IRIX Network Forums
Octane SSE - max resolution - Printable Version

+- IRIX Network Forums (//forums.irixnet.org)
+-- Forum: SGI/MIPS (//forums.irixnet.org/forum-3.html)
+--- Forum: Hardware/Triage/Repair (//forums.irixnet.org/forum-11.html)
+--- Thread: Octane SSE - max resolution (/thread-1102.html)


Octane SSE - max resolution - Jacques - 04-03-2019

I plugged my SSE card into my Octane on the weekend and can only get 1600x1200 out of it, is this the max it can do apart from the flame installed vfo of 1920x1050 (1035?)

Can it do 1920x1200 at reduced Hz?

RE: Octane SSE - max resolution - callahan - 04-04-2019

The MGRAS boards should be able to push a modern LCD 1920x1200@60 Hz. You will have to roll your own vfo file though.

Lots of old neko posts on this topic hat irix.cc has archived, mostly from recondas.

I'm actually toying with the idea of making my own vfo file for the same purpose (have a EMXI and a Dell U2412).

For reference, the limiting factor for Octane MGRAS boards (i.e. everything non-Vpro) is the max pixel clock of 160MHz. 

The Dell U2412 likes a total screen size (including blanking that overruns the screen) of 2080x1235. At 60Hz that's only a 154MHz pixel clock. If you don't have this exact monitor, you should extract the EDID info using a relatively modern (last decade) computer.

Edit: This post has an example vfo file for a Dell U2410. Although intended for a Vpro system, I think the .vfs should work as-is for MGRAS cards, but you will need to use different vfc compiler options. http://archive.irix.cc/apocrypha/nekonomicon/forum/3/16729147/1.html

If you get this working, please do share the .vfo. I'll do the same, but not sure how soon I'll get around to it.

RE: Octane SSE - max resolution - Jacques - 04-04-2019

(04-04-2019, 05:01 AM)callahan Wrote:  The MGRAS boards should be able to push a modern LCD 1920x1200@60 Hz. You will have to roll your own vfo file though.

Lots of old neko posts on this topic hat irix.cc has archived, mostly from recondas.

I'm actually toying with the idea of making my own vfo file for the same purpose (have a EMXI and a Dell U2412).

For reference, the limiting factor for Octane MGRAS boards (i.e. everything non-Vpro) is the max pixel clock of 160MHz. 

The Dell U2412 likes a total screen size (including blanking that overruns the screen) of 2080x1235. At 60Hz that's only a 154MHz pixel clock. If you don't have this exact monitor, you should extract the EDID info using a relatively modern (last decade) computer.

Edit: This post has an example vfo file for a Dell U2410. Although intended for a Vpro system, I think the .vfs should work as-is for MGRAS cards, but you will need to use different vfc compiler options. http://archive.irix.cc/apocrypha/nekonomicon/forum/3/16729147/1.html

If you get this working, please do share the .vfo. I'll do the same, but not sure how soon I'll get around to it.

Thanks, I'll try on the weekend if I have time. Last time I tried to compile a vfc on MGRAS I failed miserably.

RE: Octane SSE - max resolution - callahan - 04-05-2019

So I decided to give it a try on my Indigo2 Solid Impact (SI equivalent) today. Didn't have the best of luck. Started with a .vfs for my Dell 1905FP. All of my work is pasted below. I was able to compile a .vfo, but when it loaded there were black artifacts throughout the screen (on both the 1905FP and a U2412). I think there is a compilation problem, but everything in the vfs looks a-okay to me.

My monitor's EDID modeline:

Code:
Modeline "Mode 0" 108.00 1280 1328 1440 1688 1024 1025 1028 1066 +hsync +vsync

(The fields are "String description" Dot-Clock HDisp HSyncStart HSyncEnd HTotal VDisp VSyncStart VSyncEnd VTotal [options] )

So I made the following .vfs file:

Code:
/*
1280x1024 Format for a Dell 1905FP monitor using EDID information obtained from the monitor's firmware

*/

General {
FormatName = "1280x1024-59.5_Dell1905FP";

FieldsPerFrame          = 1;
FramesPerSecond         = 59.5;
TotalLinesPerFrame      = 1066;
TotalPixelsPerLine      = 1688;
ActiveLinesPerFrame     = 1024;
ActivePixelsPerLine     = 1280;

}
Active Line {
HorizontalBackPorch     = 248 pixels;
HorizontalSync          = 112 pixels;
HorizontalFrontPorch    = 48 pixels;
}
Field
{
Vertical Sync =
{
{
Length = 1.0H;
Low = 0.0 usec;
}
repeat 2
{
Length = 1.0H;
}
}
Initial Low;
Vertical Back Porch =
{
{
Length = 1.0H;
High = HorizontalSync;
}
repeat 37
{
Length = 1.0H;
Low = 0.0 usec;
High = HorizontalSync;
}
}
Initial High;
Active =
{
repeat 1024
{
Length = 1.0H;
Low = 0.0 usec;
High = HorizontalSync;
}
}
Vertical Front Porch =
{
repeat 1
{
Length = 1.0H;
Low = 0.0 usec;
High = HorizontalSync;
}
}
}
#if 0
postprocess
{
dump edge;
}
#endif

And, after installing vfc from the 6.5 Fondations 2 CD, compiled it using:

Code:
vfc -a ascii=1905FP.info -c board=/usr/gfx/ucode/vfc/rules/mgras.def,chip=/usr/gfx/ucode/vfc/rules/mgras_vc3.def -p "-DRSSCOUNT=1" -p "-DDB32=1" -p "-DINTERLACED=0" -p "-DSTEREO_FMT=0" -p "-DOPTIONS=0" -p "-DZBUF=1" -p "-DVERT_SERRATION=0" -o Dell1905FP.vfo

(My Indigo2 is a SI, so there is 1 RE -- SSI/MXI would use DRSSCOUNT=2). I then saved the file a .sdb file in:


Code:
/usr/gfx/ucode/MGRAS/vof

xsetmon read the new file, and properly displayed the key parameters. But when I selected it and rebooted, the black block artifacts appeared. 

When I complied the vfo file the following errors appeared, but none were fatal. My guess is these are the root of my problem, but I really have no idea how to fix them. I messed around with larger porches but still got these errors. 

Code:
"/usr/gfx/ucode/vfc/rules/mgras.def", line 311: The time you specified crosses the frame boundary and is wrapped to the end of the frame.
"/usr/gfx/ucode/vfc/rules/mgras.def", line 510: The time you specified crosses the frame boundary and is wrapped to the end of the frame.
"/usr/gfx/ucode/vfc/rules/mgras.def", line 512: The time you specified crosses the frame boundary and is wrapped to the end of the frame.
"/usr/gfx/ucode/vfc/rules/mgras.def", line 267: You specified a transition at time (line 1) + (1.11043e-07 seconds (line 0 + 111.043 nsec)) to a polarity to which the DAC_CSYNC signal was already set.  This transition was ignored.
"/usr/gfx/ucode/vfc/rules/mgras.def", line 267: You specified a transition at time (line 2) + (1.11043e-07 seconds (line 0 + 111.043 nsec)) to a polarity to which the DAC_CSYNC signal was already set.  This transition was ignored.

I'm basically giving up with the Indigo2. I may try again someday with the Octane (with the MXI and maybe see if they would compile for my V8 ... although I couldn't benefit from a custom vfo with the V8 because the U2412's preferred pixel clock would fall squarely in the V6/V8 dead zone).

Edit: I also tried to compile the example .vfs files that come with vfc, each gave the same errors as above. Looking at the mgras.def file, they appear to be related to timing of the porches (each of the error lines defines some change in status as some sync time +/- some number of pixels).

RE: Octane SSE - max resolution - Jacques - 04-05-2019

That's the same error I always came up against. Compiling for a V8 is simple, as long as you stay out of the pixel clock bug range.

RE: Octane SSE - max resolution - Trippynet - 04-05-2019

To be honest, I think 1920x1200 is going to be a major stretch for SolidIMPACT - especially in Indigo2. If I recall, even enabling the built-in 1600x1200 resolution causes problems with a lot of 3D utilities failing to render (it's been a while since I tried it mind you). MaxIMPACT may have better success as it does support more resolutions than Solid/HighIMPACT.

Of course, this relates to the Indigo2 - I don't have an Octane so haven't tried higher resolutions with the MGRAS graphics options for that.

RE: Octane SSE - max resolution - jan-jaap - 04-05-2019

(04-05-2019, 03:40 PM)Trippynet Wrote:  To be honest, I think 1920x1200 is going to be a major stretch for SolidIMPACT - especially in Indigo2. If I recall, even enabling the built-in 1600x1200 resolution causes problems with a lot of 3D utilities failing to render (it's been a while since I tried it mind you). MaxIMPACT may have better success as it does support more resolutions than Solid/HighIMPACT.

Of course, this relates to the Indigo2 - I don't have an Octane so haven't tried higher resolutions with the MGRAS graphics options for that.

Correct. Originally only a MaxIMPACT Indigo2 could do 1600x1200 I believe. It has twice the frame buffer size compared to Solid and High IMPACT and because of that it has extra features. I'm pretty sure the same is true for IMPACT on Octane, i.e. only MXI and MXE can do that.

NB: MaxIMPACT also has twice the texture RAM compared to HighIMPACT, but this is used for extra performance, not extra features.

The 1920x1030 or 1920x1035 that some SGI systems can do is an early HDTV standard. Even a PowerSeries VGX from 1990 can do that!

RE: Octane SSE - max resolution - callahan - 04-06-2019

Maybe, but I can't even get the provided-with-vfc 1280x1024x60FPS vfs template to compile without errors.