I do not know if there are more recent options, but VISTA is designed for this type of task at the level of sequence conservation: http://genome.lbl.gov/vista

Pax6 should be rather simple as it is small (~20kb human), appears to have two clear alternative transcription start sites at the beginning (first and second exons, as well as internal starts), and has several very highly conserved elements upstream of those. Because of that, I don't expect you need to do a lot of discovery (e.g., using long range dot plotting). Rather, I expect you can easily recognize the first two exons, transcription starts/promoters by sequence alignment.

Because Pax6 is well studied, it seems likely that published experimental evidence and your alignments will be more informative than promoter prediction algorithms (but it wouldn't hurt to look for things that may be novel); however, once you map the alternative first exons, you can use such programs to precisely map promoter elements/transcription start site. Ideally, you could experimentally validate any findings by 5' RACE using your cell/organism of interest.

There are genome walking kits available to find promoters of genes of interests. Its PCR based, you don't need to construct libraries into BACs or YACs, pretty easy to do, but time consuming. Thats one way to be sure if you don't have a reference genome. Here is a link to a clontech kit... http://www.clontech.com/US/Products/cDNA_Synthesis_and_Library_Construction/cDNA_and_Genomic_DNA/GenomeWalker_Kits?gclid=CjwKEAiA3vamBRDJ1Lfwt5Pckw4SJAAdhnk27_mvAi5wOKi1tfR2aNS74Tm70FvZHwuAjIym7JLwvxoCqDnw_wcB

Many thanks Carlos and Nigel.