NVIDIA GeForce GTX 970 vs NVIDIA RTX PRO 5000 72 GB Blackwell Comparison

NVIDIA
GEFORCE

NVIDIA GeForce GTX 970

CORE STATE GM204
VRAM 4 GB
CLOCK SPEED 1178 MHz
TDP 148 W
BUS WIDTH 256 bit
ARCHITECTURE Maxwell 2.0
nm
PROCESS 28 nm
LAUNCH DATE 2014
VS
NVIDIA
GEFORCE

RTX PRO 5000 72 GB Blackwell

CORE STATE GB202
VRAM 72 GB
CLOCK SPEED 2377 MHz
TDP 300 W
BUS WIDTH 384 bit
ARCHITECTURE Blackwell 2.0
nm
PROCESS 5 nm
LAUNCH DATE 2025

PERFORMANCE BENCHMARKS

3dmark_3dmark_steel_nomad_dx12
369
9,579.5
geekbench_metal
13,595
N/A
geekbench_opencl
29,982
N/A
geekbench_vulkan
20,005
N/A
passmark_directx_10
46
175
passmark_directx_11
71
219
passmark_directx_12
41
78
passmark_directx_9
143
311
passmark_g2d
764
914
passmark_g3d
9,638
24,310
passmark_gpu_compute
4,073
15,667

Analysis: NVIDIA GeForce GTX 970 vs NVIDIA RTX PRO 5000 72 GB Blackwell

The NVIDIA GeForce GTX 970 and the NVIDIA RTX PRO 5000 72 GB Blackwell represent two extremes of the GPU timeline, separated by over a decade of architectural evolution. The GTX 970, a Maxwell 2.0 product from 2014, is a legacy piece of hardware now at end-of-life, while the RTX PRO 5000 is an active, professional-grade Blackwell 2.0 part from late 2025. The benchmark data shows a complete sweep: the RTX PRO 5000 wins all eight head-to-head tests, with decisive margins in every category. Despite this, each card has a distinct purpose, and the data reveals where the older card still holds a niche and where the newer one simply dominates.

Where Each One Wins

The GTX 970 does not win a single benchmark in the head-to-head comparison, yet its profile suggests it was designed for a different era of gaming. Its strongest relative performance comes in the legacy DirectX 9 test, where it scores 143 points, compared to its scores of 46 in DirectX 10, 71 in DirectX 11, and 41 in DirectX 12. This indicates the card is best suited for running older titles and lightweight esports games from its launch period, where its 4 GB of GDDR5 memory and 224.4 GB/s bandwidth were sufficient for 1080p gameplay. The GTX 970 also holds a 39th percentile rank among all GPUs, placing it near the bottom of modern performance charts, but its nearest rivals include the AMD Radeon Vega 8 Mobile (deltaPct -0.6) and the NVIDIA GeForce GTX 750 (deltaPct -0.9), showing it remains competitive with low-end integrated and entry-level discrete solutions.

The RTX PRO 5000 72 GB Blackwell wins every category, but its most significant victories are in compute and modern graphics workloads. The card scores 15,667 in Passmark GPU Compute, which is 74% higher than the GTX 970’s 4,073, and it achieves 9579.5 in 3DMark Steel Nomad DX12, a 96.1% improvement over the older card’s 369. This makes it the clear choice for professional workloads like rendering, AI inference, and high-resolution content creation, where its 72 GB of GDDR7 memory and 66.94 TFLOPS of FP32 performance are essential. Its 37th percentile rank in the global database is misleading—this is a niche professional card, not a mainstream gaming product—and its nearest rivals include the NVIDIA GeForce GTX 580M (deltaPct 0.3) and the AMD Radeon Vega 10 Mobile (deltaPct -1.1), which are vastly inferior in absolute terms.

Architecture Differences

The architectural gap between these two cards is enormous, starting with the manufacturing process. The GTX 970 uses a 28 nm node at TSMC, packing 5,200 million transistors into a 398 mm² die, yielding a transistor density of 13.1 million per mm². The RTX PRO 5000 uses a 5 nm node, also at TSMC, but crams 92,200 million transistors into a 750 mm² die—a density of 122.9 million per mm², which is nearly ten times higher. This explains the massive performance disparity: the newer card has more than 17 times the transistors, enabling far more complex compute units.

The GTX 970 is built on Maxwell 2.0 with the GM204 chip, featuring 1,664 shading units, 104 texture mapping units, and 56 raster output units. It has no dedicated ray tracing or tensor cores, as those technologies did not exist in consumer hardware at the time. Its FP32 performance is 3.920 TFLOPS, and it has no listed FP16 capability. The RTX PRO 5000, by contrast, uses the GB202 chip on Blackwell 2.0, with 14,080 shading units, 440 TMUs, and 160 ROPs. It includes 110 ray tracing cores and 440 tensor cores, which are essential for modern graphics features like real-time ray tracing and DLSS-style AI acceleration. Its FP32 output is 66.94 TFLOPS, and it matches that exactly in FP16 with a 1:1 ratio, making it a monster for mixed-precision workloads.

Memory architecture also differs fundamentally. The GTX 970 uses 4 GB of GDDR5 on a 256-bit bus, delivering 224.4 GB/s of bandwidth at 7 Gbps effective. The RTX PRO 5000 uses 72 GB of GDDR7 on a 384-bit bus, delivering 1.34 TB/s at 28 Gbps effective—a sixfold increase in capacity and a sixfold increase in bandwidth. The RTX PRO 5000 also supports DirectX 12 Ultimate (12_2), while the GTX 970 is limited to DirectX 12 (12_1), which means the newer card supports hardware features like mesh shaders and variable rate shading that the older one cannot handle.

Head-to-Head Benchmarks

The most lopsided result is in 3DMark Steel Nomad DX12, where the RTX PRO 5000 scores 9579.5 against the GTX 970’s 369, a delta of -96.1%. This is a modern, demanding test that stresses ray tracing and DX12 Ultimate features, which the GTX 970 simply cannot execute. In Passmark G3D, the RTX PRO 5000 scores 24,310 versus 9,638, a 60.4% advantage, reflecting its superior rasterization performance in general 3D workloads.

Compute performance shows a similar gap. The RTX PRO 5000’s 15,667 in Passmark GPU Compute is 74% higher than the GTX 970’s 4,073. This is where the tensor cores and massive shading unit count come into play, making the newer card suitable for scientific simulation and machine learning tasks. In legacy DirectX benchmarks, the margins narrow but remain decisive. In DirectX 9, the RTX PRO 5000 scores 311 versus 143, a 54% lead; in DirectX 10, it scores 175 versus 46, a 73.7% lead; and in DirectX 11, it scores 219 versus 71, a 67.6% lead. Even in the 2D test, Passmark G2D, the RTX PRO 5000 wins with 914 versus 764, a 16.4% margin, suggesting better memory bandwidth for desktop compositing and 2D operations.

The smaller deltas in older APIs indicate that the GTX 970’s Maxwell architecture was well-optimized for the DirectX 9 and 11 era, but it simply cannot scale to modern requirements. The RTX PRO 5000’s lowest win margin is in DirectX 12, where it scores 78 versus 41, a 47.4% lead—still a massive gap, but the closest the two cards get in any test. This suggests that the GTX 970’s limited DX12 implementation (12_1) is its weakest point relative to the newer card’s full 12_2 feature set.

Specification Differences

The two cards differ on nearly every specification field. The GTX 970 has a base clock of 1050 MHz and a boost clock of 1178 MHz, while the RTX PRO 5000 runs at 1740 MHz base and 2377 MHz boost—a 65% higher base clock and 101% higher boost clock. Memory is the most dramatic difference: 4 GB GDDR5 versus 72 GB GDDR7, with bus widths of 256-bit versus 384-bit and bandwidth of 224.4 GB/s versus 1.34 TB/s. The shading unit count jumps from 1,664 to 14,080, TMUs from 104 to 440, and ROPs from 56 to 160. Pixel rate goes from 65.97 GPixel/s to 380.3 GPixel/s, and texture rate from 122.5 GTexel/s to 1,045.9 GTexel/s. FP32 performance rises from 3.920 TFLOPS to 66.94 TFLOPS.

Power draw and connectivity also change. The GTX 970 uses a 148 W TDP with two 6-pin power connectors and a 300 W suggested PSU, while the RTX PRO 5000 uses a 300 W TDP with a single 16-pin connector and a 700 W suggested PSU. The bus interface advances from PCIe 3.0 x16 to PCIe 5.0 x16. Display outputs shift from 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.2 to 4x DisplayPort 2.1b. Both cards share identical physical dimensions (267 mm length, 111 mm height, 40 mm width) and dual-slot cooling, and both support OpenGL 4.6 and Vulkan 1.4, but the RTX PRO 5000 adds DirectX 12 Ultimate. The GTX 970 was released on September 18, 2014, with a launch MSRP of 329 USD, while the RTX PRO 5000 launched October 20, 2025, with no listed MSRP.

FAQ

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

A: The NVIDIA RTX PRO 5000 72 GB Blackwell is dramatically faster, scoring 9579.5 compared to the GTX 970’s 369, a delta of -96.1%.

Q: Does the GTX 970 support ray tracing?

A: No, the GTX 970 has no ray tracing cores. It is based on Maxwell 2.0 and only supports DirectX 12 (12_1), whereas the RTX PRO 5000 has 110 RT cores and supports DirectX 12 Ultimate.

Q: What is the memory capacity difference?

A: The GTX 970 has 4 GB of GDDR5, while the RTX PRO 5000 has 72 GB of GDDR7. The RTX PRO 5000 also has a 384-bit bus and 1.34 TB/s bandwidth versus the GTX 970’s 256-bit bus and 224.4 GB/s.

Q: How do the two cards compare in compute performance?

A: The RTX PRO 5000 scores 15,667 in Passmark GPU Compute, which is 74% higher than the GTX 970’s 4,073, thanks to its 440 tensor cores and 66.94 TFLOPS FP32 output.

Q: Are their power requirements similar?

A: No, the GTX 970 has a 148 W TDP and requires a 300 W PSU, while the RTX PRO 5000 has a 300 W TDP and requires a 700 W PSU. The RTX PRO 5000 uses a single 16-pin connector versus the GTX 970’s two 6-pin connectors.

Q: Do both cards support the same display outputs?

A: No, the GTX 970 has 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.2, while the RTX PRO 5000 has 4x DisplayPort 2.1b.

The Verdict

The data is unambiguous: the RTX PRO 5000 72 GB Blackwell wins every single benchmark, from 2D operations to modern DX12 tests. For any professional user—engineers, 3D artists, data scientists—the choice is clear. The 72 GB memory capacity and 1.34 TB/s bandwidth are essential for large datasets, and the 66.94 TFLOPS FP32 performance handles complex simulations. The 110 RT cores and 440 tensor cores enable real-time ray tracing and AI acceleration, features the GTX 970 lacks entirely.

The GTX 970, however, is not without a purpose. Its 39th percentile ranking and closest rival being the AMD Radeon Vega 8 Mobile (-0.6% deltaPct) suggest it is still viable for very light, legacy workloads. For someone running Windows 7-era software or playing DirectX 9 titles, the GTX 970’s 143 score in that test is respectable. Its 148 W TDP and 300 W PSU requirement make it far easier to integrate into older systems, and its 4 GB memory is adequate for 1080p gaming from its era.

There is no scenario where the GTX 970 outperforms the RTX PRO 5000. The only question is whether you need the RTX PRO 5000’s capabilities at all. If your workload is strictly legacy gaming or basic office tasks, the GTX 970 remains functional. If you need modern rendering, AI, or high-resolution compute, the RTX PRO 5000 is the only option that the benchmark data supports, despite its higher power draw and newer PCIe 5.0 interface requirements. The 11 years between their release dates are evident in every metric, and the RTX PRO 5000 is the definitive choice for any contemporary demanding task.

DETAILED SPECIFICATIONS

SPECIFICATION
GTX 970
RTX PRO 5000 72 GB Blackwell
Core Specs
Shading Units
1,664
14,080 +746.2%
Shaders
1,664
14,080 +746.2%
TMUs
104
440 +323.1%
ROPs
56
160 +185.7%
SM Count
—
110
Clocks
Base Clock
1050 MHz
1740 MHz
Boost Clock
1178 MHz
2377 MHz
Memory Clock
1753 MHz 7 Gbps effective
1750 MHz 28 Gbps effective
Memory
Memory Size
4 GB
72 GB
VRAM (MB)
4,096
73,728 +1700.0%
Memory Type
GDDR5
GDDR7
Memory Bus
256 bit
384 bit
Bandwidth
224.4 GB/s
1.34 TB/s
Cache
L1 Cache
48 KB (per SMM)
128 KB (per SM)
L2 Cache
2 MB
96 MB
Performance
Pixel Rate
65.97 GPixel/s
380.3 GPixel/s
Texture Rate
122.5 GTexel/s
1,045.9 GTexel/s
FP32 (TFLOPS)
3.920 TFLOPS
66.94 TFLOPS
FP64 (TFLOPS)
122.5 GFLOPS (1:32)
1,045.9 GFLOPS (1:64)
FP16 (TFLOPS)
—
66.94 TFLOPS (1:1)
AI/RT
RT Cores
—
110
Tensor Cores
—
440
Power
TDP
148 W
300 W
TDP (W)
148
300 +102.7%
Suggested PSU
300 W
700 W
Power Connectors
2x 6-pin
1x 16-pin
Architecture
Architecture
Maxwell 2.0
Blackwell 2.0
GPU Name
GM204
GB202
Generation
GeForce 900
Blackwell PRO W (x000)
Process Size
28 nm
5 nm
Transistors
5,200 million
92,200 million
Die Size
398 mm²
750 mm²
Foundry
TSMC
TSMC
Density
13.1M / mm²
122.9M / mm²
API Support
DirectX
12 (12_1)
12 Ultimate (12_2)
OpenGL
4.6
4.6
Vulkan
1.4
1.4
OpenCL
3.0
3.0
CUDA
5.2
12.0
Shader Model
6.8
6.9
Physical
Slot Width
Dual-slot
Dual-slot
Length
267 mm 10.5 inches
267 mm 10.5 inches
Height
111 mm 4.4 inches
111 mm 4.4 inches
Outputs
1x DVI1x HDMI 2.03x DisplayPort 1.2
4x DisplayPort 2.1b
Bus Interface
PCIe 3.0 x16
PCIe 5.0 x16
Other
Launch Price
329 USD
—
Production
End-of-life
Active
Predecessor
GeForce 700
Workstation Ada
Successor
GeForce 10
—
View GeForce GTX 970 Details View RTX PRO 5000 72 GB Blackwell Details