Kevin Brosius wrote: > [EMAIL PROTECTED] wrote: > > Kevin Brosius wrote: > > > [EMAIL PROTECTED] wrote: > > > > Having just upgraded my system from Mandrake 7.2 running the > > > > original XFree86 4.1.0 binaries to the XFree86 4.2.0 provided > > > > with Mandrake 8.2 (/ partition was reformatted for this upgrade) > > > > I now have a visual problem with the XFree86 display. The card > > > > in this machine (an IBM 300PL) is a S3 Trio 3D which identifies > > > > itself to XFree86 4.2.0 as: > > [snip] > > > > The left hand side and middle of the screen are fine, but the > > > > right hand side of the display is covered with multiple copies > > > > of a strip about 180 pixels wide taken from the middle of the > > > > screen and duplicated roughly three times (in 1280x1024 > > > > resolution) on the right hand side of the screen. Every > > > > application appears to be acting normally on this setup, the X > > > > server runs happily enough without errors in the > > > > /var/log/XFree86.0.log file. This effect is not observed in > > > > 8bit colour but is seen in 15, 16 and 24bit colour > > > > depths. Moving to higher resolutions like 1400x1050 results in a > > > > similar effect with roughly five copies of that strip on the > > > > right hand side of the screen. > > > > > > A full log would be helpful. I've seen a case where missing > > > monitor config and failing DDC caused something like this. > > [snip: config] > > If you place "1024x768" in the first position under Depth 16, does it > display cleanly at that resolution? Do only the 1152x864 and 1280x1024 > modes cause trouble with the right side banding? Do you have any > documentation for the Trio 3D card which specifies the maximum memory > clock rate for your model?
Even with 1024x768, there is a stripe and a half on the right hand side of the screen. At least in 15, 16 and 24 bit colour - 8 bit is fine. I've dug around my old documentation for this PC and drawn a blank on the specs for this card. However, I do note that the figures given in the XFree86 3.3.6 output differ from the 4.2.0 output. For example, from 3.3.6: (--) SVGA: PCI: S3 Trio3D rev 1, Memory @ 0xf4000000 (--) SVGA: S3V: Trio3D rev 1, Linear FB @ 0xf4000000 (--) SVGA: Detected S3 Trio3D (--) SVGA: using driver for chipset "s3_virge" (--) SVGA: videoram: 4096k (--) SVGA: Ramdac speed: 230 MHz (--) SVGA: Detected current MCLK value of 100.227 MHz (--) SVGA: chipset: s3_virge (--) SVGA: videoram: 4096k (**) SVGA: Option "sw_cursor" (**) SVGA: Option "power_saver" (**) SVGA: Using 16 bpp, Depth 16, Color weight: 565 >From 4.2.0: (--) S3VIRGE(0): Detected current MCLK value of 100.227 MHz (II) S3VIRGE(0): Monitor0: Using hsync range of 30.00-94.00 kHz (II) S3VIRGE(0): Monitor0: Using vrefresh range of 50.00-160.00 Hz (II) S3VIRGE(0): Clock range: 20.00 to 270.00 MHz Am I right in thinking that the Clock range in the 4.2.0 output should not exceed the RAMDAC speed? I have tried taking the modeline from the 3.3.6 configuration file and rewritting it into the Mode ... EndMode section for the monitor so that I'm running exactly the same monitor timings for both 3.3.6 and 4.2.0 but the striping remains. Thanks, Toby Haynes _______________________________________________ Xpert mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/xpert
