weedeewee wrote on 2025-12-20, 20:01:

If you have the exact same card as the one for which this information was identified then, yes the information was correct, unl […]
Show full quote

bertrammatrix wrote on 2025-12-20, 19:18:

Could I get a confirmation of if r33 straps pin 100 or 97? I have a trio64 with a different pcb that I am now curious about

If you have the exact same card as the one for which this information was identified then,
yes the information was correct,
unless I was in error when counting/tracing,
and just requires you to check the datasheet and look at the traces on the pcb to confirm it.

If you want to identify which strap resistors are in use on the card you have,
then you'll have to follow the traces, either visually or by doing a continuity check using a multimeter.

That's just the thing- my card doesn't seem to have either pin 97 or 100 connected to any resistor that I can find, so if that's left floating possibly it is running 2 cycle edo (which would make sense, it looks cheap).

If I knew if it is pin 97 or 100 that gets pulled low (or high) or potentially connected together via the resistor on these other cards I could simply lift the appropriate pin and pull it high or low

bertrammatrix wrote on 2025-12-20, 20:38:

weedeewee wrote on 2025-12-20, 20:01:

If you have the exact same card as the one for which this information was identified then, yes the information was correct, unl […]
Show full quote

bertrammatrix wrote on 2025-12-20, 19:18:

Could I get a confirmation of if r33 straps pin 100 or 97? I have a trio64 with a different pcb that I am now curious about

If you have the exact same card as the one for which this information was identified then,
yes the information was correct,
unless I was in error when counting/tracing,
and just requires you to check the datasheet and look at the traces on the pcb to confirm it.

If you want to identify which strap resistors are in use on the card you have,
then you'll have to follow the traces, either visually or by doing a continuity check using a multimeter.

That's just the thing- my card doesn't seem to have either pin 97 or 100 connected to any resistor that I can find, so if that's left floating possibly it is running 2 cycle edo (which would make sense, it looks cheap).

If I knew if it is pin 97 or 100 that gets pulled low (or high) or potentially connected together via the resistor on these other cards I could simply lift the appropriate pin and pull it high or low

all I can tell you is what I already mentioned before in other comments.

seems PD3&2 pin 97 & 100 , R33 & ? , set the type of ram chips in use... not clear yet what means what . […]
Show full quote

seems PD3&2 pin 97 & 100 , R33 & ? , set the type of ram chips in use... not clear yet what means what .

edit:
ah page 47 of the S3 pdf.
If i read it correctly the 8 chip card is running in 1 cycle EDO mode, and the 4 chip card is running in 2-cycle mode.

The information about R33 is only valid for the STB powergraph 64 PCI.
if your card is not this one I can not see what resistor it could be since I'm not clairvoyant.
PD3 pin 97 & PD2 pin 100 determine the memory page mode on power-up of the card.
You could try to measure the resistance to ground or +5v? to see if these have been pulled up or down.

Just wanted to tie up some loose ends with one more post on this subject!

With the knowledge of 1 vs 2 cycle EDO and the 60Mhz and 50Mhz BIOS chips, I got to looking at my other Powergraph 64V cards. These are of a smaller, more compact type.

I have multiple different configurations of these same PCB designs. Ones with 4 soldered chips, ones with 2 soldered + 2 sockets, ones with only 2 soldered (1MB). Among those, most were clearly 2-cycle cards with sluggish performance. The others were single cycle with a 50Mhz BIOS and had middling but decent performance.

I noticed that on this particular design, the only difference between the 1-cycle and 2-cycle cards is a single resistor. R30, at the top of the row to the right of the Trio chip, just below the electrolytic cap C39. I'm guessing for simplicity of manufacturing they left the other one required for 1-cycle operation in place and just added or left this one out to accommodate different speed memory chips.

Simply adding a 10k resistor to R30 switches these cards to 1-cycle mode.

Also, I used powerstrip 2.78 to confirm that 1-cycle cards will run at 60Mhz or 50Mhz by default based on their BIOS chip (either can be overclocked with powerstrip with measurable performance increases). Also, as I suspected, when a card is configured for 2-cycle operation, it will always default to 60Mhz, even if the BIOS is set to run it at 50Mhz.

HOWEVER... When I added the resistor to a card with 60ns memory and a 60Mhz BIOS, it started to show artifacts in Windows. 60ns is apparently not fast enough to handle this configuration. Most likely, this is why the 50mhz BIOS chips exist. If the card is set to 1-cycle from the factory and they equipped it with 60ns chips, it will need the 50mhz BIOS chip to avoid memory artifacts.

One card with both 50ns and 60ns memory did not artifact at 60Mhz 1-cycle, but this could just be because the faster memory chips were used first. It's likely that artifacts would eventually show up as the the 60ns chips started to show their weaknesses... Or maybe these 60ns chips were just good enough to handle it.

The cards with all 50ns chips or faster did not artifact at 60Mhz 1-cycle, so I would say that 50ns would be the requirement for a Trio64V+ to run at "full speed" reliably... meaning, 1-cycle mode with a 60Mhz BIOS.

So, if a card had 60ns chips soldered I just left it alone. I could mod them all to "moderate performance" cards by adding R30 to switch them to 1-cycle mode and then flashing a 50Mhz BIOS, but I'm not feeling that ambitious right now. Of course swapping in all 50ns chips or better, adding R30 and using a 60Mhz BIOS would get them all running top notch... but spending hours modding my stockpile of S3 Trio cards for maximum performance is probably not the best use of time right now. 🤣

Finally... I didn't take a picture of this, but I also had a 2-chip 1MB card that was already set to 1-cycle mode and had a 50Mhz BIOS. I dropped a 60Mhz BIOS into it and it seems to be really fast though I can't compare the 32bit color speed because it lacks the VRAM. If I get real bored I might solder a couple of chips to it just to let it stretch it's legs.

It really is amazing how much more enjoyable it is to use these 1-cycle 60Mhz cards in Windows 98 vs the slower ones. It is a HUGE difference if you've got your mouse rate turned up and are running at a higher refresh rate, especially if you are showing window contents while dragging. Feels like a completely different generation of video card.

weedeewee wrote on 2025-12-20, 20:45:

all I can tell you is what I already mentioned before in other comments. […]
Show full quote

bertrammatrix wrote on 2025-12-20, 20:38:

weedeewee wrote on 2025-12-20, 20:01:

If you have the exact same card as the one for which this information was identified then, yes the information was correct, unl […]
Show full quote

If you have the exact same card as the one for which this information was identified then,
yes the information was correct,
unless I was in error when counting/tracing,
and just requires you to check the datasheet and look at the traces on the pcb to confirm it.

If you want to identify which strap resistors are in use on the card you have,
then you'll have to follow the traces, either visually or by doing a continuity check using a multimeter.

That's just the thing- my card doesn't seem to have either pin 97 or 100 connected to any resistor that I can find, so if that's left floating possibly it is running 2 cycle edo (which would make sense, it looks cheap).

If I knew if it is pin 97 or 100 that gets pulled low (or high) or potentially connected together via the resistor on these other cards I could simply lift the appropriate pin and pull it high or low

all I can tell you is what I already mentioned before in other comments.

seems PD3&2 pin 97 & 100 , R33 & ? , set the type of ram chips in use... not clear yet what means what . […]
Show full quote

seems PD3&2 pin 97 & 100 , R33 & ? , set the type of ram chips in use... not clear yet what means what .

edit:
ah page 47 of the S3 pdf.
If i read it correctly the 8 chip card is running in 1 cycle EDO mode, and the 4 chip card is running in 2-cycle mode.

The information about R33 is only valid for the STB powergraph 64 PCI.
if your card is not this one I can not see what resistor it could be since I'm not clairvoyant.
PD3 pin 97 & PD2 pin 100 determine the memory page mode on power-up of the card.
You could try to measure the resistance to ground or +5v? to see if these have been pulled up or down.

I am just trying to decipher - on YOUR card - does R33 connect one pad directly to pin97 yes/no or to pin 100 yes/no.

bertrammatrix wrote on 2025-12-23, 18:10:

I am just trying to decipher - on YOUR card - does R33 connect one pad directly to pin97 yes/no or to pin 100 yes/no.

Well,
one, It's not my card.
two. it's easily visible from the posted photos in the first comment of this thread for which the link is Some Trio64V+ Cards significantly faster in Windows 98 than others?. Just have to follow the tracks.
three, R33 pin97 yes, R33 pin100 no. pin100 R32.
also fyi, PD3-97-R33 & PD2-100-R32

edit: added link to first post of this thread which contains the relevant photos.

Just wanted to mention that this information all seems to apply for Virge and Virge DX cards as well. I just switched a Virge DX from 2-cycle to 1-cycle for a massive performance increase:
Re: What retro activity did you get up to today?

Additionally, I have noticed lots of memory timing related things in the datasheets that could possibly allow for making an "ultimate Virge" card if someone was so inclined.

Here is the Virge datasheet:
https://theretroweb.com/chip/documentation/db … e1625775493.pdf

I found these because the VGA Legacy site mentions "3.5ns" and "0ns" for various Virge/DX cards they have listed, so figured they had to be referring memory settings. Sure enough, PD 16 and PD 17 (pins 69 and 67 respectively) are used to control the CAS/OE Adjust - 0ns, 3.5ns, 5ns and 6.5ns. I don't know how much this impacts performance or what the required memory chip specs would be for the lowest latency... but it is interesting!

EDIT: Can someone who is more knowledgeable with this stuff explain what the "PD" bits and "CR" bits are referring to? A little bit of googling has me thinking that PD bits (pull down?) are what you'd change with a resistor, where the CR bits are maybe changed with BIOS or software controls? I don't understand the terminology behind what is being "pulled down" or how that equals a value of '00' or '10 ' . It probably isn't real complicated, I've just never dabbled much in this stuff.

Also, what is the significance of using a 10k ohm resistor? Will any resistor of a certain minimum resistance or higher work? Is it always the same or is it just 10k Ohm on this particular card?

The Wintune scores are very slow, for both versions. I don't think the S3 Trio drivers in Win9x or NT/2k will enable 'MTRR/fast writes/write-combining' on these old cards with a processor that supports it (P2 /K6-2 or better). People do this under DOS with fastvid but forget the drivers for these early cards don't enable it in Win9x or 2k/xp (or the built in windows drivers - if they are newer).
You can use powerstrip 2.x (3.x for 2k/xp) to enable it and performance should more than double and windows will be fast enough even with the 60mhz 2 cycle edo cards.
I know that on using a Diamond Steath 64 (S3 968), after enabling MTRR in Win2k, the speed of the card went from 19MB/s to 81MB/s.

Ozzuneoj wrote on 2026-01-31, 00:48:

EDIT: Can someone who is more knowledgeable with this stuff explain what the "PD" bits and "CR" bits are referring to? A little bit of googling has me thinking that PD bits (pull down?) are what you'd change with a resistor, where the CR bits are maybe changed with BIOS or software controls? I don't understand the terminology behind what is being "pulled down" or how that equals a value of '00' or '10 ' . It probably isn't real complicated, I've just never dabbled much in this stuff.

"PD" looks like it maps directly to physical pins on the chip. I'm not sure if it's specifically related to "pulldown", which is a term used for tying a digital pin to ground via a resistor (which causes it to be read as low, i.e. 0).
"CR" describes which internal config registers hold the settings. It's basically saying when the reset signal is triggered, PD pins 0-28 are read and copied into the internal config (or maybe control?) registers. Whether or not these registers can be changed later (or whether they are read-only) isn't clear just from the snippet you've posted.

Also, what is the significance of using a 10k ohm resistor? Will any resistor of a certain minimum resistance or higher work? Is it always the same or is it just 10k Ohm on this particular card?

10K is a very typical value used when input pins need to be constantly pulled up or down. It's a good "middle-of-the-road" value that limits current enough to avoid wasting power but still allows enough to flow for logic ICs to read the value correctly.

Ozzuneoj wrote on 2026-01-31, 00:48:

Just wanted to mention that this information all seems to apply for Virge and Virge DX cards as well. I just switched a Virge DX […]
Show full quote

Just wanted to mention that this information all seems to apply for Virge and Virge DX cards as well. I just switched a Virge DX from 2-cycle to 1-cycle for a massive performance increase:
Re: What retro activity did you get up to today?

Additionally, I have noticed lots of memory timing related things in the datasheets that could possibly allow for making an "ultimate Virge" card if someone was so inclined.

Here is the Virge datasheet:
https://theretroweb.com/chip/documentation/db … e1625775493.pdf

The attachment virge_Pinout1.png is no longer available

The attachment virge_PD_bits1.png is no longer available

The attachment virge_PD_bits2.png is no longer available

I found these because the VGA Legacy site mentions "3.5ns" and "0ns" for various Virge/DX cards they have listed, so figured they had to be referring memory settings. Sure enough, PD 16 and PD 17 (pins 69 and 67 respectively) are used to control the CAS/OE Adjust - 0ns, 3.5ns, 5ns and 6.5ns. I don't know how much this impacts performance or what the required memory chip specs would be for the lowest latency... but it is interesting!

EDIT: Can someone who is more knowledgeable with this stuff explain what the "PD" bits and "CR" bits are referring to? A little bit of googling has me thinking that PD bits (pull down?) are what you'd change with a resistor, where the CR bits are maybe changed with BIOS or software controls? I don't understand the terminology behind what is being "pulled down" or how that equals a value of '00' or '10 ' . It probably isn't real complicated, I've just never dabbled much in this stuff.

Also, what is the significance of using a 10k ohm resistor? Will any resistor of a certain minimum resistance or higher work? Is it always the same or is it just 10k Ohm on this particular card?

PD[63:0] page 21 in the datasheet.

Display Memory Pixel Data Bus Lines. Display Memory Pixel Data Bus Lines 31:0. PD[28:0] are also used as the system configuratio […]
Show full quote

Display Memory Pixel Data Bus Lines. Display
Memory Pixel Data Bus Lines 31:0. PD[28:0] are also
used as the system configuration strapping bits,
providing system configuration and setup information
upon power-on or reset. After reset, the General Data
Bus signals are multiplexed on 24 of these pins.

CR ... fairly certain this is just short for Configuration Register

one pin pulled up or down just equals one bit being set at powerup to one or zero.

In the situation you describe pin 67 & pin 69, PD17 & PD16, set bits 1 & 0 (that's bitlocation not value) of Configuration Register 68.