diff --git a/compiler/base/custom_cell_properties.py b/compiler/base/custom_cell_properties.py
index bb211842..76bb10ce 100644
--- a/compiler/base/custom_cell_properties.py
+++ b/compiler/base/custom_cell_properties.py
@@ -176,7 +176,7 @@ class cell_properties():
         self.names["col_cap_bitcell_2port"] = "col_cap_cell_2rw"
         self.names["row_cap_bitcell_1port"] = "row_cap_cell_1rw"
         self.names["row_cap_bitcell_2port"] = "row_cap_cell_2rw"
-
+        self.use_strap = False
         self._ptx = _ptx(model_is_subckt=False,
                          bin_spice_models=False)
 
diff --git a/compiler/base/design.py b/compiler/base/design.py
index 8c621742..97c288f9 100644
--- a/compiler/base/design.py
+++ b/compiler/base/design.py
@@ -48,7 +48,8 @@ class design(hierarchy_design):
             self.add_pin_indices(prop.port_indices)
             self.add_pin_names(prop.port_map)
             self.add_pin_types(prop.port_types)
-            
+
+
             (width, height) = utils.get_libcell_size(self.cell_name,
                                                      GDS["unit"],
                                                      layer[prop.boundary_layer])
diff --git a/compiler/base/hierarchy_layout.py b/compiler/base/hierarchy_layout.py
index e603bdc3..22af8373 100644
--- a/compiler/base/hierarchy_layout.py
+++ b/compiler/base/hierarchy_layout.py
@@ -41,7 +41,8 @@ class layout():
         
         self.width = None
         self.height = None
-        self.bounding_box = None
+        self.bounding_box = None # The rectangle shape
+        self.bbox = None # The ll, ur coords
         # Holds module/cell layout instances
         self.insts = []
         # Set of names to check for duplicates
@@ -1163,6 +1164,57 @@ class layout():
 
         self.bbox = [self.bounding_box.ll(), self.bounding_box.ur()]
 
+    def get_bbox(self, side="all", big_margin=0, little_margin=0):
+        """
+        Get the bounding box from the GDS
+        """
+        gds_filename = OPTS.openram_temp + "temp.gds"
+        # If didn't specify a gds blockage file, write it out to read the gds
+        # This isn't efficient, but easy for now
+        # Load the gds file and read in all the shapes
+        self.gds_write(gds_filename)
+        layout = gdsMill.VlsiLayout(units=GDS["unit"])
+        reader = gdsMill.Gds2reader(layout)
+        reader.loadFromFile(gds_filename)
+        top_name = layout.rootStructureName
+
+        if not self.bbox:
+            # The boundary will determine the limits to the size
+            # of the routing grid
+            boundary = layout.measureBoundary(top_name)
+            # These must be un-indexed to get rid of the matrix type
+            ll = vector(boundary[0][0], boundary[0][1])
+            ur = vector(boundary[1][0], boundary[1][1])
+        else:
+            ll, ur = self.bbox
+
+        ll_offset = vector(0, 0)
+        ur_offset = vector(0, 0)
+        if side in ["ring", "top", "all"]:
+            ur_offset += vector(0, big_margin)
+        else:
+            ur_offset += vector(0, little_margin)
+        if side in ["ring", "bottom", "all"]:
+            ll_offset += vector(0, big_margin)
+        else:
+            ll_offset += vector(0, little_margin)
+        if side in ["ring", "left", "all"]:
+            ll_offset += vector(big_margin, 0)
+        else:
+            ll_offset += vector(little_margin, 0)
+        if side in ["ring", "right", "all"]:
+            ur_offset += vector(big_margin, 0)
+        else:
+            ur_offset += vector(little_margin, 0)
+        bbox = (ll - ll_offset, ur + ur_offset)
+        size = ur - ll
+        debug.info(1, "Size: {0} x {1} with perimeter big margin {2} little margin {3}".format(size.x,
+                                                                                               size.y,
+                                                                                               big_margin,
+                                                                                               little_margin))
+
+        return bbox
+
     def add_enclosure(self, insts, layer="nwell", extend=0, leftx=None, rightx=None, topy=None, boty=None):
         """
         Add a layer that surrounds the given instances. Useful
@@ -1205,22 +1257,24 @@ class layout():
                              height=ymax - ymin)
         return rect
 
-    def copy_power_pins(self, inst, name, add_vias=True):
+    def copy_power_pins(self, inst, name, add_vias=True, new_name=""):
         """
         This will copy a power pin if it is on the lowest power_grid layer.
         If it is on M1, it will add a power via too.
         """
         pins = inst.get_pins(name)
         for pin in pins:
+            if new_name == "":
+                new_name = pin.name
             if pin.layer == self.pwr_grid_layer:
-                self.add_layout_pin(name,
+                self.add_layout_pin(new_name,
                                     pin.layer,
                                     pin.ll(),
                                     pin.width(),
                                     pin.height())
 
             elif add_vias:
-                self.copy_power_pin(pin)
+                self.copy_power_pin(pin, new_name=new_name)
 
     def add_io_pin(self, instance, pin_name, new_name, start_layer=None):
         """
@@ -1266,13 +1320,15 @@ class layout():
                                             width=width,
                                             height=height)
 
-    def copy_power_pin(self, pin, loc=None, directions=None):
+    def copy_power_pin(self, pin, loc=None, directions=None, new_name=""):
         """
         Add a single power pin from the lowest power_grid layer down to M1 (or li) at
         the given center location. The starting layer is specified to determine
         which vias are needed.
         """
 
+        if new_name == "":
+                new_name = pin.name
         if not loc:
             loc = pin.center()
             
@@ -1286,7 +1342,7 @@ class layout():
             height = None
             
         if pin.layer == self.pwr_grid_layer:
-            self.add_layout_pin_rect_center(text=pin.name,
+            self.add_layout_pin_rect_center(text=new_name,
                                             layer=self.pwr_grid_layer,
                                             offset=loc,
                                             width=width,
@@ -1301,7 +1357,7 @@ class layout():
                 width = via.width
             if not height:
                 height = via.height
-            self.add_layout_pin_rect_center(text=pin.name,
+            self.add_layout_pin_rect_center(text=new_name,
                                             layer=self.pwr_grid_layer,
                                             offset=loc,
                                             width=width,
@@ -1357,7 +1413,7 @@ class layout():
         [ll, ur] = bbox
 
         # Possibly inflate the bbox
-        nwell_offset = vector(self.nwell_width, self.nwell_width)
+        nwell_offset = vector(2 * self.nwell_width, 2 * self.nwell_width)
         ll -= nwell_offset.scale(inflate, inflate)
         ur += nwell_offset.scale(inflate, inflate)
 
@@ -1396,7 +1452,7 @@ class layout():
                                           to_layer="m1",
                                           offset=loc)
             else:
-                self.add_power_pin(name="gnd",
+                self.add_power_pin(name="vdd",
                                    loc=loc,
                                    start_layer="li")
             count += 1
@@ -1416,7 +1472,7 @@ class layout():
                                           to_layer="m1",
                                           offset=loc)
             else:
-                self.add_power_pin(name="gnd",
+                self.add_power_pin(name="vdd",
                                    loc=loc,
                                    start_layer="li")
             count += 1
@@ -1436,7 +1492,7 @@ class layout():
                                           to_layer="m2",
                                           offset=loc)
             else:
-                self.add_power_pin(name="gnd",
+                self.add_power_pin(name="vdd",
                                    loc=loc,
                                    start_layer="li")
             count += 1
@@ -1456,7 +1512,7 @@ class layout():
                                           to_layer="m2",
                                           offset=loc)
             else:
-                self.add_power_pin(name="gnd",
+                self.add_power_pin(name="vdd",
                                    loc=loc,
                                    start_layer="li")
             count += 1
diff --git a/compiler/base/lef.py b/compiler/base/lef.py
index ce1eef1c..1d86a63e 100644
--- a/compiler/base/lef.py
+++ b/compiler/base/lef.py
@@ -112,23 +112,25 @@ class lef:
         for pin_name in self.pins:
             pins = self.get_pins(pin_name)
             for pin in pins:
-                inflated_pin = pin.inflated_pin(multiple=1)
-                another_iteration_needed = True
-                while another_iteration_needed:
-                    another_iteration_needed = False
+                inflated_pin = pin.inflated_pin(multiple=2)
+                continue_fragmenting = True
+                while continue_fragmenting:
+                    continue_fragmenting = False
                     old_blockages = list(self.blockages[pin.layer])
                     for blockage in old_blockages:
                         if blockage.overlaps(inflated_pin):
                             intersection_shape = blockage.intersection(inflated_pin)
-                            # If it is zero area, don't add the pin
+                            # If it is zero area, don't split the blockage
                             if intersection_shape[0][0]==intersection_shape[1][0] or intersection_shape[0][1]==intersection_shape[1][1]:
                                 continue
-                            another_iteration_needed = True
+
                             # Remove the old blockage and add the new ones
                             self.blockages[pin.layer].remove(blockage)
                             intersection_pin = pin_layout("", intersection_shape, inflated_pin.layer)
                             new_blockages = blockage.cut(intersection_pin)
                             self.blockages[pin.layer].extend(new_blockages)
+                            # We split something so make another pass
+                            continue_fragmenting = True
 
     def lef_write_header(self):
         """ Header of LEF file """
diff --git a/compiler/base/pin_layout.py b/compiler/base/pin_layout.py
index e6baa4fc..f27990f5 100644
--- a/compiler/base/pin_layout.py
+++ b/compiler/base/pin_layout.py
@@ -606,7 +606,9 @@ class pin_layout:
                 # Don't add the existing shape in if it overlaps the pin shape
                 if new_shape.contains(shape):
                     continue
-                new_shapes.append(new_shape)
+                # Only add non-zero shapes
+                if new_shape.area() > 0:
+                    new_shapes.append(new_shape)
 
         return new_shapes
 
diff --git a/compiler/base/utils.py b/compiler/base/utils.py
index 2b6ef888..6bbc92e8 100644
--- a/compiler/base/utils.py
+++ b/compiler/base/utils.py
@@ -148,12 +148,15 @@ def get_gds_pins(pin_names, name, gds_filename, units):
             cell[str(pin_name)] = []
             pin_list = cell_vlsi.getPinShape(str(pin_name))
             for pin_shape in pin_list:
-                (lpp, boundary) = pin_shape
-                rect = [vector(boundary[0], boundary[1]),
-                        vector(boundary[2], boundary[3])]
-                # this is a list because other cells/designs
-                # may have must-connect pins
-                cell[str(pin_name)].append(pin_layout(pin_name, rect, lpp))
+                if pin_shape != None:
+                    (lpp, boundary) = pin_shape
+                    rect = [vector(boundary[0], boundary[1]),
+                            vector(boundary[2], boundary[3])]
+                    # this is a list because other cells/designs
+                    # may have must-connect pins
+                    if isinstance(lpp[1], list):
+                        lpp = (lpp[0], None)
+                    cell[str(pin_name)].append(pin_layout(pin_name, rect, lpp))
 
         _GDS_PINS_CACHE[k] = cell
         return dict(cell)
diff --git a/compiler/characterizer/__init__.py b/compiler/characterizer/__init__.py
index 425ad5f2..c052a0a0 100644
--- a/compiler/characterizer/__init__.py
+++ b/compiler/characterizer/__init__.py
@@ -24,7 +24,7 @@ debug.info(1, "Initializing characterizer...")
 OPTS.spice_exe = ""
 
 if not OPTS.analytical_delay:
-    if OPTS.spice_name != "":
+    if OPTS.spice_name:
         # Capitalize Xyce
         if OPTS.spice_name == "xyce":
             OPTS.spice_name = "Xyce"
@@ -45,7 +45,7 @@ if not OPTS.analytical_delay:
     if OPTS.spice_name == "ngspice":
         os.environ["NGSPICE_INPUT_DIR"] = "{0}".format(OPTS.openram_temp)
 
-    if OPTS.spice_exe == "":
+    if not OPTS.spice_exe:
         debug.error("No recognizable spice version found. Unable to perform characterization.", 1)
     else:
         debug.info(1, "Finding spice simulator: {} ({})".format(OPTS.spice_name, OPTS.spice_exe))
diff --git a/compiler/characterizer/stimuli.py b/compiler/characterizer/stimuli.py
index f5b5967f..80ddf4fc 100644
--- a/compiler/characterizer/stimuli.py
+++ b/compiler/characterizer/stimuli.py
@@ -276,8 +276,8 @@ class stimuli():
             self.sf.write(".OPTIONS MEASURE MEASFAIL=1\n")
             self.sf.write(".OPTIONS LINSOL type=klu\n")
             self.sf.write(".TRAN {0}p {1}n\n".format(timestep, end_time))
-        else:
-            debug.error("Unkown spice simulator {}".format(OPTS.spice_name))
+        elif OPTS.spice_name:
+            debug.error("Unkown spice simulator {}".format(OPTS.spice_name), -1)
 
         # create plots for all signals
         if not OPTS.use_pex:   # Don't save all for extracted simulations
diff --git a/compiler/example_configs/sky130_sram_common.py b/compiler/example_configs/sky130_sram_common.py
index 8efc8f10..a827b5a9 100644
--- a/compiler/example_configs/sky130_sram_common.py
+++ b/compiler/example_configs/sky130_sram_common.py
@@ -9,8 +9,8 @@ nominal_corner_only = True
 # Local wordlines have issues with met3 power routing for now
 #local_array_size = 16
 
-#route_supplies = "ring"
-route_supplies = "left"
+route_supplies = "ring"
+#route_supplies = "left"
 check_lvsdrc = True
 #perimeter_pins = False
 #netlist_only = True
diff --git a/compiler/gdsMill/gdsMill/vlsiLayout.py b/compiler/gdsMill/gdsMill/vlsiLayout.py
index 71e5498c..056c01c0 100644
--- a/compiler/gdsMill/gdsMill/vlsiLayout.py
+++ b/compiler/gdsMill/gdsMill/vlsiLayout.py
@@ -3,7 +3,7 @@ from datetime import *
 import numpy as np
 import math
 import debug
-
+from tech import use_purpose
 
 class VlsiLayout:
     """Class represent a hierarchical layout"""
@@ -215,9 +215,13 @@ class VlsiLayout:
         self.deduceHierarchy()
         # self.traverseTheHierarchy()
         self.populateCoordinateMap()
-
+        #only ones with text
         for layerNumber in self.layerNumbersInUse:
-            self.processLabelPins((layerNumber, None))
+            #if layerNumber not in no_pin_shape:
+                if layerNumber in use_purpose:
+                    self.processLabelPins((layerNumber, use_purpose[layerNumber]))
+                else:
+                    self.processLabelPins((layerNumber, None))
 
     def populateCoordinateMap(self):
         def addToXyTree(startingStructureName = None,transformPath = None):
@@ -903,6 +907,16 @@ def sameLPP(lpp1, lpp2):
     if lpp1[1] == None or lpp2[1] == None:
         return lpp1[0] == lpp2[0]
 
+    if isinstance(lpp1[1], list):
+        for i in range(len(lpp1[1])):
+            if lpp1[0] == lpp2[0] and lpp1[1][i] == lpp2[1]:
+                return True
+
+    if isinstance(lpp2[1], list):
+        for i in range(len(lpp2[1])):
+            if lpp1[0] == lpp2[0] and lpp1[1] == lpp2[1][i]:
+                return True
+
     return lpp1[0] == lpp2[0] and lpp1[1] == lpp2[1]
 
 
diff --git a/compiler/globals.py b/compiler/globals.py
index 1b272b98..f6929c4d 100644
--- a/compiler/globals.py
+++ b/compiler/globals.py
@@ -238,8 +238,8 @@ def setup_bitcell():
         OPTS.dummy_bitcell = "dummy_pbitcell"
         OPTS.replica_bitcell = "replica_pbitcell"
     else:
-        num_ports = OPTS.num_rw_ports + OPTS.num_w_ports + OPTS.num_r_ports
-        OPTS.bitcell = "bitcell_{}port".format(num_ports)
+        OPTS.num_ports = OPTS.num_rw_ports + OPTS.num_w_ports + OPTS.num_r_ports
+        OPTS.bitcell = "bitcell_{}port".format(OPTS.num_ports)
         OPTS.dummy_bitcell = "dummy_" + OPTS.bitcell
         OPTS.replica_bitcell = "replica_" + OPTS.bitcell
                 
@@ -607,7 +607,7 @@ def report_status():
 
     # If a write mask is specified by the user, the mask write size should be the same as
     # the word size so that an entire word is written at once.
-    if OPTS.write_size is not None:
+    if OPTS.write_size is not None and OPTS.write_size != OPTS.word_size:
         if (OPTS.word_size % OPTS.write_size != 0):
             debug.error("Write size needs to be an integer multiple of word size.")
         # If write size is more than half of the word size,
diff --git a/compiler/modules/bank.py b/compiler/modules/bank.py
index a10a4924..99ad8350 100644
--- a/compiler/modules/bank.py
+++ b/compiler/modules/bank.py
@@ -366,18 +366,13 @@ class bank(design.design):
 
         # A space for wells or jogging m2
         self.m2_gap = max(2 * drc("pwell_to_nwell") + drc("nwell_enclose_active"),
-                          3 * self.m2_pitch)
+                          3 * self.m2_pitch,
+                          drc("nwell_to_nwell"))
+
 
     def add_modules(self):
         """ Add all the modules using the class loader """
 
-        self.port_address = []
-        for port in self.all_ports:
-            self.port_address.append(factory.create(module_type="port_address",
-                                                    cols=self.num_cols + self.num_spare_cols,
-                                                    rows=self.num_rows,
-                                                    port=port))
-            self.add_mod(self.port_address[port])
 
         local_array_size = OPTS.local_array_size
 
@@ -397,12 +392,26 @@ class bank(design.design):
                                                 cols=self.num_cols + self.num_spare_cols,
                                                 rows=self.num_rows)
         self.add_mod(self.bitcell_array)
+        if self.num_spare_cols == 0:
+            self.num_spare_cols = (self.bitcell_array.column_size % (self.word_size *self.words_per_row))
+
+        self.port_address = []
+        for port in self.all_ports:
+            self.port_address.append(factory.create(module_type="port_address",
+                                                    cols=self.bitcell_array.column_size,
+                                                    rows=self.bitcell_array.row_size,
+                                                    port=port))
+            self.add_mod(self.port_address[port])
 
         self.port_data = []
         self.bit_offsets = self.get_column_offsets()
         for port in self.all_ports:
             temp_pre = factory.create(module_type="port_data",
                                       sram_config=self.sram_config,
+                                      dimension_override=True,
+                                      cols=self.bitcell_array.column_size - self.num_spare_cols,
+                                      rows=self.bitcell_array.row_size,
+                                      num_spare_cols=self.num_spare_cols,
                                       port=port,
                                       bit_offsets=self.bit_offsets)
             self.port_data.append(temp_pre)
@@ -430,7 +439,9 @@ class bank(design.design):
 
         temp.append("vdd")
         temp.append("gnd")
-
+        if 'vpb' in self.bitcell_array_inst.mod.pins and 'vnb' in self.bitcell_array_inst.mod.pins:
+            temp.append('vpb')
+            temp.append('vnb')
         self.connect_inst(temp)
 
     def place_bitcell_array(self, offset):
@@ -489,7 +500,7 @@ class bank(design.design):
                                                          mod=self.port_address[port])
 
             temp = []
-            for bit in range(self.row_addr_size):
+            for bit in range(ceil(log(self.bitcell_array.row_size, 2))):
                 temp.append("addr{0}_{1}".format(port, bit + self.col_addr_size))
             temp.append("wl_en{}".format(port))
             wordline_names = self.bitcell_array.get_wordline_names(port)
@@ -614,6 +625,10 @@ class bank(design.design):
             self.copy_power_pins(inst, "vdd", add_vias=False)
             self.copy_power_pins(inst, "gnd", add_vias=False)
 
+        if 'vpb' in self.bitcell_array_inst.mod.pins and 'vnb' in self.bitcell_array_inst.mod.pins:
+            for pin_name, supply_name in zip(['vpb','vnb'],['vdd','gnd']):
+                self.copy_power_pins(self.bitcell_array_inst, pin_name, new_name=supply_name)
+
         # If we use the pinvbuf as the decoder, we need to add power pins.
         # Other decoders already have them.
         if self.col_addr_size == 1:
@@ -1062,7 +1077,6 @@ class bank(design.design):
                                       to_layer="m2",
                                       offset=control_pos)
 
-
     def graph_exclude_precharge(self):
         """
         Precharge adds a loop between bitlines, can be excluded to reduce complexity
diff --git a/compiler/modules/dff_buf.py b/compiler/modules/dff_buf.py
index e8474fed..d979c0fe 100644
--- a/compiler/modules/dff_buf.py
+++ b/compiler/modules/dff_buf.py
@@ -109,7 +109,7 @@ class dff_buf(design.design):
         except AttributeError:
             pass
 
-        well_spacing += self.well_extend_active
+        well_spacing += 2 * self.well_extend_active
         
         self.inv1_inst.place(vector(self.dff_inst.rx() + well_spacing, 0))
 
diff --git a/compiler/modules/hierarchical_predecode.py b/compiler/modules/hierarchical_predecode.py
index f83516d2..386f9bb6 100644
--- a/compiler/modules/hierarchical_predecode.py
+++ b/compiler/modules/hierarchical_predecode.py
@@ -12,6 +12,10 @@ from vector import vector
 from sram_factory import factory
 from globals import OPTS
 from tech import layer_properties as layer_props
+from tech import layer_indices
+from tech import layer_stacks
+from tech import preferred_directions
+from tech import drc
 
 
 class hierarchical_predecode(design.design):
@@ -29,7 +33,7 @@ class hierarchical_predecode(design.design):
             self.cell_height = height
             
         self.column_decoder = column_decoder
-
+        self.input_and_rail_pos = []
         self.number_of_outputs = int(math.pow(2, self.number_of_inputs))
         super().__init__(name)
 
@@ -183,9 +187,9 @@ class hierarchical_predecode(design.design):
     def route(self):
 
         self.route_input_inverters()
-        self.route_output_inverters()
-        self.route_inputs_to_rails()
         self.route_input_ands()
+        self.route_output_inverters()
+        self.route_inputs_to_rails()    
         self.route_output_ands()
         self.route_vdd_gnd()
 
@@ -274,8 +278,46 @@ class hierarchical_predecode(design.design):
             # pins in the and gates.
             inv_out_pos = inv_out_pin.rc()
             y_offset = (inv_num + 1) * self.inv.height - self.output_layer_pitch
-            right_pos = inv_out_pos + vector(self.inv.width - self.inv.get_pin("Z").rx(), 0)
             rail_pos = vector(self.decode_rails[out_pin].cx(), y_offset)
+
+            # create via for dimensions
+            from_layer = self.output_layer
+            to_layer = self.bus_layer
+
+            cur_layer = from_layer
+            from_id = layer_indices[cur_layer]
+            to_id   = layer_indices[to_layer]
+
+            if from_id < to_id: # grow the stack up
+                search_id = 0
+                next_id = 2
+            else: # grow the stack down
+                search_id = 2
+                next_id = 0
+            
+            curr_stack = next(filter(lambda stack: stack[search_id] == cur_layer, layer_stacks), None)
+
+            via = factory.create(module_type="contact",
+                        layer_stack=curr_stack,
+                        dimensions=[1, 1],
+                        directions=self.bus_directions)
+
+            overlapping_pin_space = drc["{0}_to_{0}".format(self.output_layer)]
+            total_buffer_space = (overlapping_pin_space + via.height)
+            #FIXME: compute rail locations instead of just guessing and nudging
+            while(True):
+                drc_error = 0
+                for and_input in self.input_and_rail_pos:
+                    if and_input.x == rail_pos.x:
+                        if (abs(y_offset - and_input.y) < total_buffer_space) and (abs(y_offset - and_input.y) > via.height):
+                            drc_error = 1
+                if drc_error == 0:
+                    break
+                else:
+                    y_offset += drc["grid"]
+            rail_pos.y = y_offset
+            right_pos = inv_out_pos + vector(self.inv.width - self.inv.get_pin("Z").rx(), 0)
+            
             self.add_path(self.output_layer, [inv_out_pos, right_pos, vector(right_pos.x, y_offset), rail_pos])
 
             self.add_via_stack_center(from_layer=inv_out_pin.layer,
@@ -316,6 +358,7 @@ class hierarchical_predecode(design.design):
                                           to_layer=self.bus_layer,
                                           offset=rail_pos,
                                           directions=self.bus_directions)
+                self.input_and_rail_pos.append(rail_pos)
                 if gate_pin == "A":
                     direction = None
                 else:
diff --git a/compiler/modules/port_data.py b/compiler/modules/port_data.py
index 71c8449a..3fbb8696 100644
--- a/compiler/modules/port_data.py
+++ b/compiler/modules/port_data.py
@@ -11,6 +11,7 @@ from sram_factory import factory
 from collections import namedtuple
 from vector import vector
 from globals import OPTS
+from tech import cell_properties
 from tech import layer_properties as layer_props
 
 
@@ -20,26 +21,36 @@ class port_data(design.design):
     Port 0 always has the RBL on the left while port 1 is on the right.
     """
 
-    def __init__(self, sram_config, port, bit_offsets=None, name=""):
-
+    def __init__(self, sram_config, port, num_spare_cols=None, bit_offsets=None, name="", rows=None, cols=None, dimension_override=False):
         sram_config.set_local_config(self)
+        if dimension_override:
+            self.num_rows = rows
+            self.num_cols = cols
+            self.word_size = sram_config.word_size
+
         self.port = port
         if self.write_size is not None:
             self.num_wmasks = int(math.ceil(self.word_size / self.write_size))
         else:
             self.num_wmasks = 0
 
-        if self.num_spare_cols is None:
+        if num_spare_cols:
+            self.num_spare_cols = num_spare_cols
+        elif self.num_spare_cols is None:
             self.num_spare_cols = 0
 
         if not bit_offsets:
             bitcell = factory.create(module_type=OPTS.bitcell)
+            if(cell_properties.use_strap == True and OPTS.num_ports == 1):
+                strap = factory.create(module_type=cell_properties.strap_module, version=cell_properties.strap_version)
+                precharge_width = bitcell.width + strap.width
+            else:
+                precharge_width = bitcell.width
             self.bit_offsets = []
             for i in range(self.num_cols + self.num_spare_cols):
-                self.bit_offsets.append(i * bitcell.width)
+                self.bit_offsets.append(i * precharge_width)
         else:
             self.bit_offsets = bit_offsets
-
         if name == "":
             name = "port_data_{0}".format(self.port)
         super().__init__(name)
@@ -117,7 +128,6 @@ class port_data(design.design):
         for bit in range(self.num_spare_cols):
             self.add_pin("sparebl_{0}".format(bit), "INOUT")
             self.add_pin("sparebr_{0}".format(bit), "INOUT")
-
         if self.port in self.read_ports:
             for bit in range(self.word_size + self.num_spare_cols):
                 self.add_pin("dout_{}".format(bit), "OUTPUT")
@@ -191,14 +201,19 @@ class port_data(design.design):
         # and mirroring happens correctly
 
         # Used for names/dimensions only
-        self.cell = factory.create(module_type=OPTS.bitcell)
-
+        cell = factory.create(module_type=OPTS.bitcell)
+        if(cell_properties.use_strap == True and OPTS.num_ports == 1):
+            strap = factory.create(module_type=cell_properties.strap_module, version=cell_properties.strap_version)
+            precharge_width = cell.width + strap.width
+        else:
+            precharge_width = cell.width
         if self.port == 0:
             # Append an offset on the left
-            precharge_bit_offsets = [self.bit_offsets[0] - self.cell.width] + self.bit_offsets
+            precharge_bit_offsets = [self.bit_offsets[0] - precharge_width] + self.bit_offsets
         else:
             # Append an offset on the right
-            precharge_bit_offsets = self.bit_offsets + [self.bit_offsets[-1] + self.cell.width]
+            precharge_bit_offsets = self.bit_offsets + [self.bit_offsets[-1] + precharge_width]
+
         self.precharge_array = factory.create(module_type="precharge_array",
                                               columns=self.num_cols + self.num_spare_cols + 1,
                                               offsets=precharge_bit_offsets,
@@ -567,19 +582,32 @@ class port_data(design.design):
                 off = 1
             else:
                 off = 0
+            if OPTS.num_ports > 1:
+                self.channel_route_bitlines(inst1=self.column_mux_array_inst,
+                                            inst1_bls_template="{inst}_out_{bit}",
+                                            inst2=inst2,
+                                            num_bits=self.word_size,
+                                            inst1_start_bit=start_bit)
 
-            self.channel_route_bitlines(inst1=self.column_mux_array_inst,
-                                        inst1_bls_template="{inst}_out_{bit}",
-                                        inst2=inst2,
-                                        num_bits=self.word_size,
-                                        inst1_start_bit=start_bit)
+                self.channel_route_bitlines(inst1=self.precharge_array_inst,
+                                            inst1_bls_template="{inst}_{bit}",
+                                            inst2=inst2,
+                                            num_bits=self.num_spare_cols,
+                                            inst1_start_bit=self.num_cols + off,
+                                            inst2_start_bit=self.word_size)
+            else:
+                self.connect_bitlines(inst1=self.column_mux_array_inst,
+                                            inst1_bls_template="{inst}_out_{bit}",
+                                            inst2=inst2,
+                                            num_bits=self.word_size,
+                                            inst1_start_bit=start_bit)
 
-            self.channel_route_bitlines(inst1=self.precharge_array_inst,
-                                        inst1_bls_template="{inst}_{bit}",
-                                        inst2=inst2,
-                                        num_bits=self.num_spare_cols,
-                                        inst1_start_bit=self.num_cols + off,
-                                        inst2_start_bit=self.word_size)
+                self.connect_bitlines(inst1=self.precharge_array_inst,
+                                            inst1_bls_template="{inst}_{bit}",
+                                            inst2=inst2,
+                                            num_bits=self.num_spare_cols,
+                                            inst1_start_bit=self.num_cols + off,
+                                            inst2_start_bit=self.word_size)
 
         elif layer_props.port_data.channel_route_bitlines:
             self.channel_route_bitlines(inst1=inst1,
diff --git a/compiler/modules/precharge_array.py b/compiler/modules/precharge_array.py
index 8718dfd0..ed19b387 100644
--- a/compiler/modules/precharge_array.py
+++ b/compiler/modules/precharge_array.py
@@ -76,8 +76,8 @@ class precharge_array(design.design):
                                       size=self.size,
                                       bitcell_bl=self.bitcell_bl,
                                       bitcell_br=self.bitcell_br)
+                                      
         self.add_mod(self.pc_cell)
-
         self.cell = factory.create(module_type=OPTS.bitcell)
 
     def add_layout_pins(self):
diff --git a/compiler/modules/replica_bitcell_array.py b/compiler/modules/replica_bitcell_array.py
index 0173f1a0..baf83964 100644
--- a/compiler/modules/replica_bitcell_array.py
+++ b/compiler/modules/replica_bitcell_array.py
@@ -6,7 +6,7 @@
 
 import debug
 from bitcell_base_array import bitcell_base_array
-from tech import drc, spice
+from tech import drc, spice, cell_properties
 from vector import vector
 from globals import OPTS
 from sram_factory import factory
diff --git a/compiler/modules/sense_amp_array.py b/compiler/modules/sense_amp_array.py
index 01b74c84..713f4daf 100644
--- a/compiler/modules/sense_amp_array.py
+++ b/compiler/modules/sense_amp_array.py
@@ -8,6 +8,7 @@
 import design
 from vector import vector
 from sram_factory import factory
+from tech import cell_properties
 import debug
 from globals import OPTS
 
@@ -41,7 +42,6 @@ class sense_amp_array(design.design):
             self.en_layer = "m3"
         else:
             self.en_layer = "m1"
-
         self.create_netlist()
         if not OPTS.netlist_only:
             self.create_layout()
@@ -109,15 +109,22 @@ class sense_amp_array(design.design):
                                self.en_name, "vdd", "gnd"])
 
     def place_sense_amp_array(self):
-        if self.bitcell.width > self.amp.width:
-            self.amp_spacing = self.bitcell.width
+        cell = factory.create(module_type=OPTS.bitcell)
+        if(cell_properties.use_strap  == True and OPTS.num_ports == 1):
+            strap = factory.create(module_type=cell_properties.strap_module, version=cell_properties.strap_version)
+            precharge_width = cell.width + strap.width
+        else:
+            precharge_width = cell.width
+
+        if precharge_width > self.amp.width:
+            self.amp_spacing = precharge_width
         else:
             self.amp_spacing = self.amp.width
 
         if not self.offsets:
             self.offsets = []
             for i in range(self.num_cols + self.num_spare_cols):
-                self.offsets.append(i * self.bitcell.width)
+                self.offsets.append(i * self.amp_spacing)
 
         for i, xoffset in enumerate(self.offsets[0:self.num_cols:self.words_per_row]):
             if self.bitcell.mirror.y and (i * self.words_per_row + self.column_offset) % 2:
@@ -128,13 +135,12 @@ class sense_amp_array(design.design):
 
             amp_position = vector(xoffset, 0)
             self.local_insts[i].place(offset=amp_position, mirror=mirror)
-
         # place spare sense amps (will share the same enable as regular sense amps)
         for i, xoffset in enumerate(self.offsets[self.num_cols:]):
             index = self.word_size + i
             if self.bitcell.mirror.y and (index + self.column_offset) % 2:
                 mirror = "MY"
-                xoffset = xoffset + self.amp_width
+                xoffset = xoffset + self.amp_spacing
             else:
                 mirror = ""
 
diff --git a/compiler/options.py b/compiler/options.py
index 214eecf7..89b6c1ce 100644
--- a/compiler/options.py
+++ b/compiler/options.py
@@ -72,7 +72,9 @@ class options(optparse.Values):
     # This is the temp directory where all intermediate results are stored.
     try:
         # If user defined the temporary location in their environment, use it
+        
         openram_temp = os.path.abspath(os.environ.get("OPENRAM_TMP"))
+        
     except:
         openram_temp = "/tmp"
         
@@ -99,6 +101,7 @@ class options(optparse.Values):
     netlist_only = False
     # Whether we should do the final power routing
     route_supplies = "tree"
+    supply_pin_type = "ring"
     # This determines whether LVS and DRC is checked at all.
     check_lvsdrc = False
     # This determines whether LVS and DRC is checked for every submodule.
@@ -119,13 +122,13 @@ class options(optparse.Values):
     # Tool options
     ###################
     # Variable to select the variant of spice
-    spice_name = ""
+    spice_name = None
     # The spice executable being used which is derived from the user PATH.
-    spice_exe = ""
+    spice_exe = None
     # Variable to select the variant of drc, lvs, pex
-    drc_name = ""
-    lvs_name = ""
-    pex_name = ""
+    drc_name = None
+    lvs_name = None
+    pex_name = None
     # The DRC/LVS/PEX executable being used
     # which is derived from the user PATH.
     drc_exe = None
diff --git a/compiler/pgates/column_mux.py b/compiler/pgates/column_mux.py
index 1e8c5bf8..0dd923ba 100644
--- a/compiler/pgates/column_mux.py
+++ b/compiler/pgates/column_mux.py
@@ -56,7 +56,13 @@ class column_mux(pgate.pgate):
 
         self.place_ptx()
 
-        self.width = self.bitcell.width
+        cell = factory.create(module_type=OPTS.bitcell)
+        if(cell_props.use_strap == True and OPTS.num_ports == 1):
+            strap = factory.create(module_type=cell_props.strap_module, version=cell_props.strap_version)
+            precharge_width = cell.width + strap.width
+        else:
+            precharge_width = cell.width
+        self.width = precharge_width
         self.height = self.nmos_upper.uy() + self.pin_height
 
         self.connect_poly()
@@ -217,10 +223,15 @@ class column_mux(pgate.pgate):
         Add a well and implant over the whole cell. Also, add the
         pwell contact (if it exists)
         """
-
+        if(cell_props.use_strap == True and OPTS.num_ports == 1):
+            strap = factory.create(module_type=cell_props.strap_module, version=cell_props.strap_version)
+            rbc_width = self.bitcell.width + strap.width
+        else:
+            rbc_width = self.bitcell.width
         # Add it to the right, aligned in between the two tx
-        active_pos = vector(self.bitcell.width,
+        active_pos = vector(rbc_width,
                             self.nmos_upper.by() - 0.5 * self.poly_space)
+
         self.add_via_center(layers=self.active_stack,
                             offset=active_pos,
                             implant_type="p",
@@ -239,5 +250,5 @@ class column_mux(pgate.pgate):
         if "pwell" in layer:
             self.add_rect(layer="pwell",
                           offset=vector(0, 0),
-                          width=self.bitcell.width,
+                          width=rbc_width,
                           height=self.height)
diff --git a/compiler/pgates/precharge.py b/compiler/pgates/precharge.py
index d1999384..951fe834 100644
--- a/compiler/pgates/precharge.py
+++ b/compiler/pgates/precharge.py
@@ -30,7 +30,11 @@ class precharge(design.design):
         self.beta = parameter["beta"]
         self.ptx_width = self.beta * parameter["min_tx_size"]
         self.ptx_mults = 1
-        self.width = self.bitcell.width
+        if(cell_props.use_strap == True and OPTS.num_ports == 1):
+            strap = factory.create(module_type=cell_props.strap_module, version=cell_props.strap_version)
+            self.width = self.bitcell.width + strap.width
+        else:
+            self.width = self.bitcell.width
         self.bitcell_bl = bitcell_bl
         self.bitcell_br = bitcell_br
         self.bitcell_bl_pin =self.bitcell.get_pin(self.bitcell_bl)
diff --git a/compiler/router/grid.py b/compiler/router/grid.py
index 404a716f..843fb3ed 100644
--- a/compiler/router/grid.py
+++ b/compiler/router/grid.py
@@ -37,6 +37,8 @@ class grid:
         # This is really lower left bottom layer and upper right top layer in 3D.
         self.ll = vector3d(ll.x, ll.y, 0).scale(self.track_factor).round()
         self.ur = vector3d(ur.x, ur.y, 0).scale(self.track_factor).round()
+        debug.info(1, "BBOX coords: ll=" + str(ll) + " ur=" + str(ur))
+        debug.info(1, "BBOX grids: ll=" + str(self.ll) + " ur=" + str(self.ur))
 
         # let's leave the map sparse, cells are created on demand to reduce memory
         self.map={}
@@ -127,33 +129,47 @@ class grid:
         Side specifies which side.
         Layer specifies horizontal (0) or vertical (1)
         Width specifies how wide the perimter "stripe" should be.
+        Works from the inside out from the bbox (ll, ur)
         """
+        if "ring" in side:
+            ring_width = width
+        else:
+            ring_width = 0
+
+        if "ring" in side:
+            ring_offset = offset
+        else:
+            ring_offset = 0
+
         perimeter_list = []
         # Add the left/right columns
-        if side=="all" or side=="left":
-            for x in range(self.ll.x + offset, self.ll.x + width + offset, 1):
-                for y in range(self.ll.y + offset + margin, self.ur.y - offset - margin, 1):
+        if side=="all" or "left" in side:
+            for x in range(self.ll.x - offset, self.ll.x - width - offset, -1):
+                for y in range(self.ll.y - ring_offset - margin - ring_width + 1, self.ur.y + ring_offset + margin + ring_width, 1):
                     for layer in layers:
                         perimeter_list.append(vector3d(x, y, layer))
                 
-        if side=="all" or side=="right":
-            for x in range(self.ur.x - width - offset, self.ur.x - offset, 1):
-                for y in range(self.ll.y + offset + margin, self.ur.y - offset - margin, 1):
+        if side=="all" or "right" in side:
+            for x in range(self.ur.x + offset, self.ur.x + width + offset, 1):
+                for y in range(self.ll.y - ring_offset - margin - ring_width + 1, self.ur.y + ring_offset + margin + ring_width, 1):
                     for layer in layers:
                         perimeter_list.append(vector3d(x, y, layer))
 
-        if side=="all" or side=="bottom":
-            for y in range(self.ll.y + offset, self.ll.y + width + offset, 1):
-                for x in range(self.ll.x + offset + margin, self.ur.x - offset - margin, 1):
+        if side=="all" or "bottom" in side:
+            for y in range(self.ll.y - offset, self.ll.y - width - offset, -1):
+                for x in range(self.ll.x - ring_offset - margin - ring_width + 1, self.ur.x + ring_offset + margin + ring_width, 1):
                     for layer in layers:
                         perimeter_list.append(vector3d(x, y, layer))
 
-        if side=="all" or side=="top":
-            for y in range(self.ur.y - width - offset, self.ur.y - offset, 1):
-                for x in range(self.ll.x + offset + margin, self.ur.x - offset - margin, 1):
+        if side=="all" or "top" in side:
+            for y in range(self.ur.y + offset, self.ur.y + width + offset, 1):
+                for x in range(self.ll.x - ring_offset - margin - ring_width + 1, self.ur.x + ring_offset + margin + ring_width, 1):
                     for layer in layers:
                         perimeter_list.append(vector3d(x, y, layer))
 
+        # Add them all to the map
+        self.add_map(perimeter_list)
+
         return perimeter_list
     
     def add_perimeter_target(self, side="all", layers=[0, 1]):
diff --git a/compiler/router/router.py b/compiler/router/router.py
index dcfde6cf..d5bd4738 100644
--- a/compiler/router/router.py
+++ b/compiler/router/router.py
@@ -82,31 +82,13 @@ class router(router_tech):
         """
         Initialize the ll,ur values with the paramter or using the layout boundary.
         """
-
-        # If didn't specify a gds blockage file, write it out to read the gds
-        # This isn't efficient, but easy for now
-        # Load the gds file and read in all the shapes
-        self.cell.gds_write(self.gds_filename)
-        self.layout = gdsMill.VlsiLayout(units=GDS["unit"])
-        self.reader = gdsMill.Gds2reader(self.layout)
-        self.reader.loadFromFile(self.gds_filename)
-        self.top_name = self.layout.rootStructureName
-        
         if not bbox:
-            # The boundary will determine the limits to the size
-            # of the routing grid
-            self.boundary = self.layout.measureBoundary(self.top_name)
-            # These must be un-indexed to get rid of the matrix type
-            self.ll = vector(self.boundary[0][0], self.boundary[0][1])
-            self.ur = vector(self.boundary[1][0], self.boundary[1][1])
+            self.bbox = self.cell.get_bbox(margin)
         else:
-            self.ll, self.ur = bbox
+            self.bbox = bbox
+
+        (self.ll, self.ur) = self.bbox
 
-        margin_offset = vector(margin, margin)
-        self.bbox = (self.ll - margin_offset, self.ur + margin_offset)
-        size = self.ur - self.ll
-        debug.info(1, "Size: {0} x {1} with perimeter margin {2}".format(size.x, size.y, margin))
-        
     def get_bbox(self):
         return self.bbox
     
@@ -893,19 +875,21 @@ class router(router_tech):
         # Clearing the blockage of this pin requires the inflated pins
         self.clear_blockages(pin_name)
 
-    def add_side_supply_pin(self, name, side="left", width=2):
+    def add_side_supply_pin(self, name, side="left", width=3, space=2):
         """
         Adds a supply pin to the perimeter and resizes the bounding box.
         """
         pg = pin_group(name, [], self)
+        # Offset two spaces inside and one between the rings
         if name == "gnd":
-            offset = width + 1
+            offset = width + 2 * space
         else:
-            offset = 1
+            offset = space
         if side in ["left", "right"]:
             layers = [1]
         else:
             layers = [0]
+
         pg.grids = set(self.rg.get_perimeter_list(side=side,
                                                   width=width,
                                                   margin=self.margin,
@@ -920,39 +904,39 @@ class router(router_tech):
 
         self.new_pins[name] = pg.pins
         
-    def add_ring_supply_pin(self, name, width=2):
+    def add_ring_supply_pin(self, name, width=3, space=2):
         """
         Adds a ring supply pin that goes inside the given bbox.
         """
         pg = pin_group(name, [], self)
-        # Offset the vdd inside one ring width
+        # Offset two spaces inside and one between the rings
         # Units are in routing grids
         if name == "gnd":
-            offset = width + 1
+            offset = width + 2 * space
         else:
-            offset = 1
+            offset = space
 
         # LEFT
-        left_grids = set(self.rg.get_perimeter_list(side="left",
+        left_grids = set(self.rg.get_perimeter_list(side="left_ring",
                                                     width=width,
                                                     margin=self.margin,
                                                     offset=offset,
                                                     layers=[1]))
 
         # RIGHT
-        right_grids = set(self.rg.get_perimeter_list(side="right",
+        right_grids = set(self.rg.get_perimeter_list(side="right_ring",
                                                      width=width,
                                                      margin=self.margin,
                                                      offset=offset,
                                                      layers=[1]))
         # TOP
-        top_grids = set(self.rg.get_perimeter_list(side="top",
+        top_grids = set(self.rg.get_perimeter_list(side="top_ring",
                                                    width=width,
                                                    margin=self.margin,
                                                    offset=offset,
                                                    layers=[0]))
         # BOTTOM
-        bottom_grids = set(self.rg.get_perimeter_list(side="bottom",
+        bottom_grids = set(self.rg.get_perimeter_list(side="bottom_ring",
                                                       width=width,
                                                       margin=self.margin,
                                                       offset=offset,
diff --git a/compiler/router/supply_grid_router.py b/compiler/router/supply_grid_router.py
index f24498ab..06831299 100644
--- a/compiler/router/supply_grid_router.py
+++ b/compiler/router/supply_grid_router.py
@@ -21,7 +21,7 @@ class supply_grid_router(router):
     routes a grid to connect the supply on the two layers.
     """
 
-    def __init__(self, layers, design, margin=0, bbox=None):
+    def __init__(self, layers, design, bbox=None, pin_type=None):
         """
         This will route on layers in design. It will get the blockages from
         either the gds file name or the design itself (by saving to a gds file).
diff --git a/compiler/router/supply_tree_router.py b/compiler/router/supply_tree_router.py
index e95cdee1..282adc4c 100644
--- a/compiler/router/supply_tree_router.py
+++ b/compiler/router/supply_tree_router.py
@@ -34,7 +34,7 @@ class supply_tree_router(router):
         # The pin escape router already made the bounding box big enough,
         # so we can use the regular bbox here.
         if pin_type:
-            debug.check(pin_type in ["left", "right", "top", "bottom", "tree", "ring"],
+            debug.check(pin_type in ["left", "right", "top", "bottom", "single", "ring"],
                         "Invalid pin type {}".format(pin_type))
         self.pin_type = pin_type
         router.__init__(self,
diff --git a/compiler/sram/sram_1bank.py b/compiler/sram/sram_1bank.py
index 327ce209..828edfdd 100644
--- a/compiler/sram/sram_1bank.py
+++ b/compiler/sram/sram_1bank.py
@@ -9,6 +9,7 @@ from vector import vector
 from sram_base import sram_base
 from contact import m2_via
 from channel_route import channel_route
+from router_tech import router_tech
 from globals import OPTS
 
 
@@ -329,13 +330,31 @@ class sram_1bank(sram_base):
         # Some technologies have an isolation
         self.add_dnwell(inflate=2)
 
+        # We need the initial bbox for the supply rings later
+        # because the perimeter pins will change the bbox
         # Route the pins to the perimeter
+        pre_bbox = None
         if OPTS.perimeter_pins:
-            self.route_escape_pins()
+            rt = router_tech(self.supply_stack, 1)
+
+            if OPTS.supply_pin_type in ["ring", "left", "right", "top", "bottom"]:
+                big_margin = 12 * rt.track_width
+                little_margin = 2 * rt.track_width
+            else:
+                big_margin = 6 * rt.track_width
+                little_margin = 0
+
+            pre_bbox = self.get_bbox(side="ring",
+                                     big_margin=rt.track_width)
+
+            bbox = self.get_bbox(side=OPTS.supply_pin_type,
+                                 big_margin=big_margin,
+                                 little_margin=little_margin)
+            self.route_escape_pins(bbox)
             
         # Route the supplies first since the MST is not blockage aware
         # and signals can route to anywhere on sides (it is flexible)
-        self.route_supplies()
+        self.route_supplies(pre_bbox)
         
     def route_dffs(self, add_routes=True):
 
@@ -366,6 +385,7 @@ class sram_1bank(sram_base):
 
         if len(route_map) > 0:
 
+            # This layer stack must be different than the data dff layer stack
             layer_stack = self.m1_stack
 
             if port == 0:
@@ -375,11 +395,11 @@ class sram_1bank(sram_base):
                                    offset=offset,
                                    layer_stack=layer_stack,
                                    parent=self)
-                # This causes problem in magic since it sometimes cannot extract connectivity of isntances
+                # This causes problem in magic since it sometimes cannot extract connectivity of instances
                 # with no active devices.
                 self.add_inst(cr.name, cr)
                 self.connect_inst([])
-                #self.add_flat_inst(cr.name, cr)
+                # self.add_flat_inst(cr.name, cr)
             else:
                 offset = vector(0,
                                 self.bank.height + self.m3_pitch)
@@ -387,11 +407,11 @@ class sram_1bank(sram_base):
                                    offset=offset,
                                    layer_stack=layer_stack,
                                    parent=self)
-                # This causes problem in magic since it sometimes cannot extract connectivity of isntances
+                # This causes problem in magic since it sometimes cannot extract connectivity of instances
                 # with no active devices.
                 self.add_inst(cr.name, cr)
                 self.connect_inst([])
-                #self.add_flat_inst(cr.name, cr)
+                # self.add_flat_inst(cr.name, cr)
         
     def route_data_dffs(self, port, add_routes):
         route_map = []
@@ -422,40 +442,49 @@ class sram_1bank(sram_base):
 
         if len(route_map) > 0:
 
-            # The write masks will have blockages on M1
-            # if self.num_wmasks > 0 and port in self.write_ports:
-            #     layer_stack = self.m3_stack
-            # else:
-            #     layer_stack = self.m1_stack
+            # This layer stack must be different than the column addr dff layer stack
             layer_stack = self.m3_stack
             if port == 0:
+                # This is relative to the bank at 0,0 or the s_en which is routed on M3 also
+                if "s_en" in self.control_logic_insts[port].mod.pin_map:
+                    y_bottom = min(0, self.control_logic_insts[port].get_pin("s_en").by())
+                else:
+                    y_bottom = 0
+                    
+                y_offset = y_bottom - self.data_bus_size[port] + 2 * self.m3_pitch
+                           
                 offset = vector(self.control_logic_insts[port].rx() + self.dff.width,
-                                - self.data_bus_size[port] + 2 * self.m3_pitch)
+                                y_offset)
                 cr = channel_route(netlist=route_map,
                                    offset=offset,
                                    layer_stack=layer_stack,
                                    parent=self)
                 if add_routes:
-                    # This causes problem in magic since it sometimes cannot extract connectivity of isntances
+                    # This causes problem in magic since it sometimes cannot extract connectivity of instances
                     # with no active devices.
                     self.add_inst(cr.name, cr)
                     self.connect_inst([])
-                    #self.add_flat_inst(cr.name, cr)
+                    # self.add_flat_inst(cr.name, cr)
                 else:
                     self.data_bus_size[port] = max(cr.height, self.col_addr_bus_size[port]) + self.data_bus_gap
             else:
+                if "s_en" in self.control_logic_insts[port].mod.pin_map:
+                    y_top = max(self.bank.height, self.control_logic_insts[port].get_pin("s_en").uy())
+                else:
+                    y_top = self.bank.height
+                y_offset = y_top + self.m3_pitch
                 offset = vector(0,
-                                self.bank.height + self.m3_pitch)
+                                y_offset)
                 cr = channel_route(netlist=route_map,
                                    offset=offset,
                                    layer_stack=layer_stack,
                                    parent=self)
                 if add_routes:
-                    # This causes problem in magic since it sometimes cannot extract connectivity of isntances
+                    # This causes problem in magic since it sometimes cannot extract connectivity of instances
                     # with no active devices.
                     self.add_inst(cr.name, cr)
                     self.connect_inst([])
-                    #self.add_flat_inst(cr.name, cr)
+                    # self.add_flat_inst(cr.name, cr)
                 else:
                     self.data_bus_size[port] = max(cr.height, self.col_addr_bus_size[port]) + self.data_bus_gap
 
diff --git a/compiler/sram/sram_base.py b/compiler/sram/sram_base.py
index a7530e0b..01cf65b7 100644
--- a/compiler/sram/sram_base.py
+++ b/compiler/sram/sram_base.py
@@ -41,6 +41,14 @@ class sram_base(design, verilog, lef):
         if not self.num_spare_cols:
             self.num_spare_cols = 0
 
+        try:
+            from tech import power_grid
+            self.supply_stack = power_grid
+        except ImportError:
+            # if no power_grid is specified by tech we use sensible defaults
+            # Route a M3/M4 grid
+            self.supply_stack = self.m3_stack
+
     def add_pins(self):
         """ Add pins for entire SRAM. """
 
@@ -230,7 +238,7 @@ class sram_base(design, verilog, lef):
     def create_modules(self):
         debug.error("Must override pure virtual function.", -1)
 
-    def route_supplies(self):
+    def route_supplies(self, bbox=None):
         """ Route the supply grid and connect the pins to them. """
 
         # Copy the pins to the top level
@@ -239,29 +247,21 @@ class sram_base(design, verilog, lef):
             for inst in self.insts:
                 self.copy_power_pins(inst, pin_name, self.ext_supply[pin_name])
         
-        try:
-            from tech import power_grid
-            grid_stack = power_grid
-        except ImportError:
-            # if no power_grid is specified by tech we use sensible defaults
-            # Route a M3/M4 grid
-            grid_stack = self.m3_stack
-
         if not OPTS.route_supplies:
             # Do not route the power supply (leave as must-connect pins)
             return
         elif OPTS.route_supplies == "grid":
             from supply_grid_router import supply_grid_router as router
-            rtr=router(grid_stack, self)
         else:
             from supply_tree_router import supply_tree_router as router
-            rtr=router(grid_stack,
-                       self,
-                       pin_type=OPTS.route_supplies)
+        rtr=router(layers=self.supply_stack,
+                   design=self,
+                   bbox=bbox,
+                   pin_type=OPTS.supply_pin_type)
 
         rtr.route()
 
-        if OPTS.route_supplies in ["left", "right", "top", "bottom", "ring"]:
+        if OPTS.supply_pin_type in ["left", "right", "top", "bottom", "ring"]:
             # Find the lowest leftest pin for vdd and gnd
             for pin_name in ["vdd", "gnd"]:
                 # Copy the pin shape(s) to rectangles
@@ -283,7 +283,7 @@ class sram_base(design, verilog, lef):
                                         pin.width(),
                                         pin.height())
             
-        elif OPTS.route_supplies:
+        elif OPTS.route_supplies and OPTS.supply_pin_type == "single":
             # Update these as we may have routed outside the region (perimeter pins)
             lowest_coord = self.find_lowest_coords()
         
@@ -321,7 +321,7 @@ class sram_base(design, verilog, lef):
             # Grid is left with many top level pins
             pass
 
-    def route_escape_pins(self):
+    def route_escape_pins(self, bbox):
         """
         Add the top-level pins for a single bank SRAM with control.
         """
@@ -364,7 +364,7 @@ class sram_base(design, verilog, lef):
         from signal_escape_router import signal_escape_router as router
         rtr=router(layers=self.m3_stack,
                    design=self,
-                   margin=8 * self.m3_pitch)
+                   bbox=bbox)
         rtr.escape_route(pins_to_route)
 
     def compute_bus_sizes(self):
@@ -472,6 +472,12 @@ class sram_base(design, verilog, lef):
         self.bitcell = factory.create(module_type=OPTS.bitcell)
         self.dff = factory.create(module_type="dff")
 
+        # Create the bank module (up to four are instantiated)
+        self.bank = factory.create("bank", sram_config=self.sram_config, module_name="bank")
+        self.add_mod(self.bank)
+
+        self.num_spare_cols = self.bank.num_spare_cols
+
         # Create the address and control flops (but not the clk)
         self.row_addr_dff = factory.create("dff_array", module_name="row_addr_dff", rows=self.row_addr_size, columns=1)
         self.add_mod(self.row_addr_dff)
@@ -493,10 +499,6 @@ class sram_base(design, verilog, lef):
             self.spare_wen_dff = factory.create("dff_array", module_name="spare_wen_dff", rows=1, columns=self.num_spare_cols)
             self.add_mod(self.spare_wen_dff)
 
-        # Create the bank module (up to four are instantiated)
-        self.bank = factory.create("bank", sram_config=self.sram_config, module_name="bank")
-        self.add_mod(self.bank)
-
         # Create bank decoder
         if(self.num_banks > 1):
             self.add_multi_bank_modules()
diff --git a/compiler/tests/21_hspice_setuphold_test.py b/compiler/tests/21_hspice_setuphold_test.py
index 634b2982..9154502e 100755
--- a/compiler/tests/21_hspice_setuphold_test.py
+++ b/compiler/tests/21_hspice_setuphold_test.py
@@ -45,8 +45,8 @@ class timing_setup_test(openram_test):
                            'setup_times_HL': [0.16357419999999998],
                            'setup_times_LH': [0.1757812]}
         elif OPTS.tech_name == "sky130":
-            golden_data = {'hold_times_HL': [-0.05615234],
-                           'hold_times_LH': [-0.03173828],
+            golden_data = {'hold_times_HL': [-0.03173828],
+                           'hold_times_LH': [-0.05615234],
                            'setup_times_HL': [0.078125],
                            'setup_times_LH': [0.1025391]}
         else:
diff --git a/compiler/tests/21_ngspice_setuphold_test.py b/compiler/tests/21_ngspice_setuphold_test.py
index dab02e7d..9bda2c2c 100755
--- a/compiler/tests/21_ngspice_setuphold_test.py
+++ b/compiler/tests/21_ngspice_setuphold_test.py
@@ -45,8 +45,8 @@ class timing_setup_test(openram_test):
                            'setup_times_HL': [0.1757812],
                            'setup_times_LH': [0.1879883]}
         elif OPTS.tech_name == "sky130":
-            golden_data = {'hold_times_HL': [-0.05615234],
-                           'hold_times_LH': [-0.03173828],
+            golden_data = {'hold_times_HL': [-0.03173828],
+                           'hold_times_LH': [-0.05615234],
                            'setup_times_HL': [0.078125],
                            'setup_times_LH': [0.1025391]}
         else:
diff --git a/compiler/tests/21_xyce_setuphold_test.py b/compiler/tests/21_xyce_setuphold_test.py
index f53212f8..3aabce06 100755
--- a/compiler/tests/21_xyce_setuphold_test.py
+++ b/compiler/tests/21_xyce_setuphold_test.py
@@ -45,8 +45,8 @@ class timing_setup_test(openram_test):
                            'setup_times_HL': [0.16357419999999998],
                            'setup_times_LH': [0.1757812]}
         elif OPTS.tech_name == "sky130":
-            golden_data = {'hold_times_HL': [-0.05615234],
-                           'hold_times_LH': [-0.03173828],
+            golden_data = {'hold_times_HL': [-0.03173828],
+                           'hold_times_LH': [-0.05615234],
                            'setup_times_HL': [0.078125],
                            'setup_times_LH': [0.1025391]}
         else:
diff --git a/compiler/verify/magic.py b/compiler/verify/magic.py
index e4f0b428..454dc176 100644
--- a/compiler/verify/magic.py
+++ b/compiler/verify/magic.py
@@ -89,7 +89,10 @@ def write_drc_script(cell_name, gds_name, extract, final_verification, output_pa
     f.write("{} -dnull -noconsole << EOF\n".format(OPTS.drc_exe[1]))
     # Do not run DRC for extraction/conversion
     f.write("drc off\n")
-    f.write("gds polygon subcell true\n")
+    f.write("set VDD vdd\n")
+    f.write("set GND gnd\n")
+    f.write("set SUB gnd\n")
+    #f.write("gds polygon subcell true\n")
     f.write("gds warning default\n")
     # These two options are temporarily disabled until Tim fixes a bug in magic related
     # to flattening channel routes and vias (hierarchy with no devices in it). Otherwise,
@@ -177,6 +180,10 @@ def write_drc_script(cell_name, gds_name, extract, final_verification, output_pa
     f.write('puts "Finished drc check"\n')
     f.write("drc catchup\n")
     f.write('puts "Finished drc catchup"\n')
+    # This is needed instead of drc count total because it displays
+    # some errors that are not "DRC" errors.
+    # f.write("puts -nonewline \"Total DRC errors found: \"\n")
+    # f.write("puts stdout [drc listall count total]\n")
     f.write("drc count total\n")
     f.write("quit -noprompt\n")
     f.write("EOF\n")
@@ -244,11 +251,14 @@ def write_lvs_script(cell_name, gds_name, sp_name, final_verification=False, out
     if not output_path:
         output_path = OPTS.openram_temp
 
-    setup_file = "setup.tcl"
-    full_setup_file = OPTS.openram_tech + "tech/" + setup_file
-    if os.path.exists(full_setup_file):
+    # Copy .magicrc file into the output directory
+    setup_file = os.environ.get('OPENRAM_NETGENRC', None)
+    if not setup_file:
+        setup_file = OPTS.openram_tech + "tech/setup.tcl"
+        
+    if os.path.exists(setup_file):
         # Copy setup.tcl file into temp dir
-        shutil.copy(full_setup_file, output_path)
+        shutil.copy(setup_file, output_path)
     else:
         setup_file = 'nosetup'
 
diff --git a/technology/freepdk45/tech/tech.py b/technology/freepdk45/tech/tech.py
index 794042f3..f5decd3c 100644
--- a/technology/freepdk45/tech/tech.py
+++ b/technology/freepdk45/tech/tech.py
@@ -135,6 +135,8 @@ layer["m10"] = (29, 0)
 layer["text"]    = (239, 0)
 layer["boundary"]= (239, 0)
 
+use_purpose = {}
+
 # Layer names for external PDKs
 layer_names = {}
 layer_names["active"]  = "active"
diff --git a/technology/scn3me_subm/tech/tech.py b/technology/scn3me_subm/tech/tech.py
index 0bd12162..018a15da 100755
--- a/technology/scn3me_subm/tech/tech.py
+++ b/technology/scn3me_subm/tech/tech.py
@@ -63,6 +63,7 @@ layer["text"]           = (63, 0)
 layer["boundary"]       = (63, 0)
 layer["blockage"]       = (83, 0)
 
+use_purpose = {}
 ###################################################
 ##END GDS Layer Map
 ###################################################
diff --git a/technology/scn4m_subm/tech/tech.py b/technology/scn4m_subm/tech/tech.py
index 1812d19e..dc6cd866 100644
--- a/technology/scn4m_subm/tech/tech.py
+++ b/technology/scn4m_subm/tech/tech.py
@@ -119,6 +119,8 @@ layer["m4"]         = (31, 0)
 layer["text"]           = (63, 0)
 layer["boundary"]       = (63, 0)
 
+use_purpose = {}
+
 # Layer names for external PDKs
 layer_names = {}
 layer_names["active"]  = "active"