In This Article
Short answer: The boldest call from this old leak diary aged the worst. AMD was declared finished at the high end of desktop CPUs, and Nvidia’s next flagship was ruled out before it shipped. Both were wrong. The quieter, more careful predictions, Kepler arriving as the GeForce 600 cards and Ivy Bridge running hot, are the ones that actually came true.
Somewhere in the archives of this site is an old PC hardware diary, the kind a forum regular used to keep when leaks and engineering samples were the only window into what the chip makers were cooking up. It is full of the things that made that scene fun: insider clock speeds, die-size estimates, roadmaps swiped from a slide deck, and the occasional jab at a rival outlet for getting it wrong. It is also, with the benefit of hindsight, a fascinating record of how confidently smart people can misread where hardware is going.
What makes it worth revisiting is not the gossip. It is the scorecard. Enough time has passed that every one of those bets can now be checked against the parts that actually came out. Some of the calls were sharp. A couple were spectacularly, memorably wrong. Sorting one from the other says more about reading hardware hype than any single rumor ever could.
The AMD call that aged the worst
The headline prediction, the one the author clearly believed and kept circling back to, was that AMD was about to walk away from high-performance desktop chips for good. The FX-8350 and its Piledriver siblings on the AM3+ socket were framed as a farewell. After this, the thinking went, AMD would build nothing but APUs, the low-power chips with graphics baked in, and the enthusiast desktop crown was Intel’s to keep.
There is even a line that has aged into pure comedy: buy the FX-8350, the author wrote, because someday it will be a valued historical artifact. The chip was supposedly the end of an era. AMD had run out of road at the top, and anyone who wanted real AMD desktop power should grab the last one while it lasted.
That is not what happened. After a long, lean stretch where the FX family genuinely did struggle to keep up, AMD came back with a clean-sheet design and took the fight straight back to the high end. The Ryzen processors, built on the new Zen architecture, closed the gap on instructions per clock, then stacked on cores until the conversation flipped entirely. The company that was written off was suddenly the one setting the pace on core counts. You can trace the whole turnaround through the history of AMD’s Ryzen line on the Zen architecture, which is about as far from a farewell as a comeback gets.
What the leakers got right, and wrong, about Nvidia
Nvidia’s Kepler generation gave the diary both its best moment and one of its worst. On the upside, the basic shape of the launch was nailed. Kepler was correctly tagged as a genuinely new architecture rather than a warmed-over Fermi, and the first cards were called as the GeForce 600 series rather than the 700s that some rival sites were predicting. The GTX 680, the chip that led that launch, landed pretty much as described.
Then came the miss. The diary insisted, flatly, that the successors to the GTX 670 and 680 would not be based on the big GK110 chip. A completely different new core, it promised. That turned out to be exactly backwards. The high-end Kepler flagships were built on GK110: first the GTX Titan, then the GTX 780 and 780 Ti. The neutral record of the GK110 flagships that became the GTX Titan and 780 is the cleanest possible rebuttal.
The strange part is that the diary contradicts itself a few sections later. There is a breathless teaser about a big Kepler gaming card, wider memory bus, more cores, a serious jump over the GTX 680, that describes the GK110 path almost perfectly. The author had effectively predicted the very chip he had just sworn would never exist.
Why early Ivy Bridge ran hot
Intel’s side of the diary holds up better, even if the reasoning is only half right. Ivy Bridge, the first 22nm part and Intel’s first tri-gate consumer chip, was flagged early as a hot runner that overclocked worse than the Sandy Bridge it replaced. That reputation was real. Plenty of owners watched their shiny new chip top out lower on air than the older one it was supposed to beat, and the advice to hang on to a strong Sandy Bridge looked smart in hindsight.
Where the diary missed was the cause. It pinned everything on an immature manufacturing process that could not hold the planned low voltage. There is some truth in that, but the explanation that stuck was simpler and stranger. Intel had switched to a cheaper paste between the chip and its metal lid, and that thin layer trapped heat. Enthusiasts eventually responded by prying the lid off and replacing the paste themselves, a risky ritual that became known as delidding. The symptom was called correctly. The diagnosis was incomplete.
A snapshot of how granular the leaks got
Beyond the big bets, the diary is a time capsule of just how detailed the leak scene was. There were specific launch windows for AMD’s Richland and Kabini chips, full spec sheets for an alleged Radeon flagship that was supposed to land around twenty percent ahead on compute, and confident claims that a dual-GPU Radeon card had been quietly cancelled because nobody could build a cooler big enough to tame it.
Read now, those entries are less about whether each date was right and more about the texture of the moment. Stream processor counts, memory bus widths, idle power figures down to the watt: the whole apparatus of a release was being mapped out months before anyone could buy the thing. A few of the calls landed. Plenty drifted as roadmaps shifted under them. That gap, between a leaked plan and a shipping product, is the entire lesson.
| The prediction | What actually happened | Verdict |
|---|---|---|
| The FX chips were AMD’s last real desktop CPUs | Ryzen, on the Zen architecture, put AMD back at the high end | Wrong |
| The GTX 670 and 680 successors would skip the GK110 chip | GK110 powered the GTX Titan, then the 780 and 780 Ti | Wrong |
| Kepler launches as a new architecture in the GeForce 600 series | It did, led by the GTX 680 | Right |
| Early Ivy Bridge runs hotter and overclocks worse than Sandy Bridge | Confirmed, though mostly down to the cheaper paste under the lid, not just the process | Mostly right |
What this archive teaches about reading hardware hype
Strip away the brand names and a pattern shows up. The predictions that held were modest and close at hand: a launch that was already in motion, a chip already in reviewers’ hands, a thermal quirk people were already grumbling about. The ones that blew up were sweeping verdicts about the future, declaring a company finished or ruling a product line out years ahead of time.
A few habits travel well from that era to whatever leak is making the rounds today:
- Trust the near term over the far term. A spec for a card in final testing is worth more than a confident claim about a chip two generations out.
- Be wary of anyone declaring a company dead. The boldest, most quotable take is usually the one most likely to embarrass its author later.
- Right symptom does not mean right cause. The diary nailed that Ivy Bridge ran hot, then blamed the wrong thing entirely.
- Watch for a leaker arguing with himself. When the same source rules out a chip and then teases it a few paragraphs on, the conviction was never really there.
The takeaway
The fun of an old leak diary is not in catching the author out. It is in watching the industry refuse to behave. AMD was supposed to be done at the top and instead built one of its great comebacks. Nvidia’s big chip was ruled out and then shipped anyway, twice. The cautious, near-term calls held up fine. The grand ones did not.
That is the useful residue here, well after the parts themselves became museum pieces. Hardware rumors are at their most reliable when they describe something almost out the door and at their shakiest when they try to call the shape of the future. Read the next leak the same way, and an old diary full of misfires turns into a pretty good filter.
