diff --git a/build.zig b/build.zig
index 2127a05..70921a9 100644
--- a/build.zig
+++ b/build.zig
@@ -29,6 +29,7 @@ pub fn build(b: *std.Build) void {
     });
 
     exe.addModule("arm32", arm32.module(b));
+    exe.addModule("zba-util", b.dependency("zba-util", .{}).module("zba-util"));
     exe.addModule("zig-clap", b.dependency("zig-clap", .{}).module("clap"));
 
     exe.addAnonymousModule("bitfield", .{ .source_file = .{ .path = "lib/bitfield.zig" } }); // https://github.com/FlorenceOS/
diff --git a/src/core/Scheduler.zig b/src/core/Scheduler.zig
index 4f8ccfc..197b68b 100644
--- a/src/core/Scheduler.zig
+++ b/src/core/Scheduler.zig
@@ -20,6 +20,15 @@ pub fn push(self: *@This(), kind: Event.Kind, offset: u64) void {
     self.queue.add(.{ .kind = kind, .tick = self.tick + offset }) catch unreachable;
 }
 
+pub fn remove(self: *@This(), needle: Event.Kind) void {
+    for (self.queue.items, 0..) |event, i| {
+        if (!std.meta.eql(event.kind, needle)) continue;
+
+        _ = self.queue.removeIndex(i); // note: invalidates self.queue.items
+        return;
+    }
+}
+
 pub fn deinit(self: @This()) void {
     self.queue.deinit();
 }
@@ -58,6 +67,8 @@ pub fn handle(self: *@This(), bus_ptr: ?*anyopaque, event: Event, late: u64) voi
                 },
                 .hblank => bus.ppu.onHblankEnd(self, late),
                 .vblank => bus.ppu.onHblankEnd(self, late),
+                .sqrt => bus.io.sqrt.onSqrtCalc(),
+                .div => bus.io.div.onDivCalc(),
             }
         },
     }
@@ -68,7 +79,7 @@ pub const Event = struct {
     kind: Kind,
 
     const Kind7 = enum {};
-    const Kind9 = enum { draw, hblank, vblank };
+    const Kind9 = enum { draw, hblank, vblank, sqrt, div };
 
     pub const Kind = union(enum) {
         nds7: Kind7,
diff --git a/src/core/nds9/io.zig b/src/core/nds9/io.zig
index 1421830..e46840d 100644
--- a/src/core/nds9/io.zig
+++ b/src/core/nds9/io.zig
@@ -7,6 +7,8 @@ const Bus = @import("Bus.zig");
 const SharedIo = @import("../io.zig").Io;
 const masks = @import("../io.zig").masks;
 
+const sext = @import("zba-util").sext;
+
 const log = std.log.scoped(.nds9_io);
 
 pub const Io = struct {
@@ -19,6 +21,10 @@ pub const Io = struct {
     // Read Only
     keyinput: AtomicKeyInput = .{},
 
+    /// DS Maths
+    div: Divisor = .{},
+    sqrt: SquareRootUnit = .{},
+
     pub fn init(io: *SharedIo) @This() {
         return .{ .shared = io };
     }
@@ -31,6 +37,8 @@ pub fn read(bus: *const Bus, comptime T: type, address: u32) T {
             0x0400_0210 => bus.io.shared.ie,
             0x0400_0214 => bus.io.shared.irq,
 
+            0x0400_02B4 => @truncate(bus.io.sqrt.result),
+
             0x0410_0000 => bus.io.shared.ipc_fifo.recv(.arm9),
             else => warn("unexpected: read(T: {}, addr: 0x{X:0>8}) {} ", .{ T, address, T }),
         },
@@ -40,6 +48,9 @@ pub fn read(bus: *const Bus, comptime T: type, address: u32) T {
 
             0x0400_0180 => @truncate(bus.io.shared.ipc_fifo.sync.raw),
             0x0400_0184 => @truncate(bus.io.shared.ipc_fifo.cnt.raw),
+
+            0x0400_02B0 => @truncate(bus.io.sqrt.cnt.raw),
+
             else => warn("unexpected: read(T: {}, addr: 0x{X:0>8}) {} ", .{ T, address, T }),
         },
         u8 => switch (address) {
@@ -68,6 +79,15 @@ pub fn write(bus: *Bus, comptime T: type, address: u32, value: T) void {
             0x0400_0210 => bus.io.shared.ie = value,
             0x0400_0214 => bus.io.shared.irq = value,
 
+            0x0400_02B8 => {
+                bus.io.sqrt.param = (bus.io.sqrt.param & ~@as(u64, 0xFFFF_FFFF)) | value;
+                bus.io.sqrt.schedule(bus.scheduler);
+            },
+            0x0400_02BC => {
+                bus.io.sqrt.param = (@as(u64, value) << 32) | (bus.io.sqrt.param & @as(u64, 0xFFFF_FFFF));
+                bus.io.sqrt.schedule(bus.scheduler);
+            },
+
             0x0400_0304 => bus.io.powcnt.raw = value,
 
             else => log.warn("unexpected: write(T: {}, addr: 0x{X:0>8}, value: 0x{X:0>8})", .{ T, address, value }),
@@ -77,6 +97,11 @@ pub fn write(bus: *Bus, comptime T: type, address: u32, value: T) void {
             0x0400_0184 => bus.io.shared.ipc_fifo.cnt.raw = masks.ipcFifoCnt(bus, value),
             0x0400_0208 => bus.io.shared.ime = value & 1 == 1,
 
+            0x0400_02B0 => {
+                bus.io.sqrt.cnt.raw = value;
+                bus.io.sqrt.schedule(bus.scheduler);
+            },
+
             else => log.warn("unexpected: write(T: {}, addr: 0x{X:0>8}, value: 0x{X:0>8})", .{ T, address, value }),
         },
         u8 => switch (address) {
@@ -107,6 +132,121 @@ const PowCnt = extern union {
     raw: u32,
 };
 
+/// Divisor
+const Divisor = struct {
+    const Scheduler = @import("../Scheduler.zig");
+
+    /// DIVCNT - Division Control (R/W)
+    cnt: Cnt = .{ .raw = 0x0000_0000 },
+    /// DIV_NUMER - division numerator (R/W)
+    numerator: u64 = 0x0000_0000_0000_0000,
+    /// DIV_DENOM - division denominator (R/W)
+    denominator: u64 = 0x0000_0000_0000_0000,
+
+    /// DIV_RESULT - division quotient (R)
+    result: u64 = 0x0000_0000_0000_0000,
+    /// DIVREM_RESULT - remainder
+    remainder: u64 = 0x0000_0000_0000_0000,
+
+    const Cnt = extern union {
+        mode: Bitfield(u32, 0, 2),
+        div_by_zero: Bit(u32, 14),
+        busy: Bit(u32, 15),
+        raw: u32,
+    };
+
+    pub fn schedule(self: *@This(), scheduler: *Scheduler) void {
+        self.cnt.busy.set();
+
+        const cycle_count = switch (self.cnt.mode.read()) {
+            0b00 => 18,
+            0b01, 0b10, 0b11 => 34,
+        };
+
+        scheduler.remove(.{ .nds9 = .{.div} });
+        scheduler.push(.{ .nds9 = .{.div} }, cycle_count);
+    }
+
+    pub fn onDivCalc(self: *@This()) void {
+        self.cnt.busy.unset();
+
+        // const mask: u64 = blk: {
+        //     const value: u64 = @intFromBool(self.cnt.mode.read());
+        //     break :blk value -| 1;
+        // };
+
+        if (self.denominator == 0) {
+            log.err("TODO: deal with division error ({} / {})", .{ self.numerator, self.denominator });
+            return;
+        }
+
+        switch (self.cnt.mode.read()) {
+            0b00 => {
+                // 32bit / 32bit = 32bit , 32bit
+                const left = self.numerator & 0xFFFF_FFFFF;
+                const right = self.denominator & 0xFFFF_FFFF;
+
+                self.result = sext(u64, u32, left / right);
+                self.remainder = sext(u64, u32, left % right);
+            },
+            0b01, 0b11 => {
+                // 64bit / 32bit = 64bit , 32bit
+                const left = self.numerator;
+                const right = self.denominator & 0xFFFF_FFFF;
+
+                self.result = left / right;
+                self.remainder = sext(u64, u32, left % right);
+            },
+            0b10 => {
+                // 64bit / 64bit = 64bit , 64bit
+                const left = self.numerator;
+                const right = self.denominator;
+
+                self.result = left / right;
+                self.remainder = left % right;
+            },
+        }
+    }
+};
+
+/// Square Root Unit
+const SquareRootUnit = struct {
+    const Scheduler = @import("../Scheduler.zig");
+
+    /// SQRTCNT - Division Control (R/W)
+    cnt: Cnt = .{ .raw = 0x0000_0000 },
+
+    /// SQRT_RESULT - square root result (R)
+    result: u32 = 0x0000_0000,
+    /// SQRT_PARAM - square root paramater input (R/W)
+    param: u64 = 0x0000_0000_0000_0000,
+
+    const Cnt = extern union {
+        mode: Bit(u32, 0),
+        busy: Bit(u32, 15),
+        raw: u32,
+    };
+
+    pub fn schedule(self: *@This(), scheduler: *Scheduler) void {
+        defer self.cnt.busy.set();
+
+        scheduler.remove(.{ .nds9 = .sqrt });
+        scheduler.push(.{ .nds9 = .sqrt }, 13); // always takes 13 cycles
+    }
+
+    pub fn onSqrtCalc(self: *@This()) void {
+        defer self.cnt.busy.unset();
+
+        const mask: u64 = blk: {
+            const value: u64 = @intFromBool(!self.cnt.mode.read());
+            break :blk (value << 32) -% 1;
+        };
+
+        self.result = @truncate(std.math.sqrt(self.param & mask));
+        std.log.debug("sqrt({}) == {}", .{ self.param & mask, self.result });
+    }
+};
+
 pub const DispcntA = extern union {
     bg_mode: Bitfield(u32, 0, 2),