Aaron44126

Oh, if you have the "Cool" thermal profile set... Try with "Ultra performance" instead. I wonder if Dell is further limiting the NVIDIA GPU power level like they are with the "Quiet" profile. You could check with NVIDIA Inspector and just looking at the P-State. P8 is the lowest performance and P0 is the highest. In the "Quiet" profile it will not go to a higher performance state than P3 and that is definitely noticeable in games.

No, this a well-known thing. The virtualization support blocks access from the OS to some low-level registers that ThrottleStop needs to be able to manipulate for certain functions. (I thought that it was just having Hyper-V enabled in Windows, but maybe it is having virtualization enabled in the BIOS at all.) I've found that 4K gaming in newer titles probably isn't going to work out, except with maybe a good DLSS implementation to push it over the edge. I've been having decent results with 1440p (...and I can't really tell the difference). I published some 3DMark numbers on the second post in this thread. Not surprised if a Turing desktop card can beat it. It basically boils down to the "low" power limit for this GPU compared to the power limits that you see in desktops (especially if it has to "share" with the CPU). I’m also going to try undervolting the GPU, which I don’t expect to be transformative but will help a little bit.

Yeah, this is what has been putting me off from working with it. I think that you can do "some" things but undervolting specifically is locked out. Someone told me that you can disable virtualization, find the right voltages, and then set them with appropriate setup_var variables and it'll work even if virtualization is enabled. I haven't taken the time to figure that out yet. Also, it's not really virtualization turned on in the BIOS that messes you up here. It's just having Hyper-V enabled in Windows. (Or, that's my understanding. Also something I haven't taken the time to poke with.) If you don't need Hyper-V, then turn it off and you might be in a better spot. I'm a VMware guy, myself... but now I need to keep Hyper-V on for WSL.

I think that's dated advice from when SSDs were smaller and TBWs were less. Hibernate away.

Studio drivers added today (links in OP).

This has been normal since forever for Dell systems. The BIOS defaults to RAID on — presumably so that boot configurations that depend on RAID won't get broken if the BIOS settings are reset for whatever reason. If you don't need RAID, switching to AHCI/NVMe makes things a little simpler. You don't have to worry about loading the Intel RST driver before you can see drives while installing Windows, for example. Linux appears to tolerate RAID on just fine with individual drives, though it doesn't support the actual RAID arrays. Note that if you switch between RAID on or not and you use Windows, you must take some steps to make Windows happy or you will get a boot failure.

Audio driver update 6.0.9394.1. https://www.dell.com/support/home/en-us/drivers/driversdetails?driverid=44RR8 - Fixed the issue where there is no sound from the Dell Slim Soundbar SB521A when the system is in Battery mode. On the subject of the audio driver, I fought with that a bit over the past few days in both Precision 7560 and 7770. You need Dell Optimizer installed in order for certain audio functions to work, like being able to specify if you plugged in headphones or a headset to the 3.5mm jack (required to activate the microphone input on that jack). Also, without Optimizer installed, you can get "pop" sounds when sounds like alerts start or stop. It's like the sound chip "sleeps" while its not doing anything and then pops while it is changing power states. (This has been an issue with the Precision line since M4800/M6800, when they switched to using the Realtek sound chip.) This doesn't happen with Optimizer installed. I have found that Dell Optimizer activates certain "enhancements" for conference calling type apps (MS Teams, Webex, Zoom, etc.). I don't like them, it makes people's voices sound echo-y and weird. You can disable the enhancements in the Optimizer app, but then everything in my work conference app (Webex) is super quiet compared to other sounds on the system, so there is still some kind of filter in place even with everything audio-related in Dell Optimizer turned off. That in turn can be fixed by going to the "Sound control panel" in Windows and just disabling "audio enhancements" for the output sound device. Everything will then sound "normal" and at the appropriate volume level. However, I'm using another third-party audio filter to apply a simple -20dB filter to my headphones output. The default headphones output sound is way too loud. I have to run with the Windows volume set to "4" or "6", which means granular control of the volume level using the keyboard is not possible. (Windows changes it in increments of "2", so there are only a few settings to choose from in a reasonable volume range, often I have to decide between "a little too quiet" and "a little too loud".) With the -20dB filter in place I can run with the volume more like in the 40-60 range and I can more easily settle on a volume level that is "just right". This filter needs "audio enhancements" enabled to work. So, I need "audio enhancements" enabled, but I can't find any setting in Dell Optimizer that makes conference apps sound "normal". I ended up removing Dell Optimizer and replacing it with an older MaxxAudio application. It performs the same function as Optimizer for purposes of audio support (headset+microphone support / no audio pops / control over audio enhancements / etc.) but you can actually fully disable all of the enhancements if you like, and conference apps will sound unfiltered. To disable enhancements, just open the app, hit each of the three main sections, and turn the toggle switch at the top off. You'll have to repeat this for each output device that you use (system speakers, headphones, USB audio output, ...). URL for the MaxxAudio app is: https://apps.microsoft.com/store/detail/waves-maxxaudio-pro-for-dell-2019/9PJPPGG8TGVG?hl=en-us&gl=us

To get to the prompt to enter these commands, visit this tutorial and hop down to step 5. https://brendangreenley.com/undervolting-2020-dell-laptops-like-the-vostro-7500-and-more-tips-to-improve-thermals-battery-life-and-speed/#cpu-undervolt

No, I've had it for almost three months. It is under warranty so I can have parts replaced if needed. I will run with graphics switching enabled for now and listen out for more reports...

Tried with "hybrid graphics" disabled for a while. The system is not reliable. Sometimes I would come back after being away from the machine for a while. The laptop display would be (appropriately) powered off. However, upon moving the mouse around, the backlight would power on but the lock screen would never appear. I had to do a hard reboot to get it going again. This evening, that happened again, but after the reboot, I just got a garbled display (at the initial BIOS boot screens). I had to hook up an external display to get into the BIOS and switch hybrid graphics back on.

Pascal cards won't work with Windows because of a vBIOS/ACPI conflict. You will get BSOD at boot every time. Early engineering sample vBIOS images can work around this, but to my knowledge these have only been discovered for P3000, P4000, and P5000 GPUs. You can use Pascal GPUs with Linux on M4800. You can also use later generations (Turing+) on Windows with M4800, the issue vBIOS is specific to Pascal. You can take some extreme steps to get Windows booting with Pascal.

So my observations have been, "Ultra performance" does indeed perform a little bit better than "Optimized" (see benchmarks on second post in this thread) ... and "Quiet" will also impose limits on the NVIDIA GPU (it won't go higher than PL3 power level).

Yeah, that's what I've been saying :-P. The fans are managed by the EC and are entirely independent of anything that you can really get to from the OS/drivers/etc. Nothing to do with the Intel thermal framework or other Intel drivers that get installed (the behavior is the same whether those drivers are installed or not, or disabled or not). It is also a very Dell-specific implementation, and a new one to the 2021 systems (and later) as well. Now, the EC can be updated (there is a phase of the BIOS update that hits it) so there are going to be fan tables somewhere but getting to them or changing them, I don't know how to do that. I do believe there are likely calls that could be made from the Dell SMBIOS WMI/SMI interface that would change its behavior but I don't know how to "discover" those either. I've been content with the fan behavior, for the most part, I just wish there was a way to keep the fans from turning off on light load. (I've been managing this so far by throwing up a low-priority artificial load when the fan speed gets too low.) If I replace the heatsink and am satisfied with the Sunon fans turning on "quietly" then I might stop caring about this.

Seems like there are more parts than just that. Possibly, the pressure plate that does the "compression" of the memory module against the motherboard CAMM connector is a different thickness.

See @Dell-Mano_G's post just above...

Mine broke a tab on the right side near the USB-C port. I used a pry tool (like guitar pick) to open it the first time, but since then I’ve been able to pull from the SmartCard slot opening and just remove it entirely by hand.

This happened to me too. I’ve opened it several times since then and it hasn’t been an issue. It seems to shut and fit fine so I’m not worried about it.

Same suggestion I frequently give ..... These things can go to 100 ºC on the CPU in no time. Turn off "turbo boost" unless you really need it for your workload. It will help with both fan noise and surface temperature. (See "Turbo boost toggle" link my signature for some options to quickly/automatically enable & disable it.)

No. I've had it disabled since I got the system.

https://www.nvidia.com/en-us/geforce/news/12gb-4080-unlaunch/ https://www.theverge.com/2022/10/14/23404595/nvidia-rtx-408-12gb-unlaunch GeForce 4080 12GB had lower specs than GeForce 4080 16GB (not just in the vRAM area, but also CUDA core count / etc.). I guess they will launch it in the future with a different name (4070 makes sense...).

Regarding power use / poor battery performance, you should always check and see if the dGPU is actually on or not if you have Optimus / graphics switching enabled. If you have basically nothing happening but more than 10w power draw, it is possible that the dGPU is not powered off. Many users have reported issues with the dGPU staying powered on, even when no apps are using it, and even when the Windows system tray thing you can enable from NVIDIA control panel is showing it in the gray/off state. It's been going on since the beginning of time (or the 12ish years that Precisions have offered graphics switching) and not everyone seems to be affected by it. I have this issue on both my Precision 7560 and 7770. My Precision 7530 was spared from it, but other users of that system did report this issue. Dell has not been helpful in figuring out what is going on. To see if the dGPU is actually on or not, I use NVAPI to query the running temperature. If it gives you a value, the dGPU is still on; it'll throw an error at you if it is off. To kick the dGPU into turning off when it should not be on, I just disable/enable it in Device Manager. (I actually have a quick script for this now so I don't have to do any clicking. I fire it automatically fire it about two minutes after login... For me, it seems to behave as it should at boot but then silently power on the dGPU about one minute into a session, and then it stays powered on until I do something about it.) The disable/enable trick also works to get the dGPU to power off when applications get "stuck" on it (i.e. after docking/undocking, connecting in via RDP, using fast user switching, ...). Another trick that @Ionising_Radiation found is to run an Adobe app like Lightroom and keep it minimized. For some reason, this fixes up the dGPU power behavior.

Think about how high the warranty TBW value is and how long you might want to use the drive. Let's look at, for example, Samsung 980 Pro. The 1TB version is 600 TBW for warranty purposes. Samsung is confident that you can write 600 TB to the drive to the point that they will replace it for you if it fails before then. Let's say you think you might keep it for five years (which happens to be the warranty period). 600 TBW ÷1825 days = 336 GB per day you would have to write to the drive to hit the TBW limit before the five-year warranty is up, or about one third of the drive's total capacity. There might be days here and there when you write that much but mostly I'd imagine you'd be in the single-digit GB's per day for daily use, if even that, unless you're doing something very data intensive. (Third-party tests have shown that these drives generally work well beyond the warranty TBW limit, as well. It's not like you have to throw it out if it does hit 600 TBW.) Bottom line — Don't worry too much about your SSD drive health with regards to writes unless you're doing something extreme. It does have a limited number of writes, but that number is still so high that you're unlikely to reach it during the drive's practical lifetime. Not to say that SSDs can't fail... It's just not often from "hitting the write limit" (in my experience). Have a backup strategy for any important data. Rule of thumb for me is at least 10% free space (unless maybe it is a very large drive, then perhaps less than 10%). That's more because I don't want to be worried about running out of space by surprise/accident than because of performance considerations, but having the drive not being completely full can help the SSD do its thing (garbage collection / wear leveling / etc.) more effectively.

BIOS update 1.16.0. https://www.dell.com/support/home/en-us/drivers/driversdetails?driverid=Y2V4K - Firmware updates to address security vulnerabilities including (Common Vulnerabilities and Exposures - CVE) such as CVE-2022-34400, CVE-2022-21198, CVE-2022-26845, CVE-2022-29893, CVE-2022-27497, CVE-2022-33159, CVE-2022-26047, CVE-2021-23223, CVE-2021-37409, CVE-2021-44545, CVE-2022-21212, CVE-2022-21197, CVE-2022-21160, CVE-2022-21139, CVE-2022-21172, CVE-2022-21240, and CVE-2022-32482. I just installed 1.15.2 a couple of days ago...

Man, I have to use ≥300 nits (indoors) or the image just seems really dim to me. I do mostly work with "dark mode" applications though. (I do think that 500 nits is too high, I have to turn the brightness down from max on these newer panels. On the flipside, I think it's crazy that they are selling a 250 nits panel in one of these.) I typically work with turbo boost disabled during "productivity" periods specifically to keep fan noise under control. But I unlock it for gaming, and for that I just put on ANC headphones and the fan noise becomes irrelevant. Nice to be able to have it both ways.

I didn't mention in my post above... "blacks" in the 17.3" 4K panel are definitely "meh". It's not too at all when you are doing "computing" (web browsing, office-type work, coding, etc.) but I think it is pretty noticeable if you get into photos/videos/games. I did complain about it some pages back but I have gotten used to it and "fixed" the backlight bleed so I am good with the panel now. But, if you do creative/graphics stuff then I think it would be hard to suggest anything other than the IPS panel (unless maybe you mostly work with the laptop docked). I'm not into the creative/art space, and myself, I'd also rather have a non-glossy panel that is not covered by glass, I don't really like the glare or reflection. (I was on a system with a glossy panel from like 2008-2012 and I always wished I didn't have it.) So I do not regret choosing the 17" IPS over the 16" OLED. I guess, at least the choice is there so you can weigh it in when deciding which model to get, rather than being forced down one path or the other.