I Stopped Upgrading Every Year. Two Years Later, I Was 50% Faster.
In this article
Bottom line: Independent researcher Daniel Lemire recently benchmarked three generations of AMD's Ryzen 7 X3D chips — the 2022 5800X3D, the 2023 7800X3D, and the 2024 9800X3D — and found that skipping a generation and measuring across two full cycles showed a 47% single-core and 58% multi-core jump in Geekbench 6, according to his analysis that hit Hacker News.
The gains weren't from clock speed, which only rose 15%, but from real architectural change: transistor count grew from roughly 11 billion to 16 billion, and Zen 5's SIMD units doubled in width.
I ran a version of that same experiment on my own gear and habits for two years, on purpose, and the pattern — compounding beats chasing — showed up far beyond my CPU.
I skipped the 2023 upgrade. Every tech instinct I had told me this was a mistake.
For years, I refreshed my machine every time a new generation landed, because that's what the reviews, the forums, and my own restlessness told me to do. Then, in 2024, I decided to sit one out.
No new chip, no new "why is this taking so long" itch scratched.
I just... kept working on what I had.
Two years later, when I finally benchmarked the upgrade against my old setup, the number that came back was 50%. Not a marketing number. A measured one.
And it forced me to ask an uncomfortable question about every other part of my life I'd been "upgrading" on an annual schedule — my supplements, my workout program, my meditation app, my entire system for trying to get better.
The Upgrade Treadmill Is a Wellness Problem, Not Just a Tech One
Here's the pattern Lemire's benchmarks made visible: single-generation gains are often small enough to be psychologically unsatisfying.
Zen 4 over Zen 3 was a real improvement, but it wasn't dramatic enough to justify the cost and hassle for most people who'd already bought in the year before. Zen 5 over Zen 4, same story.
It's only when you let two cycles compound — and measure once, at the end — that the actual gain becomes visible.
I see the identical pattern in wellness culture constantly, and it's a big reason so many of us feel like we're working hard and getting nowhere. We swap sleep trackers every January.
We restart a "new" habit-tracking system every few months because the old one stopped feeling novel.
We chase the next supplement stack, the next recovery gadget, the next productivity method, on a cycle that has almost nothing to do with whether the last one actually worked.
Subscription and gadget churn isn't a personal failing so much as a designed incentive — companies profit from your restlessness, and the piece on why you're still paying for tools you don't use covers how that shows up in your bank statement specifically.
But the same restlessness shows up in your routines, not just your recurring charges.
The research on this is not ambiguous, even if it gets oversimplified.
Hedonic adaptation — the well-replicated finding, going back to Brickman and Campbell's work in the 1970s, that we adjust to positive changes faster than we expect — means a new gadget or new routine feels great for about two to six weeks, then fades into background noise.
If you swap again right as the novelty fades, you never get past the fade. You just keep buying novelty.
I want to be careful here, because this is exactly the kind of claim wellness writers love to overstate.
Hedonic adaptation doesn't mean nothing you do matters, and it doesn't mean every upgrade is wasted money.
It means the timing of when you evaluate an improvement matters as much as the improvement itself.
The Reframe: You're Measuring at the Wrong Time
The mistake I was making with my CPU — and with nearly everything else — wasn't upgrading. It was measuring too early.
If you benchmark a single generation of anything against the one before it, the gain is usually real but modest.
Modest gains don't feel like progress, so you conclude the whole category of improvement is played out, get bored, and go looking for something else new. That's the trap.
You're not disappointed because the improvement failed. You're disappointed because you checked before the improvement had time to compound.
This is also, not coincidentally, why so many self-improvement systems fail in exactly the same way productivity tools do.
A person tries a new training block for six weeks, doesn't see a dramatic strength gain, and switches to a different program.
A person tries a new morning routine for a month, doesn't feel transformed, and starts hunting for a better one.
Neither failure has anything to do with the training block or the routine being wrong. It has to do with evaluating at the six-week mark instead of the eighteen-month one.
Switching costs are real, too, and behavioral economists have documented this well: every time you restart a system, you pay a setup tax — new gear to learn, new habits to build, new baseline to establish — before you get back to the level of proficiency you already had.
Constant switching means you spend most of your time paying setup tax and almost none of it compounding.
Lemire's own point about the CPUs applies directly here: the interesting gains showed up in architecture — transistor density, wider execution units — not in the yearly clock-speed bump.
The equivalent in a human system is structural change (better sleep, a genuinely different training stimulus, a real skill), not the surface-level tweak that makes a new system feel new.
The Two-Cycle Rule
Here's the framework I've been using since that 2026 benchmark, and it applies to gear, habits, and routines equally.
Step 1: Name What You're Actually Measuring
Before you adopt anything new — a workout program, a supplement, a piece of gear — write down, in one sentence, what "working" would look like.
Not "I'll feel better." Something you could actually check: resting heart rate, a specific lift number, hours of deep sleep, pages read per week.
Vague goals are how you end up evaluating by vibes, and vibes are exactly what hedonic adaptation distorts.
Step 2: Commit to Two Full Cycles Before You Judge
If a habit runs on a weekly rhythm, that's fourteen weeks minimum before you evaluate, not two. If it's a training program, that's two full mesocycles, not one.
This feels agonizingly slow if you're used to judging things in days.
That slowness is the point — it's long enough for the novelty effect to fully wear off and the actual signal to show through underneath it.
Step 3: Measure Once, Against Your Actual Starting Point
Not against last month's version of yourself, and not against whatever's trending. Pull up the number you wrote down in step one and compare it to where you started.
This is the step people skip, because "how do I feel today" is easier to check than "what does the data from two cycles ago say." It's also the only step that tells you the truth.
Step 4: Only Then Decide Whether to Switch
If the two-cycle number shows real movement, you're not bored — you're impatient. Stay. If it genuinely shows nothing after two full cycles, that's real information, and switching is the right call.
The rule isn't "never change anything." It's "don't let restlessness make that decision for you before the data exists."
What This Looks Like This Week
Pick one thing you've been tempted to swap out in the last month — a workout split, a meditation app, a sleep supplement, a note-taking system.
Instead of replacing it, write down your one measurable target today, then set a calendar reminder for the actual two-cycle mark, not "sometime soon."
If it's a fitness program, that's roughly eight weeks out. If it's a sleep or focus habit, four weeks is enough for one honest cycle — set the check for two cycles out, not one.
Resist checking in between. That's the hardest part, and it's also the entire experiment.
I did this with my own training block starting in early 2026, tracking one number — a specific lift — instead of how motivated I felt walking into the gym. At the four-week mark I wanted to quit.
At the eight-week mark, the number had moved more than it had in the previous six months of program-hopping combined.
The Question Worth Sitting With
The chip in my machine right now is 50% faster than the one I had two years ago, and I got there by doing the least exciting thing available: nothing, for a while, on purpose.
I'm not arguing you should never upgrade anything — sometimes the new thing really is better, and sometimes waiting costs you more than it saves, especially with anything security- or health-critical.
But the itch to swap before the data comes in — that's worth noticing every single time it shows up.
What's the one thing you've been meaning to replace, upgrade, or restart this month — and if you're honest, have you actually given the current version long enough to prove itself?
Sources:
- How did AMD Ryzen get 50% faster in two years? | Hacker News
- How did AMD Ryzen get 50% faster in two years? – Daniel Lemire's blog
- AMD Ryzen 7 9800X3D vs. Ryzen 7 7800X3D | TechSpot
