phase 11: GPU-resident KV cache

- GpuKVCache: pre-allocated GPU buffers, D2D copy append at offset
- Per-head strided layout [num_kv_heads, max_seq_len, head_dim]
- Fixed critical bug: seq_len must advance AFTER all layers write
  (not inside the loop per-layer)
- GpuBuffer::copy_from_device_at for offset-based D2D copy
- Tensor::from_storage constructor for wrapping raw GPU buffers
- Exported Storage and Dims from xserv-tensor

Correctness: GPU KV cache vs CPU KV cache = 50/50 bit-identical
Performance: ~neutral (KV cache was never the main bottleneck —
reshape/merge/transpose CPU round-trips dominate for Qwen3-8B)

TTFT: 122ms, TBT: 142ms, 7.0 tok/s (marginal change from 7.3)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

crates/xserv-model/src/kv_cache.rs

@@ -0,0 +1,118 @@
+use xserv_cuda::GpuBuffer;
+use xserv_tensor::{DType, Device, Tensor};
+use crate::config::ModelConfig;
+
+/// GPU-resident KV cache. Pre-allocates max_seq_len on GPU,
+/// appends new K/V via D2D copy at offset (no CPU round-trip).
+pub struct GpuKVCache {
+    // Per layer: contiguous GPU buffer for K and V
+    // Layout: [num_kv_heads, max_seq_len, head_dim] — contiguous per head
+    k_bufs: Vec<GpuBuffer>,
+    v_bufs: Vec<GpuBuffer>,
+    seq_len: usize,
+    max_seq_len: usize,
+    num_kv_heads: usize,
+    head_dim: usize,
+    elem_size: usize,
+    dtype: DType,
+}
+
+impl GpuKVCache {
+    pub fn new(config: &ModelConfig, max_seq_len: usize, dtype: DType) -> Self {
+        let num_layers = config.num_layers();
+        let num_kv_heads = config.num_kv_heads();
+        let head_dim = config.head_dim();
+        let elem_size = dtype.size_bytes();
+        let buf_size = num_kv_heads * max_seq_len * head_dim * elem_size;
+
+        let mut k_bufs = Vec::with_capacity(num_layers);
+        let mut v_bufs = Vec::with_capacity(num_layers);
+        for _ in 0..num_layers {
+            let mut k = GpuBuffer::alloc(buf_size).expect("alloc KV cache K");
+            let mut v = GpuBuffer::alloc(buf_size).expect("alloc KV cache V");
+            k.zero().unwrap();
+            v.zero().unwrap();
+            k_bufs.push(k);
+            v_bufs.push(v);
+        }
+
+        Self { k_bufs, v_bufs, seq_len: 0, max_seq_len, num_kv_heads, head_dim, elem_size, dtype }
+    }
+
+    pub fn seq_len(&self) -> usize { self.seq_len }
+    pub fn max_seq_len(&self) -> usize { self.max_seq_len }
+
+    /// Append new K/V tensors for a given layer.
+    /// k_new, v_new: [1, num_kv_heads, new_tokens, head_dim] on GPU, contiguous.
+    /// `write_pos` is the sequence position to write at (caller manages this).
+    pub fn append(&mut self, layer: usize, k_new: &Tensor, v_new: &Tensor, new_tokens: usize, write_pos: usize) {
+        assert!(write_pos + new_tokens <= self.max_seq_len, "KV cache overflow");
+        let es = self.elem_size;
+        let hd = self.head_dim;
+        let max_s = self.max_seq_len;
+        let nh = self.num_kv_heads;
+
+        let k_src = k_new.storage().gpu_buffer();
+        let v_src = v_new.storage().gpu_buffer();
+
+        for h in 0..nh {
+            let src_off = h * new_tokens * hd * es;
+            let dst_off = (h * max_s + write_pos) * hd * es;
+            let count = new_tokens * hd * es;
+            self.k_bufs[layer].copy_from_device_at(k_src, src_off, dst_off, count).unwrap();
+            self.v_bufs[layer].copy_from_device_at(v_src, src_off, dst_off, count).unwrap();
+        }
+    }
+
+    pub fn advance_seq_len(&mut self, new_tokens: usize) {
+        self.seq_len += new_tokens;
+    }
+
+    /// Get K/V cache tensors for a layer up to `seq_len` tokens: [1, num_kv_heads, seq_len, head_dim]
+    pub fn get_kv(&self, layer: usize) -> (Tensor, Tensor) {
+        let sl = self.seq_len;
+        self.get_kv_len(layer, sl)
+    }
+
+    pub fn get_kv_len(&self, layer: usize, sl: usize) -> (Tensor, Tensor) {
+        let hd = self.head_dim;
+        let nh = self.num_kv_heads;
+        let es = self.elem_size;
+        let max_s = self.max_seq_len;
+
+        // Allocate output tensors [1, nh, sl, hd]
+        let out_size = nh * sl * hd * es;
+        let mut k_out = GpuBuffer::alloc(out_size).expect("alloc k_out");
+        let mut v_out = GpuBuffer::alloc(out_size).expect("alloc v_out");
+
+        // Copy each head's valid portion
+        for h in 0..nh {
+            let src_off = (h * max_s) * hd * es;
+            let dst_off = (h * sl) * hd * es;
+            let count = sl * hd * es;
+            k_out.copy_from_device_at(&self.k_bufs[layer], src_off, dst_off, count).unwrap();
+            v_out.copy_from_device_at(&self.v_bufs[layer], src_off, dst_off, count).unwrap();
+        }
+
+        let shape = &[1usize, nh, sl, hd];
+        let k = unsafe { tensor_from_gpu_buffer(k_out, shape, self.dtype) };
+        let v = unsafe { tensor_from_gpu_buffer(v_out, shape, self.dtype) };
+        (k, v)
+    }
+}
+
+/// Create a Tensor from a GpuBuffer (takes ownership).
+unsafe fn tensor_from_gpu_buffer(buf: GpuBuffer, shape: &[usize], dtype: DType) -> Tensor {
+    use xserv_tensor::storage::Storage;
+    use xserv_tensor::shape::contiguous_strides;
+    use smallvec::SmallVec;
+
+    let storage = Storage::cuda(buf);
+    Tensor::from_storage(
+        storage,
+        SmallVec::from_slice(shape),
+        contiguous_strides(shape),
+        0,
+        dtype,
+    )
+}