NVIDIA GeForce RTX 4070 vs NVIDIA GeForce RTX 4070 SUPER Comparison

FAQ

Q: Which card is faster in the 3DMark Steel Nomad DX12 test?

A: The NVIDIA GeForce RTX 4070 SUPER scores 4627, which is 20.1% higher than the RTX 4070's score of 3854.

Q: How do the two cards compare in average benchmark score?

A: The RTX 4070 SUPER has an average benchmark score of 43223, while the RTX 4070 averages 37648. This places the SUPER at the 83rd percentile of all GPUs, versus the 81st percentile for the RTX 4070.

Q: Do both cards use the same memory configuration?

A: Yes, both cards feature 12 GB of GDDR6X memory on a 192-bit bus with 504.2 GB/s of bandwidth.

Q: What is the largest performance gap between the two cards?

A: The largest gap appears in both 3DMark Steel Nomad DX12 and Passmark DirectX 10, where the RTX 4070 SUPER leads by 20.1% in each test.

Q: Which card has higher compute performance?

A: The RTX 4070 SUPER achieves 17108 in Passmark GPU Compute, which is 16.2% ahead of the RTX 4070's 14720.

Q: Are the physical dimensions identical?

A: No, the RTX 4070 SUPER is longer at 267 mm (10.5 inches) versus 240 mm (9.4 inches) for the RTX 4070, and slightly taller and thicker as well.

Architecture Differences

Both cards are built on the same AD104 chip using the Ada Lovelace architecture, fabricated on TSMC's 5 nm process. The die size is identical at 294 mm², and both pack 35,800 million transistors, giving the same transistor density of 121.8M per mm². The architectural foundation is therefore shared, but NVIDIA has configured the two chips differently.

The RTX 4070 SUPER enables more of the AD104 die. It has 7168 shading units, 224 texture mapping units, and 80 ROPs. The RTX 4070, by comparison, has 5888 shading units, 184 TMUs, and 64 ROPs. This represents a meaningful increase in raw execution resources for the SUPER variant.

Ray tracing and tensor core counts follow the same pattern. The RTX 4070 SUPER has 56 RT cores and 224 tensor cores, while the RTX 4070 has 46 RT cores and 184 tensor cores. The SUPER therefore carries roughly 22% more RT cores and tensor cores than the base card.

Clock behavior is mixed. The base clock of the RTX 4070 SUPER is 1980 MHz, which is 60 MHz higher than the RTX 4070's 1920 MHz. However, both cards share an identical boost clock of 2475 MHz. Memory clocks are also identical at 1313 MHz, translating to 21 Gbps effective.

The result is a substantial difference in theoretical throughput. The RTX 4070 SUPER reaches 35.48 TFLOPS FP32, while the RTX 4070 reaches 29.15 TFLOPS FP32, a gap of about 21.7%. FP16 performance scales identically at a 1:1 ratio, with the SUPER at 35.48 TFLOPS and the RTX 4070 at 29.15 TFLOPS. Fill rates also differ: the SUPER produces 198.0 GPixel/s and 554.4 GTexel/s, versus 158.4 GPixel/s and 455.4 GTexel/s for the RTX 4070.

Power consumption rises accordingly. The RTX 4070 SUPER has a TDP of 220 W, while the RTX 4070 is rated at 200 W. Both cards use a single 16-pin power connector and a suggested PSU of 550 W. Neither card has a listed game clock, and both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Head-to-Head Benchmarks

The RTX 4070 SUPER wins all ten recorded head-to-head benchmarks, but the margin varies significantly by workload. The most decisive results come in DX12 and DX10 legacy tests, where the SUPER leads by 20.1% in both 3DMark Steel Nomad DX12 and Passmark DirectX 10. In 3DMark Steel Nomad, the SUPER scores 4627 versus 3854; in Passmark DX10, it scores 167 versus 139.

Vulkan performance also shows a wide gap. The SUPER scores 205624 in Geekbench Vulkan, which is 18.1% ahead of the RTX 4070's 174152. Compute workloads are similarly favorable to the SUPER, with a 16.2% lead in Passmark GPU Compute (17108 versus 14720).

OpenCL results are closer but still decisive. The SUPER records 172795 in Geekbench OpenCL, an 11.6% improvement over the RTX 4070's 154858. Passmark G3D shows an 11.4% lead, with the SUPER at 29995 and the RTX 4070 at 26927.

Passmark DX11 and DX9 results narrow the gap further. In DX11, the SUPER leads by 11.9% (273 versus 244), and in DX9 the lead is 7.5% (344 versus 320). The smallest margin anywhere is in Passmark DX12, where the SUPER wins by only 6.8% (110 versus 103). Passmark G2D, a 2D graphics test, is nearly identical: the SUPER scores 1184 versus 1164, a 1.7% edge.

The average benchmark scores reinforce the trend. The SUPER averages 43223 across all recorded tests, while the RTX 4070 averages 37648. In the database's nearest rival comparison, the SUPER sits just 0.1% below the NVIDIA Quadro M6000 24 GB and 1% below the NVIDIA GeForce RTX 4090 Mobile. The RTX 4070, by contrast, sits 0.1% above the NVIDIA Tesla P4 and 1.3% below the NVIDIA GeForce RTX 4080 Mobile. The performance class separation is clear: the SUPER competes at a higher tier than the RTX 4070.

Specification Differences

The two cards diverge most clearly in their execution resources. The RTX 4070 SUPER has 7168 shading units versus 5888 for the RTX 4070. TMUs are 224 against 184, and ROPs are 80 against 64. RT cores number 56 versus 46, and tensor cores number 224 versus 184.

Clock specifications differ only at the base level. The SUPER boosts from 1980 MHz, while the RTX 4070 boosts from 1920 MHz; both top out at 2475 MHz boost. Memory clocks are identical at 1313 MHz (21 Gbps effective).

Throughput figures all favor the SUPER. Pixel rate is 198.0 GPixel/s versus 158.4 GPixel/s. Texture rate is 554.4 GTexel/s versus 455.4 GTexel/s. FP32 and FP16 both measure 35.48 TFLOPS versus 29.15 TFLOPS.

Power and physical specs also differ. The SUPER draws 220 W TDP versus 200 W. Its dimensions are 267 mm by 112 mm by 42 mm, while the RTX 4070 measures 240 mm by 110 mm by 40 mm. Both cards are dual-slot, use a 16-pin connector, and require a 550 W PSU. Both use the same 12 GB GDDR6X memory on a 192-bit bus with 504.2 GB/s bandwidth. Both are end-of-life products with a launch MSRP of 599 USD.

The Verdict

The benchmark data is unambiguous: the RTX 4070 SUPER outperforms the RTX 4070 in every recorded test. It wins all ten head-to-head comparisons, with margins ranging from 1.7% in Passmark G2D to 20.1% in 3DMark Steel Nomad and Passmark DX10. The average score difference is substantial, with the SUPER at 43223 versus 37648 for the RTX 4070.

The SUPER also places higher in the overall GPU percentile ranking, at the 83rd percentile compared to the 81st. This places it in a different competitive bracket among nearest rivals: the SUPER is within 1% of the RTX 4090 Mobile, while the RTX 4070 sits near the RTX 4080 Mobile and AMD Radeon RX Vega 56. The performance tier separation is consistent.

The RTX 4070's advantages are limited to power draw and physical size. It consumes 20 W less power and is 27 mm shorter, 2 mm less tall, and 2 mm thinner. For buyers constrained by small cases or strict power budgets, those differences may matter. But in raw performance, the RTX 4070 SUPER is the clear choice.

Both cards share the same launch MSRP of 599 USD, the same memory configuration, and identical boost clocks. The SUPER achieves its lead through more enabled silicon on the same AD104 die, not through higher clocks. The data indicates the RTX 4070 SUPER is the stronger product in nearly every measurable way.

Where Each One Wins

The RTX 4070 SUPER dominates across all benchmark categories. Its largest wins come in DX12-heavy workloads, where the 20.1% lead in 3DMark Steel Nomad DX12 indicates strong modern gaming performance. The 20.1% lead in Passmark DX10 and the 18.1% lead in Geekbench Vulkan suggest the SUPER also handles older APIs and Vulkan titles with a clear edge.

Compute-oriented users benefit most from the SUPER. The 16.2% lead in Passmark GPU Compute and the 11.6% lead in Geekbench OpenCL make it the better option for GPGPU workloads, rendering, and any task that stresses tensor or shading throughput. The higher RT core and tensor core counts support this interpretation.

The RTX 4070 retains a niche in compact builds. It is shorter by 27 mm, narrower by 2 mm, and thinner by 2 mm, which can ease installation in small form factor cases. Its 200 W TDP also makes it slightly less demanding on cooling, though both cards require a 550 W PSU. In 2D workloads, the gap is negligible: the SUPER leads by only 1.7% in Passmark G2D, so users focused on desktop or 2D tasks would see little difference.

For gaming, the SUPER wins every API test, from DX9 to DX12, with particular strength in DX12 and Vulkan. For compute, the SUPER wins decisively. The only scenario where the RTX 4070 makes sense is when physical size or the 20 W power difference is the primary constraint. Otherwise, the recorded data consistently favors the RTX 4070 SUPER.