RigPressWE TEARAPARTHARDWARESO YOU DON'T.
Benchmarks. Build guides. Opinions with receipts.
TESTED
IN-HOUSE
NO ADS
Why another hardware blog exists in 2026
Every hardware review reads like a spec sheet rewritten by an intern who's never touched a heatsink.
You've seen them. The review that spends six paragraphs explaining what PCIe 5.0 is, then buries the actual gaming numbers in a table at the bottom. The "build guide" that recommends parts based on affiliate commission, not component quality. The benchmark that tests at 4K Ultra when you're gaming at 1440p on a monitor that cost $280.
We've all rage-quit a review tab mid-scroll. Closed fifteen browser tabs. Ended up asking on Reddit anyway, only to get "it depends" from someone with a four-year-old flair. The information exists. It's just buried under ad impressions and editorial guidelines from brands that also happen to be advertisers.
"The GPU that 'wins' in a synthetic benchmark is rarely the one that wins in your living room at 11pm with a game you actually want to play."
— From Issue 03: The Synthetic Benchmark Problem
How we actually do this.
One apartment. One power strip. Six GPUs and a thermal camera that we didn't tell the landlord about.
We Buy It Ourselves
Every component tested at RigPress is purchased at retail, on the same sites you use, at the same prices you pay. No review units, no pre-production samples, no NDA embargoes that conveniently expire after launch-day hype peaks.
One Apartment, One Power Strip
Our test bench lives in a 700 sq ft apartment in Chicago. The ambient temperature is tracked. The power draw is logged. We test at 1080p, 1440p, and 4K because we know most of you aren't on a $3,000 display.
Six GPUs, Minimum
Every review includes at least three direct competitors and two generational comparisons. We answer 'should I upgrade from my 3070?' not just 'how fast is the new card?' The question you're actually asking is the question we answer.
The bench. Feb 2026. That's an RTX 5080 FE on the left, a 9070 XT on the right, and a borrowed oscilloscope in the back that we definitely know how to use.
logged 1.412V stable on the 9950X after 72hrs of Prime95 — not a single WHEA error. that's the number that matters.
— Bench log, Jan 14 2026
Sample the writing.
RTX 5080 vs RX 9070 XT: The $700 Question Nobody Is Asking Correctly
The benchmark sheets will tell you the 5080 wins. They're not wrong. They're just answering a question most of you aren't asking. The real question — the one that matters if you're gaming at 1440p on a 165Hz display you bought two years ago — is whether the 18% average performance lead over the 9070 XT justifies a $280 premium. Spoiler: at the settings you actually use, in the titles you actually play, that gap compresses to 11%. At that compression ratio, the math stops working for Nvidia. We ran both cards through 14 titles over 40 hours. The results are not what the press kit suggested...
"At 1440p/165Hz, the performance gap compresses to 11%. The math stops working for Nvidia."
1440p / Max Settings — Avg FPS
Cyberpunk 2077 RT Overdrive · i9-14900K · 32GB DDR5
Recent benchmark dispatches.
RTX 5080 vs RX 9070 XT: The $700 Question Nobody Is Asking Correctly
40 hours. 14 titles. One apartment. The performance gap at 1440p is not what Nvidia's press kit suggested.
7800X3D at $280: Still the Build Anchor for 2026 Mid-Range Gaming
Two years after launch. Four competitors tested. The answer hasn't changed — but the reasons have gotten more interesting.
PSU Tier List 2026: Why Your "850W Gold" Might Be Lying to You
Eleven power supplies. A kill-a-watt meter. And some genuinely embarrassing test results from brands with good marketing budgets.
Stable at 1.4V: Our 9950X Extreme OC Log
72 hours. 1,200 WHEA events. Three thermal paste applications. One stable result.
DDR5-8000 vs DDR5-6000: The Gaming Delta Is Smaller Than You Think
Eight kits tested. The sweet spot is not where the spec sheet says it is.
Mini-ITX in 2026: The Case for Going Small When Everything Runs Hot
Four cases. Two AM5 builds. One lesson about thermals in 6.5L of aluminum.
Free. No paywall. No account required.
One email. Every Tuesday.
No noise, just signal.
The week's best benchmark findings, one build recommendation, and whatever opinion we couldn't fit into the main article. Read by 84,000 people who take their hardware seriously.
No tracking pixels. Unsubscribe with one click. We've been burned by bad newsletters too.
"written every monday night, usually after the benchmarks finish at 2am"