Gaming On Nvidia's TINY Super Computer...

SUMMARY

Dawid from Dawid Does Tech Stuff unboxes and tests the Nvidia Jetson Orin Nano, attempting to install OSes and game on this ARM-based AI supercomputer despite significant technical hurdles.

STATEMENTS

The Nvidia Jetson Orin Nano is an affordable ARM-based system integrating an Ampere GPU for AI tasks like generative computing, robotics, and object recognition.
Its modular design features a swappable System on Module (SoM) resembling a RAM stick, allowing potential future upgrades.
The device includes practical I/O such as USB ports, M.2 slots for storage, and MIPI CSI connectors for external cameras.
Initial attempts to boot a standard Windows install on the NVMe drive fail due to ARM architecture incompatibility with x86 software.
Changing BIOS settings without proper guidance can brick the system, requiring hardware resets or replacements.
Nvidia's official Linux installation uses a microSD card flashed with JetPack OS via Balena Etcher, avoiding NVMe complexities.
Installing x86 Steam on ARM Linux proves impossible without advanced emulation tools like Box64 and Wine64, leading to emulation alternatives.
RPCS3 emulator enables PS2 games like Jak and Daxter to run at playable but inconsistent frame rates on the Jetson.
PS3 emulation, such as for GTA 5, fails due to crashes after prolonged shader compilation on the slow SD card boot.
Enabling ACPI in BIOS allows USB booting for Windows 11 ARM, but lacks official Nvidia drivers, resulting in posting failures.

IDEAS

Describing the Jetson Orin Nano as an "AI Raspberry Pi" highlights its compact, hobbyist-friendly form factor for advanced computing.
Marketing it solely as a generative AI tool overlooks more engaging applications like robotics, where object recognition shines.
The SoM's heat-sinked RAM-like appearance suggests clever engineering for easy upgrades in embedded systems.
Booting OS from SD card feels archaic, amplifying slowness and frustration in modern hardware setups.
Nvidia's installation flowcharts embody engineer precision, contrasting with user-friendly consumer guides.
Trial-and-error BIOS tweaks reveal how non-standard hardware punishes uninformed experimentation.
Community YouTubers like Mr B simplify convoluted processes, proving accessible tutorials outperform official docs for beginners.
Attempting x86 software on ARM equates to forcing incompatible shapes, underscoring architecture's invisible barriers.
Emulation via RPCS3 transforms ARM AI hardware into a retro gaming machine, bridging old consoles to new silicon.
Frame rates to two decimal places in emulators expose the absurdity of precision in low-performance scenarios.
Controller incompatibilities add layers of hassle, turning simple gaming into multi-step configuration puzzles.
Repurposing AI-focused GPUs for gaming uncovers untapped potential but exposes optimization gaps for entertainment.

INSIGHTS

Specialized hardware like the Jetson excels in intended AI roles but demands creative workarounds for unintended uses like gaming, revealing flexibility's limits.
Clear, step-by-step community instructions transform overwhelming technical setups into achievable tasks, emphasizing guidance over raw intelligence.
Modular designs in tiny computers promote sustainability by enabling upgrades, challenging disposable electronics norms.
ARM architecture's efficiency for AI contrasts with its friction for x86 gaming, highlighting ecosystem silos in computing evolution.
Slow storage like SD cards exacerbates compilation times in emulators, illustrating how boot media choice amplifies hardware bottlenecks.
Frustration from crashes and inconsistencies in non-native environments underscores the value of official support for cross-platform ambitions.

QUOTES

"This box from AliExpress is very exciting because in it is one of Nvidia's new AI Raspberry Pies."
"I'm probably going to do the least practical thing with it possible, which is install Windows 11 on it."
"Why are there so many flowcharts? Engineers are animals."
"I was trying to smash a square peg into a round hole by repeatedly pseudoapping at it."
"I like that they give you the frame rate to two decimal points because you know that 4.24 frames pers is playable, but 4.02 is garbage."

HABITS

Persistently troubleshooting by searching Google, Reddit, and YouTube for guides before abandoning approaches.
Starting with simple, direct methods like plugging in drives or using app stores, escalating only when they fail.
Documenting setup failures in real-time to share authentic experiences and learn from mistakes.
Switching to proven alternatives, such as emulation or official OS paths, after repeated native install frustrations.
Unlocking maximum power modes early to test hardware limits under full load.

FACTS

The Jetson Orin Nano's SoC die, labeled TE960MA1, rivals the size of an RTX 5070 die while integrating both CPU and Ampere GPU.
It features two 4GB RAM modules for a total of 8GB, optimized for embedded AI workloads.
The system draws power from a compact 45W brick, enabling portable supercomputing in a small form factor.
Dual MIPI CSI connectors support direct external camera integration for real-time object recognition.
Nvidia dedicates a community webpage showcasing diverse projects, from robotics to AI experiments, beyond official marketing.

REFERENCES

Mr B's YouTube guide for Jetson Orin Nano OS installation using Nvidia JetPack.
Mr B's video on installing Steam via Box64 and Wine64 for ARM gaming.
Nvidia's community webpage highlighting user projects in robotics and AI applications.

HOW TO APPLY

Download the JetPack OS image from Nvidia's website and use Balena Etcher to flash it onto a microSD card, ensuring compatibility with the ARM architecture.
Carefully insert the microSD card into the Jetson Orin Nano's slot, avoiding damage to the delicate connector, then reassemble the device.
Connect peripherals like a monitor via DisplayPort, keyboard, and mouse, then power on the system to initiate the Linux boot process.
Complete the initial Linux setup wizard, adjusting language, network, and user accounts as prompted for basic functionality.
Install RPCS3 emulator by downloading the AppImage, enabling executable permissions in file properties, and running it to begin shader compilation.
Select and launch a PS2 game ISO within RPCS3, monitor frame rates, and adjust settings like resolution to 720p and low shaders for stability.

ONE-SENTENCE TAKEAWAY

Gaming on the Jetson Orin Nano via emulation succeeds modestly but demands patience amid ARM setup challenges.

RECOMMENDATIONS

Rely on video tutorials from creators like Mr B for ARM-specific installations to bypass official documentation overload.
Prioritize Linux over Windows for native support, avoiding driver voids that doom cross-OS experiments.
Embrace emulation for retro gaming on AI hardware, but prepare for long compilation times on SD storage.
Test power modes immediately to maximize GPU output, ensuring hardware isn't throttled during intensive tasks.
Consult community forums before BIOS changes to prevent bricking non-standard systems.

MEMO

In a cluttered AliExpress package arrived the Nvidia Jetson Orin Nano, a pint-sized powerhouse billed as the company's most affordable generative AI supercomputer. Tech enthusiast Dawid, host of Dawid Does Tech Stuff, couldn't resist repurposing it for the unlikeliest task: gaming. This ARM-based marvel, akin to a souped-up Raspberry Pi, packs an Ampere GPU derived from high-end graphics cards, promising AI feats like object recognition or robotic navigation. Yet Dawid eyed its gaming potential, armed only with curiosity and a 45W power brick. The modular design impressed—a swappable brain module slots in like laptop RAM, hinting at future-proofing in a world of planned obsolescence. I/O abounds: USB ports, dual M.2 slots for storage, even camera connectors for vision-based projects. But beneath the cute facade lay a beast: a die the size of an RTX 5070, flanked by 8GB of RAM.

Tearing it down revealed engineering elegance mixed with quirks. The SoC, etched as TE960MA1, hummed with potential, but booting proved nightmarish. Dawid's first stab at Windows 11 on an NVMe drive crashed spectacularly—ARM's incompatible with x86 norms, looping through "synchronous exemption" errors. BIOS tinkering only worsened it; a misplaced setting bricked the board, forcing a covert Amazon replacement after hours of futile resets. Nvidia's Linux path, via microSD, felt retro: flash JetPack OS with Balena Etcher, jam the card into a finicky slot, and pray. Flowcharts overwhelmed, but Mr B's idiot-proof guide saved the day. Post-setup, Ubuntu-like interface bloomed, though SD slowness grated—color shifts plagued the display, a minor Linux hiccup.

Gaming ambitions clashed with reality. Steam install? A farce. x86 binaries rebelled on ARM soil, unmet dependencies mocking terminal commands. Enter emulation: RPCS3 for PlayStation classics. Compiling shaders dragged eternally on that SD card, akin to a device in digital purgatory. Jak and Daxter's PS2 trilogy finally launched, hitting 60fps sporadically but dipping to stuttery lows—frame times jagged, playability tenuous. A DualShock 4 tamed controls, yet Jak 3 tanked to 16fps even at low settings. PS3's GTA 5? Crashes after exhaustive PPU builds. Precision to two decimals amused: 4.24fps "playable," 4.02 "garbage." Visual glitches screamed under-optimization, underscoring the Jetson's AI roots over gaming polish.

One last Windows push via ACPI-enabled BIOS teased success, only to reboot into oblivion—Nvidia offers no drivers, sealing ARM's isolation. Dawid's odyssey exposed the Jetson's charm: versatile for hobbyists in robotics or AI slop generation, but gaming demands masochistic patience. Community ingenuity bridges gaps, from Reddit fixes to emulation hacks, proving small hardware sparks big creativity. For tinkerers, it's a gateway to embedded wonders; for gamers, a cautionary pixelated tale. As AI surges, such devices blur lines between tools and toys, inviting experimentation despite the thorns.