ponderous 'make world' times post GCC 3.3...
Harti Brandt
brandt at fokus.fraunhofer.de
Wed Aug 13 08:52:00 PDT 2003
On Wed, 13 Aug 2003, Steve Ames wrote:
SA>On Wed, Aug 13, 2003 at 05:23:56PM +0200, Harti Brandt wrote:
SA>> The problem seems to be in the kernel, not the compiler. I have rebuilt a
SA>> gcc 3.2.2. The actual kernel (from yesterday) is slightly faster when
SA>> built with 3.3.1 as opposed to 3.2.2 (in the order of 3-4%), but VERY
SA>> slow. A kernel from June 1st built with 3.2.2 has its 'normal' speed. I'm
SA>> trying now I binary search to find the victim.
SA>
SA>This may be a red herring...
SA>
SA>My first post gcc 3.3 kernel took 12 hours to build on my U5. This was a
SA>GENERIC kernel since I hadn't recompiled since January (long story). The
SA>following two kernels also took 12 hours. Then I cleaned up the kernel
SA>config and its back down to around 4.5 hours... It could be a kernel
SA>debugging option maybe? Could be that a bit of hit/miss experimenting
SA>could find it. Here's my current config (4.5 hour 'make world').
The problem is, that compiling just anything takes 3 time longer on an
actual kernel, than on one from June 1st. Same world, same kernel config,
same compiler (3.2.2).
harti
SA>
SA>-Steve
SA>
SA>/sys/sparc64/conf/SPARC
SA>-----------------------------------------
SA>machine sparc64
SA>cpu SUN4U
SA>ident GENERIC
SA>
SA>options SCHED_4BSD #4BSD scheduler
SA>options INET #InterNETworking
SA>options INET6 #IPv6 communications protocols
SA>options FFS #Berkeley Fast Filesystem
SA>options SOFTUPDATES #Enable FFS soft updates support
SA>options UFS_ACL #Support for access control lists
SA>options UFS_DIRHASH #Improve performance on big directories
SA>options NFSCLIENT #Network Filesystem Client
SA>options CD9660 #ISO 9660 Filesystem
SA>options PROCFS #Process filesystem (requires PSEUDOFS)
SA>options PSEUDOFS #Pseudo-filesystem framework
SA>options COMPAT_43 #Compatible with BSD 4.3 [KEEP THIS!]
SA>options COMPAT_FREEBSD4 #Keep this for a while
SA>options SCSI_DELAY=15000 #Delay (in ms) before probing SCSI
SA>options SYSVSHM #SYSV-style shared memory
SA>options SYSVMSG #SYSV-style message queues
SA>options SYSVSEM #SYSV-style semaphores
SA>
SA>device apb # Sun APB PCI-PCI bridge
SA>device ebus
SA>device isa
SA>device pci
SA>device sbus
SA>device central
SA>device fhc
SA>
SA>options OFW_NEWPCI
SA>
SA>device ata
SA>device atadisk # ATA disk drives
SA>device atapicd # ATAPI CDROM drives
SA>
SA>device ofw_console # OpenBoot firmware console device
SA>
SA>device genclock # Generic clock interface
SA>device eeprom # eeprom (really an ebus driver for the MK48Txx)
SA>device "mk48txx" # Mostek MK48T02, MK48T08, MK48T59 clock
SA>
SA>
SA>
SA>device miibus # MII bus support
SA>device hme # Sun HME (Happy Meal Ethernet)
SA>
SA>device random # Entropy device
SA>device loop # Network loopback
SA>device ether # Ethernet support
SA>device pty # Pseudo-ttys (telnet etc)
SA>device gif # IPv6 and IPv4 tunneling
SA>device faith # IPv6-to-IPv4 relaying/(translation)
SA>
--
harti brandt,
http://www.fokus.fraunhofer.de/research/cc/cats/employees/hartmut.brandt/private
brandt at fokus.fraunhofer.de, harti at freebsd.org
More information about the freebsd-sparc64
mailing list