add 18th iteration of classifier

1 files changed, 46 insertions, 0 deletions

diff --git a/results/classifier/118/performance/1743 b/results/classifier/118/performance/1743
new file mode 100644
index 00000000..ef710f22
--- /dev/null
+++ b/results/classifier/118/performance/1743
@@ -0,0 +1,46 @@
+x86: 0.952
+performance: 0.942
+device: 0.923
+graphic: 0.913
+PID: 0.809
+socket: 0.726
+files: 0.620
+vnc: 0.588
+network: 0.587
+register: 0.547
+permissions: 0.508
+boot: 0.507
+arm: 0.490
+ppc: 0.472
+semantic: 0.451
+debug: 0.436
+risc-v: 0.415
+mistranslation: 0.413
+architecture: 0.402
+kernel: 0.381
+TCG: 0.349
+i386: 0.258
+user-level: 0.255
+peripherals: 0.212
+VMM: 0.210
+virtual: 0.161
+hypervisor: 0.092
+assembly: 0.038
+KVM: 0.031
+
+QEm+Android emulator crashes on x86 host (but not mac M1)
+Description of problem:
+Using QEmu+Android emulator crashes when using tflite on x86 hosts (but not M1 macs).
+Steps to reproduce:
+1. Install android toolchain, including emulator (sdkmanager, adb, avdmanager etc)
+2. Start android emulator on an x86 host
+3. Follow instructions to download and run tflite benchmarking tool [here](https://www.tensorflow.org/lite/performance/measurement)
+4. Crashes with the following error
+
+```
+06-27 17:38:28.093  8355  8355 F ndk_translation: vendor/unbundled_google/libs/ndk_translation/intrinsics/intrinsics_impl_x86_64.cc:86: CHECK failed: 524288 == 0
+```
+
+We have tried with many different models and the result is always the same. The same models run fine when the emulator runs on a mac M1 host.
+Additional information:
+