From an 11th gen I get about 6 to 7 with light usage, two to three with any development. It's largely a thin client at this point. Battery health is at 92%.
I tried upgrading to the ryzen and when it was good it was really good. I was able to keep a user mode libvirt vm running for dev work and mid brightness under 5W power draw. That used slirp networking, adding a bridge or default nat nic takes up about 2w to 3w of it's own power.
But like most windows laptops the suspend mucked things up. Not even power draw while asleep, but when awaking from sleep the power minimum was 10w with it more often at 20w with similar usage. I tried several wifi cards, nvme drives, port configurations etc. Also tried Fedora, Ubuntu and Nixos.
On Linux this carries over to the discussion of tlp vs power profile daemon, and soon tund. I saw much better performance and regularity with tlp, but that seems like it's not the path forward.
The steam deck shows that suspend can be fixed and done well with decent battery life under linux.
They are all USB-C as far as the physical interface goes, but their protocol capabilities are different: the rear ones are USB4 (actually full TB4 AFAIU, except Intel refuses to certify AMD systems), the front ones are USB3.2+DP and USB3.2. The fundamental reason seems to be that a “mobile CPU” is more of a SoC, including built-in USB capability, and (unlike the Intels) the AMD ones only do two USB4 ports, which on the Framework are routed to the rear.
This, then, is done through USB4-capable retimers, which turned out to draw noticeable power when a USB-A card with a pulldown is connected and idling; but no better retimers could be substituted. The front USB3.2 ports use different retimers and don’t suffer from this. (Why do the Intel motherboards not have this problem? No idea, but if I had to guess, Intel probably makes the retimers for those and is refusing to let them be used with AMD processors.)
Also this is the type of stuff where Apple would make sure the end user would not have to worry about it. Charitably, they’d put in the money/effort to make sure all 4 USB ports are the same (notably, all USB ports on a macbook pro can charge at 100W, which is not a trivial task); or uncharitably, they’d just say fuck it and ship a laptop with only 2 USB ports.
Not just overheating, but overheating the chips to the level where they put extra performance stress on the kernel for some weird reasons. Basically I was seeing 70% CPU usage plugged in on the left and 20% usage on the right with the same task. MacOS has many issues like that, the "Apple cares more about users/design" idea is a bad meme at this point.
It has been like this for years. Apple does some things well, and there are reasons to like those things, but their following is overzealous to the point of kneeling before what is, fundamentally, still a for-profit corporation.
I feel like business schools have failed this generation of management. They look at Apple and see them do something that customers like, but it has a cost to the company, like giving up the ability to track people. So they conclude that it will be an advantage if they defect and make money from ads.
Then they see Apple do something customers hate and get away with it and the lesson they take is that they can get away with it too. But they can't, because the reason why Apple got away with it was by doing the thing the customers liked. And then they can't understand why they can't achieve Apple's margins.
You idiots, do the opposite. Then you can charge Apple's margins from the first one and take their market share with the second one.
IIRC this is not a reflection on actual CPU use. Those extra CPU usage is more like throttling. I don't remember where I read this though, worth searching about this if you're interested.
I mean, yes, Apple (even before they get to making their own silicon) can slam a wad of cash on the table and say “make us a better retimer or else” or perhaps even “make us a better SoC or else”. Not a lot of other laptop manufacturers can. (Now that I’m looking, it seems that Intel and Kandou are basically it when it comes to USB4 retimers, and the Kandou ones—whose deficiencies we’re discussing here—only recently arrived on the market.)
That said, I believe all Frameworks will accept 100W down any of the four ports, even if the batteries are only designed to charge at 55 or 61W[1] and the rest will have to be consumed by the rest of the system somehow.
Of course. I just wanted to be clear that Framework laptops have 100W-rated power circuitry that works through any of the four ports, yet it would not be entirely correct to say that they “charge at 100W”: the batteries charge at 1C, that is capacity/1h; any power over that only counts towards powering the running system.
Updated parent comment with links.
The back 2 are USB 4 while the front are usb 3.2 is one thing.
Last I read, the USB a cards don't fully enter suspend in the os. I have an 11th gen and I pop out my USB a to save battery when I don't need it. The hdmi also used to have that problem but that's since been fixed.
I recently did a deep dive on gc issues at work, and compared eden the default collector and z.
Our app was collecting pretty heavily, and an average collection time of about 50ms. With Z, and Linux huge pages it was collecting sub millisecond. This was at the cost of a bit more cpu, as z uses more threads for collection.
Using gatling for performance testing, we saw a small increase in requests per second. But the benefit was a smoother response time, with less spikes to high p9X measures.
Using flight recorder with this is a great way to get details of how the gc is performing.
Addendum: As noted below zgc uses more memory, I only found it viable on 16GB heaps or above. With it really improving performance around 32GB.
I medically can't drive, and my other half doesn't want to drive. I see this as well, a number of my circle also want walkable/transit areas. In the states it's NYC, Philly, Boston, Chicago, San Fran, Seattle and a bit more. I feel like I need to leave just to get a sane metro area.
I'm in NYC now, but keep circling back to Philly. We're renting for ~3200 now a one bedroom 650sqft in NYC, both work from home. To get the ideal separation we want, not have our joint offices in the living room, we would need to go to between $5,500 and $6,500. While in Philly we could get a trinity, small town house three floors plus basement, about 900sqft, in downtown for ~2200.
I wouldn't buy in Philly, I just don't trust the city planning at this point. They're trying, but it's an uphill battle. The tax situation as a self-employed was much more complex in Philly. Safety is not something to just shy away. I lived in Old City, two years back, and there were still shootings near my apartment. It's a problem in a lot of cities as we gut social spending and relief programs. Philly did open a safe injection site, and is making head way.
I really like Philly, close to NYC Megabus was 15$ a seat, and about 3 hours. Amtrak is even quicker. Great music and food scene, Reading is great for food / produce.
The biggest issue I've found is the job market. Locally, a lot shifted out to office parks, requiring regional rail, and walking along multi-lane roads. If you're working remote, I got cost of living adjusted like crazy. The quotes I got were 80% pay cut over my NYC rate. While local jobs, were only a 10% cut. The local tech scene is a bit behind, more legacy.
Comcast has a great VC program as well for startups. The city also has tax programs to help get startups in the area.
> I'm in NYC now, but keep circling back to Philly. We're renting for ~3200 now a one bedroom 650sqft in NYC, both work from home. ... While in Philly we could get a trinity, small town house three floors plus basement, about 900sqft, in downtown for ~2200.
Where in NYC are you and how does it compare to downtown philly? $3500 NYC rent means you're living in a hip/gentrified neighborhood. You can certainly find cheaper apartments if you are willing to go deeper into Queens. Whole houses? They exist but are in the 4-6k range as well ( friends aunt rents a home in belle harbor for something like 4-5k/month). By me (ozone park) the rent is not that high (~1600+)and we have plenty of busses and the A train. Further north is woodhaven with busses and J&Z train. You can go further east but you are now past most subways.
The weakest link in this setup is your current internet. I've found while traveling, internet is generally pretty bad under, 10Mbs. Making this a rather difficult task. American view.
At home, my "laptop" is a dumb terminal that docks to my desk monitors. But computes to servers in the closet.
At my families who have (150:down/10:up)Mbs it was no different from being at home. I connected over wire guard. I even took my zoom/client calls in vms over rdp. Passing my webcam from my local laptop -> wire guard -> server. I also used parsed to stream gaming from home.
Conversely, I was at an Airbnb recently with (5/0.512)Mbs DSL. It was a real chore. For the internet black hole, I've opted to go for a SFFpc instead. That can fit into checked baggage or carry on. Even if I can't setup a monitor and keyboard at the airbnb hotel. Having both off in lodging wifi is suitable enough.
Outside of html docs, it's purpose is similar to swagger
* Auto generate mock / containers for testing
* Generate server interfaces
* Generate models that go across the wire
* Remove boiler plate to spin up an async client, sqs, kafka etc.
That being said most of the generators are from the node sides and left a bit to be desired. I also see it falling into the same issue as open api. Most places I've been dynamically generate the open api spec.
I had written a JVM async/open api gradle plugin. That generated all the above. The nice thing was that if your server didn't implement the agreed upon spec it would fail to compile. Allowing for asynchronous development, back end/mobile/front end agree on the spec and can work on the feature at the same time.
This can also be expanded into k6/gatling. Hit this endpoint, guarantee this sla and ensure everything functions as expected. By having the contract first, you can automate a majority of the down stream tasks.
I think it has a use for contract first development, and is good in conjunction with something like avro. But it serves best in a contract first flow. Where in you define a spec, if it compiles you match the spec. Most servers don't take in a spec and define routes off operation names.
Why not dynamic async api generation? A lot of the generators reflect at run time, slowing down startup, and don't provide the most human readable definitions. Also lagging behind the latest implementation, and miss potential new features.
Been using Kotlin for ten'ish years. I feel the same, although I'm not part of the Java camp, coming more functional i.e. Scala.
I was working on a number of open source libraries for a more functional reactive Kotlin, but have largely stopped.
They really painted them selves into a corner by doubling down on Spring, Micronaut etc. There is little benefit to those and Kotlin. From the back-end side, it's Java light, and Java has largely caught up.
If they kept to Ktor, Quarkus, vertx that could compete a bit with the Node/lighter crowd. Or people who limit class usage and focus on a more functional style. But that leads into that target audience...
Is not as apt to use Jetbrains projects, ancedotal observation. When I started Jetbrains was heavily used for python, ruby, etc. I converted several teams to Kotlin pre 1.0. But now with VS Code being the primary IDE/Editor, it's a much harder pitch.
The multi platform stuff feels half baked out side of UI. Having done several run time agnostic services, I hit so many road bumps. Jetpack compose is really cool though.
Where to go? This is an honest question. I just got done a project where we tried to go to Rust, and reverted due to wider difficulty with the borrower system across teams.
On another contract I spent several days tracking down node module resolution issues, and incompatibilities between peer dependencies. The script portion of package.json feels lacking compared to gradle and dependent task trees. NRWL/nx and Microsoft/Rush help a bit here.
I still feel Kotlin is a really good language, with graal if treated as Node with a better build system. But I rarely see that, mostly it's akin to J2EE or mobile.
Zig looks promising because it has an actual leader, putting their name and skin in the game. I think a single visionary must provide a personal commitment to their language and take risks on its behalf, at least during its infancy. Or else no one person wants to be responsible for the language, leaving it overwhelmed, trying to satisfy everyone at once. Zig vs Rust in a nutshell.
New languages that lose their founder tend to flounder.
Side question, is there a better remote desktop/spice client than Microsoft Remote Desktop via the app store? I've found it's performance lagging compared to MS RDP client, or Remmina on Linux?
I really wish fabric8, and more specifically kotlin k8 dsl[2] was getting more traction.
It removes the down side of yaml all over the place. It's missing the package management features of helm. But I have several jars acting as baseline deployments, and provisioning. It works really well, and I have an entire language. So I can map over a list of services, instead of do templating. The other big down side is a java run takes a minute or two to kick off
I was resilient to k8 for a long time. Complexity was secondary to cost, but Digital ocean has a relatively cheap implementation now. This commonality and perseverance of tooling is great.
I want metrics, a simple annotation. I want a secret injected from vault, just add an annotation. It's also cloud agnostic, so this logic can be deployed any where some one provides a k8 offering.
EKS was very powerful. As running service accounts via non managed clusters. Removed the need to pass an access key pair to the application. That service account just ran with a corresponding iam role.
It's been a while since I've looked at Fabric8 but it had good java -> k8s integration and was great for writing k8s tools.
It appears though that Fabric8 is useful for solo java projects without complex dependencies on non-java projects or small java shop. It overlaps with where jenkins-x is going, which has made major strides in the last 24 months.
The original team that worked on Fabric8 lead by James Strachan all moved on from Redhat and many of them are working on Jenkins-x.
I tried upgrading to the ryzen and when it was good it was really good. I was able to keep a user mode libvirt vm running for dev work and mid brightness under 5W power draw. That used slirp networking, adding a bridge or default nat nic takes up about 2w to 3w of it's own power.
But like most windows laptops the suspend mucked things up. Not even power draw while asleep, but when awaking from sleep the power minimum was 10w with it more often at 20w with similar usage. I tried several wifi cards, nvme drives, port configurations etc. Also tried Fedora, Ubuntu and Nixos.
On Linux this carries over to the discussion of tlp vs power profile daemon, and soon tund. I saw much better performance and regularity with tlp, but that seems like it's not the path forward.
The steam deck shows that suspend can be fixed and done well with decent battery life under linux.