-
Hi CacheLib developers, In the cachebench stressor, upon a get operation cache hit, it seems like the returned item handle is not used by cachebench. My understanding is that (please correct me if I am wrong) since the item handle is not used, the item value is actually not accessed. I imagine that in a real CacheLib use case, the item value has to be accessed at least once (e.g. sent throughput the network). From a memory behavior perspective, does this means that the current cachebench is underestimating the number of memory accesses? If my understanding above is correct, what do you think would be a reasonable way to access the item values? Would a string copy/memcpy work? Note that I think that the item keys are already being accessed within Thank you for your time, |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
Hi @kevins981, Actually, the Cachebench supports accessing all bytes of found item. You can enable the option using "touchValue" Also, please note that the main use of cachebench is not to simulate the application behavior. |
Beta Was this translation helpful? Give feedback.
Hi @kevins981,
Actually, the Cachebench supports accessing all bytes of found item. You can enable the option using "touchValue"
Also, please note that the main use of cachebench is not to simulate the application behavior.