NVIDIA GeForce RTX 5090 vs NVIDIA Tesla V100 SXM2 16 GB Comparison

NVIDIA
GEFORCE

NVIDIA GeForce RTX 5090

CORE STATE GB202
VRAM 32 GB
CLOCK SPEED 2407 MHz
TDP 575 W
BUS WIDTH 512 bit
ARCHITECTURE Blackwell 2.0
nm
PROCESS 5 nm
LAUNCH DATE 2025
VS
NVIDIA
GEFORCE

Tesla V100 SXM2 16 GB

CORE STATE GV100
VRAM 16 GB
CLOCK SPEED 1530 MHz
TDP 250 W
BUS WIDTH 4096 bit
ARCHITECTURE Volta
nm
PROCESS 12 nm
LAUNCH DATE 2017

PERFORMANCE BENCHMARKS

3dmark_3dmark_steel_nomad_dx12
18,355
N/A
geekbench_opencl
334,370
87,454
geekbench_vulkan
376,728
141,336
passmark_directx_10
226
N/A
passmark_directx_11
341
N/A
passmark_directx_12
185
N/A
passmark_directx_9
395
N/A
passmark_g2d
1,413
N/A
passmark_g3d
39,650
N/A
passmark_gpu_compute
26,756
N/A

Analysis: NVIDIA GeForce RTX 5090 vs NVIDIA Tesla V100 SXM2 16 GB

NVIDIA’s Tesla V100 SXM2 16 GB and GeForce RTX 5090 represent two very different eras of GPU design. The V100 is an end-of-life compute accelerator built for datacenter workloads, while the RTX 5090 is an active consumer flagship aimed at gaming and high-performance desktop tasks. The benchmark data shows a decisive performance gap, but the story is more nuanced than raw scores alone.

Head-to-Head Benchmarks

The head-to-head results are unambiguous. In Geekbench OpenCL, the RTX 5090 scores 334,370 against the V100’s 87,454. That works out to a delta of -73.8% from the RTX 5090’s perspective, meaning the V100 trails by nearly three-quarters. In Geekbench Vulkan, the gap narrows slightly but remains massive: the RTX 5090 hits 376,728 while the V100 manages 141,336, a -62.5% delta. The RTX 5090 wins both available comparisons, and it does so by comfortable margins.

These are not close calls. The RTX 5090 more than triples the V100’s OpenCL output and nearly triples its Vulkan score. Even accounting for the V100’s age — it released in 2017, while the RTX 5090 arrived in 2025 — the sheer magnitude of the difference is striking. The V100’s average benchmark score sits at 114,395, which places it in the 95th percentile of all GPUs. The RTX 5090’s average score is 79,842, yet it only ranks in the 92nd percentile. This inverted percentile situation arises because the RTX 5090’s benchmark suite includes many more tests, including several low-scoring Passmark DirectX entries that drag its average down.

The RTX 5090’s individual test results tell a more complete story. Its Passmark G3D score is 39,650, and its Passmark GPU Compute score is 26,756. DirectX 9 and 11 results are 395 and 341 respectively, while DirectX 10 and 12 come in at 226 and 185. The 3DMark Steel Nomad DX12 score is 18,355. None of these tests exist for the V100, so direct comparisons are impossible, but they demonstrate the RTX 5090’s breadth across legacy and modern APIs.

Where Each One Wins

The V100 wins nowhere in the head-to-head data — it loses both available tests. However, its profile suggests strengths that the RTX 5090 cannot match. The V100 uses HBM2 memory with a 4096-bit bus, delivering 897.0 GB/s of bandwidth. The RTX 5090 uses GDDR7 on a 512-bit bus, reaching 1.79 TB/s. The V100’s memory architecture is older but was designed for sustained compute throughput, not bursty gaming workloads.

The RTX 5090 wins everywhere else. It has 104.8 TFLOPS of FP32 performance versus the V100’s 15.67 TFLOPS. Its FP16 output is 104.8 TFLOPS (1:1 ratio), while the V100 offers 31.33 TFLOPS (2:1). Texture rate favors the RTX 5090 at 1,636.8 GTexel/s versus 489.6 GTexel/s. Pixel rate is 423.6 GPixel/s against 195.8 GPixel/s. The RTX 5090 also adds 170 ray tracing cores and 680 tensor cores, while the V100 has 640 tensor cores and no RT cores at all.

In practical terms, the RTX 5090 is the choice for any workload that runs on modern APIs. The V100’s lack of display outputs means it cannot drive a monitor, while the RTX 5090 offers 1x HDMI 2.1b and 3x DisplayPort 2.1b. The V100 is a compute-only module, whereas the RTX 5090 is a fully featured consumer card.

Architecture Differences

The architectural gap is generational. The V100 is built on Volta using a 12 nm process from TSMC, packing 21,100 million transistors into an 815 mm² die. The RTX 5090 uses Blackwell 2.0 on a 5 nm process, also from TSMC, with 92,200 million transistors on a 750 mm² die. Transistor density tells the story: the V100 achieves 25.9M transistors per mm², while the RTX 5090 reaches 122.9M per mm² — nearly five times denser.

Clock speeds differ dramatically. The V100 runs at a 1312 MHz base and 1530 MHz boost. The RTX 5090 starts at 2017 MHz base and boosts to 2407 MHz. Memory clocks are also different: the V100’s HBM2 runs at 876 MHz (1752 Mbps effective), while the RTX 5090’s GDDR7 runs at 1750 MHz (28 Gbps effective).

Shading units scale from 5,120 on the V100 to 21,760 on the RTX 5090. TMUs jump from 320 to 680, and ROPs go from 128 to 176. The RTX 5090 supports DirectX 12 Ultimate (12_2), while the V100 only reaches DirectX 12 (12_1). Both support OpenGL 4.6 and Vulkan 1.4.

Power and physical design differ as well. The V100 is an SXM module with no power connectors and a 250 W TDP, requiring a 600 W suggested PSU. The RTX 5090 is a dual-slot card with a single 16-pin connector, a 575 W TDP, and a 950 W suggested PSU. It measures 304 mm in length, 137 mm in height, and 40 mm in width. The V100 has no listed dimensions.

FAQ

Q: Which GPU has higher raw compute performance?

A: The RTX 5090 dominates. Its FP32 output is 104.8 TFLOPS versus the V100’s 15.67 TFLOPS, and its FP16 output is 104.8 TFLOPS versus the V100’s 31.33 TFLOPS.

Q: Does the V100 support ray tracing?

A: No. The V100 has 640 tensor cores but no RT cores. The RTX 5090 has 170 RT cores and 680 tensor cores.

Q: Can the V100 be used for gaming?

A: Not practically. It has no display outputs, so it cannot connect to a monitor. The RTX 5090 offers 1x HDMI 2.1b and 3x DisplayPort 2.1b.

Q: How do their memory subsystems compare?

A: The V100 uses 16 GB of HBM2 on a 4096-bit bus with 897.0 GB/s bandwidth. The RTX 5090 has 32 GB of GDDR7 on a 512-bit bus with 1.79 TB/s bandwidth — double the capacity and nearly double the bandwidth.

Q: Which card is more power-hungry?

A: The RTX 5090 has a 575 W TDP and requires a 950 W suggested PSU. The V100 has a 250 W TDP and a 600 W suggested PSU.

Q: Are these cards from the same generation?

A: No. The V100 is from the Tesla Volta generation, released in 2017, and is end-of-life. The RTX 5090 is from the GeForce 50 series, released in 2025, and is still active.

The Verdict

The data is one-sided. The RTX 5090 wins both head-to-head benchmarks by margins of 73.8% and 62.5%. It offers more than six times the FP32 throughput, double the memory capacity, and nearly double the memory bandwidth. It supports ray tracing, modern DirectX 12 Ultimate, and has display outputs. The V100 has no path to victory in any measured metric.

The V100’s only saving grace is its efficiency profile. It draws 250 W compared to the RTX 5090’s 575 W, and it uses an SXM module design that fits into specific datacenter chassis. For legacy compute workloads that were optimized for Volta’s tensor cores, the V100 might still function adequately. But those workloads would run faster on the RTX 5090, provided the software supports Blackwell.

Who should pick the RTX 5090? Anyone building a high-end desktop system for gaming, content creation, or mixed compute tasks. Its 32 GB of GDDR7 memory and 104.8 TFLOPS FP32 output make it a universal tool. The launch MSRP is 1,999 USD. Who should pick the V100? Essentially no one buying new hardware. It is end-of-life, lacks display outputs, and loses every benchmark comparison. The only scenario favoring the V100 is a pre-existing infrastructure built around SXM modules and Volta-specific code that cannot be migrated.

Specification Differences

| Field | NVIDIA Tesla V100 SXM2 16 GB | NVIDIA GeForce RTX 5090 |

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

| Architecture | Volta | Blackwell 2.0 |

| Process Node | 12 nm | 5 nm |

| Transistors | 21,100 million | 92,200 million |

| Die Size | 815 mm² | 750 mm² |

| Transistor Density | 25.9M / mm² | 122.9M / mm² |

| Base Clock | 1312 MHz | 2017 MHz |

| Boost Clock | 1530 MHz | 2407 MHz |

| Memory Clock | 876 MHz (1752 Mbps effective) | 1750 MHz (28 Gbps effective) |

| Memory Size | 16 GB | 32 GB |

| Memory Type | HBM2 | GDDR7 |

| Memory Bus | 4096 bit | 512 bit |

| Memory Bandwidth | 897.0 GB/s | 1.79 TB/s |

| Shading Units | 5120 | 21760 |

| TMUs | 320 | 680 |

| ROPs | 128 | 176 |

| RT Cores | N/A | 170 |

| Tensor Cores | 640 | 680 |

| Pixel Rate | 195.8 GPixel/s | 423.6 GPixel/s |

| Texture Rate | 489.6 GTexel/s | 1,636.8 GTexel/s |

| FP32 | 15.67 TFLOPS | 104.8 TFLOPS |

| FP16 | 31.33 TFLOPS (2:1) | 104.8 TFLOPS (1:1) |

| TDP | 250 W | 575 W |

| Slot Width | SXM Module | Dual-slot |

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

| Suggested PSU | 600 W | 950 W |

| Bus Interface | PCIe 3.0 x16 | PCIe 5.0 x16 |

| Display Outputs | No outputs | 1x HDMI 2.1b, 3x DisplayPort 2.1b |

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

| Release Date | 2017-06-20 | 2025-01-29 |

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

| Launch MSRP | N/A | 1,999 USD |

DETAILED SPECIFICATIONS

SPECIFICATION
RTX 5090
Tesla V100 SXM2 16 GB
Core Specs
Shading Units
21,760
5,120 -76.5%
Shaders
21,760
5,120 -76.5%
TMUs
680
320 -52.9%
ROPs
176
128 -27.3%
SM Count
170
80 -52.9%
Clocks
Base Clock
2017 MHz
1312 MHz
Boost Clock
2407 MHz
1530 MHz
Memory Clock
1750 MHz 28 Gbps effective
876 MHz 1752 Mbps effective
Memory
Memory Size
32 GB
16 GB
VRAM (MB)
32,768
16,384 -50.0%
Memory Type
GDDR7
HBM2
Memory Bus
512 bit
4096 bit
Bandwidth
1.79 TB/s
897.0 GB/s
Cache
L1 Cache
128 KB (per SM)
128 KB (per SM)
L2 Cache
96 MB
6 MB
Performance
Pixel Rate
423.6 GPixel/s
195.8 GPixel/s
Texture Rate
1,636.8 GTexel/s
489.6 GTexel/s
FP32 (TFLOPS)
104.8 TFLOPS
15.67 TFLOPS
FP64 (TFLOPS)
1.637 TFLOPS (1:64)
7.834 TFLOPS (1:2)
FP16 (TFLOPS)
104.8 TFLOPS (1:1)
31.33 TFLOPS (2:1)
AI/RT
RT Cores
170
—
Tensor Cores
680
640 -5.9%
Power
TDP
575 W
250 W
TDP (W)
575
250 -56.5%
Suggested PSU
950 W
600 W
Power Connectors
1x 16-pin
None
Architecture
Architecture
Blackwell 2.0
Volta
GPU Name
GB202
GV100
Generation
GeForce 50
Tesla Volta (Vxx)
Process Size
5 nm
12 nm
Transistors
92,200 million
21,100 million
Die Size
750 mm²
815 mm²
Foundry
TSMC
TSMC
Density
122.9M / mm²
25.9M / mm²
API Support
DirectX
12 Ultimate (12_2)
12 (12_1)
OpenGL
4.6
4.6
Vulkan
1.4
1.4
OpenCL
3.0
3.0
CUDA
12.0
7.0
Shader Model
6.9
6.8
Physical
Slot Width
Dual-slot
SXM Module
Length
304 mm 12 inches
—
Height
137 mm 5.4 inches
—
Outputs
1x HDMI 2.1b3x DisplayPort 2.1b
No outputs
Bus Interface
PCIe 5.0 x16
PCIe 3.0 x16
Other
Launch Price
1,999 USD
—
Production
Active
End-of-life
Predecessor
GeForce 40
Tesla Pascal
Successor
GeForce 60
Tesla Turing
View GeForce RTX 5090 Details View Tesla V100 SXM2 16 GB Details