So I have this card, and it's pretty fast in DOS, but can't get it stable with 0 W/S. It makes all kinds of screen artifacts and freezes right away during boot. Do you think replacing the video RAM with 60ns chips would make any difference?
I don't think faster video RAM will solve your issue. Faster video RAM can remove flicker artifacts at high resolution / refresh rate, and faster video RAM can avoid corrupted images. But faster video RAM will not prevent system crashes / freezes. The freezes indicate an issue between the ARK chip and the 486 processor, which is decoupled from the interface between the ARK chip and the video RAM quite well.
The card should be very fast already at 1WS (if your mainboard doesn't add further wait states, up to 44MB/s at 33MHz FSB and up to 53MB/s at 40MHz FSB), so the bus is unlikely to be a severe bottleneck. Whether you get 0WS working (up to 66MB/s at 33MHz FSB, up to 80MB/s at 40MHz FSB, if the card can sustain these rates) depends on the FSB clock, the CPU type (Cyrix CPUs seem to be more sensitive with bus timings), the mainboard chipset and possibly also the attention to detail on signal routing.
Some mainboards have an option called "LDEV sample point" / "ELBA sample point" or something alike, with settings like "end of T1/end of T2" or "T2/T3". To make 0WS operation effective, you have to choose the lower option here. Choosing "end of T2" or "T3" (or however the bigger number is called) will force 1WS minimum. Another VL performance option on some boards is "LRDY synchronization" with options like "transparent" or "synchronize". Choosing "synchronize" here makes the system more robust, but it will generally add another wait state to VL cycles that is not controlled by the card.
stalk3r wrote on 2023-05-07, 10:00:So I have this card, and it's pretty fast in DOS, but can't get it stable with 0 W/S. It makes all kinds of screen artifacts and […]
So I have this card, and it's pretty fast in DOS, but can't get it stable with 0 W/S. It makes all kinds of screen artifacts and freezes right away during boot. Do you think replacing the video RAM with 60ns chips would make any difference?
What mainboard do you use?
Bios settings?
What CPU? 33MHz or 40MHz local bus clock?
My Ark is running fine with 0WS on the Asus VL/I-486SV2GX4 at 40MHZ.
Not possible on the PVI-486SP3 at 40MHz.
CoffeeOne wrote on 2023-05-07, 12:20:What mainboard do you use? Bios settings? What CPU? 33MHz or 40MHz local bus clock? My Ark is running fine with 0WS on the Asus […]
stalk3r wrote on 2023-05-07, 10:00:So I have this card, and it's pretty fast in DOS, but can't get it stable with 0 W/S. It makes all kinds of screen artifacts and […]
So I have this card, and it's pretty fast in DOS, but can't get it stable with 0 W/S. It makes all kinds of screen artifacts and freezes right away during boot. Do you think replacing the video RAM with 60ns chips would make any difference?
What mainboard do you use?
Bios settings?
What CPU? 33MHz or 40MHz local bus clock?
My Ark is running fine with 0WS on the Asus VL/I-486SV2GX4 at 40MHZ.
Not possible on the PVI-486SP3 at 40MHz.
It's a Shuttle HOT-419 , with an Intel 486 DX4-100. 33 MHz clock. RAM and cache timings are on the tightest possible in the BIOS.
mkarcher wrote on 2023-05-07, 10:52:I don't think faster video RAM will solve your issue. Faster video RAM can remove flicker artifacts at high resolution / refresh rate, and faster video RAM can avoid corrupted images. But faster video RAM will not prevent system crashes / freezes. The freezes indicate an issue between the ARK chip and the 486 processor, which is decoupled from the interface between the ARK chip and the video RAM quite well.
The card should be very fast already at 1WS (if your mainboard doesn't add further wait states, up to 44MB/s at 33MHz FSB and up to 53MB/s at 40MHz FSB), so the bus is unlikely to be a severe bottleneck. Whether you get 0WS working (up to 66MB/s at 33MHz FSB, up to 80MB/s at 40MHz FSB, if the card can sustain these rates) depends on the FSB clock, the CPU type (Cyrix CPUs seem to be more sensitive with bus timings), the mainboard chipset and possibly also the attention to detail on signal routing.
Some mainboards have an option called "LDEV sample point" / "ELBA sample point" or something alike, with settings like "end of T1/end of T2" or "T2/T3". To make 0WS operation effective, you have to choose the lower option here. Choosing "end of T2" or "T3" (or however the bigger number is called) will force 1WS minimum. Another VL performance option on some boards is "LRDY synchronization" with options like "transparent" or "synchronize". Choosing "synchronize" here makes the system more robust, but it will generally add another wait state to VL cycles that is not controlled by the card.
Wow, thanks, that's informative . I have two jumpers, both on default settings, CPU ADS# Signal and CPU RDY# Signal Delay . Not sure what these are, or do they matter at all. The card is quite fast already, I just wanted to check if there is an easy way to push it even further.
stalk3r wrote on 2023-05-09, 19:38:Wow, thanks, that's informative . I have two jumpers, both on default settings, CPU ADS# Signal and CPU RDY# Signal Delay . Not sure what these are, or do they matter at all. The card is quite fast already, I just wanted to check if there is an easy way to push it even further.
CPU RDY# Signal Delay is likely what newer chipsets have as BIOS option. Delay enabled is "RDY synchronized" and delay disabled is "RDY transparent". The ADS delay jumper might add another waitstate. At FSB33, you shouldn't need any of these delays, but they might help you out in case you get issues at FSB50.
If it makes you feel any better, I already tried replacing all the RAM on my card with 45ns parts (originally 70ns) and it made absolutely no difference in terms of overclocking the bus.
Which motherboard are you using your card in? Sometimes there is a jumper on the motherboard itself which controls the VLB wait states.
On my board, it was already set to 0WS, and it seemed to ignore jumper on the ARK.
"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium
stalk3r wrote on 2023-05-09, 19:33:CoffeeOne wrote on 2023-05-07, 12:20:What mainboard do you use? Bios settings? What CPU? 33MHz or 40MHz local bus clock? My Ark is running fine with 0WS on the Asus […]
stalk3r wrote on 2023-05-07, 10:00:So I have this card, and it's pretty fast in DOS, but can't get it stable with 0 W/S. It makes all kinds of screen artifacts and […]
So I have this card, and it's pretty fast in DOS, but can't get it stable with 0 W/S. It makes all kinds of screen artifacts and freezes right away during boot. Do you think replacing the video RAM with 60ns chips would make any difference?
What mainboard do you use?
Bios settings?
What CPU? 33MHz or 40MHz local bus clock?
My Ark is running fine with 0WS on the Asus VL/I-486SV2GX4 at 40MHZ.
Not possible on the PVI-486SP3 at 40MHz.It's a Shuttle HOT-419 , with an Intel 486 DX4-100. 33 MHz clock. RAM and cache timings are on the tightest possible in the BIOS.
Hello,
It is really a bit strange, that you cannot use 0WS at 33MHz.
You could double check with external 25MHz, is 0WS working fine? Or you have problems with 0 WS at 25MHz, too. Then this jumper does something really bad 😁
CoffeeOne wrote on 2023-05-10, 16:11:Hello, […]
stalk3r wrote on 2023-05-09, 19:33:CoffeeOne wrote on 2023-05-07, 12:20:What mainboard do you use? Bios settings? What CPU? 33MHz or 40MHz local bus clock? My Ark is running fine with 0WS on the Asus […]
What mainboard do you use?
Bios settings?
What CPU? 33MHz or 40MHz local bus clock?
My Ark is running fine with 0WS on the Asus VL/I-486SV2GX4 at 40MHZ.
Not possible on the PVI-486SP3 at 40MHz.It's a Shuttle HOT-419 , with an Intel 486 DX4-100. 33 MHz clock. RAM and cache timings are on the tightest possible in the BIOS.
Hello,
It is really a bit strange, that you cannot use 0WS at 33MHz.
You could double check with external 25MHz, is 0WS working fine? Or you have problems with 0 WS at 25MHz, too. Then this jumper does something really bad 😁
That's a good idea, I'll try it. By the way there is a Multi I/O card on the VL bus as well, can it be the culprit? It is strange that I tried with different VLB video cards but always got instant artifacts and freezes with 0WS.
Anonymous Coward wrote on 2023-05-10, 02:09:If it makes you feel any better, I already tried replacing all the RAM on my card with 45ns parts (originally 70ns) and it made absolutely no difference in terms of overclocking the bus.
Which motherboard are you using your card in? Sometimes there is a jumper on the motherboard itself which controls the VLB wait states.
On my board, it was already set to 0WS, and it seemed to ignore jumper on the ARK.
It's a Shuttle HOT-419. I've tried to control the wait states both on the M/B and on the card itself, it does not change the end result. But good to know a faster RAM does not make a difference..
My reading skills must be poor. I didn't notice that you already mentioned with board you have further up the thread. Your board seems to be modern and of decent quality, so it seems that should not be the reason you card can't do 0WS.
One thing I noticed which is a bit odd is that your DIP RAM chips are mixed. It looks like that's not what came from the factory. Perhaps you can try removing the DIPs and leaving the SOJs in place. I don't remember for certain, but I think this card might be able to boot off of either bank of memory. See if that lets you run 0WS.
"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium
heres the manual it explains
https://theretroweb.com/motherboard/manual/ho … 32928582485.pdf
I think I read somwhere that the more VLB cards plugged in makes the VL Bus slower? 2 is the limit? but 3 is too many or somthing not sure. I would try removing the multi IO card and see if anything changes. Id also try differnt VLB slot.
Is your L2 cache double or single banked not sure if that matters or not but I read boards with double banked cache are more stable with zero wait states.
You also might want to check what the VL bus is running at 33hmz or less.. I read these cards cannot do zero wait states at 40mhz.
After you try removing the IO card I would play with the CPU RDY# Signal Delay Configuration. and the VESA Local Bus Configuration on page 23 and 24 of the manual.
Not that it makes any difference but I have a ark 1000 on a 486 66mhz 33fsb and it runs fine at zero wait state, this on my alaris cougar board.
More VLB cards degrades signal quality and possibly causes stability problems. It shouldn't make things slower unless you add wait states. If you have a well designed board, and well designed cards you can sometimes exceed VLB specifications without issue. Isn't the OPTi 895 one of the 486 chipsets that is supposedly VLB 2.0 compliant?
Whether the L2 cache is double or single banked will affect cache timings, not your graphics card timings (that I know of at least).
0ws at 40MHz is usually not much of an issue, except for some older VRAM designs. At least of all the DRAM based cards I own (which is a lot), pretty much all of them can do it.
Some can even do 0WS at 50MHz. The ARK1000 is one of the very last VGA chipsets that works on VLB. I've owned 3 of them, and they all worked well at 50MHz 0WS.
"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium
