Hacking Sneek Speed and Cluster Size

[apoptygma]

Just wondering how people have seen the performance of Sneek shift when using larger or smaller clusters. It's been identified that there is a bit of code restricting the loading of data to one cluster at a time and so in theory, formatting an SD with the largest clusters possible should provide the best speed. The only problem here is that some compatibility issues might arise from not using standard 32k clusters.

 

[thesund0g]

You should see better performance out of large clusters based on what Crediar said. Just bear in mind you will have more "slack space" as a cluster cannot house more than one file. All that can add up with a smaller card, but it probably isn't a problem.