NVIDIA publishes exactly one performance figure for TITAN V, "110 TeraFLOPS" of deep learning throughput, and no FP32 or FP64 number at all. The 14.90 TFLOPS FP32 recorded here is derived from NVIDIA's own 5120 CUDA cores and 1455 MHz boost clock, and the 7.45 TFLOPS FP64 from the GV100 half-rate ratio that the V100 datasheet demonstrates numerically (7.8 against 15.7). Note that 110 sits below the 119 the boost clock would imply, so NVIDIA's tensor headline is the conservative figure and is stored as printed. Volta has no structured sparsity, so it is already dense. NVIDIA's live product URL now redirects to an RTX 6000 Ada page, so the specification source is an archived capture.

NVIDIA Titan V 12GB

Volta PCIe 2017 12nm Spec confidence: Official
FP32
15
TFLOPS
VRAM
12 GB
HBM2
Bandwidth
653 GB/s
memory
TDP
250 W
59.6 TFLOPS/kW

Overview

TITAN V put the GV100 die, the same silicon as the Tesla V100, into a desktop PCIe card with 12 GB of HBM2, and it is still the cheapest way to get full-rate FP64 out of an NVIDIA consumer board. That is its real distinction: 7.45 TFLOPS of double precision, half the FP32 rate, where every GeForce card of the era ran FP64 at a thirty-second or a sixty-fourth. It was also the first card outside the datacenter with tensor cores, rated by NVIDIA at 110 TFLOPS of deep learning throughput. Volta tensor cores do FP16 and nothing else, so there is no INT8, INT4, BF16 or TF32 here, and no structured sparsity, which arrived with Ampere. The 652.8 GB/s of HBM2 bandwidth still reads well against modern GDDR cards; the 12 GB capacity does not.

Performance

Peak theoretical throughput by precision type

PrecisionPeak
FP64
64-bit floating point
7.5TFLOPS
FP32
32-bit floating point
15TFLOPS
TF32
No verified data available
BF16
No verified data available
FP16
16-bit floating point
110TFLOPS
FP8
No verified data available
FP6
No verified data available
FP4
No verified data available
INT8
No verified data available

Every figure here is dense. The vendor documents no structured (2:4) sparsity mode for this part, so there is no second number to quote.

The Titan V 12GB in the GPU landscape

Peak FP16 TFLOPS (dense) against TDP, single-GPU parts tracked by Flopper

06001,2001,8002,4003,0000 W250 W500 W750 W1000 W1250 W1500 WInstinct MI355XTitan V 12GB

Higher and further left is better: more half-precision throughput for less power.

Specifications

Architecture

Volta

Form Factor

PCIe

Launch Year

2017

Process Node

12nm

Memory

12 GB HBM2

Bandwidth

652.8 GB/s

TDP

250 W

Max Power

250 W

Transistors

21.1 billion

CUDA Cores

5,120

Spec Confidence

Official

Full Specifications

Compute Engine
CUDA Cores 5,120
Tensor Cores 640 (1st Gen)
Streaming Multiprocessors 80
Base Clock 1.20 GHz
Boost Clock 1.46 GHz
Chip Design
Transistors 21.1 billion
Die Size 815 mm²
Process Node 12nm
Memory
VRAM 12 GB
Memory Type HBM2
Bandwidth 653 GB/s
Interface Width 3072-bit
Memory Clock 1.7 GT/s
Interconnect & I/O
PCIe 3.0 x16
Cache
L2 Cache 5 MB
Power & Thermal
TDP 250 W
Max Board Power 250 W
Power Connector One 6-pin, One 8-pin
Cooling Active
Enterprise Features
ECC Memory No
Sparsity No
Compute APIs CUDA, DirectX 12, Vulkan, OpenGL
Physical & Media
Width 2-slot
General
Form Factor PCIe
Architecture Volta
Process Node 12nm
Launch Year 2017

Datasheet & Resources

Data Provenance

Every figure traced to a source

Primary Source

Publisher
NVIDIA
Published
2017-12-07

Data Quality

Spec confidence
Official
Clock basis
Typical
Core precisions with figures
3 of 9
Normalization
All values in TFLOPS

Rent this GPU

$0.09 / GPU-hr

Lowest of 1 live listing across 1 provider

See all rental prices →

Price history and every provider on the pricing page. Some links are affiliate links.

Similar GPUs

NVIDIA V100S PCIe 32GB

PCIe · 2020
FP32: 16 TFLOPS
Compare vs V100S

NVIDIA V100 SXM2 32GB

SXM · 2018
FP32: 16 TFLOPS
Compare vs V100

NVIDIA V100 SXM2 16GB

SXM · 2017
FP32: 16 TFLOPS
Compare vs V100

NVIDIA V100 PCIe 32GB

PCIe · 2018
FP32: 14 TFLOPS
Compare vs V100

NVIDIA V100 PCIe 16GB

PCIe · 2017
FP32: 14 TFLOPS
Compare vs V100

Frequently Asked Questions

How many TFLOPS does the NVIDIA Titan V have?

The NVIDIA Titan V delivers 15 TFLOPS FP32 and 110 TFLOPS FP16 at peak. Flopper does not currently have a verified FP8 throughput figure for it.

What is the power consumption of the NVIDIA Titan V?

The NVIDIA Titan V has a TDP (Thermal Design Power) rating of 250 watts.

How much memory does the NVIDIA Titan V have?

The NVIDIA Titan V is equipped with 12 GB of VRAM with 652.8 GB/s of memory bandwidth.

What architecture is the NVIDIA Titan V based on?

The NVIDIA Titan V is based on the Volta architecture, launched in 2017.

Stay Updated on GPU Releases

Get notified when new GPUs are added or specifications are updated.

Loading verification...

No spam, unsubscribe anytime.

Back to GPUs