-
Notifications
You must be signed in to change notification settings - Fork 2
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
Anonymized stats through buckets #878
base: main
Are you sure you want to change the base?
Conversation
iris-mpc-gpu/Cargo.toml
Outdated
@@ -9,7 +9,7 @@ repository.workspace = true | |||
|
|||
[dependencies] | |||
bincode = "1.3.3" | |||
cudarc = { version = "0.12", features = ["cuda-12020", "nccl"] } | |||
cudarc = { git = "https://github.com/worldcoin/cudarc-fork.git", features = ["cuda-12020", "nccl"] } |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why do we need the fork again? Also if we need it this should probably be pinned to a "rev"
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The macro only defines up to 12 parameters: https://github.com/coreylowman/cudarc/blob/e768643c98ba0f62e0e1eb954a5bc3ea52c2b825/src/driver/safe/launch.rs#L356-L359
I've instead switched to the underlying implementation, but needed to export the function for that: worldcoin/cudarc-fork@b6ba8da
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We could just push upstream an impl with more params as a quick hack?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I created a PR in cudarc that extends this to 20
) -> CudaSlice<u32> { | ||
assert_eq!(len & 3, 0); | ||
let mut rand = unsafe { | ||
self.device_manager | ||
.device(idx) | ||
.alloc::<u32>(len >> 2) | ||
.unwrap() | ||
.unwrap() // TODO: fix, make this async |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hmm, same in otp_decrypt as well. Should probably be happening async, yes.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
ah thanks for the pointer, stumbled across this while implementing, some non async leftover
No description provided.