Re: slow compiling on amd64
No - I haven't tried an older version. The oldest I would go on a production machine would be 3.9. I could try 3.9, but to be honest I don't have time to test things out. I need these servers up, yesterday. I really don't want to use another OS, but might have to if I don't solve this problem quickly. Regards, Stephen On 15-Nov-06, at 10:19 PM, Brian Keefer wrote: On Nov 15, 2006, at 8:17 PM, Stephen Schaff wrote: this is my first post to the list - so please bear with me... I have 2 amd64 machines that I plan on using in production, and 1 amd64 machine at home for testing. I tried installing the amd64 openbsd on both machines and discovered that doing a make on anything goes really, really slowly. I have the i386 openbsd installed on my test system and it does everything very quickly. So, I tried installing i386 on my 2 production machines. It's still slow on both of them! When I say slow, here's what I mean. I'm compiling a new kernel with raid support. Just doing a make depend take roughly 30 seconds on my test machine and 30 minutes on the production machines. # time make depend TEST MACHINE: 0m31.36s real 0m20.64s user 0m6.32s system PRODUCTION MACHINE: 36m8.08s real 5m32.17s user 1m37.57s system Another poster and myself have been puzzling over amd64 performance problems as well. It seems that the OpenBSD/amd64 OS was fast back in 3.5, but somewhere between then and now it has slowed down dramatically. Have you tried installing older versions of OpenBSD to see if the performance is better? Brian Keefer www.Tumbleweed.com The Experts in Secure Internet Communication
Re: slow compiling on amd64
What strikes me as very bizarre is that my slower amd64 machine at home is just fine and runs really well. That one has an nvidia chipset on the A8N-SLI motherboard. The machines that aren't working properly have the A8N-VM CMS board which also uses the nvidia chipset. I just don't understand how there can be a difference factor of 10. 30 seconds for make depend on the A8N-SLI and 30 mins on the A8N-VM CMS (???) I MUST be missing something simple - has nobody else seen this? Regards, Stephen On 15-Nov-06, at 9:36 PM, Chris Kuethe wrote: Dmesg? Nvidia chipsets are dog-slow. On 11/15/06, Stephen Schaff [EMAIL PROTECTED] wrote: this is my first post to the list - so please bear with me... I have 2 amd64 machines that I plan on using in production, and 1 amd64 machine at home for testing. I tried installing the amd64 openbsd on both machines and discovered that doing a make on anything goes really, really slowly. I have the i386 openbsd installed on my test system and it does everything very quickly. So, I tried installing i386 on my 2 production machines. It's still slow on both of them! When I say slow, here's what I mean. I'm compiling a new kernel with raid support. Just doing a make depend take roughly 30 seconds on my test machine and 30 minutes on the production machines. # time make depend TEST MACHINE: 0m31.36s real 0m20.64s user 0m6.32s system PRODUCTION MACHINE: 36m8.08s real 5m32.17s user 1m37.57s system Here's the hardware: # sysctl hw TEST MACHINE: hw.machine=i386 hw.model=AMD Athlon(tm) 64 Processor 3000+ (AuthenticAMD 686-class, 512KB L2 cache) hw.ncpu=1 hw.byteorder=1234 hw.physmem=1073246208 hw.usermem=1072939008 hw.pagesize=4096 hw.disknames=wd0,cd0 hw.diskcount=2 hw.sensors.0=it0, Fan1, 5443 RPM hw.sensors.3=it0, VCORE_A, 1.41 V DC hw.sensors.4=it0, VCORE_B, 0.00 V DC hw.sensors.5=it0, +3.3V, 3.28 V DC hw.sensors.6=it0, +5V, 5.03 V DC hw.sensors.7=it0, +12V, 11.78 V DC hw.sensors.8=it0, Unused, 0.82 V DC hw.sensors.9=it0, -12V, -17.00 V DC hw.sensors.10=it0, +5VSB, 4.78 V DC hw.sensors.11=it0, VBAT, 3.06 V DC hw.sensors.12=it0, Temp 1, 35.00 degC hw.sensors.13=it0, Temp 2, 37.00 degC hw.sensors.14=it0, Temp 3, 25.00 degC hw.cpuspeed=1810 hw.setperf=100 hw.vendor=ASUSTeK Computer INC. hw.product=A8N-SLI DELUXE hw.version=1.XX hw.serialno=123456789000 hw.uuid=000fa389-5f1d-d711-9ec4-0011d84a06a8 PRODUCTION MACHINE: hw.machine=i386 hw.model=AMD Athlon(tm) 64 Processor 3500+ (AuthenticAMD 686-class, 512KB L2 cache) hw.ncpu=1 hw.byteorder=1234 hw.physmem=1005940736 hw.usermem=1005699072 hw.pagesize=4096 hw.disknames=cd0,wd0,wd1,wd2,wd3 hw.diskcount=5 hw.sensors.0=lm0, VCore A, 2.96 V DC hw.sensors.1=lm0, VCore B, 3.63 V DC hw.sensors.2=lm0, +3.3V, 3.38 V DC hw.sensors.3=lm0, +5V, 5.67 V DC hw.sensors.4=lm0, +12V, 16.32 V DC hw.sensors.5=lm0, -12V, -12.86 V DC hw.sensors.6=lm0, -5V, -5.36 V DC hw.sensors.7=lm0, Temp1, 33.00 degC hw.sensors.10=lm0, Fan3, 4017 RPM hw.cpuspeed=2211 hw.setperf=100 hw.vendor=ASUSTeK Computer INC. hw.product=A8N-VM CSM hw.uuid=c478ed80-74fe-d511-b068-749cdaa7f59a ANY ideas? This one is stumping me completely and I've wasted a week trying to sort it out. TIA! Stephen -- GDB has a 'break' feature; why doesn't it have 'fix' too?
Re: slow compiling on amd64
On 2006/11/16 01:02, Stephen Schaff wrote: I just don't understand how there can be a difference factor of 10. factor of 100. 30 seconds for make depend on the A8N-SLI and 30 mins on the A8N-VM CMS (???) I MUST be missing something simple - has nobody else seen this? softdep mount option? this will slow down creation/removal of large numbers of files. hard-drive write caching? (check under the 'Device has enabled...' section of 'sudo atactl wd0') drive write speeds will be low if it's not enabled (some drives normally enable it, some don't and you can do so in rc.local).
Re: slow compiling on amd64
On Thu, Nov 16, 2006 at 10:53:10AM +, Stuart Henderson wrote: On 2006/11/16 01:02, Stephen Schaff wrote: I just don't understand how there can be a difference factor of 10. factor of 100. 30 seconds for make depend on the A8N-SLI and 30 mins on the A8N-VM CMS (???) I MUST be missing something simple - has nobody else seen this? softdep mount option? this will slow down creation/removal of large numbers of files. hard-drive write caching? (check under the 'Device has enabled...' section of 'sudo atactl wd0') drive write speeds will be low if it's not enabled (some drives normally enable it, some don't and you can do so in rc.local). To test this, how about finding a somewhat smaller program - like, some port - and compiling it on a mfs? Provided you have enough memory to not need swap, this should mean that the hard disk will not affect the results. If the results are comparable on an mfs, tune the hard disk. If not, I have no idea... Joachim
Re: slow compiling on amd64
Thank you for your suggestions. It looks like write caching is enabled. I've pasted the results below. Stephen On 16-Nov-06, at 3:53 AM, Stuart Henderson wrote: On 2006/11/16 01:02, Stephen Schaff wrote: I just don't understand how there can be a difference factor of 10. factor of 100. yes - guess I was tired when calculating that! 30 seconds for make depend on the A8N-SLI and 30 mins on the A8N-VM CMS (???) I MUST be missing something simple - has nobody else seen this? softdep mount option? this will slow down creation/removal of large numbers of files. hard-drive write caching? (check under the 'Device has enabled...' section of 'sudo atactl wd0') drive write speeds will be low if it's not enabled (some drives normally enable it, some don't and you can do so in rc.local). sudo atactl wd0: Model: ST3250823AS, Rev: 3.03, Serial #: 5ND2CD2Q Device type: ATA, fixed Cylinders: 16383, heads: 16, sec/track: 63, total sectors: 488397168 Device capabilities: ATA standby timer values IORDY operation IORDY disabling Device supports the following standards: ATA-1 ATA-2 ATA-3 ATA-4 ATA-5 ATA-6 ATA-7 Master password revision code 0xfffe Device supports the following command sets: READ BUFFER command WRITE BUFFER command Host Protected Area feature set Read look-ahead Write cache Power Management feature set Security Mode feature set SMART feature set Flush Cache Ext command Flush Cache command Device Configuration Overlay feature set 48bit address feature set Set Max security extension commands DOWNLOAD MICROCODE command SMART self-test SMART error logging Device has enabled the following command sets/features: READ BUFFER command WRITE BUFFER command Host Protected Area feature set Read look-ahead Write cache Power Management feature set SMART feature set Flush Cache Ext command Flush Cache command Device Configuration Overlay feature set 48bit address feature set DOWNLOAD MICROCODE command
Re: slow compiling on amd64
Stuart Henderson [EMAIL PROTECTED] wrote: I just don't understand how there can be a difference factor of 10. factor of 100. (Are you really sure a minute has 100 seconds?) softdep mount option? this will slow down creation/removal of large numbers of files. Please, you are not a dog, that Pavlovian response is nonsense. hard-drive write caching? (check under the 'Device has enabled...' section of 'sudo atactl wd0') drive write speeds will be low if it's not enabled (some drives normally enable it, some don't and you can do so in rc.local). Strictly speaking, you should DISABLE write caching when using softupdates. Anyway, none of this comes remotely close to explaining the problem. -- Christian naddy Weisgerber [EMAIL PROTECTED]
Re: slow compiling on amd64
Stephen Schaff [EMAIL PROTECTED] wrote: I just don't understand how there can be a difference factor of 10. 30 seconds for make depend on the A8N-SLI and 30 mins on the A8N-VM CMS (???) What do top and systat vmstat report where the CPU is going? I MUST be missing something simple - has nobody else seen this? Ian Darwin has a laptop that is mostly busy handling an interrupt storm. -- Christian naddy Weisgerber [EMAIL PROTECTED]
Re: slow compiling on amd64
On 2006/11/16 16:25, Christian Weisgerber wrote: I just don't understand how there can be a difference factor of 10. factor of 100. (Are you really sure a minute has 100 seconds?) No I'm not, come to think of it... softdep mount option? this will slow down creation/removal of large numbers of files. Please, you are not a dog, that Pavlovian response is nonsense. Ah, I didn't realise how little was written by 'make depend'. For some operations, having WCE and softdep off does make a difference approaching this of order of magnitude. Knowing that some drives ship with WCE turned off (e.g. at least those supplied with some HP DL385 and Fujitsu-Siemens servers), it's a quick and simple thing to check which accounts for performance differences on some operations of this sort of order of magnitude. I just tried untarring ports.tar.gz on some F-S box: WCE, softdep no WCE, softdep WCE, no softdep no WCE, no softdep 0m49.29s 13m8.72s 1m51.58s 30m9.95s Strictly speaking, you should DISABLE write caching when using softupdates. Surely this applies without softupdates too, though? It also relies on the drive doing what you tell it, which isn't guaranteed, especially with consumer drives optimised for performance in benchmarks (aiui some drives always enable write-cache no matter what you tell them; with these, providing they're ATA-6 compliant, you may force it with READ VERIFY SECTOR/S).
Re: slow compiling on amd64
Sorry - of course - here's my dmesg: OpenBSD 4.0 (RAMDISK_CD) #39: Sat Sep 16 19:34:26 MDT 2006 [EMAIL PROTECTED]:/usr/src/sys/arch/i386/compile/RAMDISK_CD cpu0: AMD Athlon(tm) 64 Processor 3500+ (AuthenticAMD 686-class, 512KB L2 cache) 2.22 GHz cpu0: FPU,V86,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36, CFLUSH,MMX,FXSR,SSE,SSE2,SSE3 real mem = 1005940736 (982364K) avail mem = 910962688 (889612K) using 4256 buffers containing 50401280 bytes (49220K) of memory mainbus0 (root) bios0 at mainbus0: AT/286+(00) BIOS, date 10/05/05, BIOS32 rev. 0 @ 0xf0010, SMBIOS rev. 2.3 @ 0xf06d0 (66 entries) bios0: ASUSTeK Computer INC. A8N-VM CSM apm0 at bios0: Power Management spec V1.2 apm0: flags 30102 dobusy 0 doidle 1 pcibios0 at bios0: rev 3.0 @ 0xf/0x1 pcibios0: PCI IRQ Routing Table rev 1.0 @ 0xf8b80/288 (16 entries) pcibios0: no compatible PCI ICU found: ICU vendor 0x10de product 0x0260 pcibios0: Warning, unable to fix up PCI interrupt routing pcibios0: PCI bus #4 is the last bus bios0: ROM list: 0xc/0xec00 0xcf000/0x1000 cpu0 at mainbus0 pci0 at mainbus0 bus 0: configuration mode 1 (no bios) NVIDIA C51 Host rev 0xa2 at pci0 dev 0 function 0 not configured NVIDIA C51 Memory rev 0xa2 at pci0 dev 0 function 1 not configured NVIDIA C51 Memory rev 0xa2 at pci0 dev 0 function 2 not configured NVIDIA C51 Memory rev 0xa2 at pci0 dev 0 function 3 not configured NVIDIA C51 Memory rev 0xa2 at pci0 dev 0 function 4 not configured NVIDIA C51 Memory rev 0xa2 at pci0 dev 0 function 5 not configured NVIDIA C51 Memory rev 0xa2 at pci0 dev 0 function 6 not configured NVIDIA C51 Memory rev 0xa2 at pci0 dev 0 function 7 not configured ppb0 at pci0 dev 2 function 0 NVIDIA C51 PCIE rev 0xa1 pci1 at ppb0 bus 1 ppb1 at pci0 dev 3 function 0 NVIDIA C51 PCIE rev 0xa1 pci2 at ppb1 bus 2 ppb2 at pci0 dev 4 function 0 NVIDIA C51 PCIE rev 0xa1 pci3 at ppb2 bus 3 vga1 at pci0 dev 5 function 0 NVIDIA GeForce 6150 rev 0xa2 wsdisplay0 at vga1 mux 1: console (80x25, vt100 emulation) NVIDIA MCP51 Host rev 0xa2 at pci0 dev 9 function 0 not configured pcib0 at pci0 dev 10 function 0 NVIDIA MCP51 ISA rev 0xa2 NVIDIA MCP51 SMBus rev 0xa2 at pci0 dev 10 function 1 not configured ohci0 at pci0 dev 11 function 0 NVIDIA MCP51 USB rev 0xa2: irq 5, version 1.0, legacy support usb0 at ohci0: USB revision 1.0 uhub0 at usb0 uhub0: NVIDIA OHCI root hub, rev 1.00/1.00, addr 1 uhub0: 8 ports with 8 removable, self powered ehci0 at pci0 dev 11 function 1 NVIDIA MCP51 USB rev 0xa2: irq 3 usb1 at ehci0: USB revision 2.0 uhub1 at usb1 uhub1: NVIDIA EHCI root hub, rev 2.00/1.00, addr 1 uhub1: 8 ports with 8 removable, self powered pciide0 at pci0 dev 13 function 0 NVIDIA MCP51 IDE rev 0xa1: DMA, channel 0 configured to compatibility, channel 1 configured to compatibility atapiscsi0 at pciide0 channel 0 drive 0 scsibus0 at atapiscsi0: 2 targets cd0 at scsibus0 targ 0 lun 0: HL-DT-ST, CD-ROM GCR-8525B, 1.02 SCSI0 5/cdrom removable cd0(pciide0:0:0): using PIO mode 4, DMA mode 2 pciide0: channel 1 disabled (no drives) pciide1 at pci0 dev 14 function 0 NVIDIA MCP51 SATA rev 0xa1: DMA pciide1: using irq 5 for native-PCI interrupt wd0 at pciide1 channel 0 drive 0: ST3250823AS wd0: 16-sector PIO, LBA48, 238475MB, 488397168 sectors wd0(pciide1:0:0): using PIO mode 4, Ultra-DMA mode 5 wd1 at pciide1 channel 1 drive 0: ST3250823AS wd1: 16-sector PIO, LBA48, 238475MB, 488397168 sectors wd1(pciide1:1:0): using PIO mode 4, Ultra-DMA mode 5 pciide2 at pci0 dev 15 function 0 NVIDIA MCP51 SATA rev 0xa1: DMA pciide2: using irq 5 for native-PCI interrupt wd2 at pciide2 channel 0 drive 0: ST3250823AS wd2: 16-sector PIO, LBA48, 238475MB, 488397168 sectors wd2(pciide2:0:0): using PIO mode 4, Ultra-DMA mode 5 wd3 at pciide2 channel 1 drive 0: ST3250824AS wd3: 16-sector PIO, LBA48, 238475MB, 488397168 sectors wd3(pciide2:1:0): using PIO mode 4, Ultra-DMA mode 5 ppb3 at pci0 dev 16 function 0 NVIDIA MCP51 PCI-PCI rev 0xa2 pci4 at ppb3 bus 4 VIA VT6306 FireWire rev 0x80 at pci4 dev 5 function 0 not configured em0 at pci4 dev 9 function 0 Intel PRO/1000GT (82541GI) rev 0x05: irq 5, address 00:0e:0c:a2:de:f6 NVIDIA MCP51 HD Audio rev 0xa2 at pci0 dev 16 function 1 not configured nfe0 at pci0 dev 20 function 0 NVIDIA MCP51 LAN rev 0xa1: irq 5, address 00:13:d4:ff:0e:75 eephy0 at nfe0 phy 1: Marvell 88E Gigabit PHY, rev. 2 pchb0 at pci0 dev 24 function 0 AMD AMD64 HyperTransport rev 0x00 pchb1 at pci0 dev 24 function 1 AMD AMD64 Address Map rev 0x00 pchb2 at pci0 dev 24 function 2 AMD AMD64 DRAM Cfg rev 0x00 pchb3 at pci0 dev 24 function 3 AMD AMD64 Misc Cfg rev 0x00 isa0 at pcib0 isadma0 at isa0 pckbc0 at isa0 port 0x60/5 pckbd0 at pckbc0 (kbd slot) pckbc0: using irq 1 for kbd slot wskbd0 at pckbd0: console keyboard, using wsdisplay0 npx0 at isa0 port 0xf0/16: using exception 16 pccom0 at isa0 port 0x3f8/8 irq 4: ns16550a, 16 byte fifo fdc0 at isa0 port 0x3f0/6 irq 6 drq 2 biomask ffed netmask ffed ttymask ffef rd0: fixed, 3800