Port-powerpc archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Show stopper bugs for 5.0



-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

On May 21, 2008, at 16:54, Manuel Bouyer wrote:

On Wed, May 21, 2008 at 04:40:06PM -0400, Michael Lorenz wrote:


On May 21, 2008, at 15:43, Manuel Bouyer wrote:

On Tue, May 20, 2008 at 08:39:09PM -0400, Michael Lorenz wrote:
I didn't commit the fix yet because chuq thinks that's just a
workaround for a problem elsewhere, but I couldn't find anything
remotely related in the 7400 errata list. He's got a dual 7455 which doesn't have this problem - his CPUs have L3 cache though, no idea if

Maybe it's L1 cache have different characteristics ?

Both have 32kB instruction and 32kB data cache.

Could it happen that this part of code has already been cached in your L1 instruction cache at this point of operations, but not on chuq's ?

As far as I know the L1 caches should behave identical on both CPUs.

Do you know the line size and associativity ?

Both L1 caches are 32-byte line, 8-way set associative on both the 7400 and 7455.

have fun
Michael

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)

iQEVAwUBSDSPRcpnzkX8Yg2nAQLtVAf7BEk8TRo4hsjFz/M1xnvqQ7/RziilrJVX
GP/Y2TvJS0WdEmua64rJUX+ed2+a7u07LjaD1gMrtDmapbrqjbPgd8xRQMGHD4Wb
oEdBhNK2udFrhFGhdELDP77ze45huRxq0PjWaSzxMKD+bb7tScQ4oESfdTqwbK4X
TDFhCdvnFh09Sjrr6v0UyKI0w2DRywt45phoT8AXrik9iBA+A3Inbq1nigbqkTNZ
8X5e3sXkVISF7MpCFRs9TJyD2zBwXrXFagBD2NOvkrpbT0bMU4u6CpWoeH61GeBI
JgpLpoVRTW4nQguac0MhY67GvLRbYf8Zet5PVvbsXjUq0xjic1wuBA==
=ZzRt
-----END PGP SIGNATURE-----


Home | Main Index | Thread Index | Old Index