We really need to stress the fact that we work with version 4.1, not 4.3! 4.3 gives better results, so there is no possibility to compare both benches.
Yesterday i've tried to clean out the newcastle results, because it's really messy (indeed, if you just enter 3000+, it becomes a newcastle). It's almost done, but I'm really thankful you edited the score yourself.
Thanks
I checked validation + clockspeed
SuperPi 1M will take more time
By the way, can we search through the results and filter out the verified results ? (cfr. filter blocked results)
Just finished the last entries at 09/09 22h30. It took me about two days to check every single entrie in the 32M challenge, but finally the database is ok.
# [ SuperPi 32m ] (417 results) : 09/09/06 22h30
# [ PiFast ] (104 results) : 10/09/06 14h30
Next up: SiSandra
Well, since LOD can be used by ATI and NVIDIA cards, it's no longer considered as a cheat (ATI benchers referred to LOD before they could use it).
Like you said, almost all the top scores are with lod.