Hi. In <[EMAIL PROTECTED]>, on "Thu, 09 Nov 2000 21:21:38 +0900", Taketoshi Sano <[EMAIL PROTECTED]> wrote:
> The modification to use the new fb code was introduced on chips driver > at 4.0.1d partly, and that was very possibly the cause of the problem. > But 4.0.1e completely moves to use the new fb code and most of the old > cfb code is now removed from chips driver code. A developer in chips list explains the problem. 4.0.1d just set a register in wrong place to wrong value (not set a register in reuired place to be correct). 4.0.1e fixed it as well as other change, so this lock up problem was solved in 4.0.1e. He said: | Can you reply to the debian lists that this problem exists | and the solution. I've only seen debian e-mails second hand. So I send this. > I have checked the xStone and it shows great improvement, at least > in depth 16 where I tested both of 4.0.1c and 4.0.1d (4.0.1d was > broken and it does not work with/without "UseFB" option, so I can't > check the performance of it). I found that 4.0.1c has lower score on xbench than 3.3.6, but 4.0.1e regain it as well as this fix. (I've not tested by x11perf yet) Regards. -- Taketoshi Sano: <[EMAIL PROTECTED]>
