NVIDIA GeForce GTX 970 vs NVIDIA GeForce RTX 5070 Ti SUPER Comparison

FAQ

Q: What is the performance gap between the GTX 970 and the RTX 5070 Ti SUPER in the recorded database?

A: In the single shared benchmark, 3DMark Steel Nomad DX12, the RTX 5070 Ti SUPER scores 6269.5, while the GTX 970 scores 369. The delta is recorded as -94.1%, meaning the GTX 970 is over 94% behind the newer card in this test.

Q: How do their memory configurations compare?

A: The GTX 970 uses 4 GB of GDDR5 on a 256-bit bus, delivering 224.4 GB/s of bandwidth. The RTX 5070 Ti SUPER uses 16 GB of GDDR7 on the same 256-bit bus, but delivers 896.0 GB/s, which is four times the bandwidth of the older card.

Q: Which card has more shading units and texture units?

A: The RTX 5070 Ti SUPER has 8960 shading units and 280 texture mapping units. The GTX 970 has 1664 shading units and 104 texture mapping units. The newer card has more than five times the shading units and nearly three times the texture units.

Q: Are both cards dual-slot designs?

A: Yes, both the GTX 970 and the RTX 5070 Ti SUPER are listed as dual-slot cards. However, the RTX 5070 Ti SUPER is physically larger: 304 mm long versus 267 mm, and 48 mm wide versus 40 mm.

Q: What is the process node difference between these two GPUs?

A: The GTX 970 is built on a 28 nm process at TSMC, while the RTX 5070 Ti SUPER is on a 5 nm process, also at TSMC. The transistor density jumps from 13.1M per mm² on the older chip to 120.6M per mm² on the newer one.

Q: Do these cards support the same DirectX version?

A: No. The GTX 970 supports DirectX 12 (12_1), while the RTX 5070 Ti SUPER supports DirectX 12 Ultimate (12_2). Both support OpenGL 4.6 and Vulkan 1.4.

Architecture Differences

The two GPUs come from completely different architectural generations. The GTX 970 uses the GM204 chip based on Maxwell 2.0 architecture, part of the GeForce 900 generation. The RTX 5070 Ti SUPER uses the GB203 chip based on Blackwell 2.0 architecture, part of the GeForce 50-series.

The manufacturing process is a major differentiator. The GTX 970 uses a 28 nm process at TSMC, while the RTX 5070 Ti SUPER uses a 5 nm process at the same foundry. This directly impacts transistor counts: the GTX 970 packs 5,200 million transistors on a 398 mm² die, while the RTX 5070 Ti SUPER packs 45,600 million transistors on a slightly smaller 378 mm² die. The transistor density per square millimeter is 13.1M for the older chip versus 120.6M for the newer one.

The RTX 5070 Ti SUPER introduces dedicated hardware that the GTX 970 lacks entirely: 70 ray tracing cores and 280 tensor cores. These are absent from the Maxwell-based card. The shading unit count also scales dramatically, from 1664 on the GTX 970 to 8960 on the RTX 5070 Ti SUPER. Texture mapping units increase from 104 to 280, and raster operations units go from 56 to 96.

Memory architecture differs as well. The GTX 970 uses GDDR5, while the RTX 5070 Ti SUPER uses GDDR7. Both use a 256-bit memory bus, but the newer memory type operates at a much higher effective speed. The bus interface also changes: the GTX 970 uses PCIe 3.0 x16, while the RTX 5070 Ti SUPER uses PCIe 5.0 x16.

Display output capabilities have been updated on the newer card. The GTX 970 has 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.2. The RTX 5070 Ti SUPER has 1x HDMI 2.1b and 3x DisplayPort 2.1b, dropping the legacy DVI connector.

Power delivery is also a fundamental difference. The GTX 970 uses 2x 6-pin power connectors and has a 148 W TDP. The RTX 5070 Ti SUPER uses a single 16-pin connector and has a 350 W TDP. The older card has a suggested power supply of 300 W, while the newer card does not list a suggested PSU in the database.

Head-to-Head Benchmarks

The database contains one directly comparable benchmark between these two cards: 3DMark Steel Nomad DX12. This is a demanding modern test that stresses current-generation features.

In this test, the GTX 970 scores 369 points. The RTX 5070 Ti SUPER scores 6269.5 points. The delta percentage is recorded as -94.1%, which indicates the GTX 970 trails the newer card by that margin. This is not a close contest. The RTX 5070 Ti SUPER outperforms the GTX 970 by a factor of roughly 17 in this specific workload.

Looking at the broader benchmark picture, the GTX 970 has a much wider set of recorded results. Its average benchmark score across all tests in the database is 7157. The RTX 5070 Ti SUPER has an average benchmark score of 6270, but this is based on only the single Steel Nomad result, which is an extremely demanding test. The GTX 970's average is pulled up by older, less demanding tests like Passmark DirectX 9 where it scores 143, and Passmark G2D where it scores 764.

The GTX 970's best recorded scores come from compute-oriented tests. It scores 29982 in Geekbench OpenCL, 20005 in Geekbench Vulkan, and 13595 in Geekbench Metal. Its Passmark G3D score is 9638, and its GPU compute score is 4073. These numbers reflect the card's age and its performance profile in legacy workloads.

The RTX 5070 Ti SUPER's position in the database percentile is 36, while the GTX 970 sits at 39. This percentile is relative to all GPUs tracked in the database, and the newer card's lower percentile is heavily influenced by the fact that its only benchmark result is an extremely heavy DX12 test. The GTX 970's percentile benefits from a diverse set of test results across multiple API generations.

The nearest rivals for the GTX 970 in the database are the Intel Iris Pro Graphics P580 with an average score of 7170 and a delta of -0.2%, the NVIDIA GeForce GTX 560 SE at 7171 (-0.2%), the AMD Radeon Vega 8 Mobile at 7203 (-0.6%), and the NVIDIA GeForce GTX 750 at 7222 (-0.9%). This shows the GTX 970 sits in a cluster of older, low-to-mid-range GPUs.

For the RTX 5070 Ti SUPER, the nearest rivals are the NVIDIA GeForce RTX 4070 Ti SUPER AD102 at 6270 with a 0% delta, the NVIDIA Quadro K620 at 6282 (-0.2%), the AMD FirePro W600 at 6223 (0.8%), and the AMD Radeon R7 M350 at 6327 (-0.9%). This grouping is unusual and reflects the limited benchmark data available for the newer card.

Specification Differences

The following specifications differ directly between the two cards:

• Process Node: 28 nm (GTX 970) versus 5 nm (RTX 5070 Ti SUPER)
• Transistors: 5,200 million versus 45,600 million
• Die Size: 398 mm² versus 378 mm²
• Transistor Density: 13.1M / mm² versus 120.6M / mm²
• Base Clock: 1050 MHz versus 2295 MHz
• Boost Clock: 1178 MHz versus 2452 MHz
• Memory Clock: 1753 MHz (7 Gbps effective) versus 1750 MHz (28 Gbps effective)
• Memory Size: 4 GB versus 16 GB
• Memory Type: GDDR5 versus GDDR7
• Memory Bandwidth: 224.4 GB/s versus 896.0 GB/s
• Shading Units: 1664 versus 8960
• Texture Mapping Units: 104 versus 280
• Raster Operations Units: 56 versus 96
• Ray Tracing Cores: Not present versus 70
• Tensor Cores: Not present versus 280
• Pixel Rate: 65.97 GPixel/s versus 235.4 GPixel/s
• Texture Rate: 122.5 GTexel/s versus 686.6 GTexel/s
• FP32 Performance: 3.920 TFLOPS versus 43.94 TFLOPS
• FP16 Performance: Not listed versus 43.94 TFLOPS (1:1)
• TDP: 148 W versus 350 W
• Power Connectors: 2x 6-pin versus 1x 16-pin
• Suggested PSU: 300 W versus not listed
• Bus Interface: PCIe 3.0 x16 versus PCIe 5.0 x16
• Display Outputs: 1x DVI, 1x HDMI 2.0, 3x DisplayPort 1.2 versus 1x HDMI 2.1b, 3x DisplayPort 2.1b
• DirectX Support: 12 (12_1) versus 12 Ultimate (12_2)
• Dimensions: 267 mm x 111 mm x 40 mm versus 304 mm x 137 mm x 48 mm
• Release Date: 2014-09-18 versus 2025-12-31
• Production Status: End-of-life versus Active
• Launch MSRP: 329 USD versus 749 USD

Where Each One Wins

The GTX 970 wins in the domain of legacy software compatibility. It has Passmark scores for DirectX 9 (143), DirectX 10 (46), and DirectX 11 (71), which indicate it was designed for and performs adequately in older API environments. Its Geekbench OpenCL score of 29982 shows it can handle compute workloads from that era. The card is also far more power-efficient in absolute terms, drawing only 148 W compared to 350 W, and it uses standard 6-pin power connectors rather than the newer 16-pin connector.

The GTX 970 also has a physical advantage in size. It is shorter at 267 mm versus 304 mm, and narrower at 40 mm versus 48 mm. For small form factor builds or older cases, the GTX 970 is easier to fit. Its recommended 300 W power supply is more accessible than the unspecified requirement of the newer card.

The RTX 5070 Ti SUPER wins in nearly every measurable performance metric. It has 11.2 times the FP32 throughput (43.94 TFLOPS versus 3.920 TFLOPS). Its pixel rate is 3.6 times higher (235.4 GPixel/s versus 65.97 GPixel/s), and its texture rate is 5.6 times higher (686.6 GTexel/s versus 122.5 GTexel/s). Memory bandwidth is exactly four times higher at 896.0 GB/s versus 224.4 GB/s.

The RTX 5070 Ti SUPER is the only one of the two with ray tracing cores and tensor cores. It supports DirectX 12 Ultimate, which enables hardware-accelerated ray tracing and mesh shaders. Its 16 GB of GDDR7 memory is four times the capacity of the GTX 970, making it viable for modern high-resolution textures and large datasets.

The newer card also has a much higher pixel fill rate and texture fill rate, which directly translates to better performance in modern game engines. Its 5 nm process node allows for dramatically higher transistor density, and its PCIe 5.0 interface doubles the bandwidth available for data transfer compared to the GTX 970's PCIe 3.0.

The Verdict

The data presents a clear generational divide. The GTX 970 is an end-of-life product from 2014, built on Maxwell 2.0 with 4 GB of GDDR5, and its recorded benchmark results cluster around legacy workloads. The RTX 5070 Ti SUPER is an active product from the GeForce 50-series, built on Blackwell 2.0 with 16 GB of GDDR7, and it includes dedicated ray tracing and tensor hardware.

In the single directly comparable benchmark, 3DMark Steel Nomad DX12, the RTX 5070 Ti SUPER is over 94% faster than the GTX 970. This is the definitive metric in the database for modern DX12 gaming performance. The GTX 970 cannot compete in this workload.

For a builder considering these two cards, the choice depends entirely on the target use case. The GTX 970 should only be considered for legacy systems running older DirectX 9, 10, or 11 titles, or for builds where the 148 W power draw and compact 267 mm length are hard requirements. Its Passmark DirectX 9 score of 143 and DirectX 11 score of 71 show it still has some life in older games.

The RTX 5070 Ti SUPER is the choice for any modern workload. Its 43.94 TFLOPS FP32 performance, 896.0 GB/s memory bandwidth, and 16 GB frame buffer make it suitable for current and upcoming games. The presence of 70 ray tracing cores and 280 tensor cores enables features that the GTX 970 simply cannot execute. Its 350 W TDP and larger physical footprint are the trade-offs for this performance level.

The recorded data shows no scenario where the GTX 970 wins in modern benchmarks. The RTX 5070 Ti SUPER is the superior card for contemporary gaming and compute tasks. The GTX 970 remains relevant only for those who need to maintain compatibility with older software environments or who have strict power and size constraints that the newer card cannot meet.