Sunday, June 12, 2011

Revisiting Bulldozer ES weirdness story

You all remember my original blog post about BD ES weirdness that goes on in Far East(and probably elsewhere). Problem is/was that those chips are gimped in many ways so that competition is unable to figure out what is the true potential of the first new AMD core design since original K7.

Well we have a sequel now.

It seems some folks(obr from czech PCT website) can't keep their mouth shut.The guy in question has been known for his hatred towards AMD since ages ago,but that's beside the point.Whether he has signed NDA or not,he again published some ES (presumably B1) scores which are now surprisingly much better than B0 scores floating around on the net.As I don't want to link to original post,I will just post a summary of how much faster B1 is. B0 numbers can be found in chiphell forum and on youtube(yes on youtube...).

SPi 1m
B1 is 79%/60% faster than B0 (B0 scores 26s with C6 disabled/23.3s with C6 enabled Vs B1's 14.5s)

B1 is 56% faster than B0 (B0's 9.15s VS B1's 5.86s)

B1 is 70% faster than B0 (B1 gets 7.86pts vs B0's 4.6pts)

No data on B0,B1 is supposedly getting ~28K or 47% better than Thuban.

Now we have semi crippled B1 which is massively faster than any of those ES presented in Chinese forums. Level of gimping goes down as we approach launch date...


  1. This new number no change your firts speculation numbers?

  2. Well new numbers show us that B1 is better than B0.Whether this is the "real" B1 or just another crippled sample is yet to be seen. I still think retail is better than this B1.Maybe just a bit better ,maybe more.