Pricing

About us

Blog

Access Portal

Getting Started

Access Portal

Getting Started

SILICON MARK™

The trusted benchmark for GPU performance

The trusted benchmark for GPU performance

The trusted benchmark for GPU performance

The trusted benchmark for GPU performance

Trusted by businesses, developers, and data centers to evaluate and compare GPU performance with confidence.

Trusted by businesses, developers, and data centers to evaluate and compare GPU performance with confidence.

Trusted by businesses, developers, and data centers to evaluate and compare GPU performance with confidence.

Trusted by businesses, developers, and data centers to evaluate and compare GPU performance with confidence.

SILICON MARK™

Scoreboard

Scoreboard

View more

View more

View more

View more

SILICON MARK™

Scoreboard

View more

View more

SILICON MARK™

Scoreboard

View more

View more

BEFORE SILICONMARK

Inconsistent benchmarks and unclear choices

Stop guessing with unclear performance claims

AFTER SILICONMARK

Clarity, confidence, and smarter decisions

Start relying on independent, trusted benchmarks

AFTER SILICONMARK

Clarity, confidence, and smarter decisions

Start relying on independent, trusted benchmarks

AFTER SILICONMARK

Clarity, confidence, and smarter decisions

Start relying on independent, trusted benchmarks

AFTER SILICONMARK

Clarity, confidence, and smarter decisions

Start relying on independent, trusted benchmarks

PERFORMANCE BENCHMARKS THAT MATTER

What SiliconMark Measures

What SiliconMark Measures
What SiliconMark Measures

What SiliconMark Measures

SiliconMark evaluates five key performance metrics to give you a clear, apples-to-apples view of system capabilities. For multi-node clusters, it also measures inter-node bandwidth and latency — helping you understand how your system handles distributed workloads, which is critical for modern AI and HPC applications.

SiliconMark evaluates five key performance metrics to give you a clear, apples-to-apples view of system capabilities. For multi-node clusters, it also measures inter-node bandwidth and latency — helping you understand how your system handles distributed workloads, which is critical for modern AI and HPC applications.

SiliconMark evaluates five key performance metrics to give you a clear, apples-to-apples view of system capabilities. For multi-node clusters, it also measures inter-node bandwidth and latency — helping you understand how your system handles distributed workloads, which is critical for modern AI and HPC applications.

SiliconMark evaluates five key performance metrics to give you a clear, apples-to-apples view of system capabilities. For multi-node clusters, it also measures inter-node bandwidth and latency — helping you understand how your system handles distributed workloads, which is critical for modern AI and HPC applications.

GPU Performance

GPU Performance

GPU Performance

GPU Performance

GPU Performance

Cluster Performance

Cluster Performance

Cluster Performance

Cluster Performance

Cluster Performance

LLM Performance

LLM Performance

LLM Performance

LLM Performance

LLM Performance

Machines Inventory

Machines Inventory

Machines Inventory

Machines Inventory

Machines Inventory

CERTIFIED BENCHMARKS AND MORE

Who is SiliconMark for

Who is SiliconMark for
Who is SiliconMark for

Who is SiliconMark for

Reduce benchmarking effort
and complexity

Node #1

Node #20

Node #19

Node #17

Node #24

Node #18

Node #21

Node #22

Node #23

Node #25

Node #2

Node #3

Node #4

Node #5

Node #6

Node #8

Node #9

Node #10

Node #11

Node #12

Node #13

Node #14

Node #15

Node #16

Node #7

Build customer trust with
certified benchmarks

SiliconMark

Benchmark Certificate

TM

Job Name / ID: Database A

Client Name: AI Company

Date: 2024-12-12

This certificate is awarded to the GPU tested using the SiliconMark benchmark system. The results reflect its performance in key compute categories as of the timestamp noted below.

Field

GPU ID

FP32 Performance

FP16 Performance

Memory Bandwidth

GPU 1

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a38

7.54 TFLOPS

61.00 TFLOPS

320.45 GB/s

GPU 2

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a46

6.95 TFLOPS

66.60 TFLOPS

300.00 GB/s

GPU 3

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a99

7.74 TFLOPS

64.50 TFLOPS

322.22 GB/s

GPU 4

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a90

7.11 TFLOPS

63.22 TFLOPS

308.94 GB/s

GPU 5

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a36

8.00 TFLOPS

60.60 TFLOPS

311.34 GB/s

GPU 6

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a22

6.90 TFLOPS

61.80 TFLOPS

299.00 GB/s

GPU 7

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a33

7.770 TFLOPS

62.34 TFLOPS

310.00 GB/s

GPU 8

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a87

7.230 TFLOPS

60.30 TFLOPS

300.00 GB/s

Manufacturer's Specific

N/A

8.141 TFLOPS

65.13 TFLOPS

320.00 GB/s

Performance Results

Benchmark Insights:

1.Floating Point Operations Per Second (FP32): Measures single-precision computation capacity critical for AI and ML workloads. 

2.Floating Point Operations Per Second (FP16): Represents half-precision computation speed, optimized for deep learning tasks.

3.Memory Bandwidth: Indicates the data transfer rate between GPU memory and its processor.

Date: 2024-12-12

Signature: Authorized by SiliconMark Team



Compare GPUs across
providers and time

Reduce benchmarking effort
and complexity

Node #1

Node #20

Node #19

Node #17

Node #24

Node #18

Node #21

Node #22

Node #23

Node #25

Node #2

Node #3

Node #4

Node #5

Node #6

Node #8

Node #9

Node #10

Node #11

Node #12

Node #13

Node #14

Node #15

Node #16

Node #7

Build customer trust with
certified benchmarks

SiliconMark

Benchmark Certificate

TM

Job Name / ID: Database A

Client Name: AI Company

Date: 2024-12-12

This certificate is awarded to the GPU tested using the SiliconMark benchmark system. The results reflect its performance in key compute categories as of the timestamp noted below.

Field

GPU ID

FP32 Performance

FP16 Performance

Memory Bandwidth

GPU 1

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a38

7.54 TFLOPS

61.00 TFLOPS

320.45 GB/s

GPU 2

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a46

6.95 TFLOPS

66.60 TFLOPS

300.00 GB/s

GPU 3

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a99

7.74 TFLOPS

64.50 TFLOPS

322.22 GB/s

GPU 4

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a90

7.11 TFLOPS

63.22 TFLOPS

308.94 GB/s

GPU 5

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a36

8.00 TFLOPS

60.60 TFLOPS

311.34 GB/s

GPU 6

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a22

6.90 TFLOPS

61.80 TFLOPS

299.00 GB/s

GPU 7

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a33

7.770 TFLOPS

62.34 TFLOPS

310.00 GB/s

GPU 8

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a87

7.230 TFLOPS

60.30 TFLOPS

300.00 GB/s

Manufacturer's Specific

N/A

8.141 TFLOPS

65.13 TFLOPS

320.00 GB/s

Performance Results

Benchmark Insights:

1.Floating Point Operations Per Second (FP32): Measures single-precision computation capacity critical for AI and ML workloads. 

2.Floating Point Operations Per Second (FP16): Represents half-precision computation speed, optimized for deep learning tasks.

3.Memory Bandwidth: Indicates the data transfer rate between GPU memory and its processor.

Date: 2024-12-12

Signature: Authorized by SiliconMark Team



Compare GPUs across
providers and time

Node #1

Node #20

Node #19

Node #17

Node #24

Node #18

Node #21

Node #22

Node #23

Node #25

Node #2

Node #3

Node #4

Node #5

Node #6

Node #8

Node #9

Node #10

Node #11

Node #12

Node #13

Node #14

Node #15

Node #16

Node #7

Reduce benchmarking
effort and complexity

Build customer trust with
certified benchmarks

SiliconMark

Benchmark Certificate

TM

Job Name / ID: Database A

Client Name: AI Company

Date: 2024-12-12

This certificate is awarded to the GPU tested using the SiliconMark benchmark system. The results reflect its performance in key compute categories as of the timestamp noted below.

Field

GPU ID

FP32 Performance

FP16 Performance

Memory Bandwidth

GPU 1

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a38

7.54 TFLOPS

61.00 TFLOPS

320.45 GB/s

GPU 2

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a46

6.95 TFLOPS

66.60 TFLOPS

300.00 GB/s

GPU 3

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a99

7.74 TFLOPS

64.50 TFLOPS

322.22 GB/s

GPU 4

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a90

7.11 TFLOPS

63.22 TFLOPS

308.94 GB/s

GPU 5

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a36

8.00 TFLOPS

60.60 TFLOPS

311.34 GB/s

GPU 6

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a22

6.90 TFLOPS

61.80 TFLOPS

299.00 GB/s

GPU 7

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a33

7.770 TFLOPS

62.34 TFLOPS

310.00 GB/s

GPU 8

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a87

7.230 TFLOPS

60.30 TFLOPS

300.00 GB/s

Manufacturer's Specific

N/A

8.141 TFLOPS

65.13 TFLOPS

320.00 GB/s

Performance Results

Benchmark Insights:

1.Floating Point Operations Per Second (FP32): Measures single-precision computation capacity critical for AI and ML workloads. 

2.Floating Point Operations Per Second (FP16): Represents half-precision computation speed, optimized for deep learning tasks.

3.Memory Bandwidth: Indicates the data transfer rate between GPU memory and its processor.

Date: 2024-12-12

Signature: Authorized by SiliconMark Team



Compare GPUs across
providers and time

Node #1

Node #20

Node #19

Node #17

Node #24

Node #18

Node #21

Node #22

Node #23

Node #25

Node #2

Node #3

Node #4

Node #5

Node #6

Node #8

Node #9

Node #10

Node #11

Node #12

Node #13

Node #14

Node #15

Node #16

Node #7

Reduce benchmarking
effort and complexity

Build customer trust with certified benchmarks

SiliconMark

Benchmark Certificate

TM

Job Name / ID: Database A

Client Name: AI Company

Date: 2024-12-12

This certificate is awarded to the GPU tested using the SiliconMark benchmark system. The results reflect its performance in key compute categories as of the timestamp noted below.

Field

GPU ID

FP32 Performance

FP16 Performance

Memory Bandwidth

GPU 1

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a38

7.54 TFLOPS

61.00 TFLOPS

320.45 GB/s

GPU 2

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a46

6.95 TFLOPS

66.60 TFLOPS

300.00 GB/s

GPU 3

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a99

7.74 TFLOPS

64.50 TFLOPS

322.22 GB/s

GPU 4

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a90

7.11 TFLOPS

63.22 TFLOPS

308.94 GB/s

GPU 5

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a36

8.00 TFLOPS

60.60 TFLOPS

311.34 GB/s

GPU 6

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a22

6.90 TFLOPS

61.80 TFLOPS

299.00 GB/s

GPU 7

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a33

7.770 TFLOPS

62.34 TFLOPS

310.00 GB/s

GPU 8

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a87

7.230 TFLOPS

60.30 TFLOPS

300.00 GB/s

Manufacturer's Specific

N/A

8.141 TFLOPS

65.13 TFLOPS

320.00 GB/s

Performance Results

Benchmark Insights:

1.Floating Point Operations Per Second (FP32): Measures single-precision computation capacity critical for AI and ML workloads. 

2.Floating Point Operations Per Second (FP16): Represents half-precision computation speed, optimized for deep learning tasks.

3.Memory Bandwidth: Indicates the data transfer rate between GPU memory and its processor.

Date: 2024-12-12

Signature: Authorized by SiliconMark Team



Compare GPUs across
providers and time

Reduce benchmarking effort and
complexity

Node #1

Node #20

Node #19

Node #17

Node #24

Node #18

Node #21

Node #22

Node #23

Node #25

Node #2

Node #3

Node #4

Node #5

Node #6

Node #8

Node #9

Node #10

Node #11

Node #12

Node #13

Node #14

Node #15

Node #16

Node #7

Build customer trust with certified
benchmarks

SiliconMark

Benchmark Certificate

TM

Job Name / ID: Database A

Client Name: AI Company

Date: 2024-12-12

This certificate is awarded to the GPU tested using the SiliconMark benchmark system. The results reflect its performance in key compute categories as of the timestamp noted below.

Field

GPU ID

FP32 Performance

FP16 Performance

Memory Bandwidth

GPU 1

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a38

7.54 TFLOPS

61.00 TFLOPS

320.45 GB/s

GPU 2

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a46

6.95 TFLOPS

66.60 TFLOPS

300.00 GB/s

GPU 3

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a99

7.74 TFLOPS

64.50 TFLOPS

322.22 GB/s

GPU 4

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a90

7.11 TFLOPS

63.22 TFLOPS

308.94 GB/s

GPU 5

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a36

8.00 TFLOPS

60.60 TFLOPS

311.34 GB/s

GPU 6

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a22

6.90 TFLOPS

61.80 TFLOPS

299.00 GB/s

GPU 7

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a33

7.770 TFLOPS

62.34 TFLOPS

310.00 GB/s

GPU 8

GPU-49fa3a59-ba1b-eb51-dc8b-9a0254300a87

7.230 TFLOPS

60.30 TFLOPS

300.00 GB/s

Manufacturer's Specific

N/A

8.141 TFLOPS

65.13 TFLOPS

320.00 GB/s

Performance Results

Benchmark Insights:

1.Floating Point Operations Per Second (FP32): Measures single-precision computation capacity critical for AI and ML workloads. 

2.Floating Point Operations Per Second (FP16): Represents half-precision computation speed, optimized for deep learning tasks.

3.Memory Bandwidth: Indicates the data transfer rate between GPU memory and its processor.

Date: 2024-12-12

Signature: Authorized by SiliconMark Team



Compare GPUs across providers and time

Ready to Unlock the Future of AI Infrastructure Pricing?

Make better compute decisions today

Make better compute decisions today

Make better compute decisions today

Gain access to the world’s first benchmarks for GPU pricing and compute economics. Whether you’re a data center operator, fund manager, enterprise strategist, or fintech innovator Silicon Data Indexes give you the insights you need to act with confidence.

Terms of Use

California Privacy Notice

Privacy Notice

Newsroom

Documentation

© 2025 Silicon Data® is a registered trademark of Silicon Data Inc. All rights reserved.

Terms of Use

California Privacy Notice

Privacy Notice

Newsroom

Documentation

© 2025 Silicon Data® is a registered trademark of Silicon Data Inc. All rights reserved.

Terms of Use

California Privacy Notice

Privacy Notice

Newsroom

Documentation

© 2025 Silicon Data® is a registered trademark of Silicon Data Inc. All rights reserved.

Terms of Use

California Privacy Notice

Privacy Notice

Newsroom

Documentation

© 2025 Silicon Data® is a registered trademark of Silicon Data Inc. All rights reserved.

Terms of Use

California Privacy Notice

Privacy Notice

Newsroom

Documentation

© 2025 Silicon Data® is a registered trademark of Silicon Data Inc. All rights reserved.