metal : remove obosolete asserts (#17295)

ggerganov · web-flow · commit 416e7c7f477c · 2025-11-16T09:50:26.000+02:00
diff --git a/ggml/src/ggml-metal/ggml-metal-ops.cpp b/ggml/src/ggml-metal/ggml-metal-ops.cpp
@@ -2191,8 +2191,6 @@ int ggml_metal_op_flash_attn_ext(ggml_metal_op_t ctx, int idx) {
             ggml_metal_encoder_dispatch_threadgroups(enc, ncpsg, std::max(ne12, ne32), std::max(ne13, ne33), 32, 1, 1);
 
             need_sync = true;
-        } else {
-            assert(ggml_metal_op_flash_attn_ext_extra_pad(op) == 0);
         }
 
         if (has_mask) {
@@ -2222,8 +2220,6 @@ int ggml_metal_op_flash_attn_ext(ggml_metal_op_t ctx, int idx) {
             ggml_metal_encoder_dispatch_threadgroups(enc, nblk0, nblk1, ne32*ne33, 32, 1, 1);
 
             need_sync = true;
-        } else {
-            assert(ggml_metal_op_flash_attn_ext_extra_blk(op) == 0);
         }
 
         if (need_sync) {
@@ -2363,8 +2359,6 @@ int ggml_metal_op_flash_attn_ext(ggml_metal_op_t ctx, int idx) {
             ggml_metal_encoder_dispatch_threadgroups(enc, ncpsg, std::max(ne12, ne32), std::max(ne13, ne33), 32, 1, 1);
 
             need_sync = true;
-        } else {
-            assert(ggml_metal_op_flash_attn_ext_extra_pad(op) == 0);
         }
 
         if (need_sync) {

Original file line number	Diff line number	Diff line change
`@@ -2191,8 +2191,6 @@ int ggml_metal_op_flash_attn_ext(ggml_metal_op_t ctx, int idx) {`
`2191`	`2191`	`ggml_metal_encoder_dispatch_threadgroups(enc, ncpsg, std::max(ne12, ne32), std::max(ne13, ne33), 32, 1, 1);`
`2192`	`2192`
`2193`	`2193`	`need_sync = true;`
`2194`		`- } else {`
`2195`		`- assert(ggml_metal_op_flash_attn_ext_extra_pad(op) == 0);`
`2196`	`2194`	`}`
`2197`	`2195`
`2198`	`2196`	`if (has_mask) {`
`@@ -2222,8 +2220,6 @@ int ggml_metal_op_flash_attn_ext(ggml_metal_op_t ctx, int idx) {`
`2222`	`2220`	`ggml_metal_encoder_dispatch_threadgroups(enc, nblk0, nblk1, ne32*ne33, 32, 1, 1);`
`2223`	`2221`
`2224`	`2222`	`need_sync = true;`
`2225`		`- } else {`
`2226`		`- assert(ggml_metal_op_flash_attn_ext_extra_blk(op) == 0);`
`2227`	`2223`	`}`
`2228`	`2224`
`2229`	`2225`	`if (need_sync) {`
`@@ -2363,8 +2359,6 @@ int ggml_metal_op_flash_attn_ext(ggml_metal_op_t ctx, int idx) {`
`2363`	`2359`	`ggml_metal_encoder_dispatch_threadgroups(enc, ncpsg, std::max(ne12, ne32), std::max(ne13, ne33), 32, 1, 1);`
`2364`	`2360`
`2365`	`2361`	`need_sync = true;`
`2366`		`- } else {`
`2367`		`- assert(ggml_metal_op_flash_attn_ext_extra_pad(op) == 0);`
`2368`	`2362`	`}`
`2369`	`2363`
`2370`	`2364`	`if (need_sync) {`