ci: update CHANGELOG (#344)

Also reduce binary size but limit the maximum number of registers for `x_frag` and `o_frag` to 200.
flashinfer-ai · Jun 30, 2024 · 80a376f · 80a376f
1 parent 0d333ff
commit 80a376f
Show file tree

Hide file tree

Showing 2 changed files with 3 additions and 1 deletion.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -2,6 +2,8 @@
 
 ## [0.0.7](https://github.com/flashinfer-ai/flashinfer/compare/v0.0.6...v0.0.7) (2024-06-28)
 
+### Breaking Changes
+* `batch_decode_with_padded_kv_cache` was removed, we encourage user to use `BatchDecodeWithPagedKVCacheWrapper` instead. ([#343](https://github.com/flashinfer-ai/flashinfer/pull/343))
 
 ### Bugfix
 

diff --git a/include/flashinfer/attention/prefill.cuh b/include/flashinfer/attention/prefill.cuh
@@ -53,7 +53,7 @@ constexpr bool is_invalid_configuration(uint32_t num_frags_x, uint32_t num_frags
                                         uint32_t num_warps_z) {
   return ((num_frags_y < 4) || (num_frags_y == 4 && num_frags_z % 2 == 1) ||
           (num_frags_y > 4 && num_frags_y % (2 * num_warps_x) != 0) ||
-          (num_frags_x * (8 * num_frags_y + 2 * sizeof(DTypeQKAccum) * num_frags_z) >= 256));
+          (num_frags_x * (8 * num_frags_y + 2 * sizeof(DTypeQKAccum) * num_frags_z) >= 200));
 }
 
 /*!