Aaron44126

Well, I was wrong about the fan curves. Basically, I didn't wait long enough I guess. I tried using the "Quiet" profile for a while, and it did raise the fan speed by around 400 RPM over "Optimized" ... but when I left it on that profile for 10 minutes or so, it gradually settled down to around 200 RPM less than what "Optimized" had been running at (around 1500 RPM, same as I was at when I noticed this problem to begin with). So maybe it's not getting worse, but it's still worse than it was a month or so ago.

Yeah, I can hear it as soon as it hits 1400-1500 RPM. (It is pretty quiet in my home office.) Hearing it at that low level doesn't really bother me (though 2000+ starts to be an annoyance), what is bothering me is the fact that it is going up from what used to be the baseline and temperatures are higher than reasonable for components operating at low load. My Precision 7560 setup is pretty similar and it is running almost 20 °C cooler (also fully silent, fans hanging out at 1200 RPM and below — and they are smaller fans). I'm also using Process Lasso to switch Windows power profiles automatically (when I start a game or other demanding application) and a custom tool will notice the power profile has changed and automatically change Dell's thermal mode. I generally use ANC headphones for gaming so 3000+ RPM is inaudible in that case.

Yeah, generally I care more about low noise than low surface temperature. I don't think they don't have the "Quiet" fan curve quite right. It is louder than "Optimized" on an idle workload — at least in this situation where the CPU is running at an elevated temperature. ... But it does appropriately have a lower speed on high load (it won't easily go up to high RPMs while running on "Quiet"). And running "Quiet" will also cap dGPU performance (but I think "Cool" does that too).

Reasonably confident that it's not a software issue. Nevermind the fact that there is no CPU load contributing to this heat, and that the EC independently controls the fans (without help from a driver or other software running within the OS)... I can also reproduce it in my Windows 11 install, which I have as a dual-boot option for testing but haven't touched in months.

It's really weird, this seems to be getting worse by the day. Now around 1600-1700 RPM if I use "Optimized" profile and 2200 RPM if I use "Quiet". 100-200 RPM increase since yesterday. When I first noticed this issue a few days ago, Quiet was running 1500 RPM. ...I used to be able to run Quiet and it would alternate between ≈1000 RPM and wanting to shut the fans all of the way off. I repasted again yesterday (doing a much more careful job this time) — cleaned the old paste off of the CPU and GPU and heatsink carefully, using Arctic MX-4 (maybe not the best but up there with the good ones, I've been using it in other systems with no problems). A different tube than the one I used the other day at that. Behavior before and after the repaste was the same. Talking about idle workload again. GPU powered off, CPU running 1.5 GHz at less than 2% load, the system should be able to run silently. Instead, CPU cores are around 64 °C and the EC feels it has to work to cool them off. No dust in the fans or grills (cleaned all that out a few days ago). Somehow, the cooler is just becoming increasingly ineffective? I wonder if a heat pipe is jammed up or something? I'm at a loss as to what could be causing it. If it's not an issue with the cooler, something is going wrong actually in the system itself and generating some extra heat. (The system is definitely hot to the touch, it's not just the fans going crazy for no reason.) It's not hot here. Woke up to 7 inches of snow. New cooler can't arrive soon enough!

Crazy. They have more than half a year already to prepare. I mean, I guess there's a lot of evidence against him, but still...

BIOS update 1.20.0. https://www.dell.com/support/home/en-us/drivers/driversdetails?driverid=24CYP - Fixed the issue where the system cannot detect the Dell DA310 USB-C Multiport adapter after updating the BIOS. This issue occurs after connecting the adapter to the system. - Fixed the issue where the system charges beyond the Custom Charge Stop set in the Battery Configuration in the BIOS setup. - Fixed the issue where a small logo screen is displayed first and later switches to the full screen logo. This issue occurs when the Full Screen Logo option is enabled and you boot the system. - Fixed the issue where system cannot connect to the BIOSConnect server with all the supported WLAN cards. - Fixed the issue where the system reboots repeatedly when USB bootable devices with longer product name and serial number are connected to the system.

At least with the part delay... I should be getting a new heatsink and not one that has been sitting around in a warehouse for months. (Better chance of getting a "good" one, I'd think.)

Are you still using the factory/stock paste job since you got the "golden sample" or did you redo it? (Just looking at the screenshot there, my dGPU temp is like 15+ °C higher than yours with nothing really going on. Gotta get this fixed! 😛 I should have called in for the better heatsink swap before now.)

The new heatsink should come with new pads and factory applied paste, so hopefully it works well enough like everyone else’s newer heatsink and I’ll just hold out until then.

Repasted. Observations: Repaste brought RPM speeds down 200-300 RPM, enough to be "silent" again, but not all the way down to 1000ish RPM. I probably didn't do a great paste job (shaky hands / sort of in a hurry) but good enough for now. ........."Optimized" profile is quieter than the "Quiet" profile, huh? If I switch to "Optimized" it will drop to the 1000 RPM range, and if I switch to "Quiet" it shoots right back up. Guess I'm using "Optimized" as my low-power profile from now on? ("Cool" and "Ultra performance" run at higher fan speeds than "Quiet".) I'll have to test with it and see if it ramps up more quickly (which I do not want). Also, got a notice from Dell on the heatsink shipping. They put an estimated ship date of March 24 on it. Good thing this isn't a part that I need in a hurry?

How many devices? 40 doesn't really seem like too much. I have a family of four here... On the network are four laptops, a Mac Mini, multiple phones & iPads, at least three gaming consoles, a Blu-ray player, TV streaming boxes, a couple of printers, a handful of IoT devices, not to mention a few switches and three APs... Isn't that hard to get up to "tens" of devices.

Yes. I was going to upgrade next week (with Windows patches) like I mentioned before. But, given this issue I will probably push that off until April patching, I don't want "variables" flying around when trying to address this.

Yeah, and really, 1000-1200 RPM (at idle) has been the norm for months, no matter where I was using the system. And like I mentioned, I was running an artificial CPU load if the fan speed got too low to keep them from powering off. (That is no longer necessary.) It is still the tail end of winter time here, ambient temperatures aren't very elevated... I haven't been paying close attention, it could have been creeping up over the past few weeks and I just noticed it when it started to regularly be over 1500 RPM (which is when it becomes noticeably audible). Today I was poking around for possible alternate causes but I can't find any. dGPU is properly powered off when it should be. CPU doesn't seem to be doing any background busywork, and clock speed is low (I have turbo boost off most of the time). Yet CPU temperatures are in the mid 50's to low 60's, definitely too high for an idle workload. Since I don't know when the replacement heatsink is going to be in, I might take it apart and repaste this evening anyway, if I can find the time — mostly because I'd like to know if it is an issue with the cooling, or something else (crazy PCH load driving up temps or something?).

I ended up doing the online chat, because I though it would be easier. In the end, it was probably not easier. The rep wanted to go through my BIOS version / chipset driver version / asked a lot of questions. All due diligence, its fine, but it took like half an hour. Anyway. A new heatsink assembly is being dispatched. (Part only, no technician.) A funny thing is they gave me timeframes of "2 business days" for parts only and "3 business days" for an on-site tech. I thought this was supposed to be next business day service? Hoping to get the heatsink assembly tomorrow, but we'll see... [Edit] ...Of course there would be a parts delay.

Last year we got these "confirmed" dates from sales reps: May 10 June 7 June 16 June 21 June 23 June 28 ...The actual release date was July 7. No rep actually dropped that date until July 5. So, just a heads up — take anything the reps say regarding the release date with a grain of salt. Now, they were having supply/production difficulties last year which delayed the release. Maybe they have that sorted out and it really will launch on April 1...? [Edit] April 1 is a Saturday, which doesn't make sense as a launch day. Now, you said "by" April 1 so they could release before then. Dell typically launches systems on Tuesday or Thursday. It will definitely not fit. Get the one without the heatsink. You can use Dell's provided heatsink or get a 2mm aftermarket heatsink, that might fit.

Noticing that my idle fan speeds have been creeping up. It used to idle at 1000-1100 RPM (and I had to run an artificial CPU load to prevent the fans from periodically turning off). Today it was running at more like 1700-1900 RPM. I blew out the fans (quite dusty) but it only went down to about 1500 RPM. ...Rather than repaste again, I think that tomorrow I am finally going to call Dell and ask for a new heatsink assembly. I'm still using the original Delta version.

Yes, Chrome is weird and launches many processes that don't actually have windows associated with them, and I'd suspect that only the actual GUI processes are marked DPI aware here. (Chrome's scaling support is very good as you noted.) ...Every tab gets a process, every extension gets a process, and there are some processes that just manage GPU access, networking, storage, and so on. I think that they set it up this way just to make it harder for one process crashing to bring down the whole browser? There are probably articles about it. Anyway, you can see some of what Chrome is doing by looking at its own Task Manager (right-click the top bar, the empty space where tabs go above the URL, and select "Task Manager" from the menu). You can match the PIDs with the PIDs in Windows Task Manager to see what is what. (Not sure why you would really need to, but possibly interesting.)

AFAIK, Windows 11 and (later versions of) Windows 10 behave the same when it comes to DPI scaling. That said, I haven't done a whole lot of testing on Windows 11, so there could be improvements that I am not aware of. I am subscribed to a lot of Microsoft tech blogs where they post about the ins and outs of OS changes, so I'd think that they would have an article if they had made specific improvements to scaling in Windows 11, and I would have seen such a thing... but... maybe not. The scaling inside of applications that you reference is not really related to system-level scaling, it's mostly handled by the application itself. I think that you would find that the behavior is pretty consistent between Windows 11, and say, Windows 7. Apps can scale most things fine by just blowing up the font size and adjusting the layout accordingly. There might have been improvements to font rendering on later versions of Windows to help it look a little bit better. Rasterized content (i.e. embedded images) might look blurry scaled up, unless the image was already higher resolution than its display size. So, I haven't done anything to tweak Quicken's scaling, but I might try the "System (Enhanced)" DPI override if you think it works good. I do have a couple of comments that I can make here. The first comment is with regards to how Windows decides to scale applications that do not support native scaling. It uses bitmap scaling like I mentioned before, basically rendering the apps at 100% and then blowing them up to match the desktop's scaling ratio. The thing is, though, the bitmap scaling method used is different depending on if you are using an integer scaling ratio (200%, 300%, 400%, etc.) or not (150%, 175%, 225%, etc.). With a non-integer scaling ratio, you get a blurry bilinear filter which I think is what you are complaining about. Here, I have Quicken scaled to 175% (blurry) behind Notepad scaled to 175% (Notepad supports scaling natively so it is crisp by comparison). I zoomed it way in so the individual pixels are visible. (Click the image to zoom in if the forum shows it small.) With an integer scaling ratio, Windows uses a pixel doubling / nearest neighbor scaling method. There is no blur, it looks "blocky" instead. This is what I experience using Quicken on a 4K panel with 200% scaling. It basically looks the same as it would running on a 1080p panel at 100%; when you're not zoomed in like this the blockiness isn't noticeable unless you kind of stop and look at it closely. This isn't just true of Quicken... Any app that doesn't support native scaling behaves like this. The second comment is regarding Quicken itself, just a little performance trick that I have discovered which you might find useful. Quicken keeps everything open that you have ever looked at during the current session. Every view that you open (transaction registers, screens with graphs, etc.) basically gets opened in a window and never closed, even if you move to a different view. You might have noticed that it gets bogged down and slow if you have been bouncing around between a lot of different screens in the app, and closing and reopening it can restore performance back to "normal". The reason is that, whenever you change something (add/delete/update a transaction), it has to update not only the window that you are currently looking at, but also all of the other views that you have had open in the current session (even though you can't see them). ...And Quicken is not exactly super speedy at updating even one view in many cases. The trick is that you can close out views, so that they will not have to be updated anymore, and that will improve performance. Just hit Ctrl+F4 and the current view will be closed. I discovered this by accident one day. I'm in the habit of doing this whenever I am done with anything other than registers for accounts that I use most commonly. Also, when I first open the app, I go to the "All transactions" screen, then I go back to the "Home" screen and close it with Ctrl+F4. The "Home" screen will have to be updated on pretty much any transaction change, but I never look at it, so there's no reason to spend time waiting for it to update whenever I update or enter a transaction. ...I have to visit "All transactions" first before going to close "Home" view, because it will not let you close a view if it is the only one running. (You can also press Ctrl+Tab to switch to the most recent opened view "behind" the current one. It doesn't let you cycle between all open views, though.) I agree on the whole subscription thing, I am similarly disgruntled by the price of Quicken. I need the Home & Business version which is the most expensive one. It actually costs more than buying a personal Office 365 subscription, which seems crazy for the amount of "app" that you get. ...But there's no way around it, and the $6/month or whatever is definitely worth it compared to the thought of trying to do the same work without Quicken. ...Though I am happy that the new management is actually cranking out real features and fixing stuff from time to time. It seems like for the last several years before Intuit sold it, they would release a new version every year that was barely any different than the year before but want to charge full price. (And you had to pay up at least every few years because they would discontinue online support for the old versions after a period of time. So it was basically a subscription model already.) [Edit] One other thing I remembered. You can actually add "DPI awareness" as a column in Task Manager if you want to see what's going on with each running program. "Unaware" apps would be the blurry ones. I would like to continue discussing the ins and outs of various Linux things, but separate from this discussion... I'll try to make a "Here's everything I'm worried about for a potential switch to Linux!" thread over in the Linux subforum sometime in the next week or two.

He's apparently been failing to pay out severance — which is required by California law in some cases (since he gave short notice for the initial mass layoff) — in addition to failing to make rent payments and fulfill other financial obligations. I'm not sure what's up with this guy, it is bound to catch up to him eventually indeed as lawsuits play out, but he doesn't seem to care about "guidelines" right now. He's also been shown to fire people at the drop of a hat for basically saying things he doesn't like to hear. If I were employed by Twitter then I would certainly be treading carefully.

You have to cut a piece off that hits one of the VRMs at the "top" of the card (far side from the socket/connector). People have done it with a Dremel or similar tool.

Windows has very robust high-DPI support (since around Windows 10, version 1703) and it is probably better than any other desktop OS. But you do have to sort of get how it works in order to use it effectively. Scaling issues are virtually always the fault of the application, not the OS, often because the application is claiming a level of DPI awareness that it does not actually have, or claiming to be DPI aware but implementing it wrong. (I use some Java applications at work that claim full DPI awareness but are not actually aware, so they just render everything very tiny on a 4K display set to 200% scaling for example.) There are options in Windows to override the scaling behavior by-app so in my experience it is always possible to get a workable experience for any application (though you may have to resort to plain bitmap scaling which can make the applications appear blurry). On the flip side, some apps claim to be DPI unaware but use largely standard GUI elements so no dev work is really required for them to be scaled; such apps can be safely "forced" to run with scaling enabled and they will look better for it. Mainstream applications have come a long way in implementing proper scaling support, as high-DPI displays become more mainstream in both laptops and desktops. But, I still routinely run into lesser-used business applications that have poor scaling support. So, part of what I meant when I put that in my list for Linux is figuring out how to mange applications that do not scale properly, or override scaling behavior if I so choose. Can you force applications to run with bitmap scaling if they do not implement scaling properly? Can you run some applications at a scaling level other than what you have set at the system level? Do applications running under Wine scale properly at all? Etc. Scaling is tricky business for a GUI developer. It's way easier to implement scaling in a macOS application. macOS only supports 100% and 200% scaling levels. (All other options that they have are achieved by running applications at 200% and then using bitmap scaling to make them larger or smaller from there — the OS handles that — and it will make apps appear blurry.) So, application developers just have to provide 1× and 2× versions of various graphical assets and the work is mostly done. Windows supports any arbitrary scaling level (25% increments in the basic graphics settings screen, but there is an advanced screen where you can enter any scaling value that you like) so it can obviously be more of a pain for applications to properly implement scaling if they are using any custom GUI elements. Some Windows applications support per-monitor scaling, which I greatly appreciate as well. If you move application windows between monitors running at different scaling levels, the application will adjust the scaling level on the fly. (Applications that do not support this run at the primary monitor scaling level and are bitmap-scaled up or down when running on other monitors at a different scaling level, again adding a sort of "blur" to them.) Most Windows built-in apps, UWP apps, and web browsers support per-monitor scaling. MS Office apps do as well (except OneNote for some reason). I don't know if any Linux applications support per-monitor scaling. macOS does not support this at all, it again relies on bitmap scaling to run apps at any level other than 100% or 200%. (Windows has supported DPI scaling since at least Windows 98, which I have recently tested in VMware, but it was sort of a hidden away option so not many applications supported it properly until more recently. It's only been really usable in more recent versions of Windows, since they have added support for to changing the scaling level without having to log out and start a whole new desktop session, selecting different scaling levels for different monitors in a multi-monitor configuration, overriding the scaling behavior for particular misbehaving applications, signals needed for developers to implement per-monitor DPI awareness, etc.) There's one of my key applications that I would have to figure out how to deal with to switch to Linux. Quicken is so terrible in its own way, but every other personal finance management solution is far worse, there really is no replacement for it. 1. When I looked into this, my understanding was that KDE isn't running on Wayland yet (or at least that is not the default configuration) which means that using different scaling ratios on different connected monitors is not possible. (When working, I have my laptop display which is 4K/200% and I pair it with a larger external display which is 4K/125%.) GNOME and other DEs on Wayland can offer different scaling ratios for different monitors. (But I did not realize that GNOME has to basically use full integer scaling.....) 2. This is also normal for Windows, some apps have robust support and can switch on the fly but many just use the scaling level that was set at launch need to be restarted to switch.

Yeah, NDAs are one thing to worry about, but I was thinking more directly of the fact that anyone speaking publicly about internal turmoil will likely be outright fired by Musk as soon as he figures out who they are...

If you're asking about the article immediately above, The Verge has many links in the article pointing to information about the downtime itself and Musk's statements on the matter, but as far as Twitter's internal situation is concerned, The Verge has acquired some internal sources (Twitter employees) that can't go "on the record" for obvious reasons. So I guess it's up to individuals to decide how reliable that reporting is. (I've been reading The Verge since its launch in 2011 and have found no reason to doubt their credibility... It's my second favorite tech news outlet.)

What happens when you put one engineer on the API project. How a single engineer brought down Twitter https://www.theverge.com/2023/3/6/23627875/twitter-outage-how-it-happened-engineer-api-shut-down (...Though, apparently they have their Slack instance back.)