Squall Leonhart Posted August 12, 2017 Share Posted August 12, 2017 I was reading a review just now from a certain website, and the reviewer has blamed Aida64 for not being compatible with Threadrippers UMA mode on its less than perfect memory bandwidth. This same reviewer went on to bench a bunch of video games in UMA, just because it was the default setting completely ignoring the fact that the NUMA setting is intended for gaming appication. He might have well have been advertising for Intel with the gaming scores. IMO, its time these has beens leave the hardware journalism business if they don't even understand whats going on anymore. Quote Link to comment Share on other sites More sharing options...
Fiery Posted August 17, 2017 Share Posted August 17, 2017 That is not really ignorance, but more a failed understanding of how AIDA64 memory bandwidth benchmarks work. Even hardware manufacturers (I won't name drop here) fall into that trap from time to time. What's important to know about our memory bandwidth benchmarks is that even though they are multi-threaded, it doesn't mean that they will always utilize all available cores or logical processors (threads) to measure bandwidth. We use a clever automatic calibration process that finds the optimal number of CPU resources to attain the highest possible memory bandwidth scores. The whole UMA vs. NUMA score issue would look different if AIDA64 always utilized all 16 cores (or 32 threads) of AMD Threadripper processors. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.