Is Lunar Lake a Game Changer for Intel?
Published: 09-25-2024
This is Intel’s big swing,
Image Credit: Intel
Usually, we’re not too interested in the ultra-mobile segment of the CPU market. After all, even our mobile workstation systems are 200W plus performance monsters, so it’s not our usual thing. However, there’s no denying that power-efficient mobile systems are on the rise. There are plenty of people doing professional work on MacBooks that sip power and run for days on battery power.
Intel, on the other hand, has always struggled to bring performance and power efficiency into the same package. While x86 processors don’t have to be less power efficient than the Arm competition on paper, in practice they’ve never come close to an actual product you can buy.
Now Lunar Lake laptops are starting to appear, and hopefully this is the chip that puts Intel back on track.
Cracking the AI Performance Puzzle
There’s no doubt that AI software running locally on your computer will be a big deal sooner rather than later. LLMs and other generative AI tech will play a key role in professional workflows, and Lunar Lake promises AI processing power in spades, with an NPU rated for 120TOPs. Triple the minimum requirements for Windows CoPilot Pro.
On a traditional workstation laptop, you can simply have your beefy (yet power-hungry) GPU do this processing, but by adding it to Lunar Lake you can get away with AI tasks without the need for a massive GPU.
Speaking of GPUs...
Not too long ago, iGPUs were pretty much a joke, good for nothing more than running the Windows desktop and basic apps. Oh, how times have changed. While they’re no match for dedicated GPUs, modern iGPUs have plenty of grunt.
However, that’s largely been true for AMD APUs, which power handheld gaming PCs and ultrabooks with decent 3D performance, as well as the major gaming consoles, of course.
Sacrifices are Necessary
To get Lunar Lake into shape, Intel ditched hyperthreading, and reduced the core count. So while you’ll get good single-threaded performance, it’s a regression for multi-threaded workloads.
However, most workloads for these types of computers are not heavily multithreaded. That includes common professional apps like CAD and CAM, which traditionally only put load on one or two cores, unless you’re rendering things out. That makes the latest generation of ultrabooks more practical as professional tools than ever before, without reliance on cloud services.
Was It Worth It?
Image Credit: Asus
Intel claims that Lunar Lake needs half the power of its predecessors for the same performance level. This is in the face of a new generation of Windows Arm laptops running the Qualcomm Snapdragon Elite chips. These laptops have phenomenal battery life, but numerous performance and app compatibility issues. So if Intel can get into the same power consumption ballpark while maintaining perfect app compatibility, that’s a big deal.
We think that Lunar Lake is a big step in the right direction for Intel, and that the lessons they’ve learned here will also result in better power consumption for high-end desktop and server chips eventually. Now they just need to get that 18A process node sorted, _or ask TSMC for help again,) and we’re good to go.
|
