diff --git a/applications/CMakeLists.txt b/applications/CMakeLists.txt
index 5a75b84e1c0c3b4d4daa19381d7032489d2cbddb..f472b3d2b80421c4f18b1283a63f80a065a97435 100644
--- a/applications/CMakeLists.txt
+++ b/applications/CMakeLists.txt
@@ -1,2 +1,3 @@
 add_subdirectory(poisson_dg)
 add_subdirectory(poisson_dg_tensor)
+add_subdirectory(knl)
diff --git a/applications/knl/CMakeLists.txt b/applications/knl/CMakeLists.txt
new file mode 100644
index 0000000000000000000000000000000000000000..f935128ee68641b5bcecd6287f913d92045ac2cc
--- /dev/null
+++ b/applications/knl/CMakeLists.txt
@@ -0,0 +1,2 @@
+add_subdirectory(poisson_dg)
+
diff --git a/applications/knl/poisson_dg/CMakeLists.txt b/applications/knl/poisson_dg/CMakeLists.txt
new file mode 100644
index 0000000000000000000000000000000000000000..3fb870b93fe7503680851d173e82421629e20e18
--- /dev/null
+++ b/applications/knl/poisson_dg/CMakeLists.txt
@@ -0,0 +1,5 @@
+dune_add_formcompiler_system_test(UFLFILE poisson_dg.ufl
+                                  BASENAME app_knl_poisson
+                                  INIFILE knl_poisson_dg.mini
+                                  NO_TESTS
+                                  )
diff --git a/applications/knl/poisson_dg/knl_poisson_dg.mini b/applications/knl/poisson_dg/knl_poisson_dg.mini
new file mode 100644
index 0000000000000000000000000000000000000000..876dc1cafe657e75ed9da13211d5fb334dde6a85
--- /dev/null
+++ b/applications/knl/poisson_dg/knl_poisson_dg.mini
@@ -0,0 +1,50 @@
+__name = app_knl_poisson_dg_{__exec_suffix}
+__exec_suffix = deg{formcompiler.ufl_variants.degree}_{opcount_suffix}_level{formcompiler.instrumentation_level}
+
+opcount_suffix = opcount, nonopcount | expand opcount
+{opcount_suffix} == opcount and {formcompiler.instrumentation_level} != 4 | exclude
+
+# Calculate the size of the grid to equlibritate it to 100 MB/rank
+# Input parameters
+dim = 3
+mbperrank = 100
+ranks = 16
+floatingbytes = 8
+
+# Metaini Calculations
+memperrank = {mbperrank} * 1048576 | eval
+dofsperdir = {formcompiler.ufl_variants.degree} + 1 | eval
+celldofs = {dofsperdir} ** {dim} | eval
+cellsperrank = {memperrank} / ({floatingbytes} * {celldofs}) | eval
+cellsperdir = {cellsperrank} ** (1/{dim}) | eval | toint
+firstdircells = {ranks} * {cellsperdir} | eval
+dimminusone = {dim} - 1 | eval
+ones = 1 | repeat {dimminusone}
+otherdircells = {cellsperdir} | repeat {dimminusone}
+
+# Set up the timing identifier
+identifier = knl_poisson_dg_deg{formcompiler.ufl_variants.degree}
+
+# Setup the grid!
+extension = 1.0 | repeat {dim}
+cells = {firstdircells} {otherdircells}
+partitioning = {ranks} {ones}
+
+[wrapper.vtkcompare]
+name = {__name}
+extension = vtu
+
+[formcompiler]
+fastdg = 1
+sumfact = 1
+vectorize_quad = 1
+vectorize_diagonal = 1
+instrumentation_level = 2, 3, 4 | expand
+opcounter = 1, 0 | expand opcount
+time_opcounter = 0, 1 | expand opcount
+quadrature_order = {formcompiler.ufl_variants.degree} * 2 | eval
+architecture = knl
+
+[formcompiler.ufl_variants]
+cell = hexahedron
+degree = 1, 3, 5, 7, 9  | expand
diff --git a/applications/knl/poisson_dg/poisson_dg.ufl b/applications/knl/poisson_dg/poisson_dg.ufl
new file mode 100644
index 0000000000000000000000000000000000000000..2c0f2c34436f1c6b76817d6abad344fae3183d62
--- /dev/null
+++ b/applications/knl/poisson_dg/poisson_dg.ufl
@@ -0,0 +1,32 @@
+dim = 3
+x = SpatialCoordinate(cell)
+f = -6.
+g = x[0]*x[0] + x[1]*x[1] + x[2]*x[2]
+
+V = FiniteElement("DG", cell, degree)
+
+u = TrialFunction(V)
+v = TestFunction(V)
+
+n = FacetNormal(cell)('+')
+
+alpha = 1.0
+h_ext = CellVolume(cell) / FacetArea(cell)
+cse_gamma_ext = (alpha * degree * (degree + dim - 1)) / h_ext
+h_int = Min(CellVolume(cell)('+'), CellVolume(cell)('-')) / FacetArea(cell)
+cse_gamma_int = (alpha * degree * (degree + dim - 1)) / h_int
+
+theta = -1.0
+
+r = inner(grad(u), grad(v))*dx \
+  + inner(n, avg(grad(u)))*jump(v)*dS \
+  + cse_gamma_int*jump(u)*jump(v)*dS \
+  - theta*jump(u)*inner(avg(grad(v)), n)*dS \
+  - inner(n, grad(u))*v*ds \
+  + cse_gamma_ext*u*v*ds \
+  + theta*u*inner(grad(v), n)*ds \
+  - f*v*dx \
+  - theta*g*inner(grad(v), n)*ds \
+  - cse_gamma_ext*g*v*ds
+
+forms = [r]