Building Missing Maps with Machine Learning
By Humanity & Inclusion
8 months ago
My latest submission has been crashing with what I suspect is linked to this issue (https://github.com/tensorflow/tensorflow/issues/17411).
My code was crashing with tensorflow 1.5, and upgrading to 1.8+ seems to fix it. However I got into trouble as official tensorflow builds >1.5.0 now use optimisations such as the Intel AVX instruction sets.
I ran into the issue mentioned above on my cloud GPU instance (a P5000 at paperspace, if it matters), and since my docker image is crashing in a similar way when I submit it here, I wonder if it’s affected by the same bug.
In short, could you share what CPU (not the GPU) our submissions are run on, or specifically, what instruction sets are available (cat /proc/cpuinfo should do).
This will help decide what build of tensorflow I can use.
Or maybe I just have another bug, but I haven’t got access to the logs yet…
7 months ago |
@LS207 : This is the server where the submissions are currently running :
processor : 47
vendor_id : GenuineIntel
cpu family : 6
model : 63
model name : Intel(R) Xeon(R) CPU E5-2680 v3 @ 2.50GHz
stepping : 2
microcode : 0x3c
cpu MHz : 1215.722
cache size : 30720 KB
physical id : 1
siblings : 24
core id : 13
cpu cores : 12
apicid : 59
initial apicid : 59
fpu : yes
fpu_exception : yes
cpuid level : 15
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm epb invpcid_single ibrs ibpb stibp kaiser tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm xsaveopt cqm_llc cqm_occup_llc dtherm ida arat pln pts
bugs : cpu_meltdown spectre_v1 spectre_v2 spec_store_bypass
bogomips : 4990.08
clflush size : 64
cache_alignment : 64
address sizes : 46 bits physical, 48 bits virtual