GeekRadar / The comparison desk

COMPARE.

GPU

NVIDIA

Tesla V100 SXM2 32 GB

Graphics processor

Video memory: 32 GB^*
Memory bus: 4,096 bit^*
Rated memory bandwidth: 898 GB/s^*

No matched benchmark results are available for this entry yet.

Processing & platform

Architecture: Volta^*
Process node: 12 nm^*
GPU cores: 5,120^*
GPU base clock: 1,290 MHz^*
Reported GPU frequency: 1,290 MHz^*
Graphics interface: PCIe 3.0 x16^*
Reported chip die area: 815 mm²^*
Chip/package TDP: 250 W^*
Transistors: 21,100 million^*
GPU boost clock: 1,530 MHz^*
GPU L2 cache: 6 MB^*
Graphics processor: GV100^*
Texture mapping units: 320^*
Raster output units: 128^*
Streaming multiprocessors: 80^*
Tensor cores: 640^*
Chip foundry: TSMC^*
Card or module format: SXM Module^*
Declared GPU TDP: 250 W^*
Suggested system PSU: 600 W^*
External power connectors: None^*
DirectX / feature level: 12 (12_1)^*
OpenGL: 4.6^*
OpenCL: 3.0^*
Vulkan: 1.4^*
CUDA compute capability: 7.0^*
Shader model: 6.9^*

Memory & storage

Memory options: 32^*
Video memory: 32 GB^*
Video memory type: HBM2^*
Memory bus: 4,096 bit^*
Rated memory bandwidth: 898 GB/s^*

Power specifications

Thermal design power: 250 W^*

System configuration

Detailed specifications

Specifications 1 · 21 fields

GPU architecture — Tesla V100 SXM2 32GB — 32 GB HBM2

Volta^*

Native hardware support: BF16 tensor — Tesla V100 SXM2 32GB — 32 GB HBM2

No^*

Theoretical peak: FP16 tensor dense per GPU — Tesla V100 SXM2 32GB — 32 GB HBM2

125^* TFLOPS

Theoretical peak: FP32 non-tensor dense per GPU — Tesla V100 SXM2 32GB — 32 GB HBM2

15.7^* TFLOPS

Native hardware support: FP4 tensor — Tesla V100 SXM2 32GB — 32 GB HBM2

No^*

Theoretical peak: FP64 non-tensor dense per GPU — Tesla V100 SXM2 32GB — 32 GB HBM2

7.8^* TFLOPS

Native hardware support: FP64 tensor — Tesla V100 SXM2 32GB — 32 GB HBM2

No^*

Native hardware support: FP6 tensor — Tesla V100 SXM2 32GB — 32 GB HBM2

No^*

Native hardware support: FP8 tensor — Tesla V100 SXM2 32GB — 32 GB HBM2

No^*

Native hardware support: INT8 tensor — Tesla V100 SXM2 32GB — 32 GB HBM2

No^*

Native hardware support: TF32 tensor — Tesla V100 SXM2 32GB — 32 GB HBM2

No^*

Physical configuration — Tesla V100 SXM2 32GB — 32 GB HBM2

SXM2^*

GPU interconnect bandwidth (NVLink 2) bidirectional — Tesla V100 SXM2 32GB — 32 GB HBM2

300^* GB/s

GPU interconnect bandwidth per direction — Tesla V100 SXM2 32GB — 32 GB HBM2

150^* GB/s

Scope and conditions

Calculation: 300 / 2^*

PCIe interface declaration — Tesla V100 SXM2 32GB — 32 GB HBM2

PCIe Gen3^*

Maximum board power — Tesla V100 SXM2 32GB — 32 GB HBM2

300^* W

Peak memory bandwidth per GPU — Tesla V100 SXM2 32GB — 32 GB HBM2

900^* GB/s

Memory interface width per GPU — Tesla V100 SXM2 32GB — 32 GB HBM2

4,096^* bit

Memory capacity per GPU — Tesla V100 SXM2 32GB — 32 GB HBM2

32^* GB

Memory technology — Tesla V100 SXM2 32GB — 32 GB HBM2

HBM2^*

GPU model — Tesla V100 SXM2 32GB — 32 GB HBM2

Tesla V100 SXM2 32GB^*

Specifications 2 · 8 fields

Product Name: Tesla V100 SXM2 32 GB^*
GPU Chip: GV100^*
Released: Mar 27th, 2018^*
Bus: PCIe 3.0 x16^*
Memory: 32 GB, HBM2, 4096 bit^*
GPU clock: 1,290^* MHz
Memory clock: 877^* MHz
Shaders TMUs ROPs: 5120 / 320 / 128^*

Specifications 3 · 13 fields

Foundry: TSMC^*
Vendor: NVIDIA^*
FP16 GFLOPS: 31,330^* GFLOPS
FP32 GFLOPS: 15,670^* GFLOPS
FP64 GFLOPS: 7,834^* GFLOPS
Product: NVIDIA Tesla V100 SXM2 32 GB^*
Type: GPU^*
Release Date: 2018-03-27^*
Process Size (nm): 12^* nm
TDP (W): 250^* W
Die Size (mm^2): 815^* mm²
Transistors (million): 21,100^*
Freq (MHz): 1,290^* MHz

Specifications 4 · 29 fields

proc transistors: 21,100 million^*
proc die size: 815^* mm²
cores: 5,120^*
tensor cores: 640^*
sm count: 80^*
base clock: 1,290^*
boost clock: 1,530^*
memory clock: 877^*
ram: 32^*
name: NVIDIA Tesla V100 SXM2 32 GB^*
ram type: HBM2^*
bus width: 4,096^*
bus interface: PCIe 3.0 x16^*
ram bandwidth: 898^* GB/s
fp16 performance: 31.33^*
fp32 performance: 15.67^*
fp64 performance: 7.834^*
tdp: 250^* W
suggested psu: 600^* W
slot width: SXM Module^*
api cuda: 7^*
is retail board: No^*
gpi value: 16.61^*
vendor: nvd^*
manufacturer: NVIDIA^*
arch name: Volta^*
card release date: 2018-03-27^*
proc foundry: TSMC^*
proc process size: 12^* nm

Specifications 5 · 39 fields

CUDA compute capability

7.0^*

DirectX / feature level

12 (12_1)^*

OpenCL

3.0^*

OpenGL

4.6^*

Shader model

6.9^*

Vulkan

1.4^*

Architecture

Volta^*

Graphics interface

PCIe 3.0 x16^*

Memory bus

4,096^* bit

Reported shader processors

5,120^*

Reported theoretical FP16 peak

31.33^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Reported theoretical FP32 peak

15.67^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Reported theoretical FP64 peak

7.834^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Declared L1 cache configuration

128 KB (per SM)^*

Declared GPU L2 cache

6 MB^*

Reported display outputs

No outputs^*

Theoretical pixel fill rate

195.8 GPixel/s^*

External power connectors

None^*

Reported chip die area

815 mm²^*

Chip foundry

TSMC^*

Process node declaration

12 nm^*

Transistors

21,100 million^*

Graphics processor

GV100^*

Reported video memory

32^* GB

Rated memory bandwidth

898.0 GB/s^*

Video memory type

HBM2^*

Raster output units

128^*

Reported shading units

5,120^*

Card or module format

SXM Module^*

Streaming multiprocessors

80^*

Suggested system PSU

600 W^*

2:4 tensor sparsity

No^*

Declared GPU TDP

250 W^*

Tensor cores

640^*

Theoretical texture fill rate

489.6 GTexel/s^*

Texture mapping units

320^*

GPU cores

5,120^*

Video memory

32^* GB

Video memory type

HBM2^*

Specifications 6 · 8 fields

Product Name: Tesla V100 SXM2 32 GB^*
GPU Chip: GV100^*
Released: Mar 27th, 2018^*
Bus: PCIe 3.0 x16^*
Memory: 32 GB, HBM2, 4096 bit^*
GPU clock: 1,290^* MHz
Memory clock: 877^* MHz
Shaders / TMUs / ROPs: 5120 / 320 / 128^*

Specifications 7 · 52 fields

Brand: NVIDIA^*
Top BUS WIDTH: 4,096^* bit
Graphics Processor GPU Name: GV100^*
Graphics Processor Architecture: Volta^*
Render Config SM Count: 80^*
Graphics Features CUDA: 7^*
Render Config Tensor Cores: 640^*
Graphics Processor Process Size: 12^* nm
Graphics Processor Transistors: 21,100 million^*
Graphics Processor Die Size: 815^* mm²
Graphics Card Release Date: Mar 27th, 2018^*
Graphics Card Generation: Tesla Volta (Vxx)^*
Graphics Card Successor: Tesla Turing^*
Graphics Card Production: End-of-life^*
Name: Tesla V100 SXM2 32 GB^*
Graphics Card Bus Interface: PCIe 3.0 x16^*
Clock Speeds Memory Clock: 877 MHz 1754 Mbps effective^*
Memory Memory Size: 32^* GB
Memory Memory Type: HBM2^*
Memory Memory Bus: 4,096^* bit
Memory Bandwidth: 898^* GB/s
Render Config TMUs: 320^*
Top GRAPHICS PROCESSOR: GV100^*
Render Config ROPs: 128^*
Theoretical Performance Pixel Rate: 195.8 GPixel/s
Theoretical Performance Texture Rate: 489.6 GTexel/s
Board Design Slot Width: SXM Module^*
Board Design TDP: 250^* W
Board Design Suggested PSU: 600^* W
Board Design Outputs: No outputs^*
Graphics Features DirectX: 12 (12_1)^*
Graphics Features OpenGL: 4.6^*
Graphics Features OpenCL: 3^*
Graphics Features Vulkan: 1.3^*
source file: nvidia2018.json^*
Graphics Processor Foundry: TSMC^*
Graphics Card Predecessor: Tesla Pascal^*
Graphics Processor Density: 25.9M / mm²^*
Top TMUS: 320^*
Top ROPS: 128^*
Clock Speeds Base Clock: 1,290^* MHz
Clock Speeds Boost Clock: 1,530^* MHz
Graphics Features Shader Model: 6.8^*
Top CORES: 5,120^*
Render Config Shading Units: 5,120^*
Render Config L2 Cache: 6^* MB
Top MEMORY SIZE: 32^* GB
Theoretical Performance FP32 (float): 15.67^* TFLOPS
Theoretical Performance FP64 (double): 7.834 TFLOPS (1:2)
Render Config L1 Cache: 128 KB (per SM)^*
Theoretical Performance FP16 (half): 31.33 TFLOPS (2:1)
Top MEMORY TYPE: HBM2^*

Specifications 8 · 27 fields

manufacturer: NVIDIA^*
manufacturer: NVIDIA^*
rop: 64^*
rop: 128^*
igp: No^*
igp: No^*
bus: PCIe 3.0 x16^*
bus: PCIe 3.0 x16^*
memType: HBM2^*
memType: HBM2^*
gpuChip: GV100^*
gpuChip: Vega 10^*
productName: Tesla V100 SXM2 32 GB^*
productName: Tesla V100 SXM2 32 GB^*
releaseYear: 2,018^*
releaseYear: 2,018^*
memSize: 32^*
memSize: 32^*
memBusWidth: 4,096^*
gpuClock: 1,200^*
gpuClock: 1,290^*
memClock: 876^*
memClock: 1,000^*
unifiedShader: 5,120^*
unifiedShader: 3,584^*
tmu: 320^*
tmu: 224^*

* Provisional data