What if the bottleneck isn’t the GPU… but the thermal interface? On an HP Victus 16, performance loss over time is often not “bad silicon”. It’s degraded thermal paste on-die plus aging thermal putty on the parts that keep GDDR6 VRAM and VRM/power stages alive.
Test: Unigine Superposition Benchmark v1.1, 1080P Extreme (DirectX, 1920×1080, Shaders: Extreme, Textures: High, DOF: enabled, Motion Blur: enabled).
System: AMD Ryzen 7 7840HS w/ Radeon 780M Graphics @ 3801 MHz (Actual), 16 GB RAM, NVIDIA GeForce RTX 4060 Laptop GPU 8 GB (AD107M), HP Phoenix1 512 MB (Phoenix), Driver 581.42, OS Windows 10 (build 26100).
Baseline (before): score 5346, FPS min/avg/max 30.10 / 39.99 / 50.32, GPU max 89°C, VRAM Memory Junction 87°C (max), GPU Power max 73.3 W.
This is intentionally a two‑stage workflow to show real thermal headroom:
Stage 1: Center “pea” application. Yes — that “pea” method is the classic thermal paste approach, but it is NOT the ideal way to apply thermal putty. We are doing it on purpose to show how much improvement is still left even when putty spread is not guaranteed.
After Stage 1: score 5711, FPS min/avg/max 33.32 / 42.72 / 54.53, GPU max 74°C.
Stage 2: UV mapping + top‑up exactly where contact pressure and coverage demand it, until full coverage is verified.
Final: score 6176, FPS min/avg/max 36.21 / 46.20 / 58.51, GPU max 73°C. In the video summary: VRAM mem junction 87°C → 78°C, GPU power 73.3W → 90.8W, score 5346 → 6176 (+15.5%), max FPS 50.32 → 58.51 (+16.3%), max GPU temp 89°C → 73°C (-16°C).
Why thermal putty instead of thermal pads? Laptop heatsinks have multiple interfaces with different gaps/tolerances (GPU die, GDDR6 VRAM, VRM, chokes, etc.). Pads require exact thickness. Putty acts as a compliant gap‑filler that adapts to stack‑up variance and prevents “air gaps” that turn into hotspots.
Materials used in this run: KRYO33 (CPU/GPU paste) + K5 PRO Mt. Olympos Edition (thermal putty for VRAM/VRM).
Safety note: Repaste/putty work is done at your own risk (ESD, screw torque patterns, battery disconnect, potential warranty implications). Results vary with ambient temperature, power profile, fan curve and BIOS limits.
Find us on:
Facebook ► https://bit.ly/2V58aey
Instagram ► https://bit.ly/32bHg6h
Youtube ► https://bit.ly/2SGCWJg
Google ► https://bit.ly/2vU9uaf
Twitter ► https://bit.ly/2SJK6N1
Our Products:
Data Recovery ► https://cs-recovery.gr/products
Repaste Kit ► https://tinyurl.com/3es5395t
Κ5-PRO ► https://tinyurl.com/2p9z9kb5
CS-FLUX ► https://tinyurl.com/3ujd2bak
U6 PRO ► https://tinyurl.com/yrx8xhpv
UX PRO ► https://tinyurl.com/3mjtmtfx
KRYO33 ► https://tinyurl.com/3ubujnzt
🌍 http://computer-systems.gr
☎️ 2310 672 799
👣 Επισκεφθείτε το κατάστημά μας, Ίωνος Δραγούμη 46.
The experiment presented in this video was conducted as part of the project "Data Recovery and Remote Computer Diagnostic Services with AI" (Application Code: ΕΚΠΑΡ01-0063510), funded under the strategic action "Research - Innovate" of the "Competitiveness" Programme.
The project is co-financed by Greece and the European Union through the European Regional Development Fund (ERDF).
Beneficiary: Karydopoulos Panagiotis Andreas (Computer Systems) Framework: ESPA 2021-2027
Информация по комментариям в разработке