diff --git a/sim/CMakeLists.txt b/sim/CMakeLists.txt
index 850bdf2..5d652d8 100644
--- a/sim/CMakeLists.txt
+++ b/sim/CMakeLists.txt
@@ -41,6 +41,8 @@ target_sources(sim PRIVATE
     src/main.cpp
 )
 
+target_include_directories(sim PRIVATE inc/)
+
 
 
 list(APPEND VSOURCES ../verilog/hub75e.sv ../verilog/lineram.v)
diff --git a/sim/inc/devices.hpp b/sim/inc/devices.hpp
new file mode 100644
index 0000000..5d3354e
--- /dev/null
+++ b/sim/inc/devices.hpp
@@ -0,0 +1,150 @@
+// Project-specific cosimuluated devices.
+#pragma once
+#include "tests.hpp"
+#include "Vhub75e.h"
+
+// slices the RGB values for us.
+uint8_t rgb_slice(uint32_t rgb, uint8_t bit) {
+  if (bit > 8) {
+    // todo: panic
+    return 0;
+  }
+  uint8_t r = (rgb >> (16 + bit)) & 1;
+  uint8_t g = (rgb >> (8 + bit)) & 1;
+  uint8_t b = (rgb >> bit) & 1;
+  return (r << 2) & (g << 1) & (b << 1);
+}
+
+void rgb_unslice(unsigned int &rgb, uint8_t bits, uint8_t bitpos) {
+  if (bitpos > 7 || bits > 0b111) {
+    // TODO: panic.
+    return;
+  }
+  auto r = (bits >> 2) & 1;
+  auto g = (bits >> 1) & 1;
+  auto b = (bits >> 0) & 1;
+
+  rgb |= r << bitpos << 16;
+  rgb |= g << bitpos << 8;
+  rgb |= b << bitpos << 0;
+}
+
+class HUB75Reciever : public CosimulatedDevice {
+
+  typedef std::vector<unsigned char> row_array;
+  int xsize;
+  int ysize;
+
+  row_array row0{};
+  row_array row1{};
+
+  // the previous row values that were latched in.
+  std::vector<std::pair<row_array, row_array>> past_rows{};
+  // the pulse width for each output, in clock cycles.
+  std::vector<int> pulse_widths{};
+
+  int bit_position = 7; // the bit that is currently being shifted in
+
+  int output_period_cnt;
+  // if oe = 0, count clocks. when oe = 1, store value into
+  // pulse_widths[display_bit];
+
+  // previous latch value, used to identify when to latch.
+  unsigned char prev_latch = 0;
+  // previous display clock value, used to detect rising edge.
+  unsigned char prev_display_clk = 0;
+
+  unsigned char prev_clk = 0;
+
+  unsigned char prev_oe = 1; // assuming starting high.
+
+  // references to the panel driver signals.
+  VL_IN8(&display_clk, 0, 0);
+  VL_IN8(&out_enable, 0, 0);
+  VL_IN8(&latch, 0, 0);
+  VL_IN8(&rgb0, 2, 0);
+  VL_IN8(&rgb1, 2, 0);
+  VL_IN8(&clk, 0, 0);
+
+public:
+  HUB75Reciever(int xsize, int ysize, const Vhub75e &dut)
+      : clk(dut.clk), display_clk(dut.display_clk), out_enable(dut.out_enable),
+        latch(dut.latch), rgb0(dut.panel_rgb0), rgb1(dut.panel_rgb1) {
+    this->xsize = xsize;
+    this->ysize = ysize;
+    row0.clear();
+    prev_oe = out_enable;
+    prev_display_clk = display_clk;
+    prev_latch = latch;
+    prev_clk = clk;
+  };
+
+  // evaluates the reciever.
+  virtual void tick() override {
+
+    if (prev_display_clk == 0 && display_clk == 1) {
+      // display clock rising edge.
+      row0.push_back(rgb0);
+      row1.push_back(rgb1);
+    }
+
+    if (prev_latch == 0 && latch == 1) {
+      // latch in the data: reverse the rows, and pu
+      std::reverse(row0.begin(), row0.end());
+      std::reverse(row1.begin(), row1.end());
+      past_rows.push_back(std::pair(row0, row1));
+      row0.clear();
+      row1.clear();
+    }
+
+    if (prev_clk == 0 && clk == 1) {
+      if (out_enable == 0) {
+        if (prev_oe == 1) {
+          // falling edge.
+          output_period_cnt = 1;
+        } else {
+          output_period_cnt++;
+        }
+      } else { // out_enable == 1
+        if (prev_oe == 1) {
+          // do nothing
+        }
+        if (prev_oe == 0) {
+          // rising edge
+          pulse_widths.push_back(output_period_cnt);
+        }
+      }
+    }
+
+    // update previous values
+    prev_display_clk = display_clk;
+    prev_latch = latch;
+    prev_oe = out_enable;
+    prev_clk = clk;
+  }
+
+  const auto &get_past_rows() { return this->past_rows; }
+  const std::vector<int> &get_pulse_widths() { return this->pulse_widths; }
+
+  // return the RGB version.
+  std::pair<std::vector<unsigned int>, std::vector<unsigned int>> transpose() {
+    auto r0rgb = std::vector<unsigned int>(xsize, 0);
+    auto r1rgb = std::vector<unsigned int>(xsize, 0);
+
+    auto bitdepth = pulse_widths.size();
+
+    // TODO: use more sophisticated slicing.
+    auto slice = bitdepth - 1;
+    for (const auto &[row0slice, row1slice] : this->past_rows) {
+
+      for (int i = 0; i < row0slice.size(); i++) {
+        rgb_unslice(r0rgb[i], row0slice[i], slice);
+        rgb_unslice(r1rgb[i], row1slice[i], slice);
+      }
+
+      slice--;
+    }
+
+    return std::pair(r0rgb, r1rgb);
+  }
+};
diff --git a/sim/inc/tests.hpp b/sim/inc/tests.hpp
new file mode 100644
index 0000000..208962e
--- /dev/null
+++ b/sim/inc/tests.hpp
@@ -0,0 +1,204 @@
+// Common Verilator Simulation/test constructs.
+// Includes a simulation-fixture useful for writing tests.
+#pragma once
+#include "verilated.h"
+#include "verilated_vcd_c.h"
+#include <cstdint>
+#include <span>
+#include <memory>
+#include <queue>
+#include <vector>
+
+
+// represents a generic co-simulated device.
+// While this abstract class is very simple, it enables dynamic behavior to be added to the
+// simulation fixture.
+class CosimulatedDevice {
+public:
+  virtual ~CosimulatedDevice(){};
+
+  virtual void tick() = 0;
+};
+
+
+// Simple stimulus class used to trigger basic operations
+class PulseStimulus : public CosimulatedDevice {
+  unsigned long width;
+  unsigned char &signal;
+
+  unsigned long counter = 0;
+
+public:
+  PulseStimulus(unsigned char &signal, unsigned long width) : signal(signal) {
+    this->width = width;
+  }
+  virtual void tick() override {
+    if (counter < width) {
+      signal = 1;
+    } else {
+      signal = 0;
+    }
+    counter++;
+  }
+};
+
+class FakeBRAM : public CosimulatedDevice {
+  std::array<unsigned long, 512> ram{};
+
+  std::queue<unsigned long> addr_lookup_q;
+
+  unsigned short &addr_in;
+  unsigned long &data_out;
+
+  unsigned char &clk;
+
+public:
+  FakeBRAM(int latency, unsigned char &clk, unsigned short &addr_in,
+           unsigned long &data_out)
+      : addr_lookup_q(), addr_in(addr_in), data_out(data_out), clk(clk) {
+    for (int i = 0; i < latency; i++) {
+      addr_lookup_q.push(0);
+    }
+
+    for (int i = 0; i < ram.size(); i++) {
+      ram[i] = i + 3;
+    }
+  };
+
+  FakeBRAM(int latency, unsigned char &clk, unsigned short &addr_in,
+           unsigned long &data_out, std::array<unsigned long, 512> ram)
+      : addr_lookup_q(), addr_in(addr_in), data_out(data_out), clk(clk),
+        ram(ram) {
+    for (int i = 0; i < latency; i++) {
+      addr_lookup_q.push(0);
+    }
+  }
+
+  void tick() {
+    // we push and pop in the same tick: this way we keep the queue the same
+    // size, acting as a pipeline delay.
+    addr_lookup_q.push(addr_in);
+
+    auto addr_to_load = addr_lookup_q.front();
+    addr_lookup_q.pop();
+    data_out = ram.at(addr_to_load);
+  }
+
+  // TODO: allow accessing/setting the data
+  const std::span<uint64_t, 512> get() { return this->ram; }
+
+  void write(unsigned short addr, unsigned long data) { ram[addr] = data; }
+};
+
+
+// test fixture to reduce amount of runtime code.
+// Supports:
+// adding external modules
+// running the test
+// storing if the done flag was raised (or not)
+//
+// TODO: tracing.
+template <typename DUT> class VerilatorTestFixture {
+public:
+  enum class FinishReason { Ok, Timeout };
+
+private:
+  // we call .tick() on all of these. They are bound externally to the DUT.
+  std::vector<std::shared_ptr<CosimulatedDevice>> external_devices;
+  std::unique_ptr<VerilatedContext> ctx;
+  std::unique_ptr<DUT> dut;
+
+  unsigned long timeout =
+      1000000; // clock cycles to execute before ending the simulation.
+  unsigned long simtime = 0;
+  unsigned long posedge = 0;
+
+  // stores the termination-condition for the simulation.
+  // if false, it means we timed out. If true, our done_condition returned true.
+  enum FinishReason reason;
+
+  typedef std::function<bool(DUT &, unsigned long)> done_callback;
+  // callback function to determine execution completion.
+  // This can be used to add arbitrary finish-conditions to the execution.
+  // If the function returns true, we set the done_set boolean to true;
+  done_callback done_check;
+
+  std::unique_ptr<VerilatedVcdC> trace;
+
+public:
+  // Create a new test fixture with a given timeout. Everything else (including
+  // done-condition) can be set later.
+  VerilatorTestFixture() {
+    ctx = std::make_unique<VerilatedContext>();
+    dut = std::make_unique<DUT>(ctx.get(), "dut");
+    dut->eval(); // let values settle before adding modules.
+  }
+
+  void set_timeout(unsigned long new_timeout) { this->timeout = new_timeout; }
+
+  void set_done_callback(done_callback d) { this->done_check = d; }
+
+  void add_module(std::shared_ptr<CosimulatedDevice> device) {
+    external_devices.push_back(device);
+  }
+
+  void enable_trace(std::string name) {
+    if (!trace) {
+      ctx->traceEverOn(true);
+      trace = std::make_unique<VerilatedVcdC>();
+      dut->trace(trace.get(), 99);
+      trace->open(name.c_str());
+    }
+  }
+
+  void exec() {
+    bool done = false;
+    dut->eval(); // pre-eval.
+    while (!done) {
+      dut->clk ^= 1;
+      dut->eval();
+
+      if (trace) {
+        trace->dump(10 * simtime);
+      }
+
+      if (dut->clk == 1) {
+        posedge++;
+      }
+      if (done_check) {
+        if (done_check(*dut, posedge)) {
+          reason = FinishReason::Ok;
+          done = true;
+        }
+      }
+
+      if (posedge >= timeout) {
+        reason = FinishReason::Timeout;
+        done = true;
+      }
+      // run our external devices
+      for (auto dev : external_devices) {
+        dev->tick();
+      }
+      if (trace) {
+        trace->dump(10 * simtime + 5);
+      }
+      dut->eval(); // allow combinational logic to settle if it's being set on
+                   // the negative clock.
+      if (trace) {
+        trace->dump(10 * simtime + 6);
+      }
+      simtime++;
+    }
+
+    if (trace) {
+      // close the trace.
+      trace->close();
+    }
+  }
+
+  // return a reference to the DUT itself. Useful for bespoke tests.
+  const DUT &get() { return *dut; }
+
+  const FinishReason get_reason() { return reason; }
+};
diff --git a/sim/src/main.cpp b/sim/src/main.cpp
index a55eb4a..be0bf13 100644
--- a/sim/src/main.cpp
+++ b/sim/src/main.cpp
@@ -1,163 +1,14 @@
 #include "Vhub75e.h"
+#include "tests.hpp"
+#include "devices.hpp"
 #include "verilated.h"
 #include "verilated_vcd_c.h"
 #include <array>
 #include <catch2/catch_test_macros.hpp>
-#include <catch2/matchers/catch_matchers.hpp>
-#include <catch2/matchers/catch_matchers_all.hpp>
-#include <cstdint>
 #include <memory>
-#include <queue>
 #include <stdio.h>
 #include <vector>
 
-class CosimulatedDevice {
-public:
-  virtual ~CosimulatedDevice(){};
-
-  virtual void tick() = 0;
-};
-
-// slices the RGB values for us.
-uint8_t rgb_slice(uint32_t rgb, uint8_t bit) {
-  if (bit > 8) {
-    // todo: panic
-    return 0;
-  }
-  uint8_t r = (rgb >> (16 + bit)) & 1;
-  uint8_t g = (rgb >> (8 + bit)) & 1;
-  uint8_t b = (rgb >> bit) & 1;
-  return (r << 2) & (g << 1) & (b << 1);
-}
-
-class HUB75Reciever : public CosimulatedDevice {
-
-  typedef std::vector<unsigned char> row_array;
-  int xsize;
-  int ysize;
-
-  row_array row_upper{};
-  row_array row_lower{};
-
-  // the previous row values that were latched in.
-  std::vector<row_array> past_rows{};
-  // the pulse width for each output, in clock cycles.
-  std::vector<int> pulse_widths{};
-
-  int bit_position = 7; // the bit that is currently being shifted in
-
-  int output_period_cnt;
-  // if oe = 0, count clocks. when oe = 1, store value into
-  // pulse_widths[display_bit];
-
-  // previous latch value, used to identify when to latch.
-  unsigned char prev_latch = 0;
-  // previous display clock value, used to detect rising edge.
-  unsigned char prev_display_clk = 0;
-
-  unsigned char prev_oe = 0;
-
-  // references to the panel driver signals.
-  VL_IN8(&display_clk, 0, 0);
-  VL_IN8(&out_enable, 0, 0);
-  VL_IN8(&latch, 0, 0);
-  VL_IN8(&rgb0, 2, 0);
-  VL_IN8(&rgb1, 2, 0);
-
-public:
-  HUB75Reciever(int xsize, int ysize, Vhub75e &dut)
-      : display_clk(dut.display_clk), out_enable(dut.out_enable),
-        latch(dut.latch), rgb0(dut.panel_rgb0), rgb1(dut.panel_rgb1) {
-    this->xsize = xsize;
-    this->ysize = ysize;
-    row_upper.clear();
-  };
-
-  // evaluates the reciever.
-  virtual void tick() override {
-
-    if (prev_display_clk == 0 && display_clk == 1) {
-      // display clock rising edge.
-      row_upper.push_back(rgb0);
-      row_lower.push_back(rgb1);
-    }
-
-    if (prev_latch == 0 && latch == 1) {
-      // latch in the data: reverse the rows, and pu
-      std::reverse(row_upper.begin(), row_upper.end());
-      past_rows.push_back(row_upper);
-      row_upper.clear();
-    }
-    if (out_enable == 0) {
-      if (prev_oe == 1) {
-        // falling edge.
-        output_period_cnt = 0;
-      } else {
-        output_period_cnt++;
-      }
-    } else {
-      // rising edge: store the output
-    }
-
-    // update previous values
-    prev_display_clk = display_clk;
-    prev_latch = latch;
-    prev_oe = out_enable;
-  }
-
-  const std::vector<row_array> &get_past_rows() { return this->past_rows; }
-  const std::vector<int> &get_pulse_widths() { return this->pulse_widths; }
-
-  // return the RGB version.
-  std::vector<unsigned int> transpose() {
-    auto vec = std::vector<unsigned int>();
-
-    for (auto bits : this->row_upper) {
-    }
-
-    return vec;
-  }
-};
-
-class FakeBRAM : public CosimulatedDevice {
-  std::array<unsigned long, 512> ram{};
-
-  std::queue<unsigned long> addr_lookup_q;
-
-  unsigned short &addr_in;
-  unsigned long &data_out;
-
-  unsigned char &clk;
-
-public:
-  FakeBRAM(int latency, unsigned char &clk, unsigned short &addr_in,
-           unsigned long &data_out)
-      : addr_lookup_q(), addr_in(addr_in), data_out(data_out), clk(clk) {
-    for (int i = 0; i < latency; i++) {
-      addr_lookup_q.push(0);
-    }
-
-    for (int i = 0; i < ram.size(); i++) {
-      ram[i] = i;
-    }
-  };
-
-  void tick() {
-    // we push and pop in the same tick: this way we keep the queue the same
-    // size, acting as a pipeline delay.
-    addr_lookup_q.push(addr_in);
-
-    auto addr_to_load = addr_lookup_q.front();
-    addr_lookup_q.pop();
-    data_out = ram.at(addr_to_load);
-  }
-
-  // TODO: allow accessing/setting the data
-  std::array<uint64_t, 512> &dump() { return ram; }
-
-  void write(unsigned short addr, unsigned long data) { ram[addr] = data; }
-};
-
 void LineDriverTest(VerilatedContext &ctx) {
   // create the hub75e driver and run some basic tests
 
@@ -206,82 +57,6 @@ void LineDriverTest(VerilatedContext &ctx) {
 //   LineDriverTest(*ctx);
 // };
 
-// test fixture to reduce amount of runtime code.
-// Supports:
-// adding external modules
-// running the test
-// storing if the done flag was raised (or not)
-//
-// TODO: tracing.
-template <typename DUT> class VerilatorTestFixture {
-  // we call .tick() on all of these. They are bound externally to the DUT.
-  std::vector<std::shared_ptr<CosimulatedDevice>> external_devices;
-  std::unique_ptr<VerilatedContext> ctx;
-  std::unique_ptr<DUT> dut;
-
-  unsigned long timeout =
-      1000000; // clock cycles to execute before ending the simulation.
-  unsigned long simtime = 0;
-  unsigned long posedge = 0;
-
-  // stores the termination-condition for the simulation.
-  // if false, it means we timed out. If true, our done_condition returned true.
-  bool done_condition = false;
-
-  typedef std::function<bool(DUT &, unsigned long)> done_callback;
-  // callback function to determine execution completion.
-  // This can be used to add arbitrary finish-conditions to the execution.
-  // If the function returns true, we set the done_set boolean to true;
-  done_callback done_check;
-
-public:
-  // Create a new test fixture with a given timeout. Everything else (including
-  // done-condition) can be set later.
-  VerilatorTestFixture() {
-    ctx = std::make_unique<VerilatedContext>();
-    dut = std::make_unique<DUT>(ctx.get(), "dut");
-  }
-
-  void set_timeout(unsigned long new_timeout) { this->timeout = new_timeout; }
-
-  void set_done_callback(done_callback d) { this->done_check = d; }
-
-  void add_module(std::shared_ptr<CosimulatedDevice> device) {
-    external_devices.push_back(device);
-  }
-
-  void exec() {
-    bool done = false;
-    while (!done) {
-      dut->clk ^= 1;
-      dut->eval();
-
-      if (dut->clk == 1) {
-        posedge++;
-      }
-      if (done_check) {
-        if (done_check(dut, posedge)) {
-          done_condition = true;
-          done = true;
-        }
-      }
-
-      if (posedge >= timeout) {
-        done = true;
-      }
-      // run our external devices
-      for (auto dev : external_devices) {
-        dev->tick();
-      }
-      dut->eval(); // allow combinational logic to settle if it's being set on the negative clock.
-      simtime++;
-    }
-  }
-
-  // return a reference to the DUT itself. Useful for bespoke tests.
-  DUT &get() { return dut; }
-};
-
 TEST_CASE("Hub75 Test") {
   auto ctx = std::make_unique<VerilatedContext>();
   // setup DUT
@@ -328,9 +103,61 @@ TEST_CASE("Hub75 Test") {
 
     for (int i = 0; i < rows.size(); i++) {
       auto r = rows[i];
-      using Catch::Matchers::SizeIs;
-      CHECK_THAT(r, SizeIs(128));
     }
   }
   SECTION("Pulse width") {}
 }
+
+TEST_CASE("HUB75E Driver Test") {
+  auto fixture = VerilatorTestFixture<Vhub75e>();
+  // very simple done checker.
+  auto done_check = [](Vhub75e &dut, unsigned long time) {
+    return dut.done == 1;
+  };
+  fixture.set_done_callback(done_check);
+  const Vhub75e &dut = fixture.get();
+  auto stim = std::make_shared<PulseStimulus>(dut.write_trig, 4);
+  fixture.set_timeout(250000);
+
+  fixture.add_module(stim);
+
+  SECTION("Smoke Tests") {
+    fixture.enable_trace("testing.vcd");
+    auto bram = std::make_shared<FakeBRAM>(1, dut.clk, dut.pixbuf_addr,
+                                           dut.pixbuf_data);
+    fixture.add_module(bram);
+    auto display = std::make_shared<HUB75Reciever>(128, 64, dut);
+    fixture.add_module(display);
+
+    fixture.exec();
+
+    CHECK(fixture.get_reason() ==
+          VerilatorTestFixture<Vhub75e>::FinishReason::Ok);
+
+    auto rows = display->get_past_rows();
+    CHECK(rows.size() == 8);
+    for (int i = 0; i < rows.size(); i++) {
+      auto &[r0, r1] = rows[i];
+      CHECK(r0.size() == 128);
+      CHECK(r1.size() == 128);
+    }
+    // pulse width smoke tests.
+    auto pulses = display->get_pulse_widths();
+    REQUIRE(pulses.size() == rows.size());
+    for (int i = 1; i < pulses.size(); i++) {
+      REQUIRE(pulses[i] == pulses[i - 1] / 2);
+    }
+    auto [row0, row1] = display->transpose();
+    REQUIRE(row0.size() == 128);
+    REQUIRE(row1.size() == 128);
+    auto ram_ref = bram->get();
+
+    CAPTURE(row0);
+    CHECK(std::equal(ram_ref.begin(), ram_ref.begin() + 128, row0.begin(),
+                     row0.end()));
+  }
+  SECTION("Line Correctness") {
+    // this is the part where we validate that the line in = line out.
+    // we have to generate different values since the
+  }
+}