Power Efficiency and RISC or CISC in Efficiency Cores?

`

As far as I know, no intel or AMD based laptops are going to compete with any apple silicon devices on battery life for anything other than highly controlled benchmarks that put the x86 chips right in the middle of their efficiency curve. The apple silicon chips absolutely sip power in single core scenarios, and they’re configured to aggressively avoid leaving their efficiency sweet spot. That’s in comparison to AMD and Intel chips that will boost well above that sweet spot to get better performance, and consume double or more power.

Add to all this that the framework laptop is using socketed dram compared to the on package lpddr of a macbook, and the memory subsystem is also going to be a lot more power hungry.

Many companies will market similar battery life to apple silicon devices, but in most scenarios if you run the exact same workloads, apple silicon is just much more efficient because it’s an architecture that was built for efficiency with everything from the denser, efficiency focused pdk apple uses with tsmc n5 (compared to the frequency focused pdk AMD uses), to the more conservative approach to boosting behavior, to the more tightly integrated power monitoring compared to what windows can achieve means that no windows laptop (or linux laptop without significant tweaking/compromise) will be able to match the battery life of a macbook with the same size battery without severely compromising performance, or portability.

I’d love to see a more open platform with the mobility and practicality first design ethos the modern apple laptops seem to have picked up, but the chip makers just don’t make it very easy. I’m sure Framework looked at the compromises they’d have to make to fit a 17% bigger battery into their chassis, and decided that they’d rather have it be a bit slimmer, a bit lighter, which I think is also important for a laptop, even a 16" one. We’ll see what the battery life ends up being, but I imagine it will be pretty fantastic if you tune the tdp and boosting behavior of the onboard chip, or terrible if you let it run at max clocks with a dedicated gpu also running in the expansion bay.

5 Likes

Apples advantage comes mostly from manufacturing the whole stack, i.e. the hardware and software, as well as merging CPU, GPU and memory on a single chip. This allows for far more optimization, which is where most of the efficiency comes from. Running Linux on apple silicon is just as fast as on a x86 processor, it’s just not optimized.
I have high expectations for the framework ryzen version regarding efficiency, if tuned right we should get at least a lot closer to Apples battery runtimes.

Uhh afaik no, I’m not sure if you’re mentioning Intel (who is the only manufacturer of “Efficient” cores, besides maybe IBM), but their cores are not efficient, and definetly not RISC.
But yeah, Apple’s advantages largely come from very vertical stack allowing them to optimize far more easily.
I am hoping the 7040 series from Framework is well optimized as well :slight_smile:

2 Likes

The distinction between RISC and CISC is largely academic these days as far as I know. Both Intel and AMD are implementing most of their legacy instructions in microcode coordinating smaller instructions, and ARM has slowly accumulated quite a few extensions to the ISA over the years including very “complicated” ones like neon.

On top of that, Apples performance cores are wider than even Intel’s big cores from what’s been revealed through microbenchmarking (for example Anandtechs analysis), and are anything but simple.

There are a number of reasons Apples systems are more efficient, and I don’t to presume to have the expertise to break down the most important causes, but I am pretty sure the idea that the ISA is the primary cause is a misconception based on historical differences that just aren’t as relevant today.

3 Likes

They’re definitely not RISC, I reread on that and seem to have remembered something wrong, sry for the confusion. Removed that part from my comment because it’s untrue.

2 Likes

If x86 truly is less efficient, why do so many manufacturers handicap themselves further by using an undersized battery? It frustrates me to no end - as a former Mac user searching for the perfect lightweight Linux laptop.

Both frameworks have batteries sized roughly 15% smaller than the Mac competition. That’s the difference between not quite good enough and solid all day battery life. This seems pretty par for the course among all PC laptops.

The differences between Apple Silicon and x86 is less than we might think, the problem is we have never seen a real apples to apples comparison with comparable battery sizes.

As for the compromises required for a larger battery… well, here is a 14" Laptop with a 99Wh battery:

It weighs the same as the framework 13.

It also has user-replaceable RAM and SSD.

The things holding me back:

  • ISO-style return key, no ANSI variant
  • USB PD/thunderbolt only on the right side, not the left side.
  • I want to support framework and their mission (have the AMD on pre-order, but concerned what Linux support is going to look like once it’s out)
1 Like

It looks like a great machine, but it doesn’t have 6 swappable ports, a modular expansion bay, a modular input system, or a user upgradeable mobo/platform… The framework 13 was never the highest specced, or the most compact 13" notebook, and the framework 16 will not be the best in those areas for the exact same reasons… They are modular computers that trade off a bit of performance in some areas for that modularity and repairability. If you’d rather have the most tightly integrated notebook with the best specs, a macbook, XPS, or thinkpad probably does better.

In my mind, the appeal of the framework is that it doesn’t trade off that much, but it offers something completely unique in the form of extreme modularity and upgradeability.

With all that said, I would love to see a 100Wh battery for the framework 16, but if it follows in the same pattern as the framework 13, and the chassis stays the same size gen on gen, I doubt we’ll get a full 100Wh battery anytime soon because the battery cavity size has already been decided. Maybe we’ll get a nice 10% bump like the 61Wh battery on the 13 when better battery chemistries come around in the future, but that’s at least a year or two out, so I’m not going to hold my breath.

1 Like

I did a bit more reading because I got curious about this… There actually are a few comprehensive reviews of laptops with comparable battery sizes and displays. See this review of the gram 17 which has a 90 Wh battery: LG Gram 17 review - A super light 17-inch laptop with a modest dGPU - NotebookCheck.net Reviews

It seems like for light workloads like web browsing, these P series intel chips can match the macbook pro 16 - both last about 14 hours web browsing with moderate brightness. At idle, both use 4-5 watts of power, with most of that being due to the screen. The MBP can use up to 34 watts at idle with max screen brightness according to their benchmarks, which is wild! I guess that mini led display can really pull some juice.

I think I was overestimating the lead that apple has for light workloads, especially when you look at whole system power consumption. With that said, the 1260p is 4+8 core processor, and the m2 max they were testing is a 16+8 core processor with a much stronger igpu. That the m2 max can idle down at around the same power as intel’s much smaller chips is very impressive. There are definitely still some advantages to apple silicon.

If you compare the normal m2, it can idle at well under 2 watts total system power. See this review for that data: Apple MacBook Pro 13 2022 M2 Laptop Review – Debut for the new Apple M2 - NotebookCheck.net Reviews

I think I’m getting in over my head at this point though… I’m not electrical engineer, or anything like that, and I’m only taking a surface level view of these power consumption numbers. I guess we’ll just have to see if the framework 16 ends up being a compelling product, or if they had to make too many compromises to include the ambitious features like the modular input system and the modular back bay.

1 Like

I’m not asking for any of that. I am simply asking for a competitively sized battery, and providing some evidence that it is possible. No need to be a framework apologist, I still am planning to buy the AMD 13.

Thanks for doing that research! And that’s still with a slight battery disadvantage. My hope is that AMD chips are even better - really crossing my fingers that the AMD 13 will be efficient enough to last a full day (which for me probably means 10 hours or so).

Yeah, I think I just kind of used your comment as a jumping off point for my own curiosity. Apologies for going off topic. I hope AMD delivers decent battery this gen, and considers improving the battery capacity in the future.

Given that the processor is a higher TDP, a significant portion of the larger battery will be used for the processor rather then extending battery life.

Okay… but if they can put a higher TDP processor with a big battery at the same weight as the framework, then that proves that you can do the same with a lower TDP processor.