bracecomputerlab wrote on 2025-08-01, 17:13:

Slightly off topic about RDRAM. I have been working on a creating a replica of VIA Technologies VT5228C reference schematic main […]
Show full quote

Slightly off topic about RDRAM.
I have been working on a creating a replica of VIA Technologies VT5228C reference schematic mainboard for KiCad.
VT5228C reference schematic mainboard has Apollo Pro133A (VT82C694X) and VT82C686A "Super South" along with support for 133 MHz Slot 1 processors.
I have read the Apollo Pro133A chipset design guide and Intel 440BX chipset design guide, and both total around 140 to 160 pages, if one counts the OrCAD schematic.
Anyway, the ill fated Intel 820 chipset design guide is roughly double of the page count, and is far more detailed than the aforementioned two other design guides.
Besides that, Intel 820 chip set requires 28 ohm characteristic impedance for RDRAM traces versus 65 ohm for Apollo Pro133A and 440BX.
The recommended materials used in the 28 ohm characteristic impedance 4 layer PCB stack up is also different in terms of material thickness, and perhaps these lower volume, less familiar PCB fabrication methods might be the reason why mainboard and memory modules were so expensive back then.
I am not really an expert in PCB design, but just wanted to bring this up.

Was my stuff useful?

mockingbird wrote on 2025-07-16, 20:00:

Public technology consortiums have run their course and R&D should be returned to the private sector.

[/rant]

What in the world are you talking about? These consortiums are ran by the companies. The government is not deciding what should be in the DDR-6 spec.

A monopoly, which is what intel tried to do with rambus, always hurts consumers and is ANTI free market.

Historically, Intel worked to stymie computer advancement at every turn. Purposefully holding back technology that was already developed in order to soak consumers and biz. If it weren't for AMD we'd still be using netburst cpus. Thinking we'd be in a better alt reality if we'd willfully handed Intel a monopoly on all computer memory is the most insane, wrong, and gleefully ignorant take I've ever read on this forum.

kingcake wrote on 2025-08-18, 05:27:

What in the world are you talking about? These consortiums are ran by the companies. The government is not deciding what should be in the DDR-6 spec.

It's time to go back to competing technologies. When I say competing, I mean ugly, in a very incompatible way with everything else type of competing technologies.

There are a lot of improvements and innovations that can (but sadly won't) be made that can extend the longevity of current hardware in orders of magnitude and reduce e-waste to practically nil.

The Asians, who drive the industry (we allowed Taiwan a practical monopoly, what did this buy us? 40 years of freedom from Chinese occupation?) are never going to do this, their culture doesn't allow for it, and it never will. At first, they produced what they were told to. Then they produced what they thought we wanted. Now, they design and tell us what to buy. Nothing against the Taiwanese, but RGB LED matrices don't actually have any effect on how computers perform.

How much junk will be sold until the next deposit of rare earth is depleted and a new war is fought, who knows?

"we allowed"

Rofl

Are you allowing ASML to make the best chip printers too?
So generous.

myne wrote on 2025-08-18, 12:23:

"we allowed"

I'm not trying to detract from their ingenuity, but technical innovation came from the west, yes. Find me books in Chinese from the 1970s about operating system theory.

Oh, well in that case, the British allowed you do become a great power.

Show me the US navy in 1700

myne wrote on 2025-08-18, 13:26:

Oh, well in that case, the British allowed you do become a great power.

Show me the US navy in 1700

I wasn't aware that "Tanenbaum" and "Silberschatz" were British names.

To be honest, I have not looked at your KiCad based design (Intel 440BX uniprocessor customer reference schematic entered in KiCad) so far, but your attempt and necroware's S7 VRM project probably had an impact on deciding to do a KiCad based reference design replica project back in December 2024.
The looming illegal tariffs against importing components from overseas (particularly China) and illegally doing away with de minimis rule (a rule codified into law in 2016 that clearly states that up to $800 of small packages are exempt from tariffs) got me frantically ordering components like chipsets (Intel, VIA, and ALi) and clock synthesizers (ICS and Winbond) back in early to middle of January 2025 from Chinese sellers on eBay.
After laying low for a month starting January 20th, 2025 (I guess some people should have taken Donald Trump's historical obsession with tariffs more seriously before voting for him.) due to the mass confusion gripping the country, I started to order more from Chinese sellers on eBay for additional components like connectors (i.e., IDC 40 pin connector and AGP connector) and discrete transistors (i.e., AMS1117-3.3 and AMS1117-5.0) starting around early February.
Started to tap AliExpress starting early to middle of March since I could not procure all the components from eBay.
Became pretty hooked on AliExpress as a convenient platform for procuring missing components, often sellers selling them much cheaper than on eBay.
The confusion over the fate of de minimis rule was quite acute around the middle of April since the cut off of de minimis was May 2nd arrival at the airport customs.
I got several VIA, ALi, and SiS chipset orders getting cancelled by AliExpress sellers around April 23nd due to the looming demise of the de minimis rule (Thanks a lot Donald for screwing me, this is totally illegal.).
There was mass confusion on AliExpress around April 23rd to 24th, and I literally had half of my cart items becoming unavailable to order from the US in the early morning hours of April 23rd or 24th (saw it in real time).
AliExpress uses cost efficient Chinese logistics firms like Cai Niao and Yen Wen (I hope I got the names right.) to the west coast US.
I also ordered a BGA rework machine (IR6500) and Puhui T-962A reflow oven from AliExpress.
Puhui T-962A took 2 1/2 months to arrive, and I tried to cancel the order because it was never showing up despite the seller's assurance that it was on the way.
I started the cancellation eventually, but miraculously, it somehow show up at my door step the next day!!!
Of course, I kept the item (cancelled the order cancellation immediately).
The seller was not a fraud, and it appears that the illegal suspension of de minimis rule slowed down the order and caused heartburn on my side (and the seller from China as well since they likely ate the tariffs).
It appears the other side ate the tariffs, and I did not have to face additional costs.

I had to do some totally unrelated activity for pretty much the entire August that really burned me out.
I am now slowly recovering from my hellish August, and is slowly entering small components of VIA Technologies VT5228C Apollo Pro133A.
I worked on 2 small MOSFET parts today, and need to finish entering other small items like WOL (Wake on LAN) connector.
That's where my project stands.
I hope to start the schematic entry by late September.
I also plan to do my own version of Intel 440BX uniprocessor customer reference schematic in KiCad after VT5228C replica project, and have procured most of the components already.
That will be my second mainboard project.

myne wrote on 2025-08-17, 13:47:

bracecomputerlab wrote on 2025-08-01, 17:13:

Slightly off topic about RDRAM. I have been working on a creating a replica of VIA Technologies VT5228C reference schematic main […]
Show full quote

Slightly off topic about RDRAM.
I have been working on a creating a replica of VIA Technologies VT5228C reference schematic mainboard for KiCad.
VT5228C reference schematic mainboard has Apollo Pro133A (VT82C694X) and VT82C686A "Super South" along with support for 133 MHz Slot 1 processors.
I have read the Apollo Pro133A chipset design guide and Intel 440BX chipset design guide, and both total around 140 to 160 pages, if one counts the OrCAD schematic.
Anyway, the ill fated Intel 820 chipset design guide is roughly double of the page count, and is far more detailed than the aforementioned two other design guides.
Besides that, Intel 820 chip set requires 28 ohm characteristic impedance for RDRAM traces versus 65 ohm for Apollo Pro133A and 440BX.
The recommended materials used in the 28 ohm characteristic impedance 4 layer PCB stack up is also different in terms of material thickness, and perhaps these lower volume, less familiar PCB fabrication methods might be the reason why mainboard and memory modules were so expensive back then.
I am not really an expert in PCB design, but just wanted to bring this up.

Was my stuff useful?

Hmm.
I wasn't just referring to that,but I'm confused why you'd redo the full thing when the full schematic is already on github.

There is also a fairly complete Asus p2b in kicad. All it needs is traces and the diodes corrected.

Pretty much any Asus board from that era that leaked is converted.
Just need a board name, or somewhere to send 200mb of several hundred boards.

I am doing the PCB design project to gain experience for a more advanced project I intend to work on in the future.
Note that I have not started the 440BX Uniprocessor Customer Reference Schematics due to myself still working on the VIA VT5228C replica project.
440BX Uniprocessor Customer Reference Schematics will be done afterwards, so I have not duplicated the work, for now.
As for VT5228C replica, I have about 20 more custom symbols and 30 or so custom foot prints I need to finish before I feel like I can start the schematic entry.
Of course, PCB design will be done after the schematic entry.
I have a fairly strict rule of wanting to refer to the original specification for certain industry standards like ATX power connector pin names.
Today, I was able to find the cached version of the ATX power supply design guide.

https://web.archive.org/web/19981201200409/ht … ort.com/%7Eatx/

Wow, it was like going back on a time machine to a world more than 25 years ago.

As for collecting components to build the mainboard, I have almost everything except CK100 main clock generator, CKBF SDRAM clock generator, and Motorola MMFT3055EL.
I am still missing a few small connectors and 74AS08.
I still need one more ISA slot since I have only one on hand.
I do want a burgundy colored normal height D-SUB 25 pin parallel port connector, but have not been able to find it.
Black colored D-SUB 25 pin connector is not too hard to find, even today.
If I want to use a raised height version of the burgundy colored D-SUB 25 pin connector, several sellers on AliExpress have some inventory in China.
Other than that, I got almost all the components, including obscure connectors like Header 8 VRM connector and ITP connector, needed to fabricate the mainboard.

myne wrote on 2025-09-10, 09:11:

Hmm. I wasn't just referring to that,but I'm confused why you'd redo the full thing when the full schematic is already on github […]
Show full quote

Hmm.
I wasn't just referring to that,but I'm confused why you'd redo the full thing when the full schematic is already on github.

There is also a fairly complete Asus p2b in kicad. All it needs is traces and the diodes corrected.

Pretty much any Asus board from that era that leaked is converted.
Just need a board name, or somewhere to send 200mb of several hundred boards.