Skip to content
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

[JAX] Bug fix for distributed normalization #1366

Merged
merged 4 commits into from
Dec 12, 2024
Merged

[JAX] Bug fix for distributed normalization #1366

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
486f5fb
fix ctx.aval_out indexing for workspace
phu0ngng Dec 11, 2024
a149fb0
added cudnn init to prepare phase of norm custom calls
phu0ngng Dec 11, 2024
26ba7be
add thread_local for norm registry instance
phu0ngng Dec 11, 2024
1b961b0
formatted
phu0ngng Dec 11, 2024
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
3 changes: 1 addition & 2 deletions transformer_engine/common/normalization/common.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -287,9 +287,8 @@ class CudnnNormalizationPlan : public NormalizationPlanBase {

class NormalizationPlanRegistry {
public:
// TODO thread-safe
static NormalizationPlanRegistry& getInstance() {
static NormalizationPlanRegistry instance;
static thread_local NormalizationPlanRegistry instance;
return instance;
}

Expand Down
12 changes: 6 additions & 6 deletions transformer_engine/jax/cpp_extensions/normalization.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -147,7 +147,7 @@ def lowering(ctx, x, gamma, beta, *, zero_centered_gamma, epsilon):
batch_shape = out_shape[:-1]
batch_size = reduce(operator.mul, x_shape) // hidden_size

wkspace_aval = ctx.avals_out[-2:]
wkspace_aval = ctx.avals_out[-1]

out_types = [
ir.RankedTensorType.get(out_shape, output_type),
Expand Down Expand Up @@ -441,7 +441,7 @@ def lowering(ctx, dz, x, mu, rsigma, gamma, *, zero_centered_gamma, epsilon):

sm_margin = get_backward_sm_margin()

wkspace_aval = ctx.avals_out[-4:]
wkspace_aval = ctx.avals_out[-1]
opaque = transformer_engine_jax.pack_norm_descriptor(
batch_size,
hidden_size,
Expand Down Expand Up @@ -650,7 +650,7 @@ def lowering(ctx, x, gamma, *, epsilon):
batch_shape = out_shape[:-1]
batch_size = reduce(operator.mul, x_shape) // hidden_size

wkspace_aval = ctx.avals_out[-2:]
wkspace_aval = ctx.avals_out[-1]

out_types = [
ir.RankedTensorType.get(out_shape, x_type.element_type),
Expand Down Expand Up @@ -841,7 +841,7 @@ def lowering(ctx, dz, x, rsigma, gamma, *, epsilon):
hidden_size = reduce(operator.mul, g_shape)
batch_size = reduce(operator.mul, x_shape) // hidden_size

wkspace_aval = ctx.avals_out[-3:]
wkspace_aval = ctx.avals_out[-1]

out_types = [
ir.RankedTensorType.get(x_shape, x_type.element_type),
Expand Down Expand Up @@ -1088,7 +1088,7 @@ def lowering(
batch_shape = out_shape[:-1]
batch_size = reduce(operator.mul, x_shape) // hidden_size

wkspace_aval = ctx.avals_out[-2:]
wkspace_aval = ctx.avals_out[-1]

out_types = [
ir.RankedTensorType.get(out_shape, ir_out_dtype),
Expand Down Expand Up @@ -1394,7 +1394,7 @@ def lowering(ctx, x, gamma, amax, scale, scale_inv, *, out_dtype, epsilon):
batch_shape = out_shape[:-1]
batch_size = reduce(operator.mul, x_shape) // hidden_size

wkspace_aval = ctx.avals_out[-2:]
wkspace_aval = ctx.avals_out[-1]

out_types = [
ir.RankedTensorType.get(out_shape, ir_out_dtype),
Expand Down
24 changes: 18 additions & 6 deletions transformer_engine/jax/csrc/extensions/pybind.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -83,12 +83,24 @@ pybind11::dict Registrations() {
EncapsulateFunction(ScaledUpperTriangMaskedSoftmaxBackwardHandler);

// Normalization
dict["te_layernorm_forward_ffi"] = EncapsulateFFI(LayerNormForwardHandler);
dict["te_layernorm_forward_fp8_ffi"] = EncapsulateFFI(LayerNormForwardFP8Handler);
dict["te_layernorm_backward_ffi"] = EncapsulateFFI(LayerNormBackwardHandler);
dict["te_rmsnorm_forward_ffi"] = EncapsulateFunction(RMSNormForwardHandler);
dict["te_rmsnorm_forward_fp8_ffi"] = EncapsulateFunction(RMSNormForwardFP8Handler);
dict["te_rmsnorm_backward_ffi"] = EncapsulateFunction(RMSNormBackwardHandler);
dict["te_layernorm_forward_ffi"] =
pybind11::dict(pybind11::arg("prepare") = EncapsulateFFI(CudnnHandleInitHandler),
pybind11::arg("execute") = EncapsulateFFI(LayerNormForwardHandler));
dict["te_layernorm_forward_fp8_ffi"] =
pybind11::dict(pybind11::arg("prepare") = EncapsulateFFI(CudnnHandleInitHandler),
pybind11::arg("execute") = EncapsulateFFI(LayerNormForwardFP8Handler));
dict["te_layernorm_backward_ffi"] =
pybind11::dict(pybind11::arg("prepare") = EncapsulateFFI(CudnnHandleInitHandler),
pybind11::arg("execute") = EncapsulateFFI(LayerNormBackwardHandler));
dict["te_rmsnorm_forward_ffi"] =
pybind11::dict(pybind11::arg("prepare") = EncapsulateFFI(CudnnHandleInitHandler),
pybind11::arg("execute") = EncapsulateFFI(RMSNormForwardHandler));
dict["te_rmsnorm_forward_fp8_ffi"] =
pybind11::dict(pybind11::arg("prepare") = EncapsulateFFI(CudnnHandleInitHandler),
pybind11::arg("execute") = EncapsulateFFI(RMSNormForwardFP8Handler));
dict["te_rmsnorm_backward_ffi"] =
pybind11::dict(pybind11::arg("prepare") = EncapsulateFFI(CudnnHandleInitHandler),
pybind11::arg("execute") = EncapsulateFFI(RMSNormBackwardHandler));

// Attention
pybind11::dict fused_attn_forward_ffi;
Expand Down
Loading