惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

T
Tenable Blog
H
Heimdal Security Blog
K
Kaspersky official blog
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
S
Schneier on Security
G
GRAHAM CLULEY
U
Unit 42
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
C
CERT Recently Published Vulnerability Notes
Google DeepMind News
Google DeepMind News
罗磊的独立博客
Stack Overflow Blog
Stack Overflow Blog
阮一峰的网络日志
阮一峰的网络日志
Simon Willison's Weblog
Simon Willison's Weblog
C
Cisco Blogs
Cyberwarzone
Cyberwarzone
T
The Exploit Database - CXSecurity.com
Project Zero
Project Zero
Security Archives - TechRepublic
Security Archives - TechRepublic
www.infosecurity-magazine.com
www.infosecurity-magazine.com
博客园 - 司徒正美
Exploit-DB.com RSS Feed
Exploit-DB.com RSS Feed
V
Visual Studio Blog
博客园 - Franky
Engineering at Meta
Engineering at Meta
WordPress大学
WordPress大学
Jina AI
Jina AI
P
Proofpoint News Feed
P
Proofpoint News Feed
有赞技术团队
有赞技术团队
L
LINUX DO - 最新话题
宝玉的分享
宝玉的分享
N
News and Events Feed by Topic
cs.CV updates on arXiv.org
cs.CV updates on arXiv.org
博客园 - 聂微东
T
The Blog of Author Tim Ferriss
Spread Privacy
Spread Privacy
Application and Cybersecurity Blog
Application and Cybersecurity Blog
IT之家
IT之家
S
Security Affairs
博客园 - 叶小钗
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
小众软件
小众软件
N
News | PayPal Newsroom
Cloudbric
Cloudbric
AWS News Blog
AWS News Blog
W
WeLiveSecurity
The Last Watchdog
The Last Watchdog
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
NISL@THU
NISL@THU

LinuxJedi's /dev/null

When an Amiga A570 Repair Took a Strange Turn When the Docs Fall Short: Investigating Claude Code’s Budget Cap AI Didn’t Destroy Your Company, Your Processes Did Ditching the Debug Probe: Using a Segger J-Link with a Raspberry Pi Pico When PAL Goes Wrong: Repairing an Amiga CD32 Reviving a Roland SoundBrush: Floppy MIDI Playback Without the Computer Reviving an Amiga 600: From Dead Video to a Clean Boot The Amiga 1200 That Fought Back: The Faults I Missed the First Time Why Recapping Isn’t Always the Cure: And Amiga 1200 Repair Story
A Socket 7 Upgrade: Moving Beyond the 486
LinuxJedi · 2025-12-29 · via LinuxJedi's /dev/null

A while back I built a 486-based machine to play with some late 80s / early 90s games, as well as try out hardware of the era. I blogged about the process here, with an update here. I really enjoyed using this setup, but I wondered if I could improve it enough to run Quake well. A new journey began…

Requirements

For this upgrade, there were a few things I wanted:

  • It should be an original Pentium-class machine, ideally MMX generation
  • It should use a PCI graphics card, 2D is fine, this is going to be a DOS machine
  • It should support the sound setup I was using previously (two ISA cards with Roland sound modules connected)
  • It should use SCSI, just as my current 486 setup does

I set myself a total budget of £75 and searched for the parts needed. For spares / repairs was fine, but I’d inspect photos first to make sure it is something worth repairing.

Parts Found

Whilst going through Socket 7 motherboards, I found this motherboard/CPU combo. It is a bit overkill, the CPU is an AMD K6-2 400MHz. It is still Pentium MMX class, just a bit faster.

The Retro Web has lots of information on this board, including hidden settings.

I need a heatsink and fan for this, a Socket 370 one would technically work, but they typically have stronger springs that can break Socket 7 hooks. So, I found this one:

Finally, a graphics card. I may upgrade this later, but this is what I initially found that kept me well within budget:

I’m at £66.98, a little under-budget. I have everything else I need here already, so I’m crossing fingers that I can get it all working.

Building and testing

The motherboard arrived, and the CPU in it is an AMD K6-2/400AHX. This is a slightly unusual variant because many of that era operated at 2.2V. In fact, I couldn’t find many examples at all of a 400AHX when searching around.

I compared the jumpers on the motherboard to those on the documentation.

The CPU frequency was set to 6x 66MHz (2x and 6x are the same setting). This is correct when 100MHz is not available. But the voltage switches seemed “wrong”, they were set to 3.0 V. I changed these to 2.4 V straight away. I’m glad I checked these before powering up.

Now for testing. I popped on a blob of thermal paste and added the heatsink. One thing I didn’t buy was RAM. But that is fine. I have many SIMMs available, so popped 2x 32MB 60ns EDO SIMMs into the board. I added the video card and a POST diagnostics card, just in case it didn’t boot.

I nervously hit the power button. No video appeared and the POST card stopped on the following code:

0B is the current POST code, 0A is the previous. This code means “Test CMOS RAM Checksum”. Which makes sense, it was put in storage by the previous owner, the battery is likely dead.

I popped in a new CR2032 battery and did a CMOS reset sequence (flipping a jumper for 10 seconds and then putting it back again).

This time, it actually booted!

I dug through a pile of SCSI cards and found an Adaptec 16160 PCI card. Again, probably a large overkill, but it will work with my BlueSCSIv2 and it is what I had spare. From here I was able to run some tests:

Everything here looks correct. I haven’t tweaked the BIOS settings at this point, and I know the SIMMs are set to 70 ns by default. I should at least be able to tweak this.

Rebuilding the Machine

Now that I know the board works, I pulled my 486 board out of my case, along with the ATX -> AT power converter. This new board supports AT and ATX power, so I figured I’d use ATX. This also meant swapping the power button back to a motherboard soft power switch instead of a latching switch on the ATX -> AT converter to turn on the machine.

The new motherboard went in, along with the PCI video card and SCSI card. I also put in my two sound cards and serial port card. The motherboard has onboard serial ports, but I do not have cables for these.

I then hooked up the buttons, LEDs and speaker. It definitely isn’t tidy for now, but it functions.

The heatsink is a little small compared to even a Pentium 3 one. So, after a few minutes, I checked it with a thermal camera to make sure it wasn’t getting too hot.

It is 35°C at the hottest point, in fact the buck converters providing the CPU power were running hotter. So, nothing to worry about there.

Benchmarking

Now that the system is built, let’s run some benchmarks. I used Phil’s Computer Lab benchmarks for this. First of all, Doom in Faster PC mode:

How about Quake?

Pretty snappy! Finally some benchmarks that aren’t quite as useful…

I fully suspect that the S3 card is probably the bottleneck more than the CPU is. But I have a very nice DOS build here, that I’ll be using for several projects in the future.