*** 0.10a ***
13/06/2005: release 0.10a to fix bugs -- 0.11a should have new database for
	    some chips.

	    minor string changes in x86test.S and cpu_lookup.S.

12/06/2005: move platform tests to seperate function and add #GP checks for MSR
	    reads -- it seems only deschutes and later support platform ID.

01/06/2005: minor P6, P6T and deschutes corrections in cpu_lookup.S.

30/05/2005: ignore cpuid #1 %ebx on Centaur as for cyrix in cpuid.S.

	    cyrix 6x86/6x86L (M1) fix for loop_clock_test -- loop instruction
	    only takes 2 cycles on these chips (and they don't have a TSC).

	    workaround for AMD K5 SSA5 with APIC bit incorrectly set.

26/05/2005: minor AMD fixes in cpu_lookup.S.

	    changes to centaur checking at start of cpuid routine, including
	    support for VIA FCR MSRs. Most centaur chips still untested!

25/05/2005: added intel cache entries 0x78,7d,7f and ignore 0xf0,f1 (64/128
	    byte prefetch size).

20/05/2005: minor AMD 486 fixes.

*** 0.09a ***
15/05/2005: lots more AMD family f updates, including mobile codenames.
	    extended model 2 stuff may still be incomplete.

*** 0.08a ***
14/05/2005: updated AMD family f support, includeing correct (i think) entries
	    for Newcastle and Winchester, as well as preliminary entries for
	    San Diego/Venice/Palermo and Toledo/Manchester (the X2). Minor
	    intel updates. fixed extended model reporting in cpuid.S.

04/05/2005: improved output for cpuid_probe.

*** 0.07a ***
19/04/2005: avoid testing DIV flags after CPUID for cyrix (start of x86test.S).
	    Added preliminary intel Smithfield entry in cpu_lookup.S.

14/04/2005: increase 14.318MHz divisor limit to allow divisor of 64. This
	    should only be needed when we only know the core clock and there is
	    a half-integer multiplier e.g. I have a skt7 board with 115MHz FSB
	    as 257/32 * 14.318MHz, with a 2.5x multiplier the core clock is
	    then 1285/64 * 14.318MHz.

	    ignore %ebx output for cpuid #1 on cyrix (they return "Cyri" as for
	    cpuid #0).

05/04/2005: print proper 96-bit PSN for cpuid #3.

04/04/2005: re-wrote cpuid #0x80000005 code, including support for crusoe. Also
	    other cpuid fixes for new intel functions. added number of logical
	    (HTT) processors for cpuid #1.

*** 0.06a ***
27/03/2005: added psn command to disable serial number reporting on P3.

25/03/2005: add intel prescott N0 and dothan C-0 steppings to cpu_lookup.S.
	    updated feature flags in cpuid.S

*** 0.05a ***
06/03/2005: get loop_clock_test to use highest measurement from several ticks
	    to work around clock-stealing SMIs etc. moved setup_tramp in lilo.S
	    so grub works, also added a version string. added a cld to start of
	    realmode.S. updated INSTALL. updated Makefile.

*** 0.04a ***

02/03/2005: release 0.04a -- again only minor update, mostly bug fixes.

01/03/2005: fixed bug in cyrix_test that was trashing cpu_features after cpuid.

27/02/2005: reset SSEDIS flag in HWCR MSR for AMD K7/K8 in cpuid.S. workaround
	    AMD thunderbird A1/A2 CPUID L2 bug as for spitfire A0 in cpuid.S.
	    Added GP fault checking in apic_init -- not sure about K7 model 1.

21/02/2005: tidy-up/rearrangement/updates of cpuid.S

17/02/2005: removed VGA horizontal timing and 8/9-dot stuff from term.S to
	    improve reliability. updates to cyrix MediaGX section of
	    cpu_lookup.S. changed default IRQ0 handling in pm.S.

16/02/2005: cyrix MediaGX fix for loop_clock_test -- loop instruction only
	    takes 3 cycles on these chips.

*** 0.03a ***

09/01/2005: this is only a minor update to 0.02a -- if 0.02a didn't work then
	    chances are 0.03a won't either!

notable changes since 0.02a:
 - many additions/fixes to cpu database (cpu_lookup.S). mostly intel/AMD.
 - seperated real-mode code used at startup into realmode.S
 - added reporting of intel platform ID
 - made some functions available as user commands
 - no longer run crappy bios_test by default
 - less informational messages that users won't care about.
 - removed brand ID translation in cpuid.S.
 - numerous minor corrections and optimisations