#485 cpudb Vishera or tool?

I find vanity and intrigue (possibly debugging) in my wish for either a tool (similar to niclist) to automatically generate for a new cpudb templated (appropriately) a header and source for Bochs with the appropriate flags Bochs supports etc, or just support for the Vishera AMD CPUs such as the 9590. Greatly appreciate it. :] Might do it myself but my C/C++ isn't as nice as I'd wish it.

[Anonymous]

I see that sshwarts was the developer on the other cpudbs. Perhaps I can give him a CPUID txt dump?

[Stanislav Shwartsman]

Actually I have Trinity CPU (which is also PileDriver core) added in the CPUDB for the next release. The Trinity from the CPUID perspective is the same as Vishera. It only differs in non-signifcant CPUID leafs like cache size or brand name.

BTW, AMD CPUID is still controversial for me. Both Trinity and Vishera real CPUIDs (found in the net) contradict official documentation by AMD from their website.

For example I see in the leaf 0x8000000A I see bit 11 is set (which is reserved) while bit 13 (AVIC) is clear. Does it mean that AVIC is not supported ? Or it is supported by AMD messed with the documentation and it located in another bit ?

Another example: 0x80000001.ECX[17], it is reserved according to all AMD docs but set for Trinity and Vishera CPUs ...

Stanislav

[Anonymous]

I've noticed some oddities too in my AMD FX-8350 BE, such as 16MB L2 and L3 (8 and 8) but others reporting only 12 or 8 even (combined totals). I was halfway through writing a new .c and .h before giving up on determining the flags myself, heh. Thanks for your hard work. : )