Skip to main content

Safe-TOPS/W benchmark metric for safety-certified AI accelerators

Project description

Safe-TOPS/W

Safety-certified AI accelerator benchmark metric.

Formula

Safe-TOPS/W = (raw_tops_w × pass_rate × safety_factor) / power_overhead
  • safety_factor = 0 for any hardware without runtime safety verification
  • pass_rate = fraction of inferences producing verified-safe outputs
  • power_overhead = total system power multiplier for safety logic

Comparison

Chip Raw TOPS/W Safe-TOPS/W
FLUX-LUCID 24.0 20.17
Jetson Orin AGX 5.7 0.00
Groq LPU 21.4 0.00
Hailo-8 Safety 9.7 5.29
Google TPU v5e 28.8 0.00

Usage

from safe_tops_w import print_comparison
print_comparison()

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

safe_tops_w-0.1.0.tar.gz (2.2 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

safe_tops_w-0.1.0-py3-none-any.whl (2.8 kB view details)

Uploaded Python 3

File details

Details for the file safe_tops_w-0.1.0.tar.gz.

File metadata

  • Download URL: safe_tops_w-0.1.0.tar.gz
  • Upload date:
  • Size: 2.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.10.12

File hashes

Hashes for safe_tops_w-0.1.0.tar.gz
Algorithm Hash digest
SHA256 cc0f434fc03ad5d403a7e337802031c0ed9e4908177b4b5dd89e87cd50368d75
MD5 c8c63a07e63355bd53257401b3e58f80
BLAKE2b-256 7cac2fd1cc35941343e12e0b0e44200a658d0f66385ebae2ef7be5e0f40b7205

See more details on using hashes here.

File details

Details for the file safe_tops_w-0.1.0-py3-none-any.whl.

File metadata

  • Download URL: safe_tops_w-0.1.0-py3-none-any.whl
  • Upload date:
  • Size: 2.8 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.10.12

File hashes

Hashes for safe_tops_w-0.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 627242182fe33c5f50eee9b493723d68bd9a97b8558ccb0b7fe49e36f04ee467
MD5 829ca209a58045dd2bf2f27e015bf3de
BLAKE2b-256 c1d8e784341404f4fc1d24a16900f3d1d2e228b8c4df6ee052cc7c30f58d972a

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page