On Sun, Jun 21, 2009 at 9:26 PM, Charles Hedrick<[email protected]> wrote: > As far as I know the only build tool is Maven. I'm using the same versions. > It can't be ZFS, as I tried with everything on a ram disk. > > Yes, I see > > Jun 20 13:50:42 basel unix: [ID 954099 kern.info] NOTICE: IRQ18 is being > shared by drivers with different interrupt levels. > > No one that I know has been able to say what to do about that. Do you have > any suggestions? > > Also > > Jun 20 08:15:52 basel ata: [ID 776753 kern.info] ata: simplex controller. > DMA on channel 0 PIO on channel 1 > > > It's the same hardware for Linux. > > # isainfo -kv > 64-bit amd64 kernel modules > > # scanpci > > pci bus 0x0000 cardnum 0x00 function 0x00: vendor 0x8086 device 0x2e10 > Intel Corporation 4 Series Chipset DRAM Controller > > pci bus 0x0000 cardnum 0x01 function 0x00: vendor 0x8086 device 0x2e11 > Intel Corporation 4 Series Chipset PCI Express Root Port > > pci bus 0x0000 cardnum 0x03 function 0x00: vendor 0x8086 device 0x2e14 > Intel Corporation 4 Series Chipset HECI Controller > > pci bus 0x0000 cardnum 0x03 function 0x02: vendor 0x8086 device 0x2e16 > Intel Corporation 4 Series Chipset PT IDER Controller > > pci bus 0x0000 cardnum 0x03 function 0x03: vendor 0x8086 device 0x2e17 > Intel Corporation 4 Series Chipset Serial KT Controller > > pci bus 0x0000 cardnum 0x19 function 0x00: vendor 0x8086 device 0x10de > Intel Corporation 82567LM-3 Gigabit Network Connection > > pci bus 0x0000 cardnum 0x1a function 0x00: vendor 0x8086 device 0x3a67 > Intel Corporation 82801JD/DO (ICH10 Family) USB UHCI Controller #4 > > pci bus 0x0000 cardnum 0x1a function 0x01: vendor 0x8086 device 0x3a68 > Intel Corporation 82801JD/DO (ICH10 Family) USB UHCI Controller #5 > > pci bus 0x0000 cardnum 0x1a function 0x02: vendor 0x8086 device 0x3a69 > Intel Corporation 82801JD/DO (ICH10 Family) USB UHCI Controller #6 > > pci bus 0x0000 cardnum 0x1a function 0x07: vendor 0x8086 device 0x3a6c > Intel Corporation 82801JD/DO (ICH10 Family) USB2 EHCI Controller #2 > > pci bus 0x0000 cardnum 0x1b function 0x00: vendor 0x8086 device 0x3a6e > Intel Corporation 82801JD/DO (ICH10 Family) HD Audio Controller > > pci bus 0x0000 cardnum 0x1c function 0x00: vendor 0x8086 device 0x3a70 > Intel Corporation 82801JD/DO (ICH10 Family) PCI Express Port 1 > > pci bus 0x0000 cardnum 0x1c function 0x01: vendor 0x8086 device 0x3a72 > Intel Corporation 82801JD/DO (ICH10 Family) PCI Express Port 2 > > pci bus 0x0000 cardnum 0x1d function 0x00: vendor 0x8086 device 0x3a64 > Intel Corporation 82801JD/DO (ICH10 Family) USB UHCI Controller #1 > > pci bus 0x0000 cardnum 0x1d function 0x01: vendor 0x8086 device 0x3a65 > Intel Corporation 82801JD/DO (ICH10 Family) USB UHCI Controller #2 > > pci bus 0x0000 cardnum 0x1d function 0x02: vendor 0x8086 device 0x3a66 > Intel Corporation 82801JD/DO (ICH10 Family) USB UHCI Controller #3 > > pci bus 0x0000 cardnum 0x1d function 0x07: vendor 0x8086 device 0x3a6a > Intel Corporation 82801JD/DO (ICH10 Family) USB2 EHCI Controller #1 > > pci bus 0x0000 cardnum 0x1e function 0x00: vendor 0x8086 device 0x244e > Intel Corporation 82801 PCI Bridge > > pci bus 0x0000 cardnum 0x1f function 0x00: vendor 0x8086 device 0x3a14 > Intel Corporation 82801JDO (ICH10DO) LPC Interface Controller > > pci bus 0x0000 cardnum 0x1f function 0x02: vendor 0x8086 device 0x3a02 > Intel Corporation 82801JD/DO (ICH10 Family) SATA AHCI Controller > > pci bus 0x0000 cardnum 0x1f function 0x03: vendor 0x8086 device 0x3a60 > Intel Corporation 82801JD/DO (ICH10 Family) SMBus Controller > > pci bus 0x0001 cardnum 0x00 function 0x00: vendor 0x1002 device 0x9490 > ATI Technologies Inc RV730XT [Radeon HD 4670] > > pci bus 0x0001 cardnum 0x00 function 0x01: vendor 0x1002 device 0xaa38 > ATI Technologies Inc RV730XT Audio device [Radeon HD 4670] > > pci bus 0x0003 cardnum 0x00 function 0x00: vendor 0x14e4 device 0x1659 > Broadcom Corporation NetXtreme BCM5721 Gigabit Ethernet PCI Express
Out of curiousity, have you tried running plockstat against the jvm? I'd be curious to see if there's any significant locking activity within the native jre code. A pet peeve of mine is that the Solaris JRE's don't link against an alternate allocator, so any multithreaded code that ends up calling into native code that happens to do locking (this includes things like calling malloc()) (the networking code is a common cuplrit), your application performance nosedives unless you manually start the JRE with an interposed allocator (such as mtmalloc or libumem). I've seen it so much, that when I get performance issues with java, the first thing I do is check the LWP count, if > 1, run plockstat. 90% of the time it helps significantly (sometimes enough that no further analysis is needed), and it's such a simple and easy change. Perhaps it's just bias in the apps I deal with (lots of j2ee environments), but I really don't understand why it's not the default on *solaris. All supported versions of Solaris have at least mtmalloc. _______________________________________________ opensolaris-discuss mailing list [email protected]
