diff --git a/src/amd/compiler/aco_instruction_selection.cpp b/src/amd/compiler/aco_instruction_selection.cpp
index 1de79f01da8..195c57950c5 100644
--- a/src/amd/compiler/aco_instruction_selection.cpp
+++ b/src/amd/compiler/aco_instruction_selection.cpp
@@ -3463,8 +3463,10 @@ void emit_interp_instr(isel_context *ctx, unsigned idx, unsigned component, Temp
    Temp coord2 = emit_extract_vector(ctx, src, 1, v1);
 
    Builder bld(ctx->program, ctx->block);
-   Temp tmp = bld.vintrp(aco_opcode::v_interp_p1_f32, bld.def(v1), coord1, bld.m0(prim_mask), idx, component);
-   bld.vintrp(aco_opcode::v_interp_p2_f32, Definition(dst), coord2, bld.m0(prim_mask), tmp, idx, component);
+   Builder::Result interp_p1 = bld.vintrp(aco_opcode::v_interp_p1_f32, bld.def(v1), coord1, bld.m0(prim_mask), idx, component);
+   if (ctx->program->has_16bank_lds)
+      interp_p1.instr->operands[0].setLateKill(true);
+   bld.vintrp(aco_opcode::v_interp_p2_f32, Definition(dst), coord2, bld.m0(prim_mask), interp_p1, idx, component);
 }
 
 void emit_load_frag_coord(isel_context *ctx, Temp dst, unsigned num_components)
diff --git a/src/amd/compiler/aco_instruction_selection_setup.cpp b/src/amd/compiler/aco_instruction_selection_setup.cpp
index 558c6a7eabb..b2acc5f9cb3 100644
--- a/src/amd/compiler/aco_instruction_selection_setup.cpp
+++ b/src/amd/compiler/aco_instruction_selection_setup.cpp
@@ -1099,6 +1099,9 @@ setup_isel_context(Program* program,
 
    program->lds_alloc_granule = args->options->chip_class >= GFX7 ? 512 : 256;
    program->lds_limit = args->options->chip_class >= GFX7 ? 65536 : 32768;
+   /* apparently gfx702 also has 16-bank LDS but I can't find a family for that */
+   program->has_16bank_lds = args->options->family == CHIP_KABINI || args->options->family == CHIP_STONEY;
+
    program->vgpr_limit = 256;
    program->vgpr_alloc_granule = 3;
 
diff --git a/src/amd/compiler/aco_ir.h b/src/amd/compiler/aco_ir.h
index 92511975a69..5bbe337fe17 100644
--- a/src/amd/compiler/aco_ir.h
+++ b/src/amd/compiler/aco_ir.h
@@ -1197,6 +1197,7 @@ public:
    uint16_t min_waves = 0;
    uint16_t lds_alloc_granule;
    uint32_t lds_limit; /* in bytes */
+   bool has_16bank_lds;
    uint16_t vgpr_limit;
    uint16_t sgpr_limit;
    uint16_t physical_sgprs;