Intel UHD Graphics P750 vs NVIDIA RTX PRO 5000 72 GB Blackwell Comparison

Intel
GPU

Intel UHD Graphics P750

CORE STATE Rocket Lake
VRAM System Shared
CLOCK SPEED 1300 MHz
TDP 15 W
BUS WIDTH System Shared
ARCHITECTURE Generation 12.1
nm
PROCESS 14 nm+++
LAUNCH DATE —
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

geekbench_opencl
6,554
N/A
3dmark_3dmark_steel_nomad_dx12
N/A
9,579.5
passmark_directx_10
N/A
175
passmark_directx_11
N/A
219
passmark_directx_12
N/A
78
passmark_directx_9
N/A
311
passmark_g2d
N/A
914
passmark_g3d
N/A
24,310
passmark_gpu_compute
N/A
15,667

Analysis: Intel UHD Graphics P750 vs NVIDIA RTX PRO 5000 72 GB Blackwell

# Intel UHD Graphics P750 vs NVIDIA RTX PRO 5000 72 GB Blackwell

The Intel UHD Graphics P750 is an integrated graphics solution built into Rocket Lake processors, using Intel's Generation 12.1 architecture on a 14 nm+++ process, with a 15 W TDP and system-shared memory. The NVIDIA RTX PRO 5000 72 GB Blackwell is a discrete workstation card based on the GB202 chip, built on TSMC's 5 nm process, featuring 72 GB of GDDR7 memory on a 384-bit bus with 1.34 TB/s bandwidth, and drawing up to 300 W. The data shows these are fundamentally different classes of hardware: the P750 targets basic display and light compute, while the RTX PRO 5000 targets professional rendering and AI workloads. Their average benchmark scores are surprisingly close — 6554 for the P750 versus 6407 for the RTX PRO 5000 — but this is misleading, as the P750's single OpenCL score and the RTX PRO's mixed DirectX and compute scores measure different capabilities entirely.

FAQ

Q: What is the core difference between these two GPUs?

A: The Intel UHD Graphics P750 is an integrated GPU (IGP) with 256 shading units, 64 TMUs, and 32 ROPs, sharing system memory, while the NVIDIA RTX PRO 5000 has 14,080 shading units, 440 TMUs, 160 ROPs, 110 RT cores, and 440 tensor cores, with 72 GB of dedicated GDDR7 memory.

Q: How do their raw compute performances compare?

A: The RTX PRO 5000 delivers 66.94 TFLOPS of FP32 performance versus 665.6 GFLOPS for the P750, making the RTX PRO roughly 100 times faster in raw floating-point throughput. The RTX PRO also achieves 380.3 GPixel/s pixel rate and 1,045.9 GTexel/s texture rate, compared to 41.60 GPixel/s and 83.20 GTexel/s for the P750.

Q: Which GPU has a higher average benchmark score?

A: The Intel P750 has a higher average benchmark score of 6554 compared to the RTX PRO 5000's 6407, and it sits at the 38th percentile of all GPUs versus the RTX PRO's 37th percentile. However, these averages are not directly comparable because the P750 only has one OpenCL result, while the RTX PRO has eight different test scores.

Q: What are the memory specifications for each card?

A: The P750 uses system-shared memory with system-dependent bandwidth, while the RTX PRO 5000 has 72 GB of GDDR7 on a 384-bit bus with 1.34 TB/s bandwidth and memory clocked at 1750 MHz (28 Gbps effective).

Q: What API support do they offer?

A: Both support OpenGL 4.6 and Vulkan 1.4, but the P750 supports DirectX 12 (12_1) while the RTX PRO 5000 supports DirectX 12 Ultimate (12_2), which includes hardware ray tracing and mesh shaders.

Q: What are the physical and power requirements?

A: The P750 is an IGP with no slot width or power connectors, drawing 15 W, while the RTX PRO 5000 is a dual-slot card measuring 267 mm x 111 mm x 40 mm, requiring a 1x 16-pin power connector and a 700 W suggested PSU, drawing 300 W.

Architecture Differences

The two GPUs come from entirely different architectural lineages. The Intel UHD Graphics P750 is based on Generation 12.1 architecture, built on Intel's 14 nm+++ process, and is integrated into the Rocket Lake platform. It uses a Ring Bus interface and has no dedicated RT or tensor cores. The P750's FP16 performance is 1,331.2 GFLOPS, achieved through a 2:1 ratio relative to FP32, suggesting a more limited compute pipeline. Its memory subsystem is entirely system-shared, meaning performance depends on the host system's RAM configuration.

In contrast, the NVIDIA RTX PRO 5000 uses the Blackwell 2.0 architecture on the GB202 chip, manufactured by TSMC on a 5 nm process. This is a massive chip with 92,200 million transistors on a 750 mm² die, giving a transistor density of 122.9M per mm². It features 110 dedicated RT cores for hardware-accelerated ray tracing and 440 tensor cores for AI and deep learning workloads. The FP16 performance is 66.94 TFLOPS at a 1:1 ratio with FP32, indicating full-rate half-precision compute. The memory interface is 384-bit with GDDR7, providing 1.34 TB/s of bandwidth — a critical advantage for large datasets and high-resolution textures.

The process node difference alone — 14 nm+++ versus 5 nm — explains much of the performance gap. The RTX PRO 5000's power budget of 300 W is 20 times higher than the P750's 15 W, and its PCIe 5.0 x16 interface offers far more bandwidth than the Ring Bus connection used by the integrated P750. The P750 is end-of-life production, while the RTX PRO 5000 is active and was released on October 20, 2025, succeeding the Workstation Ada generation.

Head-to-Head Benchmarks

Direct head-to-head benchmark comparisons are limited because the two GPUs run different test suites. The Intel P750 has only one benchmark result: Geekbench OpenCL with a score of 6554. The NVIDIA RTX PRO 5000 has eight results across 3DMark and Passmark tests, but its average benchmark score of 6407 is actually lower than the P750's 6554.

Looking at the RTX PRO 5000's individual scores reveals its strengths. In Passmark G3D, it scores 24310, which is a strong result for DirectX gaming workloads. Its Passmark GPU Compute score is 15667, showing solid compute performance. However, its Passmark DirectX scores are inconsistent: 311 for DirectX 9, 175 for DirectX 10, 219 for DirectX 11, and only 78 for DirectX 12. The 3DMark Steel Nomad DX12 test gives it 9579.5, but the Passmark G2D score is 914.

The P750's nearest rivals include the AMD Radeon HD 7730M (6581, -0.4% delta), NVIDIA GeForce GTX 670M (6513, +0.6%), AMD Radeon R7 M460 (6612, -0.9%), and NVIDIA GeForce GT 555M (6493, +0.9%). These are all older mobile or entry-level discrete GPUs, and the P750 is within 1% of each. The RTX PRO 5000's nearest rivals are the NVIDIA GeForce GTX 460 SE (6389, +0.3%), NVIDIA GeForce GTX 580M (6389, +0.3%), AMD Radeon Vega 10 Mobile (6476, -1.1%), and NVIDIA Quadro M5000M (6481, -1.1%). These are also older or mobile parts, and the RTX PRO is within 1.1% of each.

The data shows that in terms of average benchmark scores, the P750 edges out the RTX PRO 5000 by 147 points (2.3%). But this is a classic case where averages hide the real story. The P750's single OpenCL score reflects integrated graphics performance, while the RTX PRO's scores include diverse tests that measure different aspects of GPU capability. The RTX PRO's Passmark G3D score of 24310 is 3.7 times higher than the P750's OpenCL score, while its DirectX 9 score of 311 is far lower, suggesting driver-level differences in legacy API support.

The Verdict

The data clearly indicates that these GPUs serve completely different purposes. The Intel UHD Graphics P750 is an integrated solution for basic display output, light productivity, and minimal compute tasks, with a 15 W power envelope and no dedicated VRAM. The NVIDIA RTX PRO 5000 is a professional workstation GPU with 72 GB of GDDR7, hardware ray tracing, tensor cores, and 66.94 TFLOPS of FP32 performance — a 100-fold compute advantage over the P750.

Who should pick the P750? Users who need a simple integrated graphics solution for office work, media playback, or as a fallback for systems without a discrete GPU. Its 256 shading units and system-shared memory are sufficient for everyday tasks, and its 15 W TDP means no additional power or cooling requirements.

Who should pick the RTX PRO 5000? Professionals working with large datasets, high-resolution 3D rendering, AI inference, or compute-intensive workloads. The 72 GB memory capacity and 1.34 TB/s bandwidth are critical for models and scenes that exceed the memory capacity of consumer GPUs. The 110 RT cores and 440 tensor cores provide hardware acceleration for ray tracing and deep learning that the P750 simply cannot match.

The average benchmark scores suggest parity, but this is an artifact of different test methodologies. The RTX PRO 5000's Passmark G3D score of 24310 versus the P750's 6554 OpenCL score tells the real story: for any modern 3D workload, the RTX PRO is overwhelmingly faster. The P750's percentile rank of 38 versus the RTX PRO's 37 reflects the different benchmark suites rather than true performance equivalence.

Specification Differences

| Specification | Intel UHD Graphics P750 | NVIDIA RTX PRO 5000 72 GB Blackwell |

|---|---|---|

| Architecture | Generation 12.1 | Blackwell 2.0 |

| Process Node | 14 nm+++ (Intel) | 5 nm (TSMC) |

| Transistors | Not specified | 92,200 million |

| Die Size | Not specified | 750 mm² |

| Base Clock | 350 MHz | 1740 MHz |

| Boost Clock | 1300 MHz | 2377 MHz |

| Memory Size | System Shared | 72 GB GDDR7 |

| Memory Bus | System Shared | 384 bit |

| Memory Bandwidth | System Dependent | 1.34 TB/s |

| Shading Units | 256 | 14080 |

| TMUs | 64 | 440 |

| ROPs | 32 | 160 |

| RT Cores | 0 | 110 |

| Tensor Cores | 0 | 440 |

| FP32 Performance | 665.6 GFLOPS | 66.94 TFLOPS |

| FP16 Performance | 1,331.2 GFLOPS (2:1) | 66.94 TFLOPS (1:1) |

| Pixel Rate | 41.60 GPixel/s | 380.3 GPixel/s |

| Texture Rate | 83.20 GTexel/s | 1,045.9 GTexel/s |

| TDP | 15 W | 300 W |

| Bus Interface | Ring Bus | PCIe 5.0 x16 |

| Power Connectors | None | 1x 16-pin |

| Suggested PSU | Not specified | 700 W |

| Slot Width | IGP | Dual-slot |

| Dimensions | Not specified | 267 mm x 111 mm x 40 mm |

| Display Outputs | Motherboard Dependent | 4x DisplayPort 2.1b |

| DirectX Support | 12 (12_1) | 12 Ultimate (12_2) |

| Production Status | End-of-life | Active |

| Release Date | Not specified | 2025-10-20 |

Where Each One Wins

Intel UHD Graphics P750 wins on: Power efficiency, with a 15 W TDP versus 300 W for the RTX PRO 5000 — a 20-fold advantage. It also has a higher average benchmark score (6554 vs 6407) and a better percentile ranking (38th vs 37th). For users with no discrete GPU slot or power budget, the P750's integrated nature is a decisive advantage. Its nearest rivals are all within 1% of its score, showing it competes with older mobile discrete GPUs like the GTX 670M and GT 555M.

NVIDIA RTX PRO 5000 72 GB Blackwell wins on: Raw performance across every measurable metric. It has 55 times more shading units (14080 vs 256), 6.9 times more TMUs (440 vs 64), and 5 times more ROPs (160 vs 32). Its FP32 compute is 100.6 times higher (66.94 TFLOPS vs 665.6 GFLOPS), and its pixel rate is 9.1 times higher (380.3 vs 41.60 GPixel/s). The 72 GB of GDDR7 memory with 1.34 TB/s bandwidth is a category-defining feature for large-scale professional workloads. The 110 RT cores and 440 tensor cores enable hardware-accelerated ray tracing and AI features that are entirely absent on the P750. The RTX PRO also wins on memory bandwidth (1.34 TB/s vs system-dependent), clock speeds (1740 MHz base and 2377 MHz boost vs 350 MHz and 1300 MHz), and API support with DirectX 12 Ultimate versus DirectX 12 (12_1). Its Passmark G3D score of 24310 demonstrates dominance in modern 3D rendering workloads.

Practical split: Choose the P750 for basic desktops, office machines, or low-power builds where integrated graphics suffice. Choose the RTX PRO 5000 for professional 3D rendering, video editing, AI model training, scientific computing, or any workload that benefits from 72 GB of VRAM and hardware-accelerated ray tracing. The data shows no scenario where the P750 outperforms the RTX PRO in actual 3D compute — the RTX PRO's 66.94 TFLOPS FP32 versus the P750's 0.6656 TFLOPS is the definitive stat for any compute-heavy use case.

DETAILED SPECIFICATIONS

SPECIFICATION
UHD Graphics P750
RTX PRO 5000 72 GB Blackwell
Core Specs
Shading Units
256
14,080 +5400.0%
Shaders
256
14,080 +5400.0%
TMUs
64
440 +587.5%
ROPs
32
160 +400.0%
SM Count
—
110
Execution Units
32
—
Clocks
Base Clock
350 MHz
1740 MHz
Boost Clock
1300 MHz
2377 MHz
Memory Clock
System Shared
1750 MHz 28 Gbps effective
Memory
Memory Size
System Shared
72 GB
VRAM (MB)
—
73,728
Memory Type
System Shared
GDDR7
Memory Bus
System Shared
384 bit
Bandwidth
System Dependent
1.34 TB/s
Cache
L1 Cache
—
128 KB (per SM)
L2 Cache
—
96 MB
Performance
Pixel Rate
41.60 GPixel/s
380.3 GPixel/s
Texture Rate
83.20 GTexel/s
1,045.9 GTexel/s
FP32 (TFLOPS)
665.6 GFLOPS
66.94 TFLOPS
FP64 (TFLOPS)
166.4 GFLOPS (1:4)
1,045.9 GFLOPS (1:64)
FP16 (TFLOPS)
1,331.2 GFLOPS (2:1)
66.94 TFLOPS (1:1)
AI/RT
RT Cores
—
110
Tensor Cores
—
440
Power
TDP
15 W
300 W
TDP (W)
15
300 +1900.0%
Suggested PSU
—
700 W
Power Connectors
—
1x 16-pin
Architecture
Architecture
Generation 12.1
Blackwell 2.0
GPU Name
Rocket Lake
GB202
Generation
HD Graphics-W (Rocket Lake)
Blackwell PRO W (x000)
Process Size
14 nm+++
5 nm
Transistors
—
92,200 million
Die Size
—
750 mm²
Foundry
Intel
TSMC
Density
—
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
—
12.0
Shader Model
6.6
6.9
Physical
Slot Width
IGP
Dual-slot
Length
—
267 mm 10.5 inches
Height
—
111 mm 4.4 inches
Outputs
Motherboard Dependent
4x DisplayPort 2.1b
Bus Interface
Ring Bus
PCIe 5.0 x16
Other
Production
End-of-life
Active
Predecessor
—
Workstation Ada
View UHD Graphics P750 Details View RTX PRO 5000 72 GB Blackwell Details