add dct coeff generation with zig-zag order at compile time

src/transform.zig

@@ -2,7 +2,63 @@ const std = @import("std");
 
 const util = @import("util.zig");
 
-pub fn quantize(source: *util.Block, target: *util.BlockQuantized) void {
-    _ = target;
-    _ = source;
+pub const dct_coeffs = gen_coeffs();
+
+inline fn dct_cos(x: usize, f: usize) f16 {
+    const x_float: f16 = @floatFromInt(x);
+    const f_float: f16 = @floatFromInt(f);
+    return @cos((2.0 * x_float + 1.0) * f_float * std.math.pi / 16.0);
+}
+
+inline fn dct_coeff(u: usize, v: usize) f16 {
+    return 0.25 * (if (u == 0) 1.0 / @sqrt(2.0) else 1.0) * (if (v == 0) 1.0 / @sqrt(2.0) else 1.0);
+}
+
+inline fn zz_conv(u: usize, v: usize) struct { u: usize, v: usize } {
+    var band_i = u + v;
+    const band_max_u = @min(7, band_i);
+    const band_max_v = @min(7, band_i);
+    var idx: usize = 0;
+    for (0..band_i) |i| {
+        idx += zz_band_len(i);
+    }
+    if (band_i % 2 == 0) {
+        idx += band_max_v - v;
+    } else {
+        idx += band_max_u - u;
+    }
+    return .{
+        .u = idx / 8,
+        .v = idx % 8,
+    };
+}
+
+inline fn zz_band_len(band_i: usize) usize {
+    return if (band_i < 8) band_i + 1 else 15 - band_i;
+}
+
+fn gen_coeffs() [8][8]@Vector(64, f16) {
+    @setEvalBranchQuota(100000);
+    var ret: [8][8]@Vector(64, f16) = undefined;
+    for (0..8) |u| {
+        for (0..8) |v| {
+            const zz_idx = zz_conv(u, v);
+            for (0..8) |x| {
+                for (0..8) |y| {
+                    ret[zz_idx.u][zz_idx.v][x * 8 + y] = dct_coeff(zz_idx.u, zz_idx.v) * dct_cos(x, zz_idx.u) * dct_cos(y, zz_idx.v);
+                }
+            }
+        }
+    }
+    @setEvalBranchQuota(1000);
+    return ret;
+}
+
+pub fn quantize(source: *util.Block, target: *util.BlockQuantized, qtable: *util.QTable) void {
+    _ = qtable;
+    _ = target;
+    var holder: @Vector(64, f16) = undefined;
+    for (0..64) |i| {
+        holder[i] = @floatFromInt(source[i]);
+    }
 }