In the tech blog digest for September 23rd, a Habr article flashed by with a bold subheading: "The Turbo Button Never Actually Sped Up Your Computer." What caught my attention wasn't the revelation itself — many people have heard about the Turbo paradox. What caught me was the phrasing: a button that three generations of users believed made things "faster" actually always did the exact opposite. It's perfect meme-archaeology: an artifact whose purpose and name diverge by 180 degrees, and yet this artifact sat on millions of cases worldwide for over ten years.
I decided to dig not head-on, but sideways: where did the 4.77 MHz mechanic itself come from, who first slapped the proud name "Turbo" on a slowdown function, and what did that seven-segment frequency display actually show — the one you couldn't miss in any computer club of the nineties.
It all started in August 1981, when IBM released its first personal computer — model 5150. The heart of the machine was an Intel 8088 processor running at 4.77 MHz. The number itself is strange — not round, not pretty. The reason for its appearance is prosaic: the 8088 ran off a bus with a clock frequency of 14.318 MHz (the same one that divided by three for the processor, while also servicing the CGA color video adapter), and 14.318 / 3 ≈ 4.77. Saving on quartz crystals — that's all the engineering elegance there was.
The problem wasn't the frequency, but how software used it. Early eighties developers perceived the IBM PC not as a long-lived platform, but as a specific hardware product. Game and utility logic tied directly to processor speed: however many cycles — that many frames, however many loops — that many milliseconds of delay. Nobody imagined that in a couple years the same architecture would be running three times faster.
Clones started overclocking almost immediately. Compaq and other manufacturers offered the same thing but cheaper, then went further — Intel 8086 processors at 8 and 10 MHz, meaning two to three times faster than the original. And then the catastrophe revealed itself: old games on new machines became unplayable. Characters ran at light speed, no human could physically react to obstacles. The first case in history where a too-fast computer turned out to be a defect.
The first solution to the problem was found before the button appeared in its famous form. Accelerator cards for old machines had a mechanical toggle on the back panel: flip it — and the computer drops to compatible 4.77 MHz. Some clones made the switching software-based, through combinations like Ctrl+Alt+Plus. But all this was labeled honestly: "fast mode" and "slow, compatible."
The marketing revolution was achieved by California's Eagle Computer in summer 1984. In the Eagle PC Turbo line, a button with the proud label "Turbo" appeared on the front panel. It worked like this: pressing it slowed the machine from 8 MHz to 4.77. Fast mode was the unpressed default state. The press appreciated the irony immediately — PC Magazine in December 1984 wrote that the machine was "so fast that Eagle had to add a button to reduce speed for compatibility purposes."
Why specifically "Turbo"? The word was already the decade's main marketing spell — from Borland's Turbo Pascal to turbocharged cars. The logic is flawless: what self-respecting company would put a big button labeled "Slow" on a top product? Eagle understood that you need to sell not a compatibility switch, but a speed button — simply because fast mode is already enabled by default. The result was an oxymoron disguised as a feature.
The idea went viral. By 1988, the Turbo button stood on virtually every IBM-compatible computer case — all manufacturers of cheap clones copied it, and the word "Turbo" became the industry standard for the function of… slowing down.
By the early nineties, when frequencies reached 33, 66, and 100 MHz, the button became mandatory: without it, you couldn't run decade-old games on new hardware. And that's when the iconic accessory appeared on cases — that two-digit red LED display with the "current processor frequency." Press Turbo — the numbers changed from 66 to 33. It looked futuristic and gave the owner a sense of control over the machine's power.
The display had one flaw, though: it measured nothing. Inside the module there wasn't a single sensor — only seven-segment indicators, a resistor, and rows of jumpers. Each segment corresponded to a group of contacts, and the jumper determined whether that segment lit up always or at a certain Turbo button position. Want it to show 66 with "turbo" and 33 with the button off? Set the jumpers according to the table in the manual. Want 99? Go ahead. Photos still circulate online of systems with "66" indicators on machines whose processors ran at 25.
The implementation variation was wild. The cheapest modules could only show LO and HI — and even that was hardwired, not measured. Some specimens were "locked" to a single number and didn't connect to the button at all — they just glowed. Brand manufacturers — IBM, Compaq, Dell, Packard Bell — rarely installed such displays; this was the realm of nameless OEM cases and, of course, the domestic market. In CIS radio markets, a system without orange digits was considered incomplete, even though inside might be a processor with a frequency completely unrelated to what glowed on the facade.
A curious detail I didn't expect to encounter: the indicators outlived their own button. With the arrival of Pentium, the Turbo function on new boards almost disappeared, but the modules on cases remained — they were now powered from the Power LED circuit of the motherboard, and they continued to proudly display a pre-programmed number on machines where there was nothing left to switch.
The most non-obvious part of the story — what happened at the twilight of the era. Developers learned over time: new software started measuring clock frequency and adjusting delays to it, and the practical need for the button faded. But the legion of old programs tied to cycles didn't go anywhere — and here it turned out that hardware slowdown didn't solve everything.
A classic example — Runtime Error 200 in Turbo Pascal 7 programs. The Crt module at startup of each program calibrated the Delay function: counted the number of loop iterations in 55 milliseconds of timer interrupt, then divided that number by 55. The counter was 16-bit — on processors faster than ~200 MHz, the division result didn't fit in a word, the processor threw "divide overflow," and Borland's runtime dutifully reported to the user the cryptic error "division by zero." The program simply refused to start. This was treated with patches to executable files, utilities like Mo'Slo and CPUKILLER — software slowdowners that came to replace the hardware button. The irony: hardware became so fast that software written to slow down hardware itself broke from speed.
By 2000, the Turbo button died out completely. Cheap cases shed the extra mechanics, ATX displaced AT along with all its attributes, and real "turbo modes" appeared later — already as honest overclocking through BIOS and overclock buttons on motherboards. Over twenty years passed before the word "Turbo" on a PC case once again meant what it promised.
Strictly speaking, calling the story a "deception" isn't entirely fair — and that's the most interesting rough edge in the whole topic. The button performed a real, critically important function: without it, a huge layer of software was unplayable. The debate about "did it speed up or slow down" still lives on retrocomputing forums, and a significant portion of users confidently remember that Turbo did exactly what the label promised. Technically both sides are right: on most machines the accelerated mode was the default state, but nobody standardized the button wiring — on some systems pressing meant "enable turbo," on others — "disable." A user pressing the button at random got the opposite effect half the time, with no way to know except by launching a game and watching the hero's speed.
A separate inconsistency I honestly couldn't close: sources disagree on how deliberate the display fakery was. The Habr breakdown of boards and manuals shows jumpers in all their glory — and that's an iron fact. But how widely did assemblers consciously inflate numbers, selling a 25-megahertz machine with a "66" label? There's no systematic data, only anecdotal evidence. I can assume that most often the numbers were simply forgotten to be reprogrammed after an upgrade — but there's probably no way to verify this anymore.
The most capacious legacy of the Turbo button — a lesson in how marketing rewrites the meaning of an engineering function. If an honest label "Compatibility Slowdown" had appeared on the case, nobody would have poked at it or remembered it. But the oxymoron with a proud name lived fifteen years, spawned an iconic visual image of the era — red digits on the facade of a beige box — and still surfaces in discussions thirty years after its death. Few buttons have been granted such posthumous fame.