Aaron44126

I think that's on point and there's also this shift is also going on because ... Each major architectural advancement costs way more than they used to (in R&D, and also in just what it takes to get chips using the latest fab tech). Each major architectural advancement also brings less benefit than they used to. (Gone are the days when performance doubles every couple of years. Now we get maybe a 20-30% performance bump or something in the same timespan, at best.) You have NVIDIA claiming giant generation-over-generation gains, but that's not all from the architecture, it is also from increasing the power draw to their chips and "playing with the numbers" by rolling in new "things" like DLSS / AI frame generation which don't really make for an apples-to-apples comparison. With gains costing so much more, it makes sense that there would be more incentive to hang on to the "current" generation and extract as much profit from it as possible.

Eh. I think we're also going to see at least some companies trying to reduce prices by providing you with crappier stuff. Systems with 8GB of RAM (or less?) when 16GB would be more appropriate. Squeeze onto tiny SSDs. Etc. People like us will be able to see and avoid these, but your non-tech family member who just goes to Costco to buy a cheap laptop is going to get something extra crappy.

It might seem counterintuitive, but try it on the standard performance mode instead of "ultra performance" and see if that makes a difference. I've found that "ultra performance" hurts more than helps in unintuitive ways before.

https://www.theverge.com/report/839506/ram-shortage-price-increases-pc-gaming-smartphones

I have a link in my signature to an article about the pros and cons of using LTSC. I personally prefer LTSC as well (…for the one Windows system that I have left that I actually use…). not only is it a more “decorated” base install… I *like* the idea of security updates only — you don’t have to worry about Microsoft shoving stuff down that you have to figure out how to disable or get rid of, and you can *decide* when you are ready to upgrade to the next version. As for “performance improvements”, I don’t see that as much of a concern. Windows is already about as refined as it can be in that area, and I see GPU drivers as the main place where that is in flux, and of course you can still install current GPU drivers on the LTSC version. Now as for the issues that @Easa is experiencing here, I don’t think that it would make that much of a difference if he was using LTSC or not. This seems like deeper platform-level stuff.

Despite what the system requirements say, you can use Windows 11 with "legacy boot". Unless something has changed recently... When I tried it (a couple of years ago), the official install media still supported legacy boot, and you just had to jump through the regular hoops to disable the secure boot, TPM, and CPU checks in the installer to get it to go. But it might be easier to do by upgrading a Windows 10 system rather than installing from scratch.

Just saw this fun headline. https://www.pcworld.com/article/2998935/ram-is-so-expensive-samsung-wont-even-sell-it-to-samsung.html

Yeah, I get the way that the rep framed the price hike to you is gross. I think it is really the Dell bean counters "exploiting" as you say, realizing that they can sell the RAM at a higher price so they're going to do it. (They would have to raise prices eventually as their upstream source for memory starts to cost more, but that hasn't happened yet for stuff they have already purchased and have in inventory.) Not just Dell, anywhere you look RAM prices are up 2×-4× since summertime.

This has been brewing for a while. More demand = higher prices. They're going to sell their inventory at the highest price possible, and AI companies are willing to pay a lot, it seems like. Also, other market shifts. https://arstechnica.com/gadgets/2025/12/after-nearly-30-years-crucial-will-stop-selling-ram-to-consumers/ New fabs are being set up to try to meet the demand, but that's a years-long process. Maybe the AI bubble will burst eventually. I'm just glad that I'm not in a spot to want to buy a new system or upgrade anything right now.

I get you. I'm in the same place. I mean, I am glad to see the "Precision" name return, "Pro Max" seems like it belongs to iPhones. But everything else about it is just confusing, especially if they keep changing it.

Precision name might be making a comeback. But what does "7" mean? https://www.notebookcheck.net/Dell-Pro-Max-no-more-Leak-reveals-Dell-Pro-Precision-7-16-laptop-with-Intel-Panther-Lake-processors.1176556.0.html CES is not that far away...

PAGE_FAULT_IN_NONPAGED_AREA ...Not a fun one. Some kernel code (probably a driver) tried to read memory from a bad address. For clues, you need to open the memory dump in windbg and run the command "!analyze -v". It should hopefully point you to what driver is to blame. (Look for "MODULE_NAME" and/or "FAULTING_MODULE" in the output.) My money is on NVIDIA, just based on past experience. The BSOD screen itself sometimes shows the faulting driver on the screen (reported as a .sys file, like nvlddmkm.sys). Look out for that if it happens again. I feel like I can't point you where to look more specifically because I'm not really sure what the BSOD screen looks like now (didn't Microsoft change the layout and make it black instead of blue?)... I am using Windows less and less these days, as little as I can possibly get away with, and it has probably been over two years since I ran into a BSOD.

Ha, that's actually pretty cool that he got Linus Torvalds on, I will watch this.

He mentions that different errors have different device IDs, this Wi-Fi card one is just one example. Looking at the timestamps in the list behind, it looks like they come in a big batch all at once. It's like the whole PCI bus threw up. 😕

You won't get cert failures if you system rolls past the cert expiration date. The signature includes the date/time at which it was signed and that is what gets checked as part of the validation process, not the current system time. (The same is true for checking the signature of other static things, like .exe's, drivers, etc.) You will have a problem if/when Microsoft pushes out a new bootloader which is only signed by certificates which your BIOS doesn't recognize. You'd have to disable secure boot in this case. (This "shouldn't" be possible because you "shouldn't" be running Windows 11 on your system, as it doesn't meet the requirements. You're already running it "unsupported", "at your own risk".) Microsoft normally doesn't touch the bootloader as part of monthly patching, but they could. But it will almost certainly be touched if a "major" upgrade to Windows 11 comes along ... if they push out their annual fall release and it is not just released as an "enablement package". You could also jump through hoops to add your own certificate to the BIOS (it does support custom certs) and then sign the bootloader with that. Actually... You could probably just manually load in Microsoft's new cert as a custom cert and then that would be fine.

I agree, don't worry if the temp is pushing a bit over 100 °C for a bit. Modern CPUs are well-tuned to throttle themselves if they get too hot, and it is expected that they will get hot under load, because running them close to the throttling point is how you get the best performance. (Not just an Intel thing. My MacBook also hits around 104 °C under full load. I've run video encode jobs that last for days and it's never been a problem.) Repasting might help raise the frequency that you can run the CPU at while hovering at the thermal throttling point, and it might help lower the not full load running temperature (really only useful to calm down the fans), if the replacement thermal transfer material or application is actually notably better. The time to worry if is the system is getting unstable under load. (Random lockup would be the most likely symptom, but you could also get apps crashing, BSOD, ...)

Mobile Xeon is dead. They haven't had one since 11th gen / 2021. It really is just a change in branding though rather than a change in functionality. Intel offers select CPUs that support the old "Xeon" pro features (i.e. ECC memory), and it appears that Dell is offering those in these systems. ...Actually looks like the entire HX lineup supports these this time. When they originally dropped the Xeon branding for 12th gen, it was split, with some CPUs supporting it (i.e. 12950HX) and some not (i.e. 12900HX). 12950HX is basically a Xeon of that generation, and they still have CPUs in the same class in 13th gen and ... whatever this new batch is called, 2nd gen Core Ultra.

Back when I used a Precision as my personal system, I used it for both work and leisure. My "work" actually isn't very GPU heavy. But, my "leisure" is gaming. I wanted a beefy GPU for gaming. NVIDIA's pro cards work fine for games, but they cost more than necessary, and games don't use any of the extra "Quadro" functionality. The GeForce option was pretty cool to have for this. This sort of "workload" might not be what Dell envisions people will be using this system for, but I know I'm not the only one who uses Dell workstation's like this.

Everyone I know who has tried this has used the NVIDIA heatsink. I think the AMD and NVIDIA GPUs have different die-height (that's the reason there are two versions of the heatsink), so you may not be able to get good contact using the AMD version of the heatsink.

There's no difference in the motherboard between systems that shipped with a fingerprint reader and systems that did not. The fingerprint reader hardware just connects via a ribbon cable from the palmrest to the motherboard. People have successfully added it in by just swapping out the palmrest. This isn't to say that a situation like this is unprecedented. I once had a Precision M4800 which would not allow ControlVault to activate, giving a device manager error, even with the fresh OS image from Dell applied. Dell couldn't figure it out, even after I shipped the system to them for service, and I guess they didn't want to start replacing parts until they determined the issue so they just upgraded me to a Precision 7510.

Download the ControlVault driver package and when you run it, pick the option for "extract" rather than "install". Find the MSI file in the stuff that drops out from there and run it.

The image on the right is the bottom side of the heastink. The bottom edge there with the blue thermal pads covers the part of the GPU card that is furthest away from the connector edge. There is a protruding bit in the center that will bump into the VRMs on the GPU card. Part of that has to be cut off. You'll have to do a physical inspection holding the card up to it to see which edge needs to be cut and how much. I haven't done this mod myself. You might want to engage with other users who have done the mod, in the thread that I linked above.

I don't have a good picture for you, but it is a part that hovers above the "top" of the GPU card (the edge furthest away from the connector) where the VRMs are. The VRMs on Maxwell and later GPUs are in a slightly different position than the ones on Kepler cards. You have to cut off a bit that is protruding "downwards" and hits one of the rearranged VRMs, so that the heatsink can fit. If you try to fit the heatsink on the card with this in mind, it will be obvious.

Please see this thread (especially the later pages). vBIOS images are here.

Important to note that the Precision 7540 and 7740 upgrades are different here. Dell added the 4000 and 5000 GPUs late for the Precision 7540, and those cards have a slightly different physical layout than the 3000-and-lower cards on that system, which is why the heatsink and power cable replacement are needed. On the 7740, all of those cards launched at the same time, and all of them have a similar physical layout, so really the only thing that you should need to replace is the card itself.