Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The ARM powered ThinkPad x13s, as a developer

dev-eloper.com

Post reply on HN
Print view View original

111–120 of 168 posts

Re: The ARM powered ThinkPad x13s, as a developer

#111

Post by LoganDark »

deng wrote: post #106

People forget that it is not just the processor that consumes power. The reason the new MacBooks are so efficient is not just because they are ARM-based, it is also because they have incredibly energy efficient screens, WiFi, RAM, SSDs, etc. They can achieve this through their sophisticated manufacturing process, with highly integrated components directly soldered onto the board, a fact that is widely (and correctly)…

Is it really that hard to design sockets that don't waste such a significant amount of energy? Honest question. Do you really save so much by soldering something without any type of ribbon, socket, or other removable connector in between?

Re: The ARM powered ThinkPad x13s, as a developer

#112

Post by LoganDark »

NikolaNovak wrote: post #15
jillesvangurp wrote: post #13

Like the author, I also do a lot of Kotlin development and use intellij and pycharm a lot. This sounds like an awful laptop to do that on. Kotlin is great but it's compiler is rather slow and you end up waiting for that a lot. So, faster is better. I noticed about a 2.5x build speed improvements when I upgraded my cheap Samsung i5 laptop (a temporary thing I used for a while) to the M1 14" Mac Book Pro. On a typical…

Silly question ; am I the only person who still uses a desktop / tower machine? Feels like I'm a huge minority these days, but when it comes to needing processing power and juice it still feels it provides budget and upgrade flexibility no laptop can match.

I actually built a desktop tower, and then built a laptop to use it from bed! Because I found that regular laptops were totally unacceptable. I tried to cope with a gaming laptop but could not. It built up so many issues over less than 2 years that it became worthless.

Anyway, I get all the "portability" of a laptop (with my display, keyboard, and trackpad(!) in bed), but all the performance of an honest-to-Arceus desktop, with a full desktop 12400F (boosting up to 5.2GHz on all 6 cores at a cool 130W) and a full desktop RTX 3060 12GB (I regret not getting the Ti for performance reasons, but for VRAM reasons this is acceptable since I do ML work).

Re: The ARM powered ThinkPad x13s, as a developer

#113

Post by LoganDark »

glitchc wrote: post #23
NikolaNovak wrote: post #15

Earlier quoted context omitted.

Silly question ; am I the only person who still uses a desktop / tower machine? Feels like I'm a huge minority these days, but when it comes to needing processing power and juice it still feels it provides budget and upgrade flexibility no laptop can match.

I have a nice big workstation with heck load of cores, memory and a 4090. It's the only way to fly. Laptops are great for emails but toys when it comes to dev machines.

Jealous. I'm still trying to save up for a slightly bigger cooler to handle the 130W of my 12400F under load. It hits 5.2Ghz.

Re: The ARM powered ThinkPad x13s, as a developer

#114

Post by michaelmrose »

jklinger410 wrote: post #82
freedomben wrote: post #74

Earlier quoted context omitted.

I have had both experiences. The base system for Linux will be great for battery life (has been ruthlessly optimized for power savings by the biggest companies in the world), but some of the desktop software on top will brutalize your battery. I've experimented a lot (mainly on Fedora) and the biggest offenders are browsers, but also sometimes Gnome gets in a bad state and eats up a lot of CPU. It's often gjs eating…

> has been ruthlessly optimized for power savings by the biggest companies in the world It has been ruthlessly optimized to save power for servers with greatly documented hardware structures. It has not been optimized for Desktop OS, for Bluetooth, Sleep, WiFi, graphics card power saving etc. I am a tab zero guy, for the record. I've been using computers since Windows 3, so I'm very sensitive towards intensive proces…

A big difference in battery life comes down to hardware acceleration in your browser which depending on hardware,browser, and version may not be enabled.

Re: The ARM powered ThinkPad x13s, as a developer

#115

Post by deng »

LoganDark wrote: post #111
deng wrote: post #106

People forget that it is not just the processor that consumes power. The reason the new MacBooks are so efficient is not just because they are ARM-based, it is also because they have incredibly energy efficient screens, WiFi, RAM, SSDs, etc. They can achieve this through their sophisticated manufacturing process, with highly integrated components directly soldered onto the board, a fact that is widely (and correctly)…

Is it really that hard to design sockets that don't waste such a significant amount of energy? Honest question. Do you really save so much by soldering something without any type of ribbon, socket, or other removable connector in between?

For RAM, yes. Soldering it down right next to the CPU allows for lower voltages and faster bus speeds. For SSDs, it's more difficult. It surely makes no difference for standard PCIe, there's really no excuse to solder that one to the board. I'm actually not sure how exactly SSDs work in a MacBook M1/M2. I think it's plain NAND storage soldered directly onto the board, with the controller in the M1/M2. I have no idea about power savings here, but it sure is fast...

Re: The ARM powered ThinkPad x13s, as a developer

#116

Post by LoganDark »

ChainOfFools wrote: post #96
faster_harder wrote: post #46

Earlier quoted context omitted.

It reminds me of the glass vs plastic phone backside arguments. Glass is more fragile than plastic and will not absorb shocks. It is also heavier and harder to replace (most plastic backsides can be removed with a guitar pick). But glass is considered premium .

I was astonished at the amount of ink, or pixels wasted by this ridiculous and completely manufactured distinction on the part of inept tech "journalists" and bloggers who had to find something non-technical to identify a pecking order among phones in order to attract readership who didn't understand the more subtle distinctions between features that affect actual usage of the phone. The only other ridiculous argumen…

I hate thin and light so, so much.

People say "vote with your feet" and "just don't buy the stuff you don't like", but you can't exactly buy nothing when there are no options you like, because what if you need a laptop anyway? So you end up having to buy something "thin and light" which is a signal to the OEM that people want more thin and light even if all they wanted was a laptop.

Re: The ARM powered ThinkPad x13s, as a developer

#117

Post by michaelmrose »

jklinger410 wrote: post #72
nine_k wrote: post #58

Earlier quoted context omitted.

What does your `top` say? What does your browser's performance tool (e.g. `about:performance`) say? I went from a constantly-overheated Linux laptop to a completely calm Linux laptop, on the same hardware, distro, and DE. Checking for unnecessary processes that eat CPU, spurious ACPI events, overly-heavyweight browser tabs goes a long way. Making sure the browser uses hardware support to play YouTube is another heavy…

So the Mac has battery saving built in to the OS. You are doing a lot of reverse engineering to achieve something other software/hardware does automatically. And to be specific, I suffer from bad hardware/software communication. On a linux laptop, no less (System76, which I would never recommend). Linux S3 sleep is not hardware sleep, I lose a lot of battery there. WiFi and Bluetooth connectivity either has to be bat…

> Linux S3 sleep is not hardware sleep

No this is completely down your hard hardware

Re: The ARM powered ThinkPad x13s, as a developer

#118

Post by LoganDark »

hospitalJail wrote: post #65

Its 2023 and I still don't understand the point of CPU speeds, even as a dev doing both single and multithreading. AI is continuing to put priority on GPUs. In the worst case, I wait an extra fraction of a second with a bad CPU. With a bad GPU, you are basically excluded from the latest technology. I'm sure ARM really wants to advertise that CPU is relevant, but it continues to be nearly irrelevant when you buy somet…

I'm sitting here with "only" 12GB of VRAM and I can't train even the bigger RWKV models, and that's a freaking RNN, it's not even a transformer! And that's highly disappointing for someone who's been working as a contributor to one of the inference libraries, because the fruits of my labor aren't even going to benefit me that much. Bleh~

I (somewhat) lie, all my experience working on rwkv.cpp does gain me knowledge and also contribute to the community. I'm just sort of sad that I don't have enough VRAM to finetune a bigger model to fit something like a chat format, or learn a certain vocabulary or whatever.

Re: The ARM powered ThinkPad x13s, as a developer

#119

Post by LoganDark »

deng wrote: post #115
LoganDark wrote: post #111

Earlier quoted context omitted.

Is it really that hard to design sockets that don't waste such a significant amount of energy? Honest question. Do you really save so much by soldering something without any type of ribbon, socket, or other removable connector in between?

For RAM, yes. Soldering it down right next to the CPU allows for lower voltages and faster bus speeds. For SSDs, it's more difficult. It surely makes no difference for standard PCIe, there's really no excuse to solder that one to the board. I'm actually not sure how exactly SSDs work in a MacBook M1/M2. I think it's plain NAND storage soldered directly onto the board, with the controller in the M1/M2. I have no idea…

For RAM I think even the trace length on the motherboard matters but this is just a reason for why the sockets are usually near the CPU, not a reason why you would have to eschew sockets altogether.
Post reply on HN