After many different works on Rowhammer the current situation is largely unclear: The prevalence numbers we have on DDR3 say 60-80% of modules are affected. What is the prevalence of DDR4? Has Rowhammer become a niche problem? With ECC and TRR bypassable in practice – which countermeasures should high-availability systems employ to minimize risks? A systematization will provide a thorough overview and answer these questions.
Taiga RIG Inventory project here