Effective Strategies for Replacing Hyperthreading and Boosting Performance

31
361

HTML Article:




Tech <a href='https://ainewsera.com/japan-govt-subsidy-for-ai-robots-to-offset-labor-shortage-nhk-world-japan-news/artificial-intelligence-news/ai-and-robotics/' title='Japan govt. subsidy for AI, robots to offset labor shortage | NHK WORLD-JAPAN News' >News</a> and Updates

Tech news isn’t just information, it’s also a state of mind

But Tech news isn’t just a state of mind, it’s also information. Let’s dive into the details about Intel’s upcoming Aerol Lake platform.

Intel’s Aerol Lake Details Leaked

Recent details about Intel’s upcoming Aerol Lake platform have leaked, lending credence to rumors that team blue plans to kill hyperthreading. Reliable Hardware leaker Yukan sure on Twitter initially published an excerpt from an alleged Intel document before removing the Tweet. According to these documents, Aerol Lake S processors will feature the same core count as the current 14th gen Raptor Lake refresh tips and will have hyperthreading disabled on performance cores. Rumors have been swirling since at least July of last year that Aerol Lake would ditch hyperthreading in favor of something called rentable units allegedly influenced by the Royal core project thought up by legendary CPU architect Jim Keller during his brief stint at Intel.

Nvidia’s RTX Remix Software in Open Beta

Nvidia’s RTX Remix software is now in open Beta, allowing almost any modder to remaster old games with AI upscaled features, ray tracing, and even DLSS compatibility. The tech is built on team Green’s Omniverse 3D Graphics platform and should work for nearly any direct x 8 or 9 game. Nvidia even partnered with Mod info site mod DB to make it the deao DB for RTX remix compatible games and community-made mods.

Thanks for reading today’s tech news updates!


31 COMMENTS

  1. Wasn’t Hyper threading always a bit of a bullshit marketing thing anyways? It never actually increased cores or compute, just added L1 cache. The value of having extra registers per core naturally diminishes as you add cores.

  2. I feel like saying the 4070 super was reviewed positively is a stretch, definitely more neutral. It brought marginally better price to performance, that was the whole review from pretty much every outlet. I don't doubt sales were lower than expected, in the US (my anecdotal experience) there is a bit of economic down turn and this card wasn't far and away better than the other options on the market, nor is it really "new", I feel like people were more excited for the potential shift in pricing through out the new and used market as a whole than the card itself.

  3. I mean it's possible for hyperthreading to go away? It originally showed up in the P4 due to the long pipeline and scheduler resulting in relatively poor utilization of the functional units on it — i.e. there was a lot of idle resources for hyperthreading to pick up on. With code from an optimizing compiler, and a good CPU scheduler, you can potentially have very few spare resources resulting in hyperthreading having little to no speedup.

    That said, I would be a little surprised if they removed it — as far as I know the 15th gen CPU makes incremental improvements over the 14th, I have an 11th gen "Tiger Lake" and that definitely gets a benefit from hyperthreading.

  4. I just dont get how Nvidia make their products… From 3060 12Gb then back to 8Gb for the 4060 and only 12Gb for 700-800€ cards… And now 6Gb for the 3050, a card nobody want even in 8Gb… It's so entertaining to watch Nvidia and AMD because both of company are totally reckless on how to set clear product line that make sense