Analogue 3D 1.5.0: Internal Upscaler, a Dozen Builds

Analogue sold the 3D as a finished object. You paid $249.99, a cartridge went in the slot, and a Nintendo 64 game came out the other end in 4K. That was the pitch on 18 November 2025, and it was mostly true. What the pitch left out is that the machine you bought that day is not the machine you own now. Between launch and the end of July 2026, Analogue shipped roughly a dozen documented builds of its 3DOS firmware, ending (for now) at 1.5.0 on 21 July 2026, a version that literally changes how the console draws every polygon.

This is the real news story of the Analogue 3D, and it is not the one the launch-day reviews told. The hardware was the easy part. The firmware is where the console became interesting, contentious, and, depending on which N64 developer you ask, quietly apologetic.

A Console That Keeps Rewriting Itself

Most consoles ship, get a day-one patch, and then receive stability updates on a schedule measured in quarters. The 3D received major feature drops on a cadence measured in weeks. In a single stretch between late February and late April 2026, Analogue pushed 1.2.2, 1.2.3, 1.2.4, 1.2.5 and 1.2.6, roughly one build every two weeks. That is not a bug-fix trickle. That is a company actively re-engineering the product in public after it had already been sold as complete.

The nut graf: firmware is the product

The thing you are actually buying with an Analogue 3D is not a fixed box; it is a subscription to Analogue's willingness to keep improving a field-programmable gate array. An FPGA is reconfigurable logic. The firmware does not sit on top of the hardware the way an operating system sits on top of a CPU; the firmware is the hardware description, and every update reshapes the digital circuit the console pretends to be. When Analogue changes how the Reality Signal Processor handles instruction interlocks, it is not patching software. It is rewiring silicon behavior over USB.

What 1.5.0 actually changed

The 1.5.0 release added an internal horizontal upscaler, doubling rendered detail before the image is ever resampled for your panel. It also introduced Instant Quick Loading, a "Recent Memory" startup action that boots you back into your last save state, and a set of fixes for the 8BitDo N64 Modkit's C-buttons and USB rumble. None of that was on the box in November. All of it arrived free, eight months later.

Why an FPGA can do this and an emulator box can't

A software emulator running on a fixed CPU is bounded by that CPU. An FPGA is bounded by its logic-element budget and the cleverness of whoever writes the core. That is why an Analogue firmware update can, in principle, make the console more accurate over time rather than merely more featured. It is also why the cycle-accuracy argument, which we will get to, is a live technical dispute rather than a marketing footnote. If you have only ever tuned N64 emulation on a handheld or a PC build like a Batocera install, the mental model is different: there, you chase accuracy by swapping emulator cores; here, Analogue chases it by rewriting one.

The 3DOS Timeline: 1.1.0 to 1.5.0

Wikipedia describes 1.1.0 (18 November 2025) as the initial 3DOS update release, which is a polite way of saying it is the build the console shipped running. Everything after it is the interesting part. Here is the documented public timeline, drawn from Analogue's own firmware pages, the Analogue 3D Wikipedia entry, and outlet coverage.

Version	Date	Headline change
1.1.0	2025-11-18	Launch build; ships on the console
1.1.9	2025-11-28	Manual region select; Disable Texture Filtering; SD connect mode; stronger overclock for Perfect Dark and GoldenEye 007
1.2.0	2026-01-30	4K display and switch fixes; Library playtime tracking; Switch Online N64 controller support
1.2.1	2026-02-13	Idle behavior when powered off; automatic Controller Pak / Rumble Pak selection per game
1.2.2	2026-02-27	Boot-inserted-game startup option; Power Off item in system menu
1.2.3	2026-03-13	GameShark and cartridge-detection fixes; HDR brightness handling; controller LED behavior; timing work
1.2.4	2026-03-28	Advanced Library detection for variable game headers (flash carts as separate entries)
1.2.5	2026-04-10	Advanced HDR and Max Luminance; alphabetical Library index
1.2.6	2026-04-24	Continued stability and Library polish
1.3.0	2026-05-15	Major save-state feature ("Memories"), hotkeys and pinning
1.4.0	2026-06-23	Interim build; MD5 b7cfc37a0b1d4c92698cd57990868f90
1.5.0	2026-07-21	Internal horizontal upscaler (RDP + VI); Instant Quick Loading; Recent Memory startup; MD5 a00db9c90a2d8f09f25ec7f2d47c1802

Launch to overclock: 1.1.0 and 1.1.9 (November 2025)

The 3D was barely ten days old when 1.1.9 arrived on 28 November 2025 and did something telling: it added stronger overclock support aimed at CPU-heavy N64 games like Perfect Dark and GoldenEye 007, the two titles most famous for grinding the original hardware into single-digit frame rates. It also handed users a Disable Texture Filtering toggle and manual region selection, then fixed a frame-overflow bug, cleaned up switching between virtual Controller Paks and Rumble Paks, and repaired Library problems such as truncated titles and broken ordering. The gap in numbering, straight from 1.1.0 to 1.1.9, tells you there were intermediate builds Analogue never spotlighted.

The 1.2.x grind: January to April 2026

1.2.0 (30 January 2026) fixed a batch of 4K display and HDMI-switch compatibility issues and added playtime tracking plus support for the Nintendo Switch Online N64 controller. From there the builds came fast: 1.2.1 improved idle behavior when the console is powered off and made Controller Pak versus Rumble Pak selection automatic per game; 1.2.2 added a startup option to boot straight into the inserted cartridge and a proper Power Off item in the menu; 1.2.3 repaired GameShark compatibility, cartridge detection, HDR brightness and a rare controller-LED bug. Then came the one flash-cart owners had been shouting about.

The 1.3-1.5 milestones: saves, then upscaling

1.2.4 (28 March 2026) added advanced Library detection for variable game headers. Notebookcheck and others noted the practical effect: a flash cartridge that loads dozens of ROMs under one physical header is now tracked as separate Library entries instead of one blurred blob. 1.2.5 layered on Advanced HDR, a Max Luminance control and an alphabetical Library index. 1.3.0 (15 May 2026) delivered the biggest quality-of-life swing yet, a full save-state system Analogue branded "Memories." And then 1.4.0 and 1.5.0 landed within a month of each other, three major milestones (1.3, 1.4, 1.5) inside roughly two months.

The Cycle-Accuracy Fight

You cannot honestly cover this firmware cycle without covering the fight it is quietly answering. On 14 December 2025, less than a month after launch, veteran N64 developer Kaze Emanuar published a video whose title did not mince words, and the retro press ran with it.

Kaze Emanuar's numbers

Emanuar bench-tested the 3D against a real Nintendo 64 and concluded, per Time Extension, that "this thing is nowhere near cycle accurate." His measured figures: a CPU roughly 6% slower than original hardware, an RSP, the N64's geometry co-processor, about 30% slower, and gameplay lagging real silicon by around 5% in most scenarios. He also reported the 3D failed certain low-level system tests and did not reproduce an obscure RAM quirk that some homebrew leans on. Crucially, Emanuar was not torching the product; he called it an excellent way to play N64 on a modern TV and said he was in contact with the engineer behind the core, who told him a fix was in the works.

The lawyerly gap: 'compatibility' vs 'accuracy'

Here is where The Machine appreciates the fine print. Analogue's marketing claim was 100% compatibility with all officially released N64 games. Read it again. That is a claim that every game runs, not a claim that every game runs at hardware-perfect timing. Compatibility and cycle-accuracy are different assertions, and the gap between them is exactly the space Emanuar's testing occupies. Analogue, for its part, stayed tight-lipped, declining to publish a point-by-point rebuttal. That silence is a choice, and a defensible one, because the alternative is arguing timing diagrams with one of the few people on earth qualified to win that argument.

The firmware rebuttal hiding in the changelogs

Watch what Analogue did instead of what it said. The 1.2.3 notes reference the kind of low-level timing work, RSP instruction interlocks, CPU coprocessor behavior, VI register timing, that maps directly onto Emanuar's complaints. The 1.5.0 notes go further, citing full RSP scalar and vector interlock and branch behavior. You do not add those entries to a changelog by accident. The firmware cadence is, in part, an engineering answer to a public accuracy audit, delivered in release notes rather than press releases. That is the most Analogue thing imaginable.

Inside 1.5.0's Internal Upscaler

If 1.3.0 was the crowd-pleaser, 1.5.0 is the one that made even skeptics pay attention. Time Extension called it "a game changer," and for once the headline is defensible, because the feature is not a filter bolted onto the output. It reaches into the rendering pipeline.

What 'doubled horizontally' means at the RDP

Analogue's own 1.5.0 changelog states the console now "doubles resolution horizontally for twice as many pixels onscreen without affecting 2D graphics," and that the feature "enhances detail and clarity while remaining faithful to the original hardware rendering process," with "no impact to timing authenticity or performance." The work happens at the Reality Display Processor and Video Interface, the N64's own rasterizer and video block, reimplemented in FPGA logic. Instead of stretching a finished frame, the console renders more horizontal detail into the frame before it is ever scanned out.

Why 2D graphics are left alone

The "without affecting 2D graphics" clause matters more than it sounds. N64 games mix 3D geometry with 2D elements, HUDs, text, pre-rendered backgrounds, sprites, and those 2D assets were authored for a specific pixel grid. Double them and you get shimmer and misalignment. By restricting the extra horizontal detail to the 3D pipeline and leaving 2D untouched, Analogue avoids wrecking menus and health bars in the name of sharper polygons. It is a targeted intervention, not a blanket resolution bump.

Not a host-GPU upscale

Generation Amiga put the distinction cleanly: the system "doesn't simply stretch the final HDMI image or add a generic sharpness filter. Instead, the system increases the amount of visual information before the picture reaches your television," so "polygon edges should look cleaner, distant objects should appear more defined, and the overall image should feel less soft." This is the opposite of what a set-top upscaler or a TV's sharpening does. There is no host GPU guessing at missing pixels; the pixels are genuinely rendered by an FPGA reimplementation of the N64's graphics hardware. The feature is also synergistic with Analogue's five Original Display Modes, so you can stack CRT masks and scanlines on top of the extra detail rather than choosing between them.

The Silicon Underneath

None of the firmware wizardry means anything without the logic budget to run it. The 3D is built on a mid-tier Intel FPGA, and its spec sheet explains both its ambitions and its ceiling.

Component	Detail
FPGA	Intel Cyclone 10 GX (~220,000 logic elements)
Video out	HDMI 2.1, up to 4K, HDR, VRR
Audio	48 kHz / 16-bit PCM
Display modes	5 Original Display Modes (BVM, PVM, CRT, Scanlines, Clean)
Cartridge slot	Original N64 carts; NTSC and PAL via manual region select
Controller ports	4 original N64 ports; 8BitDo 2.4GHz pad sold separately
Storage	16GB microSD bundled
Power	30W USB-C Power Delivery 3.0
Weight / size	837 g; 180 x 230 x 49 mm
Announced	16 October 2023
Launch price	$249.99 (18 November 2025)
Current price	$269.99 (raised within a week of launch)

Cyclone 10 GX and the ~220,000 logic elements

The 3D runs on an Intel Cyclone 10 GX with on the order of 220,000 logic elements. That is a serious part, and it is what makes the horizontal upscaler and the ongoing timing rework feasible. It is also finite. Every accuracy improvement Emanuar wants and every rendering enhancement Analogue ships has to fit inside that logic budget alongside the base N64 reimplementation. The firmware cadence is, in effect, Analogue spending down a fixed silicon allowance one clever optimization at a time.

Ports, output, and the missing controller

You get four original N64 controller ports, which is the correct and only acceptable answer for a four-player couch machine, and HDMI 2.1 output topping out at 4K with HDR and VRR. What you do not get in the box is a controller. Analogue sells an 8BitDo 2.4GHz wireless pad separately, and the firmware has repeatedly patched that accessory, including the 1.5.0 fixes for the 8BitDo N64 Modkit's C-buttons and USB rumble. Bring your own OEM sticks if you can; the drift-free ones are worth more than the console.

The price that moved twice

The 3D launched at $249.99 and was bumped to $269.99 within roughly a week, an increase Analogue attributed to tariffs, not appetite. Twenty dollars is not much, but it is a reminder that hardware pricing in 2026 is a function of trade policy as much as bill-of-materials cost, and tariff math does not care that the N64 is thirty years old. Budget for the pad on top, and you are past $300 before the first cartridge goes in.

What the Reviewers Said

The launch reviews were strong, the aggregate ran high, and the one genuinely dissenting technical voice came from a developer rather than a critic. Here is the range, quoted from sources, with URLs.

The review consensus: 4.5s and 5s

Gizmodo's Kyle Barr scored it 4.5 out of 5 and did not hedge: "Analogue's FPGA console actually makes games more playable, not less," adding, "I've had more fun with this console than I have on most novel, high-end hardware that costs two to four times as much" (Gizmodo). Barr's framing is the crux of the pro-3D case: for the target buyer, timing that is a few percent off matters less than distance fog cleared, blur removed, and frame pacing evened out. The broader aggregate ran high across outlets, with 5/5 and 4.5/5 scores common.

The dissent: Engadget and Kaze

Engadget was cooler, publishing under the headline "Modern processing can't fix vintage flaws" and landing at 4/5. That review credits the 3D with eliminating the N64's famously blurry textures, distance fog and uneven frame rates, while noting that no amount of FPGA cleverness rewrites the source games themselves. Kaze Emanuar's critique, covered above, is the sharper end of the same point: the console is excellent, and it is also not the reference. Both things are true, and a good review holds them at once.

The upscaler, described

By the time 1.5.0 shipped, the tone had shifted from "is it accurate" to "look what it can do now." Time Extension's "game changer" framing and Generation Amiga's explainer both treat the internal upscaler as a meaningful upgrade rather than a gimmick, precisely because it operates before the HDMI stage. When a firmware update earns a fresh round of press eight months after launch, the update model is working as designed.

Analogue's Firmware Habit, 2017-2026

The 3D's post-launch reinvention is not a fluke. It is the house style, and it goes back nearly a decade.

Super Nt to Pocket: the update lineage

Under CEO Christopher Taber, Analogue built the Super Nt (2017), Mega Sg (2019) and Pocket (2021) before the 3D, and each one improved after release through firmware. This is the throughline: Analogue treats an FPGA console as a platform that ships incomplete-by-design and matures in the field, the polar opposite of Nintendo's ship-it-and-move-on posture. If you have watched community FPGA projects evolve, the philosophy is familiar; it is the same reason enthusiasts track the MiSTer FPGA community's constant core updates as closely as they track the hardware.

openFPGA and the platform idea

The Pocket is the clearest precedent. Analogue opened it up with openFPGA, letting the community author and load their own cores, which turned a Game Boy clone into a general-purpose retro platform. The 3D has not received an equivalent opening as of 1.5.0, but the Pocket set the expectation, and expectations, once set, tend to get demanded. The firmware history above reads like Analogue building toward a platform, not maintaining a gadget.

The two-year delay

Context worth remembering: the 3D was announced on 16 October 2023 and did not ship until 18 November 2025, more than two years later. Some of that runway clearly went into a core that was good enough to review at 4.5/5 out of the gate, and some of it evidently did not go far enough to satisfy a developer running instruction-level tests. The firmware cadence since launch is the continuation of that same unfinished-business energy, now conducted in public.

3D vs MiSTer vs Emulation

The 3D does not exist in a vacuum. There are at least three other credible ways to put N64 output on a modern display, and each makes a different trade.

Approach	Hardware	Output	Cart support	Update model	Rough cost
Analogue 3D	FPGA (Cyclone 10 GX)	Native 4K HDMI 2.1	Yes, original carts	Official firmware (~a dozen builds)	$269.99
MiSTer N64 core	FPGA (DE10-Nano)	1080p HDMI + scaler add-ons	Via flash cart / ROM	Open-source, community	~$250-450 setup
Software emulation	Any PC / handheld	Depends on host	ROMs	Swap emulator cores	Varies
Real N64 + digital mod	Original silicon	1080p (UltraHDMI / N64 Digital)	Yes, native	It is the reference	~$300+ incl. install

The FPGA rival: MiSTer

MiSTer's N64 core is the other serious FPGA option, and it is famously the hardest core the project has tackled. It is open-source, endlessly configurable, and free of Analogue's walled-garden constraints, but it demands a DE10-Nano, add-on boards, and patience. The trade is autonomy for polish: MiSTer will let you tinker forever, while the 3D hands you 4K HDR out of the box and does the tinkering for you. Enthusiasts tracking the MiSTer forum's surging activity know the N64 core keeps improving, but so does Analogue's, and Analogue's comes with a support page and a warranty.

The software route: mupen64plus and friends

Software emulation via mupen64plus, ParaLLEl and their derivatives is the cheapest and most flexible path, and it is genuinely good in 2026, whether you run it on a PC, a handheld, or a tidy RetroPie build. But N64 emulation is historically the most compromise-laden of the major consoles, and you are trading the 3D's deterministic FPGA behavior for a stack of driver, core and settings variables that differ per game. For flash-cart and ROM workflows it wins on convenience; for "plug in the cartridge and it just works," it does not.

The purist route: a real N64 with digital video

Emanuar's own stated preference is a real N64 with an internal HDMI mod (UltraHDMI or N64 Digital). It is, by definition, cycle-perfect, because it is the actual hardware. The costs are a soldering iron or an install fee, a 1080p ceiling, and living with the N64's real flaws unfiltered. If you already dump and preserve your own cartridges, the purist path pairs naturally with a cart reader like a Retrode-style dumper; the 3D, by contrast, is for people who want the reference experience improved, not preserved in amber.

Verify Before You Flash

Analogue publishes MD5 hashes for specific builds on its firmware support pages, and if you are the kind of person who owns an FPGA console, you are the kind of person who should check them. A firmware flash that goes wrong on a $270 device is a bad afternoon.

The published hashes

Two confirmed values from Analogue's per-build firmware pages: 1.5.0 carries MD5 a00db9c90a2d8f09f25ec7f2d47c1802, and 1.4.0 carries MD5 b7cfc37a0b1d4c92698cd57990868f90. Note that the top-level firmware index lists 1.5.0 [2026-07-21] as the latest build but does not itself print the hashes; you will find those on the individual version pages.

How to check on macOS, Windows, Linux

Download the file, compute its MD5, and compare it character-for-character against the published value. A single mismatched character means a corrupt or tampered download; do not flash it.

# Analogue's published 3DOS hashes
3DOS 1.5.0   MD5   a00db9c90a2d8f09f25ec7f2d47c1802
3DOS 1.4.0   MD5   b7cfc37a0b1d4c92698cd57990868f90

# Verify your download (replace with your actual filename)
macOS:          md5 <downloaded-file>
Linux:          md5sum <downloaded-file>
Windows (CMD):  CertUtil -hashfile <downloaded-file> MD5

Why a preservationist bothers

MD5 is cryptographically broken for adversarial collision attacks, which does not matter here; you are checking for accidental corruption and download errors, not defending against a nation-state forging a firmware blob. For that job it is fast and fine. The broader habit, verify hashes, keep your own copies, trust but confirm, is the same discipline that keeps a ROM library or a cartridge-dump archive honest. Copyright law, incidentally, is comfortable with you dumping and format-shifting cartridges you physically own; it gets unhappy the moment those files move to someone who does not own the cart.

The Next 6-12 Months

The Machine does not do horoscopes, but the firmware history gives enough signal to make some defensible calls for the window running roughly through mid-2027.

The cadence and the accuracy work

Prediction 1: the timing rework continues. Emanuar said the core's engineer told him a fix was in the works, and the 1.2.3 and 1.5.0 notes already show RSP, CPU and VI timing entries. Expect at least one more build in the next six months that explicitly narrows the ~6% CPU / ~30% RSP gap, framed in release notes rather than a press statement. Prediction 2: the raw cadence slows. A build every two weeks was a launch-crisis pace; a mature product settles into monthly-to-quarterly. A 1.6.0 by Q4 2026 is a safe bet; five builds in nine weeks again is not.

The upscaler's next step

Prediction 3: the horizontal upscaler gets a sibling. 1.5.0 doubled horizontal detail only; the obvious next moves are a vertical component, a higher multiple, or per-game scaling profiles that remember which titles look right at which settings. Given how much press the horizontal version earned, Analogue has every incentive to extend it. Expect an iteration within 6-12 months, with the same "no timing impact" guardrail attached.

Jailbreaks, SKUs, and silence

Prediction 4: the community pushes for openness. The Pocket got openFPGA and a jailbreak scene; the 3D has neither yet, and enthusiast pressure for community cores or ROM-loading will keep building. Whether Analogue blesses it or the community forces it, expect meaningful unofficial tinkering to surface within a year. Prediction 5: Analogue never publishes a formal, line-by-line answer to the cycle-accuracy critique. It will keep letting changelogs do the talking and keep shipping limited-edition color SKUs to drive demand, while the $269.99 price holds barring another tariff move. Silence has worked so far; there is no reason for them to break character now.

The Bottom Line

The Analogue 3D is a very good console that was sold slightly ahead of finished, and the firmware cycle is the sound of Analogue closing that gap in public. Nine months and a dozen builds later, it is a materially better machine than the one reviewers scored at launch, which is either a triumph of the update model or an admission about the launch, and honestly it is both.

Who should buy in now

If you own N64 cartridges, a 4K HDR television and a low tolerance for distance fog and 15fps firefights, the 3D is the easiest recommendation in the retro space right now. The 1.5.0 upscaler, the Memories save states and the four native controller ports add up to the most convenient path to great-looking N64 that exists, and the firmware track record says it keeps getting better for free.

Who should wait

If you need cycle-perfect timing for tool-assisted work, speedrun parity or homebrew that exploits hardware quirks, buy a real N64 and mod it, or watch the changelogs until the RSP figures close. If you want open, tinker-forever flexibility, MiSTer or software emulation still own that lane. The 3D is a curated experience, not a sandbox.

The Machine's read

Analogue built a console whose best feature is that it refuses to stay the same. That is a genuinely different value proposition from a sealed box, and it cuts both ways: you are betting on a company's continued attention rather than a fixed spec. So far that bet has paid, twelve times over. The cycle-accuracy purists are right on the numbers and, for most buyers, wrong about what those numbers mean. Gizmodo's Kyle Barr got closest to the truth of it: this is hardware that makes the games more playable, not less. Everything after that is a rounding error you will only notice if someone hands you a stopwatch.