Ozzuneoj wrote on 2025-12-27, 23:09:

MattRocks wrote on 2025-12-27, 22:58:

Ozzuneoj wrote on 2025-12-27, 22:42:

Still hoping someone can chime in on this one. To summarize: PNY FX 5200 128MB (128bit) card reporting only 64MB (64bit) in BIOS […]
Show full quote

Still hoping someone can chime in on this one. To summarize: PNY FX 5200 128MB (128bit) card reporting only 64MB (64bit) in BIOS and in Windows. No visible damage or defects of any kind on it and it otherwise works flawlessly, though obviously with 64bit performance.

Interestingly, I actually found another identical PNY card that properly reports 128MB and get this: There are no obvious differences to the PCB layout or components. In fact, the one that reports the full 128MB actually has a couple of tiny SMD caps snapped off the back from being in a scrap lot. Those apparently don't affect anything at all because the card works perfectly and is clearly faster than the one using only half its memory. And before you ask... yes, they are clearly broken off and not intended to be missing.

The BIOS files also seem to be identical between the cards.

Anyone have any ideas? I have used my microscope and inspected the card quite thoroughly but I can't find any broken components or any visible damage.

I'm wondering if maybe a solder joint on one of the memory chips has broken and it is one that doesn't introduce any errors but it disconnects the other 64MB... I have no idea if this is possible.

Here is the datasheet for the memory chips.
https://www.alldatasheet.com/datasheet-pdf/vi … 281622DT-5.html
I have looked it over and nothing obvious is jumping out at me like "if this is disconnected you lose megabytes" 😅. If anyone is feeling brainy and wants to offer any input, I'd love some theories at least.

Try Linux kernel logs; Linux might reinitialise the GPU and repeat whatever the VBIOS did at POST. The big difference being one has logs, one doesn't. Good news is you can generate two sets of logs with your two cards and diff them 😀

I haven't used Linux in 20+ years (unless you count Android 😋).

I kind of doubt that OS logs would be able to identify anything that runs deeper than the card BIOS. It is likely a physical defect of some kind, not a software or firmware issue.

If the VROM image is identical, and the physical VRAM is identical, then the fault has to be something between VBIOS and VRAM - such as resistors?

There was more battery corrosion on that motherboard than I thought! My desoldering tip! What have I done to you?

Just when I think I'm starting to understand memory configurations on old video cards, I get something like this...

I have an ASUS TNT2 Vanta, just like the one in this listing.

It is a 16MB card, and like any other TNT2 M64 or Vanta, it has a 64bit memory bus.

So, how does that work with eight of these memory chips?
https://www.alldatasheet.com/datasheet-pdf/vi … 48LC1M16A1.html

Is there a way to tell based on this datasheet that eight of these 16bit chips would only provide a 64bit total bus width, or is this just due to the way to card is wired (since the Vanta chip itself only has connections for a 64bit bus)? Does the "2 banks" part of the chip's configuration have anything to do with it? It says it is either a 512K x 16bit x 2 banks or a 1M x 16bit part.

Looking at the VGA Museum, I see lots of this 8-chip TNT2 Vanta and M64 cards.
https://www.vgamuseum.info/index.php/cpu/item … 34-nvidia-vanta
https://www.vgamuseum.info/index.php/cpu/item … a-riva-tnt2-m64

... so the configuration isn't that rare. I think the rest of the 64bit TNT2 cards I own all have 4 memory chips though, so it struck me as odd.

Is there any benefit to using a PCB design with eight chips for a low cost card like this? Is it just that they could choose to use 8 of these or 4 higher density chips on the same PCB depending on what was available?

The performance is as dismal as one would expect of any Vanta, of course.

Some of these have 4 chips using 32 bits chips (two chips each), with 8 chips solder pads. No mystery about this for grouping for 16 bits ICs forming two groups of 4 chips each. Either way.

Cheers,

Ozzuneoj wrote on 2025-12-28, 20:06:

So, how does that work with eight of these memory chips? https://www.alldatasheet.com/datasheet-pdf/vi … 48LC1M16A1.html […]
Show full quote

So, how does that work with eight of these memory chips?
https://www.alldatasheet.com/datasheet-pdf/vi … 48LC1M16A1.html

Is there any benefit to using a PCB design with eight chips for a low cost card like this? Is it just that they could choose to use 8 of these or 4 higher density chips on the same PCB depending on what was available?

The performance is as dismal as one would expect, of course.

$
1mb chips probably cost less than half 2mb chips.

They work the same as the secondary slots in a 4 slot motherboard.
One pin is chip select, which is low for the first bank, and high for the second (when accessing the second bank)
It's not faster, but I don't believe it's slower.

Following in the footsteps of llama98, someone has released an AI doohicky for CP/M. https://github.com/HarryR/z80ai

I downloaded GUESS.COM to my Z280 board and ran it there. I asked some questions, but ultimately lost interest before guessing the object. It takes a while for each response.

Ozzuneoj wrote on 2025-12-27, 16:24:

henk717 wrote on 2025-12-27, 15:43:

Been expanding my vmodem a bit, I think its good enough now for public usage. You can get it here : https://archive.org/details/ […]
Show full quote

Been expanding my vmodem a bit, I think its good enough now for public usage.
You can get it here : https://archive.org/details/serial-modem
Its based on someone elses PR to vmodem from the protoweb project, but wrapped in a VM and edited a bit to make it handle better on Windows 3.11.
I added a dialup sound wav to it when your dialing, added latency emulation and even added a basic http proxy for protoweb available by dialing number 11 (Dialing 1 and manually setting the proxy does produce better results).

Hope this is fun for the rest of you, it can be used with USB serial adapters, physical serial ports or com2com virtual serial ports in your VM's.

I'm feeling a bit dense. I have read your archive.org page and the original github page... but I can't tell what this is used for. 😅

If I'm the average retro enthusiast with a collection of PCs and parts, what do I do with this?

Its a VM that pretends to be a dial-up modem. So on my physical hardware I can connect it over serial and then use dialup to get connected. It adds realism to my Windows 3.11 install as well as Windows 98.

So you hook it up over serial like a real dialup modem and dial the number of the internet provider. The machine will tell the modem over serial to dial that number, it plays a dialup sound and then acts like an internet provider. At that point you are online without the need for a network card and optionally with latency simulation of real dial-up.

It can be used combined with stuff like dosbox to and its an alternative to having two machines connected with real modems and phone line emulator boxes. Its also an alternative to needing real landlines to experience it.

It may be a VM, but my retro PC does not know that.

henk717 wrote on 2025-12-29, 19:46:

Its a VM that pretends to be a dial-up modem. So on my physical hardware I can connect it over serial and then use dialup to get connected. It adds realism to my Windows 3.11 install as well as Windows 98.

I'll have to imagine Win95, because I never actually used Win98 on diallup. It was 99 before I had the use of a 98 system and had moved to a place that happened to be a pilot area for cable internet, so 1Mbps baybeeeeeee. So I dunno if that's a deficit or not, but I put my 5+ years in on diallup.

pete8475 wrote on 2025-12-27, 04:49:

RetroBus wrote on 2025-12-27, 02:08:

bought a box of random hardware on facebook marketplace for $20, seller didnt know what it was and neither did I, just computer hardware thrown into a barely held together box 🤣

https://www.youtube.com/watch?v=w-44JpS2Sqc

Watching now, your videos are always a good time.

Thankyou 😀 Very much appreciated

Burned the factory recovery discs for my model of Compaq Presario and took it to a stock configuration on a 16GB flash card (previous condition was fresh out of my parents attic with no hdd, so this is a nice improvement).

Had wanted to do this a week ago, but the SanDisk flash card I bought couldn't be treated as a fixed drive, so I had to get another. Wish they'd tell you that on the packaging, larger ones like the one I bought are a little pricey for a casual buy.

Just did a review on the ATI rage 128 pro, man these cards were everywhere, especial the normal ATI rage cards were embedded in a lot of systems

https://www.youtube.com/watch?v=11dPAocYoUY

RetroBus wrote on 2025-12-30, 01:12:

Just did a review on the ATI rage 128 pro, man these cards were everywhere, especial the normal ATI rage cards were embedded in a lot of systems

https://www.youtube.com/watch?v=11dPAocYoUY

Thanks for sharing another interesting review - everything you said about 3D gaming on Rage cards aligns with memory.

But from a general PC user's perspective, I remember the Rage cards slightly differently. As you showed, Rage were reviewed in magazines alongside 3Dfx. Rage were also reviewed alongside other cards, in other contexts. My recollection is that the market was far more fragmented: 2D performance, 3D CAD performance, 3D gaming performance, video playback performance, and even video capture performance. That meant a lot of different magazines, different aspirations, different benchmarks, and each focussing on one or two specific use cases.

3D gaming? Sure, the Voodoo series comfortably led on FPS but the image was different and games were often developed against Voodoo-specific shortcuts.

3D CAD? That's mission-critical work where shortcuts are intolerable - you'd need "retained mode" rendering. You'd probably want 3DLabs, but Rage was an option.

2D? Your eyes would thank you for choosing Matrox, but you'd be checking that Rage wasn't equally good.

Video playback? You'd be comparing a Rage against discrete MPEG decoder/accelerator cards.

Video capture? There was a Rage for that, so you'd be comparing Rage against Hauppauge capture cards.

Backwards compatibility? This is always a contemporary problem, and the Rage would be weighed against an S3.

The Rage (series) demanded high retail prices because they showed one card was good enough for many different tasks. The reason system integrators loved Rage is because new PC buyers never really know what they want, and the Rage had more known unknowns covered. And, the reason Rage don't demand the same high prices today is because typical retro usage is only pressing a fraction of an era's actual use cases - not general computing.

There is one sanity check we can fall on when we are unsure about any component: Did Apple pair it with their prosumer machines? The answer tells you a lot.

I happen to love the Rage 128 for that it is. However, my biggest complaint with ATI from this era was their terrible 16-bit dithering. You have either pattern or random. Some cards with certain drivers have an option to disable dithering altogether, but I don't know which ones those are. The latest drivers removed that feature, and it's definitely not available on any Radeon.

NVIDIA had excellent dithering support up through the FX series. 3dfx always did it best, of course. I don't know about other brands.

You're not wrong but 3Dfx dithering (the capitalisation shows my age) was more about the perceptual reconstruction than physical representation.

To do what 3Dfx did, they deliberately exploited CRT phosphor blending, scanline bleed, and our human visual processing. The result is that a 3Dfx 16-bit "motion" via CRT is interpreted by our brains as being closer to 24-bit. That’s also why digital screen capture can never reproduce how good Voodoo cards feel.

ATI Rage, by contrast, rendered what you might call “mathematically honest” 16-bit. The still shot of an ATI output comes much closer to representing what was actually emitted from a screen surface. There's a solid reason too - the Rage pipeline was capable of supporting mission critical 3D with auditable deterministic retained-mode rendering (crash simulation, medical scanning, etc.) where analog screen tricks were dangerous liabilities.

Other than recalling the Riva TNT2 Ultra was very fast, I don't remember much about Nvidia in "DX6 vs Glide" era and I have not looked into how any Riva achieved its outputs. My perception at the time was that the Ultra relied on binning and brute force clock rates making them rare hot statements.

But Radeons (excluding RV100) and GeForces (including FX) are later products working with the assumptions of a totally different philosophical and technical era: hardware T&L, LCD, digital output. Even if they briefly coexisted, there's almost no technical or philosophical overlap between T&L GPUs and what a 3Dfx was conceived to do or what Rage/TNT/Savage had competed with.

And, that is why 3Dfx struggled - the entire ecosystem around graphics cards fundamentally changed in many ways all at once, and the problems 3Dfx solved were increasingly irrelevant. It only got worse because early "4:3 CRT replacement LCDs" had a sort of phosphor emulation behaviour that later LCDs drop. The climate was relentlessly hostile to 3Dfx analog glide tricks - I recall having a fleeting speculation that 3Dfx would pivot into making screen technology.

But this strays far from what RetroBus has done in pairing Rage 128 Pro with an LCD - that's perfectly historically defensible for the CRT/LCD transition era, and it stands entirely separate from what earlier 3Dfx Voodoo and later Nvidia GeForce were each doing.

Great analysis! thanks for the detailed write up it definitely shines more of a light on the Rage and its time and place.

Ultima VI – Italian Translation (MS-DOS)
Amateur project for the complete translation of the game texts

What is it about?
I'm working on a complete Italian translation of Ultima VI: The False Prophet (1990), MS-DOS version.

The work involves exclusively the original in-game texts and is carried out using U6Edit, a tool that allows you to safely unpack and edit the game files containing:

• the literatures (books, scrolls, documents)
• the dialogues of all NPCs (game actors)

The goal is to achieve a faithful, consistent, and, above all, 100% compatible translation with the original engine, without introducing bugs or anomalous behavior.

Working Method
The translation follows strict operational rules, already field-tested and confirmed:

• Complete NPC Block
  We always work on the entire character block (identity / look / converse / greeting / answers / system strings), even if it is very long.
• Sections not to be modified
  They are NEVER modified:
  • @ % [ ] { } < > & *
  • Numbers and IDs
  • Control markers
  • Parser keywords
  • Logical flow (JUMP, ask, answers, keywords, wait, etc.)
• Accents
  Apostrophes only: e' a' i' o' u'
• Space
  Translation of equal length: compact sentences, short synonyms, no unnecessary prolongations.
• Vocabulary
  Absolute consistency with established terms (e.g., Code, Singularity, Vortex, Moonstones, etc.).
• Proper nouns
  Unchanged.
  Places or titles are translated only if they are already established in the text or are clearly generic.

Current Project Status

Literatures
• Total Blocks: 128 (ID 000–127)
• Translated: 128 / 128
• Completion: 100%

NPC Dialogues
• Total Characters: 223
• Translated: 50 / 223
• Completion: ~22%

Overall Progress
Total Work Blocks: 351
Completed Blocks: 178

Overall Project Progress: ~51%

Work is progressing steadily. systematic, with particular attention to game engine stability and internal linguistic consistency.

Those familiar with Ultima VI know how integral the text is to the experience: the goal is to finally make it accessible in Italian, my language, without distorting it. This translation and my other related project, Ultima VI Map Notes - Interactive “Google Maps” style map to annotate Britannia:

will be of great help to me in not "losing the thread" of the adventure, when I'll be able to play it as it deserves.

glide2cif wrapper testing: Re: Glide/CIF wrapper

Connected my legacy laptop (Thinkpad T60) to the home network via an Ethernet cable.

Background: I noticed that despite sitting literally under the optic fiber modem and main router, the laptop was connected via mediocre WiFi. That's because all the Ethernet ports on the router were taken. However, there was an unused port on the MoCA adapter, also sitting on top of the laptop. Took the shortest CAT5e cord in my box, and didn't even have to undo the twisty ties to plug both ends. Voilà - 10 times faster speeds and a more stable connection, both to the LAN and the internet.

I stumbled upon these books last night and *instantly* ordered them, can't wait for them to arrive - I've been coincidentally wondering for a while now if there were ever any books about Ultima saga...

Through the Moongate: The Story of Origin Systems, Richard Garriott, and the Ultima Saga

First Published: 2019 (Part I), 2021 (Part II)
Publisher: self-published
Genre: video game history essay
Subject: video game history, RPG, software development, PC culture

What is "Through the Moongate"?

Through the Moongate is a fundamental book for understanding the The birth, evolution, and historical impact of the Ultima saga and the legendary Origin Systems.
This is not a simple chronology of games, but a true behind-the-scenes journey that intertwines technical development, creative vision, and gaming culture of the 1980s and 1990s.

The title symbolically recalls the Moongates of Britannia: portals connecting different worlds, just as the book connects players, developers, and historical context.

Main Contents

• The Origins of Ultima
  From Richard Garriott's early experiments on the Apple II to the definition of a new way of understanding computer role-playing games.
• The Birth of Origin Systems
  The creation of an independent studio that would raise the bar for quality, ambition, and immersion in video games.
• Analysis of the saga chapters
  From the first Ultima to the more mature episodes, with attention to:
  • Evolution of the game world
  • Player freedom of action
  • Introduction of the Virtues system
  • Emerging narrative
• Firsthand accounts
  Interviews and stories from developers, collaborators, and key figures of the time, with often little-known technical and creative anecdotes.
• Cultural impact
  Ultima as a milestone in the history of Western RPGs and a source of inspiration for many modern games.

It is written from the player's perspective, but with historical rigor: it explains how and why Ultima was revolutionary, not just what it was.
It chronicles an era when few developers could disrupt the medium with bold ideas, perfect for those who love classic RPGs, the history of PC gaming, and artisanal video game development.

Connection to Ultima VI

For those, like me, who consider Ultima VI: The False Prophet one of the pinnacles of the series, Through the Moongate helps us understand the creative and technical journey that led to that level of narrative depth, freedom, and coherence in the game world, without giving away spoilers.

I just watched this episode of The Computer Chronicles from 1993 where they talk about the Pentium processor in detail.

About 13 minutes in, they show a server which had 64MB RAM, and could host up to 1GB at maximum. Didn't realize that was even a thing back then. When asked who is buying these systems, the salesman answers: people who are working on neural networks and AI. I had to do a double take on that.