John Martin wrote: > Chris Ridd wrote: >> Alan Coopersmith wrote: >> >>> The root and /usr pages were merged into a single package for IPS >>> for most packages in Solaris, since IPS doesn't need to keep them >>> seperate for zones/diskless support as SVR4 packages did, so the >>> IPS NVDAgraphics includes the contents of both the NVDAgraphicsr >>> and NVDAgraphics SVR4 packages. >>> >> >> Is there anything straightforward (ie included in IPS somewhere :-) >> that I can run to give some performance figures before and after the >> NVDAgraphics downgrade? >> >> > Your original complaint was moving windows around showed the underneath > windows repainting very slowly. X11 performance tests may not show the > same problem (and are run for every driver release). The best test is > to see > if the desktop behavior changes.
Yes, that's true. I was just hoping there was some more objective way to measure this, instead of "jiggling things around a bit" :-) Anyway, after trying to create another BE with a downgraded driver, I rebooted back into "opensolaris-2", aka snv_91, and the graphics performance is good again. Looking at the Xorg.0.log file from the previous snv_91 boot, I do see some differences: Blame the hardware. < (==) Log file: "/var/log/Xorg.0.log", Time: Tue Jul 8 11:38:39 2008 --- > (==) Log file: "/var/log/Xorg.0.log", Time: Mon Jun 30 09:38:06 2008 36c36 < (II) PCI: stages = 0x03, oldVal1 = 0x8000fd04, mode1Res1 = 0x80000000 --- > (II) PCI: stages = 0x03, oldVal1 = 0x00000000, mode1Res1 = 0x80000000 104c104 < (--) PCI:*(5:0:0) nVidia Corporation NV43GL [Quadro FX 540] rev 162, Mem @ 0xd8000000/26, 0xd0000000/27, 0xde000000/24, BIOS @ 0xdfc00000/17 --- > (--) PCI:*(5:0:0) nVidia Corporation NV43GL [Quadro FX 540] rev 162, Mem @ 0xd8000000/27, 0xd0000000/27, 0xde000000/24, BIOS @ 0xdfc00000/17 258c258 < [8] -1 0 0xd8000000 - 0xdbffffff (0x4000000) MX[B](B) --- > [8] -1 0 0xd8000000 - 0xdfffffff (0x8000000) MX[B](B) 290c290 < [8] -1 0 0xd8000000 - 0xdbffffff (0x4000000) MX[B](B) --- > [8] -1 0 0xd8000000 - 0xdfffffff (0x8000000) MX[B](B) 329c329 < [12] -1 0 0xd8000000 - 0xdbffffff (0x4000000) MX[B](B) --- > [12] -1 0 0xd8000000 - 0xdfffffff (0x8000000) MX[B](B) 567c567 < [12] -1 0 0xd8000000 - 0xdbffffff (0x4000000) MX[B](B) --- > [12] -1 0 0xd8000000 - 0xdfffffff (0x8000000) MX[B](B) 593a594,635 > (WW) ****INVALID MEM ALLOCATION**** b: 0xd8000000 e: 0xdfffffff correcting [...] > (II) window fixed: > [0] -1 0 0xd8000000 - 0xdfcfffff (0x7d00000) MX[B] > Requesting insufficient memory window!: start: 0xd8000000 end: 0xdfcfffff size 0x8000000 > (EE) Cannot find a replacement memory range It looks quite like the card didn't get initialized cleanly before and ran in a downgraded mode? One difference from this morning's test and the 90->91 update was that I physically unplugged the power cable from the box this morning, whereas the 90->91 update only did an init 6 (or equivalent) and kept the machine powered all the way through. Cheers, Chris
