llama_cpp 0.8.0 → 0.9.1

data/ext/llama_cpp/src/ggml-metal.metal

@@ -125,9 +125,17 @@ kernel void kernel_mul_row(
 }
 
 kernel void kernel_scale(
+        device const float * src0,
+        device       float * dst,
+        constant     float & scale,
+        uint tpig[[thread_position_in_grid]]) {
+    dst[tpig] = src0[tpig] * scale;
+}
+
+kernel void kernel_scale_4(
         device const float4 * src0,
         device       float4 * dst,
-        constant     float & scale,
+        constant     float  & scale,
         uint tpig[[thread_position_in_grid]]) {
     dst[tpig] = src0[tpig] * scale;
 }
@@ -176,36 +184,73 @@ kernel void kernel_soft_max(
         constant   int64_t & ne00,
         constant   int64_t & ne01,
         constant   int64_t & ne02,
-        uint3 tgpig[[threadgroup_position_in_grid]],
-        uint3 tpitg[[thread_position_in_threadgroup]],
-        uint3   ntg[[threads_per_threadgroup]]) {
-    const int64_t i03 = tgpig[2];
-    const int64_t i02 = tgpig[1];
-    const int64_t i01 = tgpig[0];
+        threadgroup float  * buf [[threadgroup(0)]],
+        uint  tgpig[[threadgroup_position_in_grid]],
+        uint  tpitg[[thread_position_in_threadgroup]],
+        uint  sgitg[[simdgroup_index_in_threadgroup]],
+        uint  tiisg[[thread_index_in_simdgroup]],
+        uint    ntg[[threads_per_threadgroup]]) {
+    const int64_t i03 = (tgpig) / (ne02*ne01);
+    const int64_t i02 = (tgpig - i03*ne02*ne01) / ne01;
+    const int64_t i01 = (tgpig - i03*ne02*ne01 - i02*ne01);
 
     device const float * psrc0 = src0 + i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00;
     device       float * pdst  = dst  + i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00;
 
     // parallel max
-    float lmax = tpitg[0] < ne00 ? psrc0[tpitg[0]] : -INFINITY;
-    for (int i00 = tpitg[0] + ntg[0]; i00 < ne00; i00 += ntg[0]) {
+    float lmax = tpitg < ne00 ? psrc0[tpitg] : -INFINITY;
+
+    for (int i00 = tpitg + ntg; i00 < ne00; i00 += ntg) {
         lmax = MAX(lmax, psrc0[i00]);
     }
-    const float max = simd_max(lmax);
+
+    float max = simd_max(lmax);
+    if (tiisg == 0) {
+        buf[sgitg] = max;
+    }
+
+    threadgroup_barrier(mem_flags::mem_threadgroup);
+
+    // broadcast, simd group number is ntg / 32
+    for (uint i = ntg / 32 / 2; i > 0; i /= 2) {
+       if (tpitg < i) {
+           buf[tpitg] = MAX(buf[tpitg], buf[tpitg + i]);
+       }
+    }
+
+    threadgroup_barrier(mem_flags::mem_threadgroup);
+
+    max = buf[0];
 
     // parallel sum
     float lsum = 0.0f;
-    for (int i00 = tpitg[0]; i00 < ne00; i00 += ntg[0]) {
+    for (int i00 = tpitg; i00 < ne00; i00 += ntg) {
         const float exp_psrc0 = exp(psrc0[i00] - max);
         lsum += exp_psrc0;
         // Remember the result of exp here. exp is expensive, so we really do not
-        // whish to compute it twice.
+        // wish to compute it twice.
         pdst[i00] = exp_psrc0;
     }
 
-    const float sum = simd_sum(lsum);
+    float sum = simd_sum(lsum);
+    if (tiisg == 0) {
+        buf[sgitg] = sum;
+    }
+
+    threadgroup_barrier(mem_flags::mem_threadgroup);
 
-    for (int i00 = tpitg[0]; i00 < ne00; i00 += ntg[0]) {
+    // broadcast, simd group number is ntg / 32
+    for (uint i = ntg / 32 / 2; i > 0; i /= 2) {
+       if (tpitg < i) {
+           buf[tpitg] += buf[tpitg + i];
+       }
+    }
+
+    threadgroup_barrier(mem_flags::mem_threadgroup);
+
+    sum = buf[0];
+
+    for (int i00 = tpitg; i00 < ne00; i00 += ntg) {
         pdst[i00] /= sum;
     }
 }
@@ -216,37 +261,73 @@ kernel void kernel_soft_max_4(
         constant   int64_t & ne00,
         constant   int64_t & ne01,
         constant   int64_t & ne02,
-        uint3 tgpig[[threadgroup_position_in_grid]],
-        uint3 tpitg[[thread_position_in_threadgroup]],
-        uint3   ntg[[threads_per_threadgroup]]) {
-    const int64_t i03 = tgpig[2];
-    const int64_t i02 = tgpig[1];
-    const int64_t i01 = tgpig[0];
+        threadgroup float  * buf [[threadgroup(0)]],
+        uint  tgpig[[threadgroup_position_in_grid]],
+        uint  tpitg[[thread_position_in_threadgroup]],
+        uint  sgitg[[simdgroup_index_in_threadgroup]],
+        uint  tiisg[[thread_index_in_simdgroup]],
+        uint    ntg[[threads_per_threadgroup]]) {
+    const int64_t i03 = (tgpig) / (ne02*ne01);
+    const int64_t i02 = (tgpig - i03*ne02*ne01) / ne01;
+    const int64_t i01 = (tgpig - i03*ne02*ne01 - i02*ne01);
 
     device const float4 * psrc4 = (device const float4 *)(src0 + i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00);
     device       float4 * pdst4 = (device       float4 *)(dst  + i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00);
 
     // parallel max
-    float4 lmax4 = tpitg[0] < ne00/4 ? psrc4[tpitg[0]] : -INFINITY;
-    for (int i00 = tpitg[0] + ntg[0]; i00 < ne00/4; i00 += ntg[0]) {
+    float4 lmax4 = tpitg < ne00/4 ? psrc4[tpitg] : -INFINITY;
+
+    for (int i00 = tpitg + ntg; i00 < ne00/4; i00 += ntg) {
         lmax4 = fmax(lmax4, psrc4[i00]);
     }
-    float lmax = MAX(MAX(lmax4[0], lmax4[1]), MAX(lmax4[2], lmax4[3]));
 
-    const float max = simd_max(lmax);
+    const float lmax = MAX(MAX(lmax4[0], lmax4[1]), MAX(lmax4[2], lmax4[3]));
+    float max = simd_max(lmax);
+    if (tiisg == 0) {
+        buf[sgitg] = max;
+    }
+
+    threadgroup_barrier(mem_flags::mem_threadgroup);
+
+    // broadcast, simd group number is ntg / 32
+    for (uint i = ntg / 32 / 2; i > 0; i /= 2) {
+       if (tpitg < i) {
+           buf[tpitg] = MAX(buf[tpitg], buf[tpitg + i]);
+       }
+    }
+
+    threadgroup_barrier(mem_flags::mem_threadgroup);
+
+    max = buf[0];
 
     // parallel sum
     float4 lsum4 = 0.0f;
-    for (int i00 = tpitg[0]; i00 < ne00/4; i00 += ntg[0]) {
+    for (int i00 = tpitg; i00 < ne00/4; i00 += ntg) {
         const float4 exp_psrc4 = exp(psrc4[i00] - max);
         lsum4 += exp_psrc4;
         pdst4[i00] = exp_psrc4;
     }
-    float lsum = lsum4[0] + lsum4[1] + lsum4[2] + lsum4[3];
 
-    const float sum = simd_sum(lsum);
+    const float lsum = lsum4[0] + lsum4[1] + lsum4[2] + lsum4[3];
+    float sum = simd_sum(lsum);
+    if (tiisg == 0) {
+        buf[sgitg] = sum;
+    }
+
+    threadgroup_barrier(mem_flags::mem_threadgroup);
+
+    // broadcast, simd group number is ntg / 32
+    for (uint i = ntg / 32 / 2; i > 0; i /= 2) {
+       if (tpitg < i) {
+           buf[tpitg] += buf[tpitg + i];
+       }
+    }
+
+    threadgroup_barrier(mem_flags::mem_threadgroup);
+
+    sum = buf[0];
 
-    for (int i00 = tpitg[0]; i00 < ne00/4; i00 += ntg[0]) {
+    for (int i00 = tpitg; i00 < ne00/4; i00 += ntg) {
         pdst4[i00] /= sum;
     }
 }
@@ -266,7 +347,7 @@ kernel void kernel_diag_mask_inf(
         dst[i02*ne01*ne00 + i01*ne00 + i00] = -INFINITY;
     } else {
         dst[i02*ne01*ne00 + i01*ne00 + i00] = src0[i02*ne01*ne00 + i01*ne00 + i00];
-     }
+    }
 }
 
 kernel void kernel_diag_mask_inf_8(
@@ -980,6 +1061,45 @@ kernel void kernel_alibi_f32(
     }
 }
 
+static float rope_yarn_ramp(const float low, const float high, const int i0) {
+    const float y = (i0 / 2 - low) / max(0.001f, high - low);
+    return 1.0f - min(1.0f, max(0.0f, y));
+}
+
+// YaRN algorithm based on LlamaYaRNScaledRotaryEmbedding.py from https://github.com/jquesnelle/yarn
+// MIT licensed. Copyright (c) 2023 Jeffrey Quesnelle and Bowen Peng.
+static void rope_yarn(
+    float theta_extrap, float freq_scale, float corr_dims[2], int64_t i0, float ext_factor, float mscale,
+    thread float * cos_theta, thread float * sin_theta
+) {
+    // Get n-d rotational scaling corrected for extrapolation
+    float theta_interp = freq_scale * theta_extrap;
+    float theta = theta_interp;
+    if (ext_factor != 0.0f) {
+        float ramp_mix = rope_yarn_ramp(corr_dims[0], corr_dims[1], i0) * ext_factor;
+        theta = theta_interp * (1 - ramp_mix) + theta_extrap * ramp_mix;
+
+        // Get n-d magnitude scaling corrected for interpolation
+        mscale *= 1.0f + 0.1f * log(1.0f / freq_scale);
+    }
+    *cos_theta = cos(theta) * mscale;
+    *sin_theta = sin(theta) * mscale;
+}
+
+// Apparently solving `n_rot = 2pi * x * base^((2 * max_pos_emb) / n_dims)` for x, we get
+// `corr_fac(n_rot) = n_dims * log(max_pos_emb / (n_rot * 2pi)) / (2 * log(base))`
+static float rope_yarn_corr_factor(int n_dims, int n_orig_ctx, float n_rot, float base) {
+    return n_dims * log(n_orig_ctx / (n_rot * 2 * M_PI_F)) / (2 * log(base));
+}
+
+static void rope_yarn_corr_dims(
+    int n_dims, int n_orig_ctx, float freq_base, float beta_fast, float beta_slow, float dims[2]
+) {
+    // start and end correction dims
+    dims[0] = max(0.0f,         floor(rope_yarn_corr_factor(n_dims, n_orig_ctx, beta_fast, freq_base)));
+    dims[1] = min(n_dims - 1.0f, ceil(rope_yarn_corr_factor(n_dims, n_orig_ctx, beta_slow, freq_base)));
+}
+
 typedef void (rope_t)(
         device const    void * src0,
         device const int32_t * src1,
@@ -1003,8 +1123,13 @@ typedef void (rope_t)(
         constant         int & n_past,
         constant         int & n_dims,
         constant         int & mode,
+        constant         int & n_orig_ctx,
         constant       float & freq_base,
         constant       float & freq_scale,
+        constant       float & ext_factor,
+        constant       float & attn_factor,
+        constant       float & beta_fast,
+        constant       float & beta_slow,
         uint  tiitg[[thread_index_in_threadgroup]],
         uint3 tptg[[threads_per_threadgroup]],
         uint3 tgpig[[threadgroup_position_in_grid]]);
@@ -1033,8 +1158,13 @@ kernel void kernel_rope(
         constant         int & n_past,
         constant         int & n_dims,
         constant         int & mode,
+        constant         int & n_orig_ctx,
         constant       float & freq_base,
         constant       float & freq_scale,
+        constant       float & ext_factor,
+        constant       float & attn_factor,
+        constant       float & beta_fast,
+        constant       float & beta_slow,
         uint  tiitg[[thread_index_in_threadgroup]],
         uint3 tptg[[threads_per_threadgroup]],
         uint3 tgpig[[threadgroup_position_in_grid]]) {
@@ -1044,19 +1174,22 @@ kernel void kernel_rope(
 
     const bool is_neox = mode & 2;
 
+    float corr_dims[2];
+    rope_yarn_corr_dims(n_dims, n_orig_ctx, freq_base, beta_fast, beta_slow, corr_dims);
+
     device const int32_t * pos = src1;
 
     const int64_t p = pos[i2];
 
-    const float theta_0 = freq_scale * (float)p;
+    const float theta_0 = (float)p;
     const float inv_ndims = -1.f/n_dims;
 
     if (!is_neox) {
         for (int64_t i0 = 2*tiitg; i0 < ne0; i0 += 2*tptg.x) {
 
             const float theta = theta_0 * pow(freq_base, inv_ndims*i0);
-            const float cos_theta = cos(theta);
-            const float sin_theta = sin(theta);
+            float cos_theta, sin_theta;
+            rope_yarn(theta, freq_scale, corr_dims, i0, ext_factor, attn_factor, &cos_theta, &sin_theta);
 
             device const T * const src = (device T *)((device char *) src0 + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
             device       T * dst_data  = (device T *)((device char *)  dst + i3*nb3  + i2*nb2  + i1*nb1  + i0*nb0);
@@ -1071,9 +1204,12 @@ kernel void kernel_rope(
         for (int64_t ib = 0; ib < ne0/n_dims; ++ib) {
             for (int64_t ic = 2*tiitg; ic < n_dims; ic += 2*tptg.x) {
 
-                const float theta = theta_0 * pow(freq_base, inv_ndims*ic - ib);
-                const float cos_theta = cos(theta);
-                const float sin_theta = sin(theta);
+                // simplified from `(ib * n_dims + ic) * inv_ndims`
+                const float cur_rot = inv_ndims*ic - ib;
+
+                const float theta = theta_0 * pow(freq_base, cur_rot);
+                float cos_theta, sin_theta;
+                rope_yarn(theta, freq_scale, corr_dims, cur_rot, ext_factor, attn_factor, &cos_theta, &sin_theta);
 
                 const int64_t i0 = ib*n_dims + ic/2;