VOGONS


Newly made ISA CL-GD5434

Topic actions

Reply 40 of 138, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

I'm going to have to test this out.
Are you running the original GX version?

"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

Reply 41 of 138, by Grzyb

User metadata
Rank Oldbie
Rank
Oldbie
Anonymous Coward wrote on 2020-06-23, 14:36:

I'm going to have to test this out.
Are you running the original GX version?

Yes, ATI Graphics Pro Turbo, ISA variant, the chipset is marked: 210888GX00
Also, see the "mach64 Accelerator Programmer's Guide" - http://vgamuseum.info/index.php/companies/ite … fc08c4fb21b13e2

Żywotwór planetarny, jego gnijące błoto, jest świtem egzystencji, fazą wstępną, i wyłoni się z krwawych ciastomózgowych miedź miłująca...

Reply 42 of 138, by dirkmirk

User metadata
Rank Oldbie
Rank
Oldbie

https://www.youtube.com/watch?v=laMzPYhFSHE&t=5s

Thats a video I made 8 years ago with 2meg Diamond Speedstar64 on my 386DX-40, Dell U2412M monitor, the picture is bloody good on the lcd.

Also have the STB Nitro 64 2meg.

Would be a good card to recreate.

Reply 43 of 138, by canthearu

User metadata
Rank Oldbie
Rank
Oldbie
386SX wrote on 2020-06-22, 13:54:

It sounds like a nice complex project and impressive. I suppose the chip choice has been made but what about older (than Trio) S3 based chips? The ones that were used on the VLB bus should or not compete with the GD5434?
Anyway I've the GD5429 on ISA with 1MB 60ns DRAM (even if my 386 mainboard seems to not like to run it stable I'm trying to understand why) and sure were fast cards but not that impressive on the vga output quality imho. I've switched it to a Wonder XL 28000-5 ATI card and the vga quality filters are much better on a LCD that usually are more sensible to the analog input signals than CRT were back in those times. Not to mention obviously PCI cards like the Millennium, so I'd imagine a similar design for the vga quality target.
Also maybe I'll say something obvious but I'll look on the PCB final thickness to have a perfect no-force installation on ISA connectors. Many cards I've tried had all different sizes of PCB layers and some were even almost difficult to install like it would maybe risk to "open" the whole connector, while others cards got installed as easily as a PCI one... so I suspect there was a margin of different number of layers and final sizes and I'd not make it too much thick nor too much thin...

On my 386DX40, the GD5428 1meg card I bought didn't run stable during all windows 3.1 benchmarks unless I disabled 0 wait state operation.

Because the card did not come with a jumper to toggle this, I actually used a bit of plastic in the slot to forcibly disconnect pin B7 (which it uses to operate zero state) from the ISA bus on this card. This fixed my stability issue with this card at least

Reply 44 of 138, by 386SX

User metadata
Rank l33t
Rank
l33t
canthearu wrote on 2020-06-27, 14:54:
386SX wrote on 2020-06-22, 13:54:

It sounds like a nice complex project and impressive. I suppose the chip choice has been made but what about older (than Trio) S3 based chips? The ones that were used on the VLB bus should or not compete with the GD5434?
Anyway I've the GD5429 on ISA with 1MB 60ns DRAM (even if my 386 mainboard seems to not like to run it stable I'm trying to understand why) and sure were fast cards but not that impressive on the vga output quality imho. I've switched it to a Wonder XL 28000-5 ATI card and the vga quality filters are much better on a LCD that usually are more sensible to the analog input signals than CRT were back in those times. Not to mention obviously PCI cards like the Millennium, so I'd imagine a similar design for the vga quality target.
Also maybe I'll say something obvious but I'll look on the PCB final thickness to have a perfect no-force installation on ISA connectors. Many cards I've tried had all different sizes of PCB layers and some were even almost difficult to install like it would maybe risk to "open" the whole connector, while others cards got installed as easily as a PCI one... so I suspect there was a margin of different number of layers and final sizes and I'd not make it too much thick nor too much thin...

On my 386DX40, the GD5428 1meg card I bought didn't run stable during all windows 3.1 benchmarks unless I disabled 0 wait state operation.

Because the card did not come with a jumper to toggle this, I actually used a bit of plastic in the slot to forcibly disconnect pin B7 (which it uses to operate zero state) from the ISA bus on this card. This fixed my stability issue with this card at least

Interesting! Can you make a picture of that modding and wasn't B7 the -12V voltage? How much was the performance impact? I remember having problems on older mobo but I had in the bios the wait state options to tweak so I found the fastest ones even if it weren't 0-0 values. With this mobo I've no many option to tweak and with this specific vga, a very late December 1995 well built one with a very clean pcb layout, all smd components, with the blue vga connector like the modern ones, I've Windows 3.1 strange crashes with mouse freezing, num lock led remaining on etc..
Tested memtest+2.01 no problems, tested checkit and both memory and video tests pass; I might try changing the cache with lower speed ones but I don't think it's the problem.
It seems that reordering the list of config.sys steps improved something at least on dos side cause it happened also to crash the boot process when loading emm386 or before loading the CLMODE vga app..
But also did not exist some app to unlock advanced values in the bios? I also suspect the VESA modes and the high resolution ones seems to stress the stability even more and in this card there's no jumper to disable the vesa onboard bios like the Oak I have.

Reply 46 of 138, by matze79

User metadata
Rank l33t
Rank
l33t

I run GD5420 1Mb ISA and can‘t complain about Performance issues.
CPU is AM5x86 133Mhz

All VGA Resolution Titles are Fine Doom 1, Duke Nukem 3D, Blood...

A 486Sx 16 is Slower then 386DX 25... at lest my experience on ncr Computer

https://www.retrokits.de - blog, retro projects, hdd clicker, diy soundcards etc
https://www.retroianer.de - german retro computer board

Reply 47 of 138, by mkarcher

User metadata
Rank l33t
Rank
l33t
386SX wrote on 2020-06-27, 16:37:
canthearu wrote on 2020-06-27, 14:54:

On my 386DX40, the GD5428 1meg card I bought didn't run stable during all windows 3.1 benchmarks unless I disabled 0 wait state operation.

Because the card did not come with a jumper to toggle this, I actually used a bit of plastic in the slot to forcibly disconnect pin B7 (which it uses to operate zero state) from the ISA bus on this card. This fixed my stability issue with this card at least

Interesting! Can you make a picture of that modding and wasn't B7 the -12V voltage?

You are right, B7 is -12V, but B8 is /0WS. I'm quite sure canthearu just mistyped or misremembered the number.

Reply 48 of 138, by watlers_world

User metadata
Rank Newbie
Rank
Newbie

Newer 16-bit ISA slots that work at enhanced speeds (faster than specified in the ISA standard) should be referred to as EISA.

The lower speeds of ISA hardware should give accelerators such as the GD5434 a chance to shine.

Reply 49 of 138, by dirkmirk

User metadata
Rank Oldbie
Rank
Oldbie
watlers_world wrote on 2020-06-27, 18:53:

After testing my 486sx/16 with a GD5422, I found it to be particularly slow.
Do you think that an accelerated cirrus chip like the GD5434 would be a better choice for such a machine?

Possibly but thats a slow CPU

For what its worth in my testing their is no performance disadvantage in any situation vs the renowned ET4000 but hardly any of those were 2D accelerated, as you can see in my youtube video its not a stuttery mess in windows like an unaccelerated card so it has all of the benefits of the ET4000 but the advantage of 2D acceleration, can't fault it as an ISA card and theirs not many options, you have to get lucky to buy a good ISA accelerator card for a fair price these days.

I think occasionally you'll see a Cirrus Logic 5428/29/30 pop up and these are still a good option, otherwise very rarely a ET4000 W32/i/p or the ATI mach 32/64s equally as good(I think).

The other issue if like me you want a 2meg card not always possible.

I tried an earlier S3 924 its performance is crap in dos unfortunately but does accelerate 2D, 928 might be an improvement not sure.

I would say the WDC chip cards would be okay but may not find 2meg.

And matrox pretty rare and expensive I've got no experience with those ones.

Reply 50 of 138, by canthearu

User metadata
Rank Oldbie
Rank
Oldbie
386SX wrote on 2020-06-27, 16:37:
Interesting! Can you make a picture of that modding and wasn't B7 the -12V voltage? How much was the performance impact? I remem […]
Show full quote

Interesting! Can you make a picture of that modding and wasn't B7 the -12V voltage? How much was the performance impact? I remember having problems on older mobo but I had in the bios the wait state options to tweak so I found the fastest ones even if it weren't 0-0 values. With this mobo I've no many option to tweak and with this specific vga, a very late December 1995 well built one with a very clean pcb layout, all smd components, with the blue vga connector like the modern ones, I've Windows 3.1 strange crashes with mouse freezing, num lock led remaining on etc..
Tested memtest+2.01 no problems, tested checkit and both memory and video tests pass; I might try changing the cache with lower speed ones but I don't think it's the problem.
It seems that reordering the list of config.sys steps improved something at least on dos side cause it happened also to crash the boot process when loading emm386 or before loading the CLMODE vga app..
But also did not exist some app to unlock advanced values in the bios? I also suspect the VESA modes and the high resolution ones seems to stress the stability even more and in this card there's no jumper to disable the vesa onboard bios like the Oak I have.

Sorry, I meant pin B8. You just locate the pin on the card, and being between -12V and +12V, you will probably find that the surrounding 2 pins are not used, making the process easier. I also cut a thin strip of plastic (5mm or so) from a disposable drink container.

I simply installed the card into the PC, then pushed the plastic strip into the slot between the slot pins and the ISA card for Pin B8. Since the card doesn't use +12V or -12V, it didn't matter if I push those pins off the card too.

This will disable 0-wait state operation of the ISA card. You can make most of the lost performance back by cranking up the ISA bus speed.

Reply 51 of 138, by canthearu

User metadata
Rank Oldbie
Rank
Oldbie
watlers_world wrote on 2020-06-27, 18:53:

After testing my 486sx/16 with a GD5422, I found it to be particularly slow.
Do you think that an accelerated cirrus chip like the GD5434 would be a better choice for such a machine?

Did you correctly benchmark the system and find it to be slow compared to systems of similar vintage?

A 486SX/16 should perform about as well as a 386DX/33, depending on the application. If you are asking it to do more than a 386DX/33 could do, you would find performance quite bad.

In regards to the GD5422, it should perform just fine. In my testing, my GD5424 in zero wait state mode had similar performance to a ET4000AX, so they were not generally slow cards at all. Of course, individual cards may perform differently, so it is still possible you have a poorly designed example.

In windows 3.1/95, accelerated cards definitely have a huge edge over their accelerated examples. Features like color expansion (express an 8/16bit color using only 1-bit on the ISA bus, good for solid fills and font rendering), and BitBlt (move blocks of memory around the video card and to/from the framebuffer from system memory) greatly improve performance in light of extremely limited ISA bandwidth.

Reply 52 of 138, by canthearu

User metadata
Rank Oldbie
Rank
Oldbie
dirkmirk wrote on 2020-06-27, 23:52:
I think occasionally you'll see a Cirrus Logic 5428/29/30 pop up and these are still a good option, otherwise very rarely a ET40 […]
Show full quote
watlers_world wrote on 2020-06-27, 18:53:

After testing my 486sx/16 with a GD5422, I found it to be particularly slow.
Do you think that an accelerated cirrus chip like the GD5434 would be a better choice for such a machine?

I think occasionally you'll see a Cirrus Logic 5428/29/30 pop up and these are still a good option, otherwise very rarely a ET4000 W32/i/p or the ATI mach 32/64s equally as good(I think).

The other issue if like me you want a 2meg card not always possible.

I tried an earlier S3 924 its performance is crap in dos unfortunately but does accelerate 2D, 928 might be an improvement not sure.

I would say the WDC chip cards would be okay but may not find 2meg.

It would be extremely rare to find 2meg on an ISA card. When ISA cards were the thing, high resolution SVGA monitors cost an astounding amount of money.

I've got a S3 801 based ISA card on the way, will let you know how it compares to ET4000AX, GD5424, GD5428 and maybe an Oak 067 if I can be assed pulling it out of the 286.

Reply 53 of 138, by 386SX

User metadata
Rank l33t
Rank
l33t
canthearu wrote on 2020-06-28, 00:24:
dirkmirk wrote on 2020-06-27, 23:52:
I think occasionally you'll see a Cirrus Logic 5428/29/30 pop up and these are still a good option, otherwise very rarely a ET40 […]
Show full quote
watlers_world wrote on 2020-06-27, 18:53:

After testing my 486sx/16 with a GD5422, I found it to be particularly slow.
Do you think that an accelerated cirrus chip like the GD5434 would be a better choice for such a machine?

I think occasionally you'll see a Cirrus Logic 5428/29/30 pop up and these are still a good option, otherwise very rarely a ET4000 W32/i/p or the ATI mach 32/64s equally as good(I think).

The other issue if like me you want a 2meg card not always possible.

I tried an earlier S3 924 its performance is crap in dos unfortunately but does accelerate 2D, 928 might be an improvement not sure.

I would say the WDC chip cards would be okay but may not find 2meg.

It would be extremely rare to find 2meg on an ISA card. When ISA cards were the thing, high resolution SVGA monitors cost an astounding amount of money.

I've got a S3 801 based ISA card on the way, will let you know how it compares to ET4000AX, GD5424, GD5428 and maybe an Oak 067 if I can be assed pulling it out of the 286.

Thanks, also the S3 801 seems like a nice card but I think there were newer chips too with the ISA connector. Too bad that the GD5429 I have seems to be a very late not well supported ISA card while it was already a "fast" VLB oriented chip maybe where there were bios revisions or newer drivers. I've found the specific chip drivers cause it seems that this chip beside a better BitBLT engine report being memory mapped I/O and probably differs from the previous bios. In fact I've found a newer bios for all the GD542x up to the 5428 (called 1.30 bios version I think it was) but not for the GD5429 that has a 1994 1.00 version. Considering the card is built in the last month of 1995 it sure sound like one of the latest built vga ISA card around (or do you know later isa vga built after?). I'd love to see a GD5434 running but their prices are absurd not to mention the Mach64 ones and it's seems clear the ISA bus is the bottleneck anyway no matter what.
The Oak OTI077 suprised me to be faster than the TVGA9000A and in fact is a 1993 chip/card while the second is dated around 1991. Still both need some utility cause on modern LCD they have freqs/hz problem with the standard values.
The GD5429 drivers came with a nice Win tool to modifify these. But I begin to suspect the 0 wait state in not so compatible with the UMC 1991 chipset of my board..

Reply 54 of 138, by 386SX

User metadata
Rank l33t
Rank
l33t
canthearu wrote on 2020-06-28, 00:07:

Sorry, I meant pin B8. You just locate the pin on the card, and being between -12V and +12V, you will probably find that the surrounding 2 pins are not used, making the process easier. I also cut a thin strip of plastic (5mm or so) from a disposable drink container.

I simply installed the card into the PC, then pushed the plastic strip into the slot between the slot pins and the ISA card for Pin B8. Since the card doesn't use +12V or -12V, it didn't matter if I push those pins off the card too.

This will disable 0-wait state operation of the ISA card. You can make most of the lost performance back by cranking up the ISA bus speed.

Update: I've found that in the card the pin B8 end up in a place where there was (and never seen) a JP1 mark on the PCB but already factory opened without a jumper socket that if would have been shorted would end up into the Quadtel chip close to the vga connector.
I've read now that that should be exactly for the reason suggested above, but in the technical manual of the chip it says something about specifically the GD5429 that seems to not read the bios for the zero wait option: "Bios Read: CF(1)=0 (not on the GD5429)" In this case I suspect the PCB was supporting the pin compatbile older version but looks like the GD5429 might have zero-wait state by default and the only solution would be the bios of the mainboard. In the Help for the default settings it talks about the default register values of each option but if I disable to automatic configuration I don't have more options to modify.
I've also tried to lower down the AT bus to 6Mhz and still it hangs in Windows. Also tried changing the I/O controller card with a UMC based who knows maybe something related to the Goldstar Prime 2C I was thinking but nothing changed.
This card will make me crazy.. wasn't there some apps to unlock AMI bios hidden options?

Reply 55 of 138, by Grzyb

User metadata
Rank Oldbie
Rank
Oldbie
386SX wrote on 2020-06-28, 17:48:

This card will make me crazy.. wasn't there some apps to unlock AMI bios hidden options?

AMIsetup - ftp://ftp.elf.stuba.sk/pub/pc/utildiag/amis2990.zip

Żywotwór planetarny, jego gnijące błoto, jest świtem egzystencji, fazą wstępną, i wyłoni się z krwawych ciastomózgowych miedź miłująca...

Reply 56 of 138, by 386SX

User metadata
Rank l33t
Rank
l33t
Grzyb wrote on 2020-06-28, 18:02:
386SX wrote on 2020-06-28, 17:48:

This card will make me crazy.. wasn't there some apps to unlock AMI bios hidden options?

AMIsetup - ftp://ftp.elf.stuba.sk/pub/pc/utildiag/amis2990.zip

Obvious question maybe but is it dangerous to use? Cause I still didn't find a flasher utility to backup the bios of the mobo (the older ones requires a 486, the uniflash too).

Reply 57 of 138, by Grzyb

User metadata
Rank Oldbie
Rank
Oldbie

AMIsetup doesn't modify the ROM BIOS, only its configuration stored in the CMOS RAM.
In case of problems, it should be still possible to enter the normal CMOS SETUP, and load the default configuration.

Żywotwór planetarny, jego gnijące błoto, jest świtem egzystencji, fazą wstępną, i wyłoni się z krwawych ciastomózgowych miedź miłująca...

Reply 58 of 138, by 386SX

User metadata
Rank l33t
Rank
l33t

The original values of the bios config were on the conservative side: hidden options were the DRAM wait state and it was @ 2 (I suppose for both W/R). I tried to put it @ 1 and everything is as stable (or unstable depending on this GD5429) as before. There're also two hidden options called Fast cache read hit and write hit and was disabled. I read online I should put it to enable I don't know why. Wait state did a serious jump in the system ram speed.. from 9MB/s to almost 13MB/s in cachechk and from 12 to 15MB/s in Speedsys. Also I lowered down the I/O Recovery delay from the slowest default option BLCK 12 to BCLK 4 (I still can go down to 2) and I've seen serious hard disk improvement (from 660KB/s to 800KB/s in checkit).
The stability of the Windows still isn't there imho.. I begin to suspect it's a driver problem. This chip specifically the GD5429 is the only one different from all the earlier because it had something called "memory mapped I/O BitBLT" and had higher core/mem freqs and also the zero wait state seems to work differently probably I imagine zero @ default in the vga registers or similar.But like the GD5434 were VLB and PCI oriented chips and maybe the only specific GD5429 generic driver (K5429_D1.exe, K5429_D2.exe) hangs in Windows (while acceleration is definetely working windows moves fast as never seen on other cards, Winbench3.0 gave something like 10.000.000 score compared to the default 2.000.000) maybe because the drivers were for the VLB config?
I tried the GD5434 drivers and things seems to be more stable, slower benchmarks interesting but acceleration still works but make crash some DLL (CR***.dll) during bench obviously I was already surprised it booted the GUI ok.
So I need to find the specific ISA drivers or the manufacturer one. The card is called JA8228E/V4 and made in usa.

EDIT: the board was manufactured by Jaton Corp. that I suppose worked with nvidia long time ago.

Reply 59 of 138, by 386SX

User metadata
Rank l33t
Rank
l33t

Maybe SOLVED 😁:

I was probably suspecting correctly the problem were the drivers. I installed on a separate disk Win95 in this 386 and was running (beside the slow installation) impressively fast, almost incredible.. the GD5429 BitBLT engine gave a boost to the GUI that would make me remember why these card were indeed "accelerator".
So there was no reason why Win3.1 should not work correctly. Looking around I didn't find the original Jaton Corp. drivers, in thei old webpages didn't even listed this model strangely. But I found that STB built the Horizon ISA card that's mostly similar, change only the type of bios (one classic chip instead of the dual Quadtel plus another chip) and it still was one of the few GD5429 ISA card around. So I found its drivers and manually installed with the oemsetup.inf because the setup.exe will crash like its windows utility tool, probably looking for the real card obviously. But the oemsetup manual installation seems (!) to have solved the stability problems. No crash until now after many Winmark 3.1 tests that gave a score (fast bench) of around 9.000.000 number a bit slower than the GD5429 generic drivers but I suspect they were for the VLB cards not the ISA. This seem like a good result! Acceleration is working ok. Only problem I've to manually change vertical freq of the resolution. Any generic app for that?

Anyway, to not go off topic too much, this may be interesting for the thread itself if will be considered that probably an ISA GD5434 would need some specific drivers and not just the generic online. 😉