Technologies for data compression include a computing device having multiple search agents. Each search agent searches a history of an input stream in parallel for a match to the input stream starting at a position based on an index of the search agent. Each search agent generates in parallel a weight value associated with the corresponding match. The weight value is indicative of a length associated with the match and an encoded length associated with the match. The encoded length is indicative of a number of bits to encode the match. The computing device selects a match based on the weight values. The computing device may output a token for the selected match and encode the token using a Huffman coding. Each search agent may be embodied as a hardware component or a software component such as a thread or process. Other embodiments are described and claimed.