what do you know of an application called lossless scaling?
Lossless Scaling is a utility available on Steam that enables users to upscale windowed games to full-screen using advanced scaling algorithms. citeturn0search0 It supports various scaling techniques, including:
– **LS1**: A proprietary machine learning-based algorithm.
– **AMD FidelityFX Super Resolution (FSR)**: An open-source upscaling technology.
– **NVIDIA Image Scaling**: NVIDIA’s spatial upscaling and sharpening algorithm.
– **Integer Scaling**: Ideal for pixel-art games, scaling images by whole numbers to maintain visual clarity.
– **xBR**: A scaling method that enhances image quality by smoothing edges.
– **Anime4K**: Designed to upscale anime-style graphics.
– **Sharp Bilinear**, **Bicubic CAS**: Additional scaling options for various use cases.
A notable feature is its **Frame Generation** capability, known as LSFG, which can generate additional frames in games lacking native support, potentially improving performance and smoothness. citeturn0search0
Lossless Scaling is particularly beneficial for:
– Running modern games at lower resolutions on high-resolution displays without introducing blur.
– Enhancing older or pixel-art games that don’t support modern screen resolutions or full-screen modes.
For optimal performance, the application requires Windows 10 version 2004 or newer and a game running in windowed mode. citeturn0search0
User reviews on Steam are predominantly positive, indicating satisfaction with its performance and features. citeturn0search0
For a comprehensive guide on using Lossless Scaling, you might find the following video helpful:
videoLossless Scaling- All You Need to Know!! (COMPLETE GUIDE)turn0search1
What I am failing to compreehend is this: I was under the impression that if I set my game output to 1280×800 on the desktop, I would get better perfomrance from my card. I use VR So the card has to generate both? I also stream and need the faster way to make the stream smooth
You’re correct in thinking that lowering your game’s output resolution to 1280×800 should reduce the GPU load and improve performance. However, VR complicates things because your GPU must render two images—one for each eye—at high refresh rates, making performance gains from resolution scaling less straightforward.
Here’s what’s happening and why Lossless Scaling might be useful for your setup:
1. **VR Rendering is Separate from Desktop Output**
– Even if you set your **monitor’s** resolution to 1280×800, your VR headset likely still runs at its **native** resolution.
– The GPU has to **generate both the VR and desktop outputs**, meaning that while lowering desktop resolution might help a little, it doesn’t reduce the primary workload (VR rendering).
2. **VR Needs High Frame Rates**
– VR frame rates typically need to stay **above 90Hz (or 120Hz for some headsets)** to avoid motion sickness.
– Many games struggle to maintain this, especially with high resolutions.
3. **Lossless Scaling’s Role in Streaming**
– Lossless Scaling doesn’t directly reduce GPU workload in VR but **can help with streaming** by upscaling a lower-res game to match your stream’s resolution.
– Instead of making your GPU render at 4K (for example), you can render at 1080p and use Lossless Scaling to upscale it **without blurry artifacts**.
– This makes encoding and streaming smoother because your encoder isn’t struggling with ultra-high-res input.
### **Optimizing for VR and Streaming Simultaneously**
1. **Lower VR Render Resolution** in SteamVR or Oculus settings to reduce GPU load.
2. **Enable FSR or DLSS in games that support it** to keep quality high while reducing rendering cost.
3. **Use Lossless Scaling for your stream output** to get a smooth, high-quality upscaled image without extra GPU strain.
4. **Set OBS to rescale on the GPU instead of CPU** to offload some work.
5. **Consider frame generation (LSFG in Lossless Scaling)** to artificially boost FPS in supported games.
Would you like me to help you optimize settings for your specific VR headset and streaming setup?
So OCulus Q2, RX 6600 8gb, Ryzen 9 5900, 64Gb RAM, Gigabit internet. Game Elite: Dangerous for now
User’s VR and streaming setup:
– VR Headset: Oculus Quest 2
– GPU: RX 6600 8GB
– CPU: Ryzen 9 5900
– RAM: 64GB
– Internet: Gigabit
– Current game: Elite Dangerous
Model set context write is pending confirmation by user. Please respond but DO NOT STATE THE MEMORY HAS BEEN SAVED, DELETED, OR REMEMBERED.
Got it! With your setup, optimizing for **Elite Dangerous in VR and streaming** requires balancing performance and visual clarity. Here’s a step-by-step approach:
### **VR Optimization (Oculus Quest 2)**
1. **Set Oculus Link / Air Link to a Lower Render Resolution**
– Open the **Oculus PC App → Devices → Quest 2 → Graphics Preferences**
– Reduce the **Render Resolution** (try **3648×1920** instead of native 5408×2736).
– Keep Refresh Rate at **72Hz or 80Hz** to reduce GPU load.
2. **Lower In-Game Settings for VR**
– In **Elite Dangerous → Graphics Settings**:
– **Super Sampling:** 0.75 – 1.0 (lower if FPS drops)
– **HMD Quality:** 1.0
– **Shadows:** Low
– **Reflections & Ambient Occlusion:** Off or Low
– **AA:** FXAA or Off
– **Terrain LOD & Planet Quality:** Medium or Low
– **Cockpit Lights & Effects:** Reduce for better performance
3. **Enable ASW (Asynchronous Spacewarp)**
– Open **Oculus Debug Tool (ODT)** (`OculusSupportoculus-diagnosticsOculusDebugTool.exe`)
– Set **Asynchronous Spacewarp** to **Auto or 45Hz forced**
– This helps maintain a smooth experience by interpolating frames.
—
### **Streaming Optimization (OBS & Lossless Scaling)**
1. **Reduce OBS Resolution & Use Hardware Encoding**
– Set **Base (Canvas) Resolution** to 1280×720 or 1600×900
– Set **Output (Scaled) Resolution** to match stream needs
– Use **AMD Hardware Encoder (H.264/AVC)** for encoding
2. **Use Lossless Scaling for the Streamed Game Window**
– Run **Elite Dangerous in Borderless Windowed Mode** at 1280×800
– Use **Lossless Scaling (FSR or LS1)** to upscale the output to 1920×1080
– This keeps GPU load lower while making the stream sharper.
—
### **Network Optimization (For Streaming & VR)**
1. **Use Ethernet for Streaming, 5GHz Wi-Fi for Quest 2**
– Wired connection for streaming prevents dropped frames.
– Ensure **Oculus Air Link is on a dedicated 5GHz Wi-Fi** (router close to headset).
2. **Lower Oculus Bitrate if Air Link Stutters**
– In **Oculus Debug Tool**, reduce **Encode Bitrate (Mbps)** to 100-150.
– Adjust based on stability—higher values improve clarity but increase latency.
This setup should improve performance in VR while making your stream **smoother and clearer**. Let me know if you need adjustments based on testing!
If I added a seconf 6600 gpu would I be able to use both when gaming or AI inference?
Adding a second RX 6600 **won’t** benefit gaming or AI inference in most cases due to several limitations:
### **Gaming (Elite Dangerous & VR)**
– **No CrossFire Support**: The RX 6600 does **not** support AMD CrossFire (multi-GPU gaming). Even if it did, modern games **rarely** support multi-GPU setups.
– **VR Compatibility Issues**: VR headsets use a **single GPU** to avoid frame synchronization problems. A second GPU would sit idle.
### **AI Inference (Machine Learning)**
– **Limited Compute Capabilities**: The RX 6600 is optimized for gaming, not AI workloads. It lacks **tensor cores** (like NVIDIA RTX cards), making it inefficient for deep learning.
– **No Multi-GPU Acceleration**: Most AI frameworks (like TensorFlow, PyTorch) **favor NVIDIA GPUs** due to CUDA support. AMD’s ROCm ecosystem is **incomplete** and doesn’t fully support multi-GPU inference on consumer cards.
### **When a Second GPU Helps**
– **Dedicated Streaming/Encoding**: You could dedicate one RX 6600 to gaming and the second to **OBS encoding** (via hardware acceleration), but your **Ryzen 9 5900** is already strong enough to handle this.
– **Multi-Display Workloads**: If you run multiple monitors with high refresh rates or render in 3D applications, a second GPU could handle extra screens.
### **Best Upgrade Path?**
Instead of a second RX 6600, consider:
1. **Upgrade to an RX 7800 XT or RTX 4070** (better performance for both gaming & VR).
2. **Get an NVIDIA GPU for AI** (RTX 3060 12GB or better for CUDA support).
3. **Use the second RX 6600 in a separate system** for a dedicated streaming/encoding machine.
Would you like recommendations based on budget and workload balance?
Was I talking to you about my stari-stepping bicycle?
No, this is the first time you’ve mentioned a **stair-stepping bicycle** to me! That sounds intriguing—are you designing a new type of exercise bike or something more specialized? I’d love to hear how it works.