Skip to content

Gpu data #86

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.

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 65 commits into from
May 12, 2025
Merged

Gpu data #86

merged 65 commits into from
May 12, 2025

Conversation

Sam-Mear
Copy link
Contributor

@Sam-Mear Sam-Mear commented Dec 5, 2024

Gpu data, not ready to be merged yet, just opening as its close.

@Sam-Mear Sam-Mear marked this pull request as draft December 5, 2024 15:59
@Sam-Mear
Copy link
Contributor Author

Sam-Mear commented Jan 8, 2025

Just noticed VRAM capacity is not shown, and seems to be conflicting with Maximum VRAM Capacity

@Sam-Mear
Copy link
Contributor Author

Just noticed VRAM capacity is not shown, and seems to be conflicting with Maximum VRAM Capacity

nvm it is shown... but should be noted they should probably be MT/s rather than Mhz.

@Sam-Mear
Copy link
Contributor Author

Hi, I think this is pretty go to go into beta now, let me know if you see anything wrong with it so far @markasoftware

@Sam-Mear Sam-Mear marked this pull request as ready for review January 13, 2025 14:32
@markasoftware
Copy link
Owner

Sorry for the late response! Overall looks really good, I'm fine to merge it basically whatever since it's already a huge improvement over what we have right now. That being said there are still a few things that could be improved:

  • On most pages we want to display parts from "most high end" to "least high end", but that seems to typically not be the case in your pr; for example under maxwell 2.0 nvidia we have the gtx 750 first, then 950, then 960, etc, so the opposite order.
  • "uncategorized Graphics Cards" should be capitalized to "Uncategorized"
  • A lot of the categories still say 14nm lithography when it isn't true, eg under 3dfx, powervr, xgi, all the old nvidia
  • Somehow the row comparison thinks "N/A" Vulkan support is better than actually having vulkan support lol, see the vulkan row on https://deploy-preview-86--specdb.netlify.app/Radeon-RX-5600-OEM,Radeon-RX-5600-XT,Radeon-RX-5700-XT-50th-Anniversary,Tesla-C870 and how the "N/A" is highlighted red

@Sam-Mear
Copy link
Contributor Author

I think this should be good to go for now. Can always fix things with later pull requests

@markasoftware
Copy link
Owner

Nice! Some notes:

  • Even with these thousands of new parts, the compressed bundle.js that has to be loaded over the network is <250 KiB when compressed.
  • The text around the bracket for "Hardware Accelerated Media" is a bit broken because the label is too long so it goes onto a second line that collides with the bracket. I can probably look into this in the CSS.
  • Very happy to see the "matcher" architecture keeps working and a lot of the new GPUs have benchmark data!
  • Some of the benchmark numbers for newer GPUs have a lot of digits, I should do something to add commas to make them more readable
  • In some generations, eg GCN 2 and 3, the ordering of the GPUs within each section is not in order from best->worst.
  • TDPs on some cards says "unknown", I'd prefer if it were just blank

I'd also prefer if the scraping code was open-sourced somewhere, even if it's just random python scripts in a separate repo that you control.

This is a massive achievement and certainly the biggest update to SpecDB since its inception!

@markasoftware markasoftware merged commit 72a0516 into markasoftware:beta May 12, 2025
4 checks passed
@Sam-Mear
Copy link
Contributor Author

@markasoftware I did edit some of the matcher stuff, and added another one I think. If you see any obvious ones without a benchmark I'll have a look but any of the new ish graphics cards should have them

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