Video summary

Valve Steam Machine Review: GPU & CPU Benchmarks, SteamOS Test, Thermals, Noise, and Price

Main summary

Key takeaways

Product Review

Product Reviewed

Valve Steam Machine — a SteamOS-based living-room / console-style PC, including:

  • Performance testing on the stock setup
  • Commentary-driven testing of front-plate / fan-duct modifications (called “Downpour” steam vents)

Key Features Mentioned

Two SKUs / Pricing & Reservation

  • Reservation starts

    • $1,50 for the 512 GB model (Subtitles appear to omit a “1” or show formatting errors; other mentions suggest an ~$1,350-ish order total range for 2TB.)

  • $1,350 for the 2 TB model

  • Purchase system uses a randomized reservation queue to reduce scalping advantages
  • Controller bundles are mentioned but not detailed

Hardware Platform (Semi-Custom / Power-Limited)

CPU

  • 6-core / 12-thread Zen 4
  • Advertised boost up to 4.8 GHz
  • 30W TDP — described as the major limiter

GPU

  • RDNA 3
  • 28 compute units
  • Max advertised sustained clock: 2.45 GHz
  • 110W TDP
  • 8 GB GDDR6

RAM

  • 16 GB DDR5-5600
  • Shipped as either:
    • single 16 GB stick, or
    • two 8 GB sticks (supply-dependent)

Storage / I/O

  • 512 GB or 2 TB NVMe SSD
  • Ports:
    • 4× USB-A
    • 1× USB-C
    • microSD
    • HDMI
    • DisplayPort
    • 1 GbE Ethernet (stated as 1 GB/s)
  • No discrete audio outputs mentioned beyond display / USB usage

Audio Output Limitation

  • No dedicated audio ports; audio via HDMI / DisplayPort / USB only

SteamOS & Software

  • SteamOS 3.8
    • Compatible with some non-Valve hardware as long as it’s not Nvidia
    • Linux Nvidia driver support is described as problematic
  • Uses Proton
  • Includes Deck-related UI/compat mechanisms (often described as messy; see cons)

Performance Results (GPU-Bound Gaming)

General Positioning vs Windows-Style Baselines

The Steam Machine is described as performing closest to:

  • RX 6600
  • Intel i5-? B570 (subtitles show “Intel B570”)
  • RTX 3060

Additional notes:

  • Occasional match: RX 7600 (one instance)
  • Often ahead of older/lesser cards:
    • RX 7600 and RTX 5050 described as “most immediately ahead” modern options
  • It’s “not particularly powerful” primarily because Valve power-limits the system (PSU capacity + thermals)

Representative Game Benchmarks (1080p-ish)

Where applicable, settings are described as “Ultra/Max.”

  • Resident Evil 4: 94 FPS average

    • Leads RTX 2070 by ~6%
    • Leads RTX 2060 by ~33%
    • Behind RX 7600 (~112 FPS)

  • Black Myth: Wukong: 43.9 FPS average (SteamOS)

    • Compared to RX 7600: 47 FPS

  • Starfield (1080p Ultra): 41 FPS average

    • ~8% ahead of RX 6600

  • Oblivion Remastered (1080p Ultra, no upscaling): 27 FPS average

    • ~comparable to RX 7600
    • Steam Machine only ~1.9% better

  • Kingdom Come: Deliverance 2: 50 FPS average

  • Dragon’s Dogma 2: 52 FPS average

  • Cyberpunk 2077: Phantom Liberty (1080p Ultra): 59 FPS average

    • RTX 3060 slightly behind
    • RX 7600 ~68 FPS (+16.6% note)

Ray Tracing

Native ray tracing is described as largely not playable.

  • Cyberpunk RT (medium): 21 FPS average

    • After a last-minute Valve patch: 20.7 → 22.8 FPS (~+10%)

  • Another example:

    • Black Myth: Wukong RT: 13.8 → 16.7 FPS (~+21%)
    • Still very low

CPU / Computational Benchmarks

General CPU-Class Alignment

Performance is described as aligning with mid-generation Ryzen-class parts:

  • Stellar simulation: around Ryzen 7 3700X / Ryzen 5 3600 level
  • 7-Zip compression: 55,000 MIPS
    • close to R5 3600 (just below; ~7% behind)
  • Decompression testing: near Ryzen 5 2600 level

Blender (Cycles)

  • Positioned between:
    • 3700X and 5600X3D, with subtitles summarizing “broadly ahead of 5600X”
  • Example render time:
    • 24.4 minutes per frame, tied with R5 5600X / 5600X3D class

Sustained CPU Behavior

  • Clocks under sustained load described as:
    • mixed initially
    • then dropping to a steady state
  • Likely intended thermal limiting behavior

Thermals & Noise (User Experience)

Stock Behavior (Solid Front Plate)

GPU Thermals (Auto Fan ~1252 RPM)

  • Edge: 42°C over ambient
  • Junction: 57°C over ambient
  • VRAM: 46°C over ambient
  • Noise: 23.5 dBA at 1 meter

Fan Behavior

  • Idle is described as effectively inaudible in the reviewer’s acoustic chamber setup (noise-floor level for first minutes)

CPU Thermals

  • CPU runs: ~50°C over ambient (about 70°C with ambient)
  • Described as “fine”

“Downpour” Steam Vent Plate / Modifications (3D-Printed)

  • They tested many plate/mod variations:
    • different vent patterns
    • dust filters
    • magnets
  • Claimed improvements:
    • thermals improved roughly +6 to 7°C (lower temps) depending on configuration
  • Example normalized at 1252 RPM:
    • Downpour improves GPU/VRM thermals by several degrees

Trade-Off: Loudness at Noise-Normalized RPM

  • Downpour can be louder when normalized by RPM:
    • Example: at similar RPM, downpour measured ~27 dBA vs 23.5 dBA stock
    • Attributed to noise optimization trade-offs

Dust Filtration

  • Downpour plates include a dust filter
  • Plate airflow/thermal outcomes depend on design choices

Practical Heat-Flow Observations (Recirculation Risk)

  • Exhaust airflow appears to form a cone that jets away quickly
  • Warning: keep clearance from the wall so warm air doesn’t ricochet back
  • Carpet is a concern:
    • reducing intake air (e.g., blocking the bottom) worsens thermals, including SSD temps

Power Consumption

  • Gaming loops: ~173–176W during runs (drops between loops)
  • Gaming range: ~170–180W
    • low 200s for full torture load, with the caveat that it drops after startup
  • Idle power:
    • ~13W for SteamOS desktop during their workload “lead-up”
    • standby: < 1W
  • Conclusion:
    • PSU choice and Valve’s limits constrain headroom
    • possible “extra” headroom only around 40–50W before concerns

Repairability / Upgrades

  • Some upgrade paths exist:
    • RAM and SSD
  • Beyond that:
    • CPU/GPU are not socketable
    • limited internal mod headroom
    • no OC / TDP bypass mentioned

SteamOS User Experience & Issues (Major Cons)

Controller-First Interface

  • Keyboard/mouse UX is awkward in gaming mode
  • Menus show controller glyphs
  • Clicking text boxes triggers on-screen keyboard, even without controller

Limited BIOS

  • No memory speed/timing controls in BIOS
  • No overclocking or TDP/PPT-like bypass options

Game Compatibility / Linux Quirks

  • Some anti-cheat failures:
    • F1 games unable to run due to anti-cheat incompatibility
  • Steam Deck auto-detection artifacts:
    • per-game toggle: “disable Steam Deck auto detection”
    • some games mis-handle features (ray tracing disabled or wrong Proton choices)
  • Resolution cap confusion:
    • games default capped at 1920×1080 (even on a 4K display)
    • reviewer notes Valve says 1080p is the system default for verified testing, and UI updates may come

Overall Software Tone

  • Reviewer generally likes SteamOS direction and Proton improvements
  • Criticizes Valve’s desktop execution as messy/unpolished

Comparisons vs Competitors / DIY Value

DIY vs Steam Machine Pricing (Paper Specs Caveat)

  • The reviewer compares DIY using “paper specs,” but stresses it’s not a fair like-for-like:
    • Valve parts are semi-custom
    • and power-limited

Reported premium over DIY:

  • 512 GB: $71 premium (~7.3%)
  • 2 TB: $211 premium (~18.5%)

But real-world performance is described as much worse for Steam Machine due to:

  • heavy power limiting
  • semi-custom constraints

Example Competitor

  • HP Omen 16L (Ryzen 5 8400F + RX 7600 + 16GB + 512GB) at $1,350
  • Steam Machine 512GB described as $300 cheaper than HP’s comparable option (though “power unbound” wording is unclear in subtitles)

Value Conclusion

  • Not called “massively overpriced,” but not a performance value savior
  • Often described as not worth it for pure FPS/$ versus DIY in current conditions

Quantified Thermals / Noise Highlights (Unique Numeric Claims)

Noise (Stock, Auto RPM)

  • 23.5 dBA @ 1m at ~1252 RPM

Noise (Stock Without Front Panel)

  • 37 dBA at ~1164 RPM
  • About a 5–6° improvement, but much louder (described as “room to improve”)

Noise (Downpour)

  • Example at noise-normalized RPM: 27 dBA vs 23.5 dBA stock
  • Max fan:
    • Solid plate @ 2000 RPM: 35.4 dBA and 42°C junction
    • Downpour @ 2000 RPM: 38–38.7 dBA, with less “wind whooing noise”

GPU Temps at Solid Panel Auto

  • Edge/junction/VRAM: 42 / 57 / 46°C over ambient

SSD Thermals

  • Stock plate SSD temp: 9.9°C over ambient
  • With plate at 1252 RPM: ~12.9°C over ambient
    • “worse by a few degrees,” but far under limits

Pros (As Stated Across the Review)

  • Very quiet at idle (effectively inaudible)
  • Good thermals for its size; stock not described as thermally throttling severely with the solid plate
  • SteamOS/Proton are improving, enabling Linux gaming to work better over time
  • Good direction in software/driver behavior (when it works, better pacing/consistency)
  • Prebuilt living-room console form factor, easy entry for non-DIY users
  • Some repair/maintenance potential (iFixit parts availability mentioned generally)

Cons (As Stated Across the Review)

  • Performance limited by power/TDP constraints, not competitive on FPS/$
  • RAM configuration uncertainty:
    • single 16GB stick vs dual 8GB sticks
    • single-stick may reduce performance via memory channeling
  • SteamOS messy for desktop use:
    • controller-first UI even with mouse/keyboard
    • confusing resolution cap (default 1080p full screen)
    • workarounds required for Steam Deck detection and Linux feature handling
  • Some games fail due to anti-cheat (F1 cited)
  • Ethernet is 1 GbE (called out as a cost-down vs more common 2.5GbE)
  • Audio limitations: no dedicated audio ports; must use HDMI/DP/USB
  • Limited upgradability beyond RAM/SSD; no meaningful OC headroom

Overall Verdict / Recommendation

  • Recommended for: people who want a small, prebuilt living-room gaming PC, value SteamOS/Proton and low noise, and/or enjoy tinkering with SteamOS/mods.

  • Not recommended for: buyers seeking the best gaming performance per dollar. The review concludes it’s not a performance value winner due to strict power limiting and weaker real-world performance vs comparable DIY parts.

  • Net message: “Looks good overall, but isn’t a savior”—best justified by the fun/unique SteamOS console-like experience, not maximum FPS/$.

Speaker-Specific / Contributor Viewpoints (From Subtitles)

  • Review/testing host (main voice)

    • Deep dives into GPU/CPU benchmarks, thermals, noise measurements, power draw, and SteamOS usability
    • Emphasis on airflow engineering and trade-offs between front panels

  • “Downpour / plates” creator or store/promo voice (integrated throughout)

    • Detailed description of 3D-printed vent plates (design rationale, dust filter, colors, 7-year warranty)
    • Claims improved thermals (often cooler but sometimes louder)
    • Includes legal/practical justification for making accessories

  • Valve policy / purchasing & pricing narrator (within the review)

    • Explains Valve’s pricing/Q system
    • Scalping mitigation via randomized queue
    • SteamOS expansion stance

Original video