Skip to content

[hardware] add flops count support for A3 device #3053

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Aug 14, 2025
Merged

[hardware] add flops count support for A3 device #3053

Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
9779c14
Update flops_counter.py
codemayq Aug 14, 2025
005a723
split two kinds of device
codemayq Aug 14, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion verl/utils/flops_counter.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -86,7 +86,7 @@ def unit_convert(number, level):
flops = 119.5e12
elif "H20" in device_name:
flops = 148e12
elif "910B" in device_name:
elif "910B" in device_name or "Ascend910" in device_name:
flops = 354e12
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot Aug 14, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

high

The FLOPS value for Ascend 910 and 910B devices seems to be incorrect, and they should not be grouped together as they have different performance characteristics. According to public documentation, Ascend 910 provides 256 TFLOPS (FP16) and Ascend 910B provides 320 TFLOPS (FP16). Using a single, potentially incorrect, value of 354 TFLOPS for both could lead to inaccurate MFU calculations. Please consider using separate, accurate values for each device.

Suggested change
elif "910B" in device_name or "Ascend910" in device_name:
flops = 354e12
elif "910B" in device_name:
flops = 320e12
elif "Ascend910" in device_name:
flops = 256e12

elif "RTX 3070 Ti" in device_name:
flops = 21.75e12
Expand Down