NVIDIA GeForce RTX 4070 SUPER vs NVIDIA RTX A6000 Comparison

The NVIDIA RTX A6000 and NVIDIA GeForce RTX 4070 SUPER occupy an unusual position in the benchmark database: their aggregate scores are nearly identical, yet their architectural philosophies and performance profiles diverge sharply. The RTX A6000, a workstation Ampere part from 2020, and the RTX 4070 SUPER, a GeForce 40-series Ada Lovelace card from 2024, land within 0.2% of each other in average benchmark score (42653 vs 42576, respectively). Both sit at the 84th percentile among all GPUs. However, the head-to-head data reveals a decisive sweep: the RTX 4070 SUPER wins all nine compared tests, often by substantial margins. This is a case where the aggregate numbers hide a lopsided competitive reality.

Head-to-Head Benchmarks

The RTX 4070 SUPER does not merely edge out the A6000; it dominates across every single benchmark category recorded. The largest gap appears in Passmark DirectX 11, where the 4070 SUPER scores 273 against the A6000's 191, a 30% advantage. The pattern repeats in DirectX 9, with the 4070 SUPER posting 344 versus 245, a 28.8% lead. These legacy API results suggest the Ada Lovelace architecture is substantially more efficient at translating its compute resources into rasterization performance in older workloads.

The margin narrows somewhat in more modern or API-agnostic tests, but the 4070 SUPER still holds a clear edge. In Passmark G3D, the 4070 SUPER scores 29995 against 22577, a 24.7% advantage. The Passmark G2D test shows a 22.9% lead (1184 vs 913), indicating the newer card also handles 2D operations more efficiently. Compute workloads tell a similar story: Passmark GPU Compute sees the 4070 SUPER at 17108 versus 14110, a 17.5% win for the GeForce card.

Interestingly, the smallest deltas appear in the Geekbench tests, which are often considered more representative of general compute throughput. In Geekbench OpenCL, the 4070 SUPER scores 192684 against 180149, a 6.5% lead. The Vulkan test shows a 7.7% advantage (179271 vs 165451). Even in DirectX 12, the modern API benchmark, the 4070 SUPER leads by 20.9% (110 vs 87). The data suggests that the A6000's advantage in raw hardware resources—more shading units, more memory bandwidth—does not translate into superior real-world performance against the newer architecture. The 4070 SUPER's wins range from 6.5% to 30%, with the average across all tests sitting around 20%. This is not a marginal victory; it is a systematic performance gap.

Architecture Differences

The fundamental explanation lies in the generational leap between these two designs. The RTX A6000 uses the GA102 chip built on Samsung's 8 nm process, while the RTX 4070 SUPER employs the AD104 chip fabricated by TSMC on a 5 nm node. This process shrink is dramatic: the A6000's die measures 628 mm² and contains 28,300 million transistors, yielding a density of 45.1 million transistors per mm². The 4070 SUPER's die is less than half the size at 294 mm², yet packs 35,800 million transistors—a density of 121.8 million per mm². The newer card crams 26% more transistors into a 53% smaller area.

This transistor density advantage translates directly into architectural efficiency. The A6000 boasts higher raw specs: 10,752 shading units, 336 texture mapping units, 112 ROPs, 84 RT cores, and 336 tensor cores. The 4070 SUPER counters with 7,168 shading units, 224 TMUs, 80 ROPs, 56 RT cores, and 224 tensor cores—roughly two-thirds the resources. Yet the 4070 SUPER still wins every benchmark. The clock speeds explain part of the story: the A6000 runs at 1410 MHz base and 1800 MHz boost, while the 4070 SUPER operates at 1980 MHz base and 2475 MHz boost. That 37.5% higher boost clock on the newer card compensates for the reduced core count.

The FP32 compute figures reflect this trade-off: the A6000 delivers 38.71 TFLOPS, while the 4070 SUPER produces 35.48 TFLOPS. The older card technically has higher peak compute, but the benchmark results indicate it cannot sustain that advantage in practice. Memory configurations also diverge significantly. The A6000 features 48 GB of GDDR6 on a 384-bit bus, providing 768.0 GB/s of bandwidth. The 4070 SUPER has 12 GB of GDDR6X on a 192-bit bus, yielding 504.2 GB/s. The A6000's 52% bandwidth advantage is real, but it does not overcome the architectural efficiency gap in the tested workloads.

FAQ

Q: Why does the RTX 4070 SUPER win every benchmark despite the A6000 having more cores and memory bandwidth?

A: The 4070 SUPER's 5 nm TSMC process allows for dramatically higher clock speeds (2475 MHz boost vs 1800 MHz) and greater transistor density (121.8M/mm² vs 45.1M/mm²), which compensates for its lower core count. The benchmark data shows the newer architecture is simply more efficient at executing the same work.

Q: Is the RTX A6000's larger memory capacity reflected in any performance advantage?

A: No. The A6000's 48 GB of GDDR6 with 768.0 GB/s bandwidth does not provide a win in any of the nine compared benchmarks. The 4070 SUPER's 12 GB of GDDR6X with 504.2 GB/s bandwidth is sufficient to outperform it across all tests, though the A6000's capacity may benefit workloads not represented in these benchmarks.

Q: How close are these cards in overall performance?

A: Their average benchmark scores are nearly identical: the A6000 averages 42653, while the 4070 SUPER averages 42576, a difference of only 0.2%. Both occupy the 84th percentile among all GPUs, and each lists the other as its nearest rival.

Q: Which card has better DirectX 11 performance?

A: The RTX 4070 SUPER is 30% faster in Passmark DirectX 11, scoring 273 versus the A6000's 191. This is the largest single-test margin in the head-to-head comparison.

Q: Do the cards support the same modern APIs?

A: Yes, both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The performance differences are not due to API compatibility but rather architectural execution efficiency.

Q: What is the power consumption difference?

A: The A6000 has a TDP of 300 W with a suggested 700 W PSU, while the 4070 SUPER has a 220 W TDP with a 550 W suggested PSU. The newer card delivers better performance at 73% of the power draw.

Specification Differences

The two cards diverge across nearly every technical specification. The A6000 uses the GA102 chip on Samsung's 8 nm process, while the 4070 SUPER uses AD104 on TSMC's 5 nm node. Transistor count differs: 28,300 million for the A6000 versus 35,800 million for the 4070 SUPER. Die size is 628 mm² versus 294 mm², and transistor density is 45.1M/mm² versus 121.8M/mm².

Clock speeds favor the 4070 SUPER with a 1980 MHz base and 2475 MHz boost, against the A6000's 1410 MHz base and 1800 MHz boost. Memory configurations are completely different: the A6000 has 48 GB of GDDR6 on a 384-bit bus with 768.0 GB/s bandwidth, while the 4070 SUPER has 12 GB of GDDR6X on a 192-bit bus with 504.2 GB/s bandwidth. The A6000's memory runs at 2000 MHz (16 Gbps effective), while the 4070 SUPER's runs at 1313 MHz (21 Gbps effective).

Compute resources are higher on the A6000: 10,752 shading units, 336 TMUs, 112 ROPs, 84 RT cores, and 336 tensor cores, versus 7,168 shading units, 224 TMUs, 80 ROPs, 56 RT cores, and 224 tensor cores. Pixel and texture rates are also higher on the A6000 (201.6 GPixel/s and 604.8 GTexel/s) compared to the 4070 SUPER (198.0 GPixel/s and 554.4 GTexel/s). FP32 and FP16 both favor the A6000 at 38.71 TFLOPS versus 35.48 TFLOPS.

Power draw favors the 4070 SUPER at 220 W TDP with a 550 W suggested PSU, versus 300 W and 700 W for the A6000. Display outputs differ: the A6000 has 4x DisplayPort 1.4a, while the 4070 SUPER has 1x HDMI 2.1 and 3x DisplayPort 1.4a. The power connector is an 8-pin EPS on the A6000 and a 1x 16-pin on the 4070 SUPER. The A6000 measures 267 mm by 112 mm with unspecified width, while the 4070 SUPER is 267 mm by 112 mm by 42 mm. Release dates are October 2020 for the A6000 and January 2024 for the 4070 SUPER.

Where Each One Wins

Based strictly on the benchmark data, the RTX 4070 SUPER wins in every single compared category. It is faster in DirectX 10, 11, and 12, in DirectX 9, in 2D operations, in 3D rendering, and in compute workloads. The margins range from 6.5% in Geekbench OpenCL to 30% in Passmark DirectX 11. There is no benchmark category where the A6000 achieves a victory.

The A6000's strengths are theoretical rather than demonstrated. It has more memory (48 GB vs 12 GB), higher memory bandwidth (768.0 GB/s vs 504.2 GB/s), more shading units (10,752 vs 7,168), and higher peak FP32 compute (38.71 TFLOPS vs 35.48 TFLOPS). These specifications suggest the A6000 would be better suited for workloads that are not represented in the benchmark suite—particularly those requiring very large memory capacities or maximum raw bandwidth. The 4070 SUPER, conversely, wins on every measured performance metric while consuming less power.

The Verdict

The data presents a clear picture: the RTX 4070 SUPER outperforms the RTX A6000 in every benchmark test recorded, with margins from 6.5% to 30%. The newer card achieves this with fewer cores, less memory, and lower power consumption, thanks to its 5 nm TSMC process and higher clock speeds. The A6000's 48 GB memory capacity and superior peak compute specifications do not translate into any measured performance win.

For users prioritizing raw benchmark performance in the tested categories—DirectX, OpenCL, Vulkan, 2D, 3D, and compute—the RTX 4070 SUPER is the clear choice. It delivers identical aggregate scores to the A6000 (42576 vs 42653, a 0.2% difference) while winning all head-to-head comparisons. The A6000 retains appeal only for scenarios requiring its unique hardware features: 48 GB of VRAM and 768.0 GB/s of bandwidth, which could benefit workloads like large model inference or high-resolution texture datasets that are not covered by these benchmarks. However, for any workload represented in the available data, the RTX 4070 SUPER is the superior performer. The aggregate scores may be close, but the head-to-head results are not.