkan, where the Arc A380 posts 36,736 against 9,231 for the GTX 960, a -74.9% delta. In Geekbench OpenCL, the Arc A380 achieves 38,224 compared to 18,925, a -50.5% delta. These are not close contests; the Arc A380 wins all three head-to-head benchmarks decisively.

The GTX 960, however, holds its own in the broader benchmark suite. Its average benchmark score is 9,273, which is actually higher than the Arc A380’s 8,558 average. This discrepancy suggests that the GTX 960 performs relatively better in older or different workloads not covered by the head-to-head tests. The GTX 960 also has a slightly higher percentile ranking at 45 versus the Arc A380’s 44, indicating that in the full database of GPUs, the NVIDIA card edges out the Intel card by one percentile point. The GTX 960’s nearest rival, the GeForce GTX 465, scores 9,294 with a -0.2% delta, showing the 960 is essentially tied with that older card. The Arc A380’s nearest rival, the AMD FirePro W5170M, scores 8,595 with a -0.4% delta, placing the Intel card in a similar performance tier.

Architecture Differences

The architectural gulf between these two cards is immense. The GTX 960 uses the GM206 chip built on NVIDIA’s Maxwell 2.0 architecture, fabricated on a 28 nm process at TSMC. It contains 2,940 million transistors on a 228 mm² die, yielding a transistor density of 12.9 million per mm². The Arc A380, by contrast, uses the DG2-128 chip on Intel’s Xe-HPG architecture, built on a 6 nm process also at TSMC. It packs 7,200 million transistors into a smaller 157 mm² die, achieving a transistor density of 45.9 million per mm², nearly four times denser than the GTX 960.

Memory configurations differ significantly. The GTX 960 ships with 2 GB of GDDR5 on a 128-bit bus, delivering 112.2 GB/s of bandwidth and running at 1,753 MHz (7 Gbps effective). The Arc A380 offers 6 GB of GDDR6 on a narrower 96-bit bus, but its memory clock of 1,937 MHz (15.5 Gbps effective) yields 186.0 GB/s of bandwidth, a 66% advantage over the GTX 960. Both cards have 1,024 shading units, 64 texture mapping units, and 32 ROPs, but the Arc A380 also includes 8 ray tracing cores, which the GTX 960 lacks entirely.

Clock speeds and compute throughput also favor the Intel card. The GTX 960 runs at a 1,127 MHz base and 1,178 MHz boost, producing 2.413 TFLOPS of FP32 performance, a 37.70 GPixel/s pixel rate, and 75.39 GTexel/s texture rate. The Arc A380 operates at 2,000 MHz base and 2,050 MHz boost, delivering 4.198 TFLOPS FP32 and 8.397 TFLOPS FP16 (2:1 ratio), with a pixel rate of 65.60 GPixel/s and texture rate of 131.2 GTexel/s. The Arc A380 also supports DirectX 12 Ultimate (12_2), while the GTX 960 is limited to DirectX 12 (12_1). Both cards support OpenGL 4.6 and Vulkan 1.4.

Power and connectivity further separate the two. The GTX 960 has a 120 W TDP, requires a 300 W PSU, and uses a single 6-pin power connector. The Arc A380 is more efficient on paper with a 75 W TDP and 250 W PSU requirement, but it uses a single 8-pin connector. The GTX 960 connects via PCIe 3.0 x16, while the Arc A380 uses PCIe 4.0 x8. Display outputs differ as well: the GTX 960 offers 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.2, whereas the Arc A380 provides 1x HDMI 2.1 and 3x DisplayPort 2.0.

The Verdict

The data points to a straightforward conclusion for modern workloads: the Intel Arc A380 is the superior card. It wins all three head-to-head benchmarks by massive margins, offers triple the VRAM (6 GB versus 2 GB), supports newer DirectX 12 Ultimate features, includes ray tracing hardware, and delivers more than double the FP32 throughput. The Arc A380’s 3DMark Steel Nomad score of 808 versus 162 for the GTX 960 is a 5x advantage that cannot be ignored. For anyone running current DirectX 12 or Vulkan titles, the Arc A380 is the only rational choice between these two.

However, the GTX 960’s higher average benchmark score of 9,273 versus 8,558 for the Arc A380 tells a different story. This suggests that in the aggregate of all benchmarks, including older DirectX 9, 10, and 11 workloads, the GTX 960 holds up remarkably well. Its nearest rival delta of -0.2% against the GTX 465 shows it is essentially tied with that card, while the Arc A380’s -0.4% delta against the FirePro W5170M places it in a similar tier. The GTX 960 also has a higher percentile ranking (45 versus 44), meaning it outperforms a slightly larger fraction of the GPU database.

For users with legacy software or who prioritize compatibility with older APIs, the GTX 960 remains a viable option despite its age. Its 2 GB VRAM and 112.2 GB/s bandwidth are modest by modern standards, but its OpenGL 4.6 and Vulkan 1.4 support ensure it can run contemporary APIs, albeit with far lower performance. The Arc A380’s PassMark scores reveal a weakness: its DirectX 9 score is 73, DirectX 10 is 37, DirectX 11 is 38, and DirectX 12 is 35, all low numbers that suggest driver overhead or architectural inefficiency in legacy titles. The GTX 960, with its Maxwell architecture and mature drivers, likely handles these older APIs more gracefully.

FAQ

Q: Which GPU has better raw compute performance?

A: The Intel Arc A380 delivers 4.198 TFLOPS FP32 versus 2.413 TFLOPS for the GTX 960, and it also offers 8.397 TFLOPS FP16 where the GTX 960 has no listed FP16 capability.

Q: How do the memory specifications compare?

A: The Arc A380 has 6 GB of GDDR6 on a 96-bit bus with 186.0 GB/s bandwidth, while the GTX 960 has 2 GB of GDDR5 on a 128-bit bus with 112.2 GB/s bandwidth.

Q: Which card has ray tracing support?

A: Only the Intel Arc A380 includes ray tracing cores (8 of them). The NVIDIA GTX 960 has no RT cores in its specifications.

Q: What are the power requirements for each card?

A: The GTX 960 has a 120 W TDP and requires a 300 W PSU with a single 6-pin connector. The Arc A380 has a 75 W TDP and requires a 250 W PSU with a single 8-pin connector.

Q: Which card scores higher in the overall benchmark average?

A: The GTX 960 has an average benchmark score of 9,273, which is higher than the Arc A380’s 8,558. The GTX 960 also has a percentile rank of 45 versus 44 for the Arc A380.

Q: How do the cards compare in DirectX 12 performance?

A: In 3DMark Steel Nomad DX12, the Arc A380 scores 808 versus 162 for the GTX 960, a -80% delta. The Arc A380 also supports DirectX 12 Ultimate (12_2), while the GTX 960 is limited to DirectX 12 (12_1).

Head-to-Head Benchmarks

The 3DMark Steel Nomad DX12 test is the most lopsided contest in this comparison. The Arc A380 scores 808 points against the GTX 960’s 162, producing a -80% delta from the NVIDIA card’s perspective. This means the Arc A380 is approximately five times faster in this modern DirectX 12 workload. The test likely stresses the Arc A380’s higher texture rate (131.2 GTexel/s versus 75.39 GTexel/s) and greater memory bandwidth (186.0 GB/s versus 112.2 GB/s), both of which are roughly double the GTX 960’s figures.

Geekbench Vulkan shows a similar pattern, with the Arc A380 hitting 36,736 versus 9,231 for the GTX 960, a -74.9% delta. This fourfold advantage in Vulkan performance aligns with the Arc A380’s architecture, which was designed for modern APIs from the ground up. The GTX 960’s Maxwell 2.0 architecture predates widespread Vulkan adoption, and its 1,178 MHz boost clock cannot compensate for the Arc A380’s 2,050 MHz boost.

Geekbench OpenCL is the closest of the three head-to-head tests, though the Arc A380 still wins decisively. The Intel card scores 38,224 against 18,925 for the GTX 960, a -50.5% delta. This halving of the performance gap compared to Vulkan suggests that OpenCL workloads are less dependent on the architectural advantages of the Arc A380, but the Intel card’s higher FP32 throughput (4.198 TFLOPS versus 2.413 TFLOPS) still gives it a clear edge. The GTX 960’s Geekbench Metal score of 8,773 and its OpenCL score of 18,925 show it is competent in general compute, but the Arc A380’s numbers are simply in a different league.

The GTX 960’s only benchmark victories come outside the head-to-head set. Its average benchmark score of 9,273 is 8.4% higher than the Arc A380’s 8,558, and it edges out the Intel card in percentile ranking (45 versus 44). The GTX 960 also has a closer relationship to its nearest rivals, with deltas ranging from -0.2% to 0.6% against cards like the GTX 465 and Radeon Vega 8. The Arc A380’s nearest rival deltas span -0.4% to 1.4%, showing a slightly wider performance spread around its average. These aggregate numbers suggest the GTX 960 is more consistent across a broad range of workloads, even if it loses decisively in the modern API tests.