Skip to content

Conversation

LAhmos
Copy link

@LAhmos LAhmos commented Sep 7, 2025

update A100

@LAhmos LAhmos requested a review from JRPan September 8, 2025 02:13
Copy link

@JRPan JRPan left a comment

Choose a reason for hiding this comment

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

Where are the new latency numbers from?

# clock domains
#-gpgpu_clock_domains <Core Clock>:<Interconnect Clock>:<L2 Clock>:<DRAM Clock>
-gpgpu_clock_domains 1410:1410:1410:1512
-gpgpu_clock_domains 1410:1410:1512:6048
Copy link

Choose a reason for hiding this comment

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

Really? 6048 seems a bit high.
Also do you know how this affects sim speed?

Copy link
Author

Choose a reason for hiding this comment

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

opps that was one of the expermintes

-gpgpu_kernel_launch_latency 5000
-gpgpu_kernel_launch_latency 5000
-gpgpu_TB_launch_latency 0
-gpgpu_max_concurrent_kernel 128
Copy link

Choose a reason for hiding this comment

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

Is this removed on purpose?

Copy link
Author

Choose a reason for hiding this comment

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

nope

Copy link

Choose a reason for hiding this comment

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

Oh wait I forgot this. Can you add back this line and open another PR? I'll force merge it. I don't even think we regress A100.

@LAhmos
Copy link
Author

LAhmos commented Sep 8, 2025

what latency numbers ?
@JRPan

@LAhmos
Copy link
Author

LAhmos commented Sep 8, 2025

oh just re ran tunner to get those numbers, I think those give better error

@JRPan JRPan requested a review from Copilot September 8, 2025 02:59
Copy link

@Copilot Copilot AI left a comment

Choose a reason for hiding this comment

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

Pull Request Overview

Updates the gpgpusim.config file for SM80_A100 to reflect more accurate A100 GPU specifications and performance parameters.

  • Removes concurrent kernel limit configuration
  • Updates memory and cache configurations for improved A100 simulation accuracy
  • Adjusts timing parameters including latencies and clock domains

Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.

@JRPan JRPan merged commit 86ad347 into accel-sim:dev Sep 8, 2025
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants