diff --git a/app/src/main/cpp/skyline/services/nvdrv/devices/nvhost/host1x_channel.cpp b/app/src/main/cpp/skyline/services/nvdrv/devices/nvhost/host1x_channel.cpp
index ebc3db4a..66f86108 100644
--- a/app/src/main/cpp/skyline/services/nvdrv/devices/nvhost/host1x_channel.cpp
+++ b/app/src/main/cpp/skyline/services/nvdrv/devices/nvhost/host1x_channel.cpp
@@ -3,7 +3,6 @@
 
 #include <soc.h>
 #include <services/nvdrv/devices/deserialisation/deserialisation.h>
-#include <gpu.h>
 #include "host1x_channel.h"
 
 namespace skyline::service::nvdrv::device::nvhost {
@@ -39,10 +38,12 @@ namespace skyline::service::nvdrv::device::nvhost {
         for (size_t i{}; i < syncpointIncrs.size(); i++) {
             const auto &incr{syncpointIncrs[i]};
 
+            u32 max{core.syncpointManager.IncrementSyncpointMaxExt(incr.syncpointId, incr.numIncrs)};
+
+            // Increment syncpoints on the CPU to avoid needing to pass through the emulated nvdec code which currently does nothing
             for (size_t j{}; j < incr.numIncrs; j++)
                 state.soc->host1x.syncpoints[incr.syncpointId].Increment();
 
-            u32 max{core.syncpointManager.IncrementSyncpointMaxExt(incr.syncpointId, incr.numIncrs)};
             if (i < fenceThresholds.size())
                 fenceThresholds[i] = max;
         }
@@ -56,7 +57,8 @@ namespace skyline::service::nvdrv::device::nvhost {
             Logger::Debug("Submit gather, CPU address: 0x{:X}, words: 0x{:X}", gatherAddress, cmdBuf.words);
 
             span gather(reinterpret_cast<u32 *>(gatherAddress), cmdBuf.words);
-       //     state.soc->host1x.channels[static_cast<size_t>(channelType)].Push(gather);
+            // Skip submitting the cmdbufs as no functionality is implemented
+            // state.soc->host1x.channels[static_cast<size_t>(channelType)].Push(gather);
         }
 
         return PosixResult::Success;