Aaron44126

There is no downgrade without a clean install. LTSC 2021 is not a "service pack". It is a whole new version of Windows (despite sharing a name with its predecessor). I have nothing to offer here other than what I have already said. I know it is not the "popular opinion" but I routinely move Windows installs around in such a way and do not have any performance issues after cleanup.

My comments in another thread about "upgrading" to Enterprise from Pro only apply to upgrading to "regular" Enterprise, not "LTSC" Enterprise. Upgrading to LTSC Enterprise requires a separate procedure which I have described in the top post of this thread. You will need a Windows 10 Enterprise LTSC product key for the specific version that you are upgrading to, and the install media as well. If you are upgrading from a current version of Windows 10, LTSC 2021 will be the only choice. You can't "upgrade" to an older version of Windows.

There is no reason for performance loss as long as you get the driver situation cleaned up properly. This might also mean uninstalling or disabling services associated with drivers on your old system that are no longer needed. To turn Windows 10 Pro into Enterprise, all you have to do is visit Settings and pick the option for "change the product key", and put in a product key for Windows 10 Enterprise.

My experience has been you can just take a drive out of one computer and put it into another computer, boot it up, and it will work fine. Windows will notice that it’s in a new computer and do a device discovery on the first boot. You’ll have to figure out the drivers and stuff (as you would with a new Windows install), and reactivate Windows, but that’s all there is to it. I’ve actually done this several times. If these are Dell systems, go to BIOS setup on the target system and make sure that it is NOT set in RAID mode for the disk controller. This method will fail if it is, because the system being moved over will not have the proper RAID driver installed. You'd get a BSOD with "inaccessible boot device". If you don’t want to “move” the drive, you can use cloning software to clone all of the data to a new drive and put that in the target system; effectively the same thing. This method did not work prior to Windows 8. On old systems you’d just get a BSOD if you did this without special preparation.

Quadro 2000M stock clock is, what, 550 MHz? It is a GF106 GPU chip from NVIDIA. That same chip is clocked at around 800 MHz in desktop GeForce cards. I think that it is safe to overclock with +200 MHz with no worries, as long as the thermals are staying under control. You might be able to go even higher with a modded vBIOS. These older Fermi/Kepler laptop GPUs had a lot of overclocking potential. NVIDIA was known to artificially limit the GPU speed back then.

I have had issues with NVPCF in the past, including BSOD with newer NVIDIA drivers... But right now it is working fine with BIOS 1.23.0 (which I installed last week). The NVPCF device is actually not showing in Device Manager at all. If I view "hidden devices" then it does show in the list, but the status says that it is "Currently, this hardware device is not connected to the computer. (Code 45)".

Do you have graphics switching turned off in the BIOS? Does it do the same thing if you turn it on? When I did Quadro M5000M + Precision M6700, it would have reliability issues along these lines unless I just kept Optimus on all of the time. (The same issue did not occur with Kepler GPUs.)

Enough is enough. SBF is in jail. https://arstechnica.com/tech-policy/2023/08/sam-bankman-fried-is-going-to-jail/ https://www.theverge.com/2023/8/11/23829004/ftx-sam-bankman-fried-bail-revoked-vpn-signal [Edit] Better write-up. https://www.npr.org/2023/08/11/1191362886/ftx-sam-bankman-fried-sbf-crypto-fraud

I think @Reciever is referring to your multiple back-to-back posts minutes apart (which appear to have been cleaned up), not necessarily your post about the fans more than a day later. This tool can control fans in Linux using the same mechanism. https://github.com/clopez/dellfan You can manually set EC fan control off and set a specific fan speed. I'm unaware of any tool on Linux which bridges this method with a fan curve offering like SpeedFan.

Unfortunately, the only things you can do with the fan "manually" are set the fan to off/low/high. Consistency mode is a "hack" I put in place to lock the speed at a slightly lower setting than what you get if you just select "low". The thermal settings (optimized/cool/quiet/performance) are Dell settings. This tool can only set them on newer model laptops. However you should be able to set them in the "Dell Power Manager" app that you can download directly from Dell. If you want to play with a custom fan curve, you can use SpeedFan. I have a post on setting that up here. It refers to an older tool "DellFanCmd" but you can use Dell Fan Management in place of that. It supports the same command line options, or you can use the GUI to turn off automatic EC fan control which is the only requirement for SpeedFan to work. Note that SpeedFan is constrained to the same rule, there are only three speeds it can set the fan to (off/low/high).

Consistency mode has specific requirements. Its purpose is to lock the fan speed at a low level, but you have to wait for the EC to "decide" to turn the fans on before they can be locked. The reason is to prevent the fans from turning off (I detest the constant on/off cycling). They will automatically be unlocked if the temperature rises above the threshold set. The status bar will show the current status of this mode.

Instructions given to the embedded controller to modify the fan behavior will persist past a reboot. A full shutdown is needed to fully reset it. Dell Fan Manager will remember its previous settings but they will not be "applied" to your system until you actually run the app.

No. If the program causes trouble (which is unlikely), fully power off the laptop and then turn it back on. There is nothing that it does that survives a full shutdown.

...It is linked in my post immediately above. As an aside, I'm having difficulties with the ML-1210 printer now. The driver doesn't work with Windows on ARM. There doesn't seem to be a good way to get it working with ARM macOS, either. Also, the printer has started picking up multiple sheets at a time from the input tray and cleaning it didn't help to stop this behavior. I'll probably be replacing it later in the year.

Eh? Being based on Chromium, I fully expect Microsoft to move to the same release schedule as Google in short order, as well as other browsers that run off of Chromium (Opera, Brave, Vivaldi, etc.). They won't want to be seen as "behind" on security. ...At least updates are handled in a mostly transparent fashion and most users won't notice (other than the occasional prompt to restart your browser, if you don't already shut your machine down daily). The weekly updates are only bugfixes. "Major" updates will continue at the four-week cadence.

Yes, it applies to one of the ports but not the other.

Found this picture. Its the checkbox on the right (which should be available to click if switchable graphics is enabled). If it is not there on your system, don't know what to suggest. (Despite the name of the option and the help text, it will permanently attach that display to the iGPU.)

It is listed as being under the "Video" section, which is where I found it on the M4800. https://www.dell.com/support/manuals/en-us/precision-m6800-workstation/precm6800om/system-setup-options?guid=guid-fafcef15-7de3-4df2-a3f1-00e66aba9d3a&lang=en-us

Fortunately in this case, the keyboard is cheap to purchase and easy to replace. That "easy to replace" part at least is no longer the case with their newer 2022/2023 systems.....

The option is something like "Enable dock Display Port through Integrated Graphics". If enabled, it will move one of the DVI/DP ports on the dock to the iGPU. (I used to use this with my Precision M4800.)

Check video section in BIOS setup. There is a checkbox that will move one of the ePort outputs to the iGPU. I think it is normal for the right-click GPU selector to be missing these days. It is “supposed” to automatically pick the appropriate GPU, but you can override in either NVIDIA control panel or Windows advanced graphics settings. Unlike with Windows, you should not have to do any modding to get the driver to load in Linux. Ideally, your distro has an NVIDIA GPU driver package in the standard repos… install that and you’re good to go?

It is around that spot. It will be pretty obvious. You need to make it "flat" so that it is in line with the other parts that cover VRMs on the card. Cutting/grounding down is the obvious way to do that, but "bending" would be fine if you can make it work.

You need to flip it over. The part that needs to be removed is on the "bottom" side (the side that makes contact with the GPU card). It is around that area and it will be pretty obvious.

I don't know anything with regards to the AMD GPUs or GPU heatsink.

Missed the "action" but some notes. Tesla M6 has the same specs as a GeForce 980M or Quadro M5000M, and there have been a number of cases of people using it in this system without issue, though you do have to forego direct video output and use it with Optimus. This also means that the HDMI and DisplayPort on the laptop will be non-functional. I believe that the display outputs are physically disabled on these cards so you can't get them back by flashing a different GPU's vBIOS on. (VGA port will work, it is hooked up to the iGPU. If you have an ePort dock, you can also configure one of the DVI/DisplayPort ports on the dock to attach to the iGPU, in BIOS setup.) Pascal GPUs need a vBIOS flash before booting Windows will work. With regular commercial vBIOS, you will get an ACPI BSOD when booting Windows (even if you are just trying to boot Windows PE or the installer). You have to flash an early "engineering sample" vBIOS on to get around this. We only have working vBIOS'es for Quadro P3000, P4000, and P5000; no GeForce cards. (Somehow, this is not an issue under Linux and you can boot it up with any Pascal GPU, without messing with the vBIOS.) ...But Turing GPUs will work with Windows without any issue from the vBIOS, if you can find one that you can physically mount. My recent experience with Optimus on Linux was that I had no problems at all with it rendering games out through Optimus. I tried both native Linux games/emulators, and Windows games running through Steam+Proton. No trouble at all. Things have improved a lot over the past several years and it is now basically "plug-and-play". However, I was working with a GeForce 3080 Ti GPU (Ampere) and I am not sure if NVIDIA has done the same work with earlier GPUs. I saw lots of driver notes about certain things only working with Turing and up. As noted, depending on which cooler you have in your system, you may have to dremel off a bit of metal in order to have it fit with a newer GPU. This is true on NVIDIA GPUs in particular, anything Maxwell and later has a VRM on the board in a different spot than it was on the Kepler GPUs and it will be "in the way" of a bit of the heatsink that sticks down off the bottom. I'm not sure what the situation is on AMD GPUs. There is a version of the NVIDIA heatsink that is flat on the bottom and won't need any modification. It is not very common and it has the same part number as the normal one that you do have to mod, so you can't tell without physically inspecting it. Just sort of luck of the draw which one you got.