"The Future Is Fusion"... is AMD right?


Recommended Posts

I've recently built a Media Center/Gaming PC to go with my new HDTV, and decided to give AMD's "Trinity" platform a chance with its A10 5700. At first, I didn't think this PC could do much better than run old games at low resolutions, but Trinity has blown me away. Dark Souls? 30-40fps at 1080p - better than on PS3. Trine 2? 45fps at 1080p, high settings. Recent, GPU-bound titles running at full HD, no significant compromises in graphical quality, and enjoyable framerates.

That an integrated GPU can offer what only a discrete solution could since hardware accelerated graphics began, is a revolution.

First let's look at the current state of things. For the most powerful all-around and specifically gaming machine, things are still as they have been for the past 8 years or so: the latest Intel CPU paired with a power-hungry graphics card with a monstrous cooler, either AMD Radeon or NVIDIA Geforce. Llano didn't change this, Trinity doesn't change this, and Richland won't change this either.

What AMD's APUs bring to the table, however, is the possibility of dropping the discrete card entirely, and for a fraction of the price, power and thermal envelope, getting still pretty darn good performance in most games, even at high resolutions. It's a significant tradeoff, sure, but one that can make sense for real gaming, something unthinkable just a year ago. For HTPCs, laptops, and basically every form factor where price, power draw, noise and heat are critical concerns, APUs are bound to make discrete graphics cards irrelevant very quickly.

As for desktop PCs where these considerations matter less, the future of discrete cards is not as clear as it was either. Of course, in their current iterations (Trinity/Richland), APUs cannot compete with any recent 100$+ graphics card, a relatively small investment for vastly superior performance. Yet, the two next-generation consoles, PS4 and Xbox One, are confirmed/rumored respectively to be AMD APUs, sporting 8GB of unified memory and performance in the 2 TFlops range. As absurd at this may sound, the next generation of consoles runs on integrated graphics, based on technology very similar to the 130$ Trinity running the HTPC in my living room. This makes my brain want to explode.

There are many advantages to this solution. Unified memory means more flexibility in how it is shared in each application, less overhead in GPU-CPU communication, and a vastly simplified programming model for developers. If AMD can make this work at state-of-the-art performance level on next-generation consoles, can't we hope for this to happen on desktop PC in the forseeable future?

AMD's slogan has been "The Future is Fusion" for nearly 5 years now, and I'm starting to believe it. Intel is quickly catching up in this territory as well, with Haswell promising performance similar to the Geforce GT 650M. It's an interesting take on things to see Intel playing catch-up on AMD when AMD has been largely perceived as the underdog in the past few years. Holding the CPU performance crown may be nice, but it never gave anyone playable framerates, and I'd venture to say that AMD CPUs are still fast enough for most uses in most circumstances. At least, that's what Intel is apparently thinking, with Haswell focusing essentially on power consumption (vs ARM) and graphics performance (vs AMD).

Oh yeah, and John Carmack seems to agree:

The current generation Fusion parts are really a separate CPU and separate GPU connected on the die by better or worse interconnects, but their vision is integrating them much more tightly such that they share cache hierarchies, address space, and page tables. I think its almost a foregone conclusion that its going to be the dominant architecture in the marketplace because there are these strong forces about how we are getting more shrinks on the dies, [and] we can stick more things on there. It?s going to just pay to integrate that, and you aren't going to be able to put as many transistors towards that if its a dedicated chip. You'll pick up a lot from this tight integration with cost benefits.
Link to comment
https://www.neowin.net/forum/topic/1153754-the-future-is-fusion-is-amd-right/
Share on other sites

Isn't that the new AMD APU they are going to put on the XBOX One?

Reports are that APUs in next-gen consoles are custom designs, based on Kaveri (perhaps?) which is Richland's successor. In any case it's much more advanced than what's currently on the market for PCs.

This being said (and true), I don't understand why Intel didn't pack Iris Pro in more of their upcoming CPUs. Just like it is with IB - the low end, where IGP is most likely to be used, got measly HD2500, but high end got HD4000, mostly *in vain*, because those people often opted for superpowered discrete GPU anyway.

Does Intel think that people will opt for more expensive knowing the difference? Some would. But they *don't* know the difference - they simply fully expect the new series to be better!

This being said (and true), I don't understand why Intel didn't pack Iris Pro in more of their upcoming CPUs. Just like it is with IB - the low end, where IGP is most likely to be used, got measly HD2500, but high end got HD4000, mostly *in vain*, because those people often opted for superpowered discrete GPU anyway.

Does Intel think that people will opt for more expensive knowing the difference? Some would. But they *don't* know the difference - they simply fully expect the new series to be better!

Well, the i3225 has the HD4000 on it.

Anyway, yes, I've always thought AMD was ahead of the game in this field, it's no surprise Sony and MS went with them for their new consoles.

I think the main drawback in this discussion for me would be "choice" and tied into a set solution. I like many others sway in the winds of change as the power heads of the technology world fight it out. Right now I have an Intel Machine with an Nvidia Graphics Card and could I live with a AMD CPU + Integrated future GPU? - sure! Do I want to have choice taken away from me? Answer: NO, does everyone think like me? Answer: NO

I think it is a good way to get a machine up and running with little cost but for some - this is just an extra cost and not a set future unless people love everything to be uniform.

This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
  • Posts

    • Motrix Next 3.9.6 by Razvan Serea Motrix Next is a modern, open-source cross-platform download manager built as the official next-generation successor to the original Motrix project. It has been completely rewritten using Tauri 2, Vue 3, TypeScript, and Rust, while still relying on the powerful Aria2 download engine for high-speed multi-protocol transfers. The app supports HTTP, HTTPS, FTP, BitTorrent, ED2K and magnet links, offering advanced features like multi-connection acceleration, task scheduling, bandwidth control, and batch download management. With a significantly reduced install size (around 20MB), it focuses on being lightweight, fast, and resource-efficient compared to traditional Electron-based download tools. Designed for Windows, macOS, and Linux, Motrix Next delivers a clean, modern UI inspired by Material Design 3 principles, with smooth animations and a minimal workflow. It improves usability through better download organization, system tray integration, and enhanced torrent handling including selective file downloads and tracker management. Motrix Next features: Multi-protocol downloads — HTTP, FTP, BitTorrent, Magnet, .torrent, ED2K, and Metalink tasks BitTorrent — Selective file download, DHT, peer exchange, encryption controls, metadata caching, GeoIP peer flags, and tracker probing Browser extension integration — Embedded Extension API with independent authentication, download confirmation, smart auto-submit, filename hints, referer/cookie forwarding, and real-time controls (Chrome Web Store · Edge Add-ons) Safe filename handling — Content-Disposition, RFC 2047, non-UTF-8, percent-encoded, and extensionless URL resolution with path traversal sanitization Download organization — Favorite and recent folders, optional file-type categorization, stale-record cleanup, and completed history backed by SQLite Concurrent downloads — Independent controls for active tasks, HTTP connections per server, segments per file, and BT peer limits Speed control — Global and per-task upload/download limits with day-of-week and time-of-day scheduling System integration — Tray operation, optional tray speed display, macOS Dock badge/progress, protocol handlers for magnet://, thunder://, and motrixnext:// Lightweight mode — Destroys the WebView on minimize-to-tray while Rust keeps the engine, task monitor, notifications, history, and extension routing alive Notifications and power options — Native task start/complete/failure notifications, keep-awake during downloads, and optional shutdown after completion Network controls — Scoped proxy support for downloads, app updates, and tracker updates, plus system proxy detection Auto-update channels — Stable, Beta, and Latest Across Channels policies with separate download and install phases Diagnostics — Structured logs, exportable diagnostic ZIPs, database integrity checks, automatic DB rebuild, and Linux GPU rendering fallback Personalization — Light/dark/system theme, 10 color schemes, 26 languages, and first-launch system language detection Motrix Next 3.9.6 changelog: New Features Clipboard management — App-owned copy actions no longer trigger the Add Task auto-detect popup. aria2 input compatibility — Multi-line aria2-style task input is supported for URLs with per-task options such as out=. BitTorrent IPv6 DHT — Added IPv6 DHT support and related configuration. File category URL patterns — File category rules can match URL patterns with validation and localized hints. Task status tags — Added clearer waiting and sharing states for task cards. Download event bridge — Added an aria2 WebSocket event bridge for faster download notifications. Improvements Improved task list transitions and preserved task state during tab switches. Kept RPC origin access enabled for local integrations. Restored AppImage stripping in release builds after beta validation. Added localized preference guidance across supported languages. Download: Motrix Next 64-bit | ARM64 | macOS ~20.0 MB (Open Source) Links: Website | macOS / Linux | Screenshot Get alerted to all of our Software updates on Twitter at @NeowinSoftware
    • Segra 1.6.2 by Razvan Serea Segra is a free, open-source OBS-powered game recorder offering fast gameplay capture, instant clips, AI highlights, deep game integration, and seamless uploads—perfect for gamers, streamers, and content creators. Lightweight, fast, zero bloat. Segra key features: Automatic Game Recording: Begin capturing gameplay the moment your game launches, with zero manual setup. Instant Clipping: Save important moments instantly using a customizable hotkey—perfect for highlights, montages, or quick shares. Segra AI Highlights: Let Segra automatically detect kills, assists, deaths, and key events to generate polished highlight reels without manual editing. Gameplay Uploads: Upload recordings and clips directly to Segra.tv for fast sharing and cloud access. Deep Game Integration: Enjoy advanced game-data tracking across hundreds of supported titles, enabling smart highlight generation and stat-informed clipping. High-Performance Capture: Record up to 4K at 144 FPS using OBS-powered technology with minimal performance impact, supporting NVENC, AMD VCE, and custom quality controls. Segra Editor: Edit recordings easily with timeline controls, segment management, and event-based navigation to build the perfect clip. Customization Options: Adjust hotkeys, output formats, storage paths, codecs, capture quality, and performance settings for a tailored recording experience. Segra 1.6.2 changelog: UI: Improved the transition from the loading skeleton to the real content card. Security: Added Segra.dll code signing and automatic VirusTotal upload. Settings: Fixed the settings header to highlight Account when scrolled to the top. Recording: Updated OBSKit.NET to 1.4.1. Download: Segra 1.6.2 | 74.5 MB (Open Source) View: Segra Homepage | Github | Screenshot Get alerted to all of our Software updates on Twitter at @NeowinSoftware
    • Hey Google, these are the Gemini features I want in 2026 by Aditya Tiwari Google Gemini has been around for over three years. The AI chatbot started its journey back in 2023 (as Bard) when ChatGPT was already a talk of the town. However, it quickly attracted criticism after misrepresenting facts about the James Webb Space Telescope. The search giant spent a year fine-tuning Bard before rebranding the chatbot and its underlying generative AI model to Gemini, drawing inspiration from NASA's first human spaceflight program. Note that Bard was initially powered by LaMDA and PaLM 2; Google has since added several new features and integrations to Gemini. That said, there is scope for improvement and a gap for new features. I have been using Gemini for a while now and have realized that the chatbot lacks several features, making it harder for me to research across topics. These are mostly function-over-form updates that can improve the overall experience. Delete individual messages from a conversation Image via DepositPhotos.com One good thing about Gemini is that it can maintain context throughout the conversation. But things might get chaotic when you want to ask a related question, but don't want it to be part of your conversation in the long run. You can't ask that related question in a fresh chat because Gemini will lose the active conversation context of what you're trying to research. If Google allowed you to delete individual question/answer pairs, you could simply ask about a sub-topic and remove it from the conversation to create a smooth flow of important stuff. Offline mode Image via DepositPhotos.com A big pain of using Gemini daily is that everything loads from the cloud. It takes time for your chats to appear, and you can't view your conversation history while offline. To get a better idea, you can open the Gemini app and see how it looks without an internet connection. While Gemini models run in the cloud, it wouldn't hurt if Google could store chats (at least the text part) on the device so we can refer to them when offline. Google can also offer a lightweight version of its AI model to help with basic drafting, summarization, and other tasks. It has the Gemini Nano model, which can perform on-device processing on Google Pixel, Samsung, and some other Android brands, but it's a system feature and not related to the cloud-based Gemini app. Make temporary chats permanent I can't thank Google enough for taking the time and effort to add incognito mode or temporary chat mode to the Gemini app. It lets you have conversations without worrying that the topics will end up in your chat history or used for model training (at least on paper). Google claims that it doesn't use your temporary chats to "personalize your Gemini experience or train Google’s AI models." However, the data is stored "up to 72 hours to respond to you and to process any feedback you choose to provide." That said, I often start researching something in a temporary chat, only to realize the chatbot's answer is good enough to refer to later. Sadly, Gemini doesn't have an option to make such temporary chats permanent. In other words, I won't be able to follow up on it if I close the temporary chat. I'm left with alternatives like copying the answers into notes or another app. My digital life will get a lot better if Gemini gets a button to make temporary chats permanent. Collapse answers for a cleaner view You're heavily invested in your research game and suddenly feel the need to go up in the chat to recall something. This is when the conversation thread starts to feel like an overwhelming, unending wall of questions and answers. What if Google added a way to collapse Q&A pairs in the Gemini chat thread? It would look quite clean and easy to navigate. You'll quickly get an overview of everything you have discussed with the chatbot. Add buttons to jump between messages Suggested mockup of the feature. This reminds me of a small but useful Gemini feature that Google could add to its chatbot: the ability to hop between prompts in a conversation. Just add simple up- and down-arrow buttons, similar to YouTube Shorts, so people can quickly scroll through the messages. A table of contents or Chat Overview It's hard to get a bird's-eye view of everything you have discussed with the chatbot during a lengthy conversation. This is where a table of contents, or Chat Overview, displayed at the top of the screen, possibly in a drop-down button, might come in handy. You'll be able to get an overview of the chat and jump between messages, serving as an alternative to the up/down arrow buttons. Temporary mode for Gemini Live Image: Google You can use Gemini Live to have real-time conversations with the chatbot, which feels like you're talking to someone in the same room. However, a downside is that Gemini Live doesn't work in Temporary Chat mode, so all your conversations end up in the chat history. Google should consider expanding the temporary chat mode to include Gemini Live. Default to a specific chat One thing that feels somewhat annoying to me is that Gemini always opens in a new chat, whether on web or mobile. Sometimes, you want to return to your last chat. Google can take cues from web browsers, which let you choose whether you want to go to a new tab or a specific web page(s). Gemini can also have options to default to a specific chat when reopened. That said, generative AI chatbots have endless possibilities given the vagueness of their work. You can mold them the way you want by attaching different connectors, adding custom instructions, and including source files. It remains to be seen what Google has in store for future updates and whether anything from this wishlist gets the green light. The search giant released a stream of new Gemini updates in recent months, including Gemini 3.5 Flash and Gemini Omni Spark, adding that it now has 13 products with more than a billion users each. What do you want to see in the Gemini app? Tell us in the comments.
    • Thank you for the post. Just a FYI that links to an outside site or promoting specific software is considered spamming here. Asking general questions is fine.
  • Recent Achievements

    • Conversation Starter
      sumytbe earned a badge
      Conversation Starter
    • One Year In
      B4dM1k3 earned a badge
      One Year In
    • One Year In
      DarkWun earned a badge
      One Year In
    • Dedicated
      Almohandis earned a badge
      Dedicated
    • Dedicated
      JuvenileDelinquent earned a badge
      Dedicated
  • Popular Contributors

    1. 1
      +primortal
      507
    2. 2
      +Edouard
      181
    3. 3
      PsYcHoKiLLa
      86
    4. 4
      Michael Scrip
      78
    5. 5
      Steven P.
      76
  • Tell a friend

    Love Neowin? Tell a friend!