I'm not at all sure that CPU cache will have much effect for DF. A cache is only useful if you access the same data often enough that it is still in the cache since the last time you accessed it. If you cycle sequentially through 5MB of data and only have a 4MB cache, the data will keep getting overwritten before you need it again. The biggest CPU caches seem to be 12MB, which is pretty small compared to the amount of data used by DF. I'm not sure one way or the other if any CPU has enough cache for DF to get good use from it.
Do you have data to back up your assertion that CPU cache is important for DF?
Memory latency is surely a larger limmiting factor, but since cache is in the memory hiarchy too, it should have a greator effect than other CPU features, like clock speed, core number, or some CPU benchmark results. Toady does optimize the game, and the number one optimisation is likely to be to duplicated frequently accessed data out of large arrays, into small arrays, and other forms of consolidating data. These optimisations would be most effective with a large cache. When these smaller arrays fit or nearly fit into a cache block, or at least the cache as a whole, they are much more effective. Specifically, the L3 cache or the intel Smart Cache should be the most relevant metric.
High proformance RAM is more important though, and you want to make sure your RAM, motherboard, and CPU support the fastest memory speeds you can find for all 3 (the lowest of the three wins out). And you want the lowest latancy RAM you can find. Be aware though that while it is easy enough to determine which memory the best of a given set, the differences in memory are not very great. If the question comes up whether it's worth spending a little more on memory, or save money and get a cheaper computer, the cheaper computer may well be the better option.
Note: none of this speculation is based on actual testing.
Author
Topic: Intel Core i3 or Core i5 vs AMD A8 for Dwarf Fortress (Read 3814 times)