NeoMagic video -- request for feedback

[BarryK]

Anyone reading this who has a PC with NeoMagic video, could you kindly run a little test for me.

Boot Puppy, open a rxvt terminal, or, if X won't start, no problem.
Run this:

Code: Select all

# cat /proc/pci | grep "VGA"

Then report back to this forum what you get back from that.

I'm working on framebuffer support for 1.0.4, need this info.

Umm, it would be good to have the same info for i810 video also.

[Flash]

My FIC Sabre 815t is not a laptop but it uses the chipset in question:

# cat /proc/pci | grep "VGA"
VGA compatible controller: Intel Corp. 82815 CGC [Chipset Graphics Controller] (rev 4).
#

(Is there any way to copy from an rxvt window so I can paste into this one instead of transcribing?)

[Guest]

to copy, select the text with the mouse
to paste, click the middle mouse button

(select text by holding down the left mouse button and moving the mouse cursor over the text)

[BarryK]

Yes, rxvt is a special case -- it's on the FAQ page.
If you just drag the pointer over any text in the rxvt window, it automatically goes into the clipboard, you don't do any ctrl-c.

Thanks for the info Flash.

One other thing I need to know, is do the NeoMagic chipsets identify themselves with the text "NeoMagic" when the command "cat /proc/pci | grep "VGA"" is run?

Flash, your i815 video chip, how does that work with Puppy's Xvesa?
I know there were discussions about the i810 having limited modes.
I've got a framebuffer kernel driver module for the Intel video, but it is designed for these chips:
Intel(R) 830M/845G/852GM/855GM/865G chipset
...so, seems to be leaving the older ones out in the cold, so to speak.

[Flash]

<snip>
Flash, your i815 video chip, how does that work with Puppy's Xvesa?
<snip>

It works with any version of Puppy except bladehunter's Puppy 1.0.3 with Xfree86-4.4.0. That didn't work for me. I didn't try anything besides booting it. When it didn't work for me I didn't try to make work it because the regular Puppy works for me.

However, with regular Puppy 1.0.3, the maximum resolution the video wizard offers is 800x600x16; considerably less than the maximum available when I'm running Windows 2000 on the same machine. That doesn't bother me. I prefer a clean and simple look with big text. Still, I suppose it does indicate that there is a problem somewhere.

[Guest]

Yep it does, identify itself (should EvilMagic), but I tried of a couple NeoMagic drivers and none except XFree's and Xorg's seem to work on my Dell, could be a buggy BIOS or something or maybe NeoMagic chips are Evil....

[BarryK]

Intel(R) 830M/845G/852GM/855GM/865G chipset
...so, seems to be leaving the older ones out in the cold, so to speak.

The driver I'm referring to is the "intelfb.o" kernel module.
Although typing "modinfo intelfb" (Puppy 1.0.3 doesn't have this module, 1.0.4 does) gives the above info, I found some other documentation on the Internet that indicates it will work with i810 and i815
...so, we shall see.

The good thing about the framebuffer module is you can load it like this:

Code: Select all

# modprobe intelfb mode=1024x768-16@75

where the 75 is the refresh frequency.

...which leads to my next question... those refresh frequencies, what are the limitations? I mean, for the above example of 1024x768x16, can I only choose certain frequencies, like 60, 75 ...? why not 72 for example? ...or 95 if my monitor isn't going to explode.
...I'm a bit ignorant in this area, well a lot ignorant.

[Guest]

Yeah, I wonder about the warnings. What engineer would design a monitor, intended to be sold for use with unknown computers, that could execute a command which could damage it? It would be a simple enough matter to design in limits for protection.

When you select a refresh rate, in effect you are specifying the pixel clock frequency. The pixel clock frequency is implied by the combination of refresh rate and horizontal and vertical resolutions. I suppose the pixel clock frequency generator(s) and scan counters are on the video driver board rather than in the monitor itself. In principle at least, the pixel clock could be set at whatever frequency is necessary to give the desired refresh rate, but other considerations could make it impractical for the pixel clock to be set at just any arbitrary value within the physical limits of the D-A converters and video amplifiers. I can't imagine that it would instantly damage the D-A converters on the board, or the video amplifiers inside the monitor, to clock the system outside its design limits, but if you kept it up long enough they might fail because of prolonged overheating. If the horizontal scan circuit, which usually generates the accelerating voltage in the flyback transformer, is driven at a high enough frequency it might generate enough power to burn something up fairly quickly.

Having said that, I gather that it is possible to instantly destroy a floppy disk drive by telling it to go to an address outside its range.

[tempestuous]

Barry,
My Pentium300 Sony laptop returns -
VGA compatible controller: Neomagic Corporation NM2160 [MagicGraph 128XD] (rev 1).

With full Linux distributions (certainly Fedora and Debian, at least) the full XFree86 X server installs and works without any tweaking.
The XF86Config (or xorg.conf in Fedora) has -
Section "Device"
Driver "neomagic"

Xvesa gives weird colours and can barely be viewed. So I was forced to make a Puppy-Unleashed with a different X server. I decided to skip xfbdev since it gives me painfully slow page scrolling in Abiword (I know this from DamnSmallLinux) and go for SVGA. This works great.

Even if you get framebuffer happening in 1.0.4, I will certainly stay with SVGA - even with its 2MB size penalty ... though I am a little intrigued by bladehunter's cutdown Xfree86-4.4.0 version.

[tempestuous]

I assumed by framebuffer support, you meant xfbdev?

[Guest]

Sound s like your sony is the same spec as my dell.......XFree rocks on it for graphics....well as well as can be expected from 2 meg hahahahaha


That's with the xvesa module\driver gunna experiment more with the neomagic one

[Guest]

I decided to skip xfbdev since it gives me painfully slow page scrolling in Abiword (I know this from DamnSmallLinux) and go for SVGA.

I wonder if Damn Small is only using the generic "vesafb" kernel framebuffer driver rather than the "neofb" kernel driver?

[pakt]

Sorry about the delay - been away from the forum a few days. Hope this info can still be useful.

I have feedback for you from two laptops. The first is from an IBM ThinkPad 240, the second from a Dell Inspiron 510m with i810 video.

# cat /proc/pci | grep "VGA"
VGA compatible controller: Neomagic Corporation NM2160 [MagicGraph 128XD] (rev 1).


# cat /proc/pci | grep "VGA"
VGA compatible controller: Intel Corp. 82852/855GM Integrated Graphics Device (rev 2).


Oh, yes. I nearly forgot - here's feedback from a desktop with an Intel D815EFV motherboard (i810 graphics):

# cat /proc/pci | grep "VGA"
VGA compatible controller: Intel Corp. 82815 CGC [Chipset Graphics Controller] (rev 2).

[Guest]

I have feedback for you from an IBM ThinkPad 600E.

# cat /proc/pci | grep "VGA"

VGA compatible controller: Neomagic Corporation NM2200 [MagicGraph 256AV] (rev 18).

[TSloth]

# cat /proc/pci | grep VGA
VGA compatible controller: Trident Microsystems TGUI 9660/938x/968x (rev 211).


- TSloth

[Lobster]

...which leads to my next question... those refresh frequencies, what are the limitations? I mean, for the above example of 1024x768x16, can I only choose certain frequencies, like 60, 75 ...? why not 72 for example? ...or 95 if my monitor isn't going to explode.
...I'm a bit ignorant in this area, well a lot ignorant.

I thought someone was going to try and deliberaetly blow up their monitor (to see if it is possible) I suspect it is not - with very early monitors it might have been . . .

the max res on my monitor is 75mz - I am running with Xorg (as it suggested) at 79mz - seems OK

There must be some tests or test labs who have done this?

[keenerd-not-logged-in]

I've never heard (first or second hand) of a monitor being damaged by synv rates.

On the other hand, I've had to run an old IBM vga monitor (640x480) at much higher resolutions. It was just usable enough to find and lower the display properties.

I believe everything was ghosted about eight times, and shaded red.

[Flash]

I assume the warning is based on the idea that CRT monitors develop their accelerating voltage the same way as TVs do, from the "flyback" of the horizontal scan coils (or is it the vertical scan?) If that's the way they work, then I suppose running them at too high a scan frequency could dump too much power into the high voltage section and cause something to break down very quickly.

However, even if that's correct it only applies to CRTs. LCDs don't work the same way. They are backlit by a flourescent lamp assembly the power for which is independent of the pixel driving circuitry. I don't think an LCD display could be damaged by briefly driving it at a higher resolution/refresh rate combination than it was designed for, just don't keep doing it for a long time after it is plainly pointless.

If I ever get Xorg to work on my computer I'll be happy to try to blow up my cheapo LCD monitor.

[Ted Dog]

On a quiet Tuesday morning at a Dilbert styled mega insurance company, a loud gunshot sound broke the quiet hum of a thousand programmers' keyboard clicks. Those with miliary training hit the deck. It took me awhile to break the deep programmer haze that only C++ debugging can bring. Someone shouted out 'Its just a monitor that blew' I peared over the blue half hight 5x4 box I called home M-Thursday, to see a column of the blue/gray haze rise to the false drop ceiling.
It took me about 15 minutes to get back into the zone, so yes it can happen.