NVIDIA Tests 700W+ RTX 5090 Ti Prototype
🏠#cuda-cores#high-tdp#blackwell-dieFreshcollected in 4h

NVIDIA Tests 700W+ RTX 5090 Ti Prototype

PostLinkedIn
🏠Read original on IT之家

💡700W+ prototype reveals NVIDIA's power-hungry GPU future for AI training rigs.

⚡ 30-Second TL;DR

What changed

22,848 CUDA cores, 5% more than RTX 5090 but short of full GB202's 24,576

Why it matters

This highlights diminishing returns in GPU scaling, pushing AI infrastructure toward efficiency over raw power. High TDP could strain data center power supplies, influencing procurement decisions for training clusters.

What to do next

Benchmark your current RTX 5090 setup's power limits before planning Blackwell upgrades.

Who should care:Developers & AI Engineers

🧠 Deep Insight

Web-grounded analysis with 6 cited sources.

🔑 Key Takeaways

  • NVIDIA has been testing RTX 5090 Ti/Blackwell Titan prototype since mid-2025, with 22,848 CUDA cores (5% more than RTX 5090's 21,760), not using full GB202 die (24,064-24,576 cores).[1][2][3]
  • Power draw is 700-750W under normal conditions, exceeding 1000W in unlocked prototypes, requiring massive cooling solutions.[1][2]
  • Performance uplift is modest at 10-20% over RTX 5090, potentially due to higher clocks despite fewer cores than pro variants like RTX 6000.[1][2][3]
📊 Competitor Analysis▸ Show
FeatureNVIDIA RTX 5090 Ti/Blackwell TitanAMD Equivalent (Speculative)Intel Arc (Current Flagship)
CUDA Cores22,848N/A (RDNA 4 ~20,000+ est.)N/A (Xe-cores)
Power (TDP)700-750W~400-600W est.~300-450W
Performance vs 5090+10-20%N/A (no direct competitor)Significantly lower

🛠️ Technical Deep Dive

  • Uses modified GB202 chip, missing one GPC compared to full die in RTX 6000 (24,064 cores, 128MB L2 cache); RTX 5090 has 21,760 cores, 96MB L2.[1][3][4]
  • Blackwell architecture: 4th-gen RT cores, 5th-gen Tensor cores, Multi Frame Generation (MFG) as key new feature over Ada Lovelace.[2][5]
  • Expected 32GB GDDR7 on 512-bit bus (same as 5090), PCIe 5.0, DisplayPort 2.1b/HDMI 2.1a; potential dual-channel DDR5 IO limitations.[3][5]
  • Higher power enables modest gains via clock speeds, not core count; pro variants like RTX 6000 show limited gaming uplift despite more cores.[2]

🔮 Future ImplicationsAI analysis grounded in cited sources

High power demands (700W+) challenge PSU/cooling standards, may limit mainstream adoption; modest gains question value at premium pricing ($2500+), positioning as enthusiast/AI hybrid until RTX 60-series (2028); reflects NVIDIA's strategy amid no direct gaming competition.[1][2][6]

⏳ Timeline

2025-01
RTX 50-series (5070/5080/5090) announced at CES and released, based on Blackwell architecture with up to 21,760 cores and 32GB GDDR7.
2025-06
NVIDIA begins testing RTX 5090 Ti/Blackwell Titan prototypes.
2025-12
Driver 591.44 restores 32-bit PhysX support; reports of ROP anomalies in some RTX 50 cards resolved.

📎 Sources (6)

Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.

  1. notebookcheck.net
  2. tech4gamers.com
  3. youtube.com
  4. pcgamer.com
  5. en.wikipedia.org
  6. tomshardware.com

NVIDIA has tested a prototype GPU dubbed RTX 5090 Ti or Blackwell Titan since mid-2025, featuring 22,848 CUDA cores but not the full GB202 chip. Power draw reaches 700-750W normally and over 1000W unlocked, requiring massive cooling. Performance is only 10-20% above RTX 5090, and release is unlikely.

Key Points

  • 1.22,848 CUDA cores, 5% more than RTX 5090 but short of full GB202's 24,576
  • 2.Normal power 700-750W, unlocked prototype exceeds 1000W
  • 3.Requires enormous cooler, described as 'handheld with both hands'
  • 4.Performance gain ~10% typical, up to 20% in ideal conditions
  • 5.Prototype exists but likely never to release like past Titans

Impact Analysis

This highlights diminishing returns in GPU scaling, pushing AI infrastructure toward efficiency over raw power. High TDP could strain data center power supplies, influencing procurement decisions for training clusters.

Technical Details

Uses partial GB202 die closest to RTX PRO 6000's 24,064 cores. Compared to MSI RTX 5090 Lightning at 1000W yielding 13% FPS uplift in games.

📰

Weekly AI Recap

Read this week's curated digest of top AI events →

👉Read Next

AI-curated news aggregator. All content rights belong to original publishers.
Original source: IT之家