IBM granite/granitemoe architecture support (#6760)

* fix(ext_server): Port llama.cpp sampling refactors to ext_server

This was a fairly large changeset. I closely followed the changes here:
https://github.com/ggerganov/llama.cpp/commit/df270ef74596da8f1178f08991f4c51f18c9ee82



Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(server.cpp): Refactor server.cpp logging for llama.cpp overhaul

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* feat: Bump llama.cpp to the latest master with `granite` support

This does not yet have granite MoE support, but that can come in a
follow up PR

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(patches): Update all patches (except solar-pro) to work with bumped llama.cpp

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(solar): Update solar patch for llama.cpp bump

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* feat(llama.cpp): Bump llama.cpp for granitemoe support

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* feat(llama.cpp): Bump llama.cpp for granitemoe support

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(solar): Update the solar-pro patch for latest llama.cpp bump

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* feat(llama.cpp): Bump to the latest master of llama.cpp

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(patches): Update all patches for latest bump

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* feat(llama): Always run sync.sh from the right directory

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(llama/patches): Update llama patches

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* feat(llama)!: Rough sync with llama.cpp submodule

There are a number of changes that will need to be propagated to llama.go
before any of this works!

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(llama/patches): Add a patch and update for missing ggml-impl.h include

This include is where the ggml_cgraph struct is defined. It is included in
many of the .c files to define the forward declartion in ggml.h. It seems
that with the subset of code included here, the import was somehow lost (or
out-of-order) when building, so adding this include to llama.cpp fixes the
missing definition.

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(llama/sync): Add missing ggml-cpu-impl.h copy-over in sync.sh

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(llama): Add missing log.cpp

This was added as part of the logging overhaul done in llama.cpp

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(llama): Overhaul use of sampling module for llama.cpp changes

The changes here reflect the changes made in the big llama.cpp sampling PR
https://github.com/ggerganov/llama.cpp/pull/9294



The sampling functionality is now broken into the base interface
(llama_sampler) and the generation implementation (gpt_sampler). The
changes here reflect that. Since the sampling.h/sampling.cpp code uses c++
STL headers, the sampling_ext.[h|cpp] wrapper is maintained to allow go to
access a pure-C interface.

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(llama): Fix the impl of SampleTokenGreedy for new sampling

I don't think this method is currently used, so it could probably just be
removed so that all sampling goes through the GPT interface, but in the
interest of doing no harm, this should keep the method working as expected.

Branch: IBMGraniteArchitectureSupport

* fix(llama): Remove unused SampleTokenGreedy

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(sync): Remove bash-specific change to sync.sh

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* chore(gofumpt): Format on llama.go to pass linting

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(llm): Fix missing <thread> include in ext_server

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(llama): Remove TODO about grammar_first

This feature was not used/needed previously so should be fine without
plumbing it through now.

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(llama): Better naming for sampling wrapper and args

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(llama): Fix patch 05 to use new wrapper api and re-sync

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* runner: Flush pending responses before returning

If there are any pending reponses (such as from potential stop
tokens) then we should send them back before ending the sequence.
Otherwise, we can be missing tokens at the end of a response.

Fixes #6707

* fix(llama/sampling): Use gpt_sampler with a forward declaration

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(llama): Remove unnecessary patch for gguf impl header

This was caused by an earlier mistake in the embeddings patch that was
dereferencing the pointer instead of using the wrapper API.

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(llm): Remove use of deprecated --log-disable flag

Branch: IBMGraniteArchitectureSupport
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

---------
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

llama/ggml-cuda/diagmask.cu

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/diagmask.cuh

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/dmmv.cu

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/dmmv.cuh

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/fattn-common.cuh

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/fattn-tile-f16.cu

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/fattn-tile-f16.cuh

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/fattn-tile-f32.cu

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *
@@ -70,13 +70,17 @@ static __global__ void flash_attn_tile_ext_f32(
         const int ne1,
         const int ne2,
         const int ne3) {
+#ifndef FLASH_ATTN_AVAILABLE
+    NO_DEVICE_CODE;
+    return;
+#endif // FLASH_ATTN_AVAILABLE
     // Skip unused kernel variants for faster compilation:
     if (use_logit_softcap && !(D == 128 || D == 256)) {
         NO_DEVICE_CODE;
         return;
     }
 
-    //In this kernel Q, K, V are matrices while i, j, k are matrix indices.
+    // In this kernel Q, K, V are matrices while i, j, k are matrix indices.
 
     const int ic0 = (blockIdx.x / parallel_blocks) * ncols; // Index of the Q/QKV column to work on.
     const int ip  =  blockIdx.x % parallel_blocks; // Index in group of blocks running for the same column in parallel.

llama/ggml-cuda/fattn-tile-f32.cuh

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/fattn-vec-f16.cuh

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/fattn-vec-f32.cuh

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/fattn-wmma-f16.cuh

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/fattn.cu

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *
@@ -178,7 +178,7 @@ static void ggml_cuda_flash_attn_ext_wmma_f16(ggml_backend_cuda_context & ctx, g
     }                                                                       \
 
 static void ggml_cuda_flash_attn_ext_vec_f16(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
-    ggml_tensor * Q = dst->src[1];
+    ggml_tensor * Q = dst->src[0];
     ggml_tensor * K = dst->src[1];
     ggml_tensor * V = dst->src[2];
 
@@ -253,7 +253,7 @@ static void ggml_cuda_flash_attn_ext_vec_f16(ggml_backend_cuda_context & ctx, gg
     }                                                                       \
 
 static void ggml_cuda_flash_attn_ext_vec_f32(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
-    ggml_tensor * Q = dst->src[1];
+    ggml_tensor * Q = dst->src[0];
     ggml_tensor * K = dst->src[1];
     ggml_tensor * V = dst->src[2];
 
@@ -340,7 +340,7 @@ void ggml_cuda_flash_attn_ext(ggml_backend_cuda_context & ctx, ggml_tensor * dst
     }
 
     if (!fast_fp16_available(cc)) {
-        if (Q->ne[1] <= 8) {
+        if (Q->ne[1] <= 8 || Q->ne[0] == 256) {
             ggml_cuda_flash_attn_ext_vec_f32(ctx, dst);
         } else {
             ggml_cuda_flash_attn_ext_tile_f32(ctx, dst);

llama/ggml-cuda/fattn.cuh

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/getrows.cu

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/getrows.cuh

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/im2col.cu

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *
@@ -95,7 +95,6 @@ void ggml_cuda_op_im2col(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
     float * dst_d = (float *)dst->data;
     cudaStream_t stream = ctx.stream();
 
-    GGML_ASSERT(src0->type == GGML_TYPE_F16);
     GGML_ASSERT(src1->type == GGML_TYPE_F32);
     GGML_ASSERT( dst->type == GGML_TYPE_F16 || dst->type == GGML_TYPE_F32);
 

llama/ggml-cuda/im2col.cuh

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/mma.cuh

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/mmq.cu

 /**
- * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ * llama.cpp - commit 3f1ae2e32cde00c39b96be6d01c2997c29bae555 - do not edit this file
  *
  * MIT License
  *
@@ -52,7 +52,11 @@ void ggml_cuda_op_mul_mat_q(
     // nrows_dst == nrows of the matrix that the kernel writes into
     const int64_t nrows_dst = id == ctx.device ? ne0 : row_diff;
 
-    const mmq_args args = {src0_dd_i, src1_ddq_i, dst_dd_i, ne00, row_diff, stride00, src1_padded_row_size, src1_ncols, ne11, nrows_dst};
+    // The stream-k decomposition is only faster for recent NVIDIA GPUs.
+    // Also its fixup needs to allocate a temporary buffer in the memory pool.
+    // There are multiple parallel CUDA streams for src1_ncols != ne11 which would introduce a race condition for this buffer.
+    const bool use_stream_k = compute_capability >= CC_VOLTA && compute_capability < CC_OFFSET_AMD && src1_ncols == ne11;
+    const mmq_args args = {src0_dd_i, src1_ddq_i, dst_dd_i, ne00, row_diff, stride00, src1_padded_row_size, src1_ncols, ne11, nrows_dst, use_stream_k};
 
     switch (src0->type) {
         case GGML_TYPE_Q4_0: