From ee62cffb78a337afa27b45f880ae35bacf163316 Mon Sep 17 00:00:00 2001
From: Seonghee Lee <seongheel@nvidia.com>
Date: Tue, 25 Aug 2026 11:12:56 -0700
Subject: [PATCH 1/2] docs: scaffold tensor parallelism optimization blog post

Adds a placeholder page and Tech Blogs navigation section so the TP
optimization write-up has a home before the content lands.
---
 docs/blog/tensor-parallelism-optimization.md | 27 ++++++++++++++++++++
 docs/index.yml                               |  6 +++++
 2 files changed, 33 insertions(+)
 create mode 100644 docs/blog/tensor-parallelism-optimization.md

diff --git a/docs/blog/tensor-parallelism-optimization.md b/docs/blog/tensor-parallelism-optimization.md
new file mode 100644
index 000000000..341dcc9ea
--- /dev/null
+++ b/docs/blog/tensor-parallelism-optimization.md
@@ -0,0 +1,27 @@
+# Optimizing Tensor Parallelism in JAX and XLA
+
+> **Draft** — this page is a placeholder. Content is in progress.
+
+## Background
+
+<!-- Why TP matters: model sizes past single-GPU memory, the compute/communication
+     tradeoff versus FSDP and pipeline parallelism. -->
+
+## Where tensor parallelism costs performance
+
+<!-- All-gather / reduce-scatter on the critical path, exposed communication,
+     small GEMM tiles at high TP degree, NVLink versus inter-node bandwidth. -->
+
+## Optimizations
+
+<!-- One subsection per optimization, with the mechanism and how to enable it. -->
+
+## Results
+
+<!-- Model, hardware, TP degrees swept, throughput and step-time deltas. -->
+
+## Reproducing
+
+<!-- Container tag, framework config flags, launch command. -->
+
+## Takeaways
diff --git a/docs/index.yml b/docs/index.yml
index ccfac3e37..ce319a1d5 100644
--- a/docs/index.yml
+++ b/docs/index.yml
@@ -55,6 +55,12 @@ navigation:
       - page: nsys-jax
         path: nsys-jax.md
 
+  # ==================== Tech Blogs ====================
+  - section: Tech Blogs
+    contents:
+      - page: Optimizing Tensor Parallelism in JAX and XLA
+        path: blog/tensor-parallelism-optimization.md
+
   # ==================== Resiliency ====================
   - section: Resiliency
     contents:

From ecbac5f04b8402623f1ade976eabfd39ee4a37cb Mon Sep 17 00:00:00 2001
From: Seonghee Lee <seongheel@nvidia.com>
Date: Tue, 25 Aug 2026 11:17:55 -0700
Subject: [PATCH 2/2] docs: use MDX comment syntax in blog placeholder

Fern parses .md pages as MDX, so HTML comments fail to parse and blank
the page. No other docs page uses HTML comments.
---
 docs/blog/tensor-parallelism-optimization.md | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/docs/blog/tensor-parallelism-optimization.md b/docs/blog/tensor-parallelism-optimization.md
index 341dcc9ea..a1a88b408 100644
--- a/docs/blog/tensor-parallelism-optimization.md
+++ b/docs/blog/tensor-parallelism-optimization.md
@@ -4,24 +4,24 @@
 
 ## Background
 
-<!-- Why TP matters: model sizes past single-GPU memory, the compute/communication
-     tradeoff versus FSDP and pipeline parallelism. -->
+{/* Why TP matters: model sizes past single-GPU memory, the compute/communication
+    tradeoff versus FSDP and pipeline parallelism. */}
 
 ## Where tensor parallelism costs performance
 
-<!-- All-gather / reduce-scatter on the critical path, exposed communication,
-     small GEMM tiles at high TP degree, NVLink versus inter-node bandwidth. -->
+{/* All-gather / reduce-scatter on the critical path, exposed communication,
+    small GEMM tiles at high TP degree, NVLink versus inter-node bandwidth. */}
 
 ## Optimizations
 
-<!-- One subsection per optimization, with the mechanism and how to enable it. -->
+{/* One subsection per optimization, with the mechanism and how to enable it. */}
 
 ## Results
 
-<!-- Model, hardware, TP degrees swept, throughput and step-time deltas. -->
+{/* Model, hardware, TP degrees swept, throughput and step-time deltas. */}
 
 ## Reproducing
 
-<!-- Container tag, framework config flags, launch command. -->
+{/* Container tag, framework config flags, launch command. */}
 
 ## Takeaways