diff --git a/dpsynth/discrete_mechanisms/swift.py b/dpsynth/discrete_mechanisms/swift.py
index 1962dd6..fc644f0 100644
--- a/dpsynth/discrete_mechanisms/swift.py
+++ b/dpsynth/discrete_mechanisms/swift.py
@@ -29,6 +29,7 @@
 import dataclasses
 import functools
 import math
+import time
 import typing
 
 from absl import logging
@@ -169,37 +170,71 @@ def __call__(
         errors, candidates, domain, self.max_clique_size, budget_remaining
     )
 
+    ########################################################
+    # Precompile MirrorDescent + synth while measuring.    #
+    ########################################################
+    closed_oracle = functools.partial(
+        mbi.marginal_oracles.message_passing_stable, jtree=jtree
+    )
+    estimator = mbi.estimation.MirrorDescent(marginal_oracle=closed_oracle)
+    rows = int(mbi.estimation.minimum_variance_unbiased_total(measurements))
+
+    pgm_future, synth_future = None, None
+    try:
+      pgm_future = estimator.precompile(
+          domain, measurements, extra_cliques=list(selected)
+      )
+      synth_future = mbi.extensions.precompile(domain, list(jtree.nodes), rows)
+      logging.info('[SWIFT] Started precompilation of MirrorDescent + synth.')
+    except Exception as e:  # pylint: disable=broad-exception-caught
+      logging.warning('[SWIFT] Precompile failed (non-fatal): %s', e)
+
     ##########################################
     # Measure the selected marginal queries. #
     ##########################################
+    logging.info('[SWIFT] Starting measurements.')
     new_measurements, _ = _measure_selected_marginals(
         rng, answers, selected, budget_remaining
     )
     measurements.extend(new_measurements)
+    logging.info('[SWIFT] Finished measurements.')
 
     ########################################################
     # Estimate the model using all measurements            #
     ########################################################
-
-    closed_oracle = functools.partial(
-        mbi.marginal_oracles.message_passing_stable, jtree=jtree
-    )
+    if pgm_future is not None:
+      t0 = time.time()
+      try:
+        pgm_future.result()
+      except Exception as e:  # pylint: disable=broad-exception-caught
+        logging.warning('[SWIFT] PGM precompile failed (non-fatal): %s', e)
+      logging.info('[SWIFT] PGM precompile wait: %.2fs', time.time() - t0)
 
     callback_fn = mbi.callbacks.default(measurements, domain=domain)
-    model = mbi.estimation.MirrorDescent(
-        marginal_oracle=closed_oracle,
-    ).estimate(
+    final_model = estimator.estimate(
         domain,
         measurements,
         iters=self.pgm_iters,
         potentials=potentials,
         callback_fn=callback_fn,
     )
+    assert isinstance(final_model, mbi.MarkovRandomField)
     logging.info('[SWIFT] Estimated final model.')
 
+    if synth_future is not None:
+      t0 = time.time()
+      try:
+        synth_future.result()
+      except Exception as e:  # pylint: disable=broad-exception-caught
+        logging.warning('[SWIFT] Synth precompile failed (non-fatal): %s', e)
+      logging.info('[SWIFT] Synth precompile wait: %.2fs', time.time() - t0)
+
+    syn = mbi.extensions.synthetic_data(final_model, rows)
+    logging.info('[SWIFT] Generated %d synthetic records.', rows)
+
     return common.DiscreteMechanismResult(
-        model=model,
-        synthetic_data=model.synthetic_data(),
+        model=final_model,
+        synthetic_data=syn,
         measurements=measurements,
     )