The dirt-cheap ASUS laptop I wrote about a while back served it’s purpose well over the last few months. This time around I was looking for a laptop that fit a few specific requirements:
- is Linux compatible
- chance of working with macOS (hackintoshable)
- handles casual Windows gaming
- upgradable RAM, ideally accessible internals
- 1080p screen minimum
The Dell brand came to mind here. Linux compatibility tends to be good on Dell (some of their XPS line even *comes* with Ubuntu), they tend to gave a decent chance of hackintoshing well (and no WiFi whitelist shinnanigans), and when it comes to upgrading, Dell has repair/disassembly/service manuals for every model right on their website.
I ended up settling on the Inspiron 5570 with an Intel i5 8250u, 1080p screen, 8GB RAM, and a 256GB SSD.
Why not the equivalent AMD Ryzen Dell Inspiron 5575?
I mentioned casual games above, and the Ryzen 2500U is admittedly a great option. The integrated graphics have nearly twice the performance of Intel’s integrated HD which makes it twice the machine for Windows gaming. And it tends to be a little cheaper (with a spinning-rust hard drive – if manually upgrading to an SSD it’s about the same). I’ve seen complaints regarding constant fan noise and lower battery life which may point to higher idle power on Ryzen, but in terms of bang-for-your-buck it’s a solid option.
Unfortunately where AMD hasn’t established a great track record (yet) is with Linux support – specifically on the GPU end. Don’t get me wrong, they’ve made great strides with AMDGPU and have been doing a lot of work in recent times to add features/support to Linux. But GPU instability on Ryzen was a problem for at least the first half of this year and it seems they’re often playing “catch up” rather than getting ahead of the curve.
I absolutely hope the AMD situation on Linux continues to improve (I know people are working hard on it), but for now, Intel’s slow-but-stable iGPU just made a little more sense for me. Intel’s habit of early kernel upstreams (even at the expense of leaking upcoming product details) and long history of heavy involvement with Linux open source development in general just gives them a big edge.
The one other aspect that came into play was that Intel CPUs can often be hackintoshed. This isn’t AMD’s fault by any means – I’m sure they’d love to sell Apple CPUs. It just hasn’t ever been in the cards.
A few observations between initial use and disassembly:
- The Canadian bi-lingual keyboard is physically a US-keyboard (a plus). That means no wonky enter key, and a regular-sized shift key.
- The 1080p screen is… well.. basic. It’s crisp enough, but color has a bit of that “washed out” look to it. Nobody would ever describe it as “vibrant”. HWInfo shows it as being a 6-bit BOE06A9 NT15N41 manufactured in early 2015 which would make it over 3 years old. Perhaps Dell got a deal on some old stock or has excess stock from a large purchase years ago. That said, it works and is very usable. Note that 1080p packed into a 15.6″ display results in much of the interface looking quite tiny, so if you struggle to read small text you may want to either go with a lower resolution machine or step up to a 17.3″ laptop.
- Touchpad is textured and has that “sticky” feel to it. Some people might prefer this, but I found that it makes it very tough to move the cursor by 1 pixel at a time for those moments I need careful precision. I have to slightly roll my finger if I want to move the cursor over by a pixel.
- The SSD included uses the M.2 connector slot. So you *can* add a physical 2.5″ drive (and I did).
- Onboard ethernet is 10/100 (RealTek RTL8101/2/3). Not gigabit… Okay, so most people connect to wifi anyway and plugging in is rare. The combination of plugging in and using more than 100Mbps of throughput is probably even more rare. But still… unless there was some substantive power savings to be gained by using this chip (or something else I’m missing), this just seems like pinching pennies. I mean… you’d almost have to work to *find* non-gigabit stuff these days.
- Disassembly was much harder than it should be. I’ll go into details next.
(above: Canadian Keyboard – physically a US-keyboard, but with additional characters labelled on the keys – click for larger image)
I popped open the repair manual on Dell’s site just in case their disassembly was exotic/tricky in some way. The short version to access the “guts” of the machine is this:
- Remove the screws from the bottom.
- Optical drive slides out.
- Couple screws that were previously covered by the optical drive can now be removed.
- Pry the base cover off (starting from a specific corner).
(note that you shouldn’t follow my short instructions – follow the repair manual as it has good detail and helpful images)
The reality of disassembly was that Dell hadn’t put in a screw all the way, which prevented the optical drive from coming out. After that fight, removing the bottom panel required a lot of prying, and extra care/coaxing to ensure the HDMI and USB ports weren’t ripped off the motherboard.
Disassembly was much harder than it should have been, and carried more risk of damage than it needed to. Dell really needs to clean things up in this area.
On to something a little more positive… Once the cover is off, you have:
- Great access to the Wi-Fi adapter (Intel AC 3165)
- Great access to the M.2 SATA hard drive (Samsung PM871b)
- Great access to the RAM (1x Micron 8GGB DDR4-2400 17-17-17-39) and additional empty ram slot
- Great access to the 2.5″ empty SSD slot, with the housing, connector, and 4 screws already included and taped down
- Great access to the CPU cooler + fan
- Great access to the battery
So unlike many laptops out there, Dell makes it easy to access all the main components. No need to pull off circuit boards or the motherboard to access something hidden on the other side. Heck, I think everything *except* the keyboard can be accessed/removed with the motherboard still in. A huge plus.
(above: RAM, overview (blurry – sorry), m.2 SATA, Wifi Card – click for larger images)
You could argue that a lot of Acer machines do a better job of making it easy to access the common bits (small panel to get at storage + SSD in the cheap Aspire 5’s at least), but Dell lets you get at everything and even gives an online manual with the details. If only that bottom panel came off more easily…
Linux (Ubuntu 18.04)
Not much to say here except that… it works swimmingly…? Ubuntu does work out of the box with most laptops these days anyway (especially if relying on Intel’s integrated graphics), but there is some hardware out there that has hiccups.
I’ve spent a good bit of time in Ubuntu on the Inspiron 5570, and no unexpected issues or problems. The fact that Dell has true Linux support in some of their XPS laptops (and thus obviously acknowledges that Linux usage is a thing) gives me continued hope for the future too.
Update: I did hit an oddity… the screen backlight was always full brightness on restart. The fix is to pop into the BIOS and change the brightness settings (both for power and battery) – just change them to *something* else. Then when you restart, Ubuntu starts to remember this setting correctly. A little digging showed that this may be a glitch in the Dell BIOS where it doesn’t really have a correct value stored in the BIOS somewhere on a new machine and it won’t write a correct value until you actually change the brightness in the BIOS. Once a correct value exists, Ubuntu starts remembering screen backlight intensities properly on reboot.
Hackintoshing (macOS 10.14 Mojave)
I only did some preliminary work here and since (a) hackintoshing is a niche area, (b) Mojave beta only recently got Coffee Lake (and by extension kaby-lake-r) support, and (c) Mojave support via standard hackintosh tools/kexts is in it’s infancy and any steps given will undoubtedly change, I’m just going to gloss over this area with some points for others experimenting:
- Minimum Mojave beta 4. Old versions will KP when you try to enable native (non-fake-id) CPU/GPU as they only support up to the original Kaby Lake series (7xxx processor) support.
- I used MacBookPro14,1 ProductName. Possible the 15.x ones may be a better match due to Coffee Lake models perhaps being closer to Kaby Lake refresh.
- The Inspiron 5570 has a DVMT Pre-alloc of 32M. Thus you need to enable a workaround or manually set the BIOS to 64M via the dangerous BIOS setup_var stuff (I went the dangerous route).
- Using common config.plist laptop stuff (ie Rehabman’s 620 plist.config) which already enable “FixHeaders” etc, the Dell SMBIOS fix has to be enabled.
- FakeSMC.kext and VoodooPS2Controller.kext for basic functionality.
- Lilu kext, WhateverGreen.kext to get graphics working (must use latest versions… if using older Lilu+IntelGraphicsFixup must use -lilubetaall)
- Latest acidanthera/AppleALC.kext + config.plist Audio Inject “13” for working onboard audio
- RealtekRTL8100.kext for onboard ethernet.
- ACPIBatteryManager.kext for battery status/indicator.
- SSDT-PNLF.aml + AppleBacklightInjector.kext for basic backlight control in System Preferences (hotkeys require more work).
Things untested or not working:
- WiFi card swap (required for WiFi), untested.
- HDMI output wasn’t working. May require manual work.
- MicroSD card slot not tested (RealTek RTS5129 if you’re interested in messing around and/or trying)
- Touchpad is recognized as a mouse (System Preferences does not see a touchpad). Appears to be I2C so likely requires manual work.
- GPU Power Management (checked via AppleIntelInfo.kext) does not appear to be working (only min/max states).
- CPU Power Management does seem to work (many states), but min frequency seen was 1.3Ghz (600-800Mhz can be found in Linux/Windows).
Short version: Very functional/usable but obviously needs some time/effort. May give it a try again once Mojave is out of beta. Then again, there’s a strong chance Apple’s going to ARM desktop/laptops within a few years and whether they go full ARM or dual product line Intel/ARM (which may keep hackintoshing viable if they don’t move too much to the T2-equivalent chip) is hard to say. If they go full-on ARM there may not be many years left for macOS on Intel, similar to what happened when PowerPC was dropped for Intel over a decade ago.
Casual Gaming (WoW)
The UHD 620 included in the Kaby Lake Refresh here should perform similarly to the HD530 and HD630 found in previous desktop chips (including mine) and in some really brief testing it seems fairly close. Bumping down the settings to “3” in the BFA Pre-patch I was looking at approx 30fps in a class hall (update: above 60fps in a Garrison). I didn’t do anything heavy, but based on performance of my HD530 I’d expect a pretty smooth experience most of the time at these lower settings with the exception of massive stuff (ie 40v40 AV turtle when everyone shows up).
So expect most of the games that run on recent non-Iris Intel HD graphics to run here, including the games mentioned in my previous ASUS Pentium N4200 writeup.
Temperatures, fans, and throttling!
At idle you can’t hear the fan. According to the sensors the RPM is 0, which would mean the laptop is passively cooled at this point (update: I ran the laptop without the bottom case and verified the fan is indeed off)
At idle, temps tend to sit between 40-50C.
Run a CPU-intensive task and the turbo kicks up to 3.4Ghz. After a few seconds the fan starts spinning up. The CPU gets into the 90-100C range and thermally throttles. A few seconds later the fans are at max speed (which is quite loud) and the thermal throttling ends. Another few second pass and the laptop begins TDP throttling.
At this point, 2.8-2.9Ghz is about the max that could be held under high load.
Take away the load and the fans drop in speed, becoming less loud. Once enough time has passed at idle, it cools down enough for the fans to become silent again.
I did try undervolting (Intel’s XTU did work). Note that this is a little risky because XTU saves your settings to the BIOS and if by chance the machine doesn’t boot anymore you could be in for trouble.
At -0.110v the machine would bluescreen moments after a stress test was started. Dell’s watchdog did catch it and revert the settings to stock, but that’s not something to rely upon.
At -0.100v the machine worked and survived some heavy stress testing. This had a slight positive impact when it came to delay-before-throttling, and also allowed a steady 3.1-3.2Ghz clock speed during the TDP-throttling.
I settled at -0.080v. You can see benchmark results at my “undervolting Kaby Lake R” writeup. After all the effort that went into it, I wouldn’t necessarily recommend chasing an undervolt for that improvement unless doing a lot of intense heavily multi-threaded tasks. Note that if you ever eat a bluescreen in Windows (or sleep/suspend in Linux from the sounds of it), there’s a good chance Dell’s watchdog service will remove your undervolt anyway.
As a final note, I did pop off the CPU cooler, cleaned the TIM and put on some fresh old Arctic Silver I had laying around. Idle temperatures dropped a few degrees, but the thin material over the CPU means that load temperatures bounce around enough (in the 90C+ range) to make a solid comparison there a little hard.
The only notable mention when replacing the TIM is that the 2 dies on the CPU have slightly different heights (or did on mine) which results in the cooler not mating perfectly flush – so if you reproduce this, re-attach then re-detach the cooler from the CPU after putting on your thermal paste to ensure you have 2 die-shaped rectangles on the cooler… if there’s a clean spot left on the cooler where there should be a paste-colored rectangle, you need a little more paste on that spot to ensure you don’t get hotspots.
BIOS modding / IFR (warning: dangerous)
I decided to change the DVMT Pre-Allocated for hackintosh testing, and figured I may as well save the undervolt I had settled on in a more permanent fashion. This requires delving into the IFR.
(above: writing variables directly to the EFI. Not for the feint of heart)
It’s also highly dangerous, so I don’t recommend it. Few reasons:
- Dell hid these settings, and do not expect/intend for you to modify these. Even if everything goes correctly they could change something in a BIOS update that has the side effect of your old setting now doing something different and killing your system.
- A typo may brick your system.
- The wrong setting may brick your system.
- You really need to get the IFR from your own machine – the models all have different IFRs (for example the XPS one is different, I verified this).
- Using a setting from an incorrect IFR will probably brick your system.
- If your system won’t boot, you can’t undo the change. You *could* try disconnecting the battery and pulling the small cell battery, then hope the BIOS resets. You also *could* try one of the Dell BIOS recovery methods via a file on USB or hard drive. But if neither work, your machine is now dead.
That said, for those curious about it, I’ll paste some details in case you want to obtain your own IFR from your BIOS and compare to mine (and the notes I made for mine). Again, delve into this at your own risk.
Getting your own IFR from your BIOS (this guide was for an XPS, but while values are different the process to obtain the IFR is the same):
My own IFR, just for curiosity and comparative purposes – delete this after reading through it so you don’t accidentally mix up mine with your own IFR, write the wrong variables, and destroy your system. Note that I myself almost made this mistake when I was comparing IFR locations/values between 2 IFRs:
My own notes from my own IFR are below. Again, just for informational purposes – there could even be typos. Don’t use it. Use the one you create from your own machine:
Inspiron 5570 Bios 1.1.6 - i5-8250u, 1080p, 256GB SSD, 8GB RAM DVMT Pre-Allocated, set to 32MB (default): setup_var 0x795 0x1 DVMT Pre-Allocated, set to 64MB: setup_var 0x795 0x2 If you look through the IFR, you will notice other options here (multiples of 4 from 4-60MB) via differently formatted values. I would only use 32MB or 64MB here, as the other values are different enough that it might do something else, and setting *any* multiple of 4 is uncommon on most PCs that allow setting via BIOS anyway. DVMT Total Gfx Mem, set to 256M (default): setup_var 0x796 0x2 DVMT Total Gfx Mem, set to 128M: setup_var 0x796 0x1 DVMT Total Gfx Mem, set to MAX: setup_var 0x796 0x3 CFG Lock, Enabled (default): setup_var 0x4ED 0x1 CFG Lock, Disabled: setup_var 0x4ED 0x0 DID NOT TEST XTU INTERFACE, SINCE INTEL XTU WORKED ALREADY XTU Interface, disabled: setup_var 0x65D 0x0 XTU Interface, enabled: setup_var 0x65D 0x1 OVERCLOCKING FEATURE MUST BE ENABLED FOR UNDERVOLTNG TO WORK OverClocking Feature, Disable (default): 0x65C 0x0 OverClocking Feature, Enable: 0x65C 0x1 MUST SET THE NEGATIVE OFFSET FLAG IN EVERYTHING BELOW - DEFAULT IS POSITIVE WHICH WILL OVERVOLT Core Voltage Offset Negative: 0x664 0x1 Core Voltage Offset: 0x662 0x19 (should result in 25 aka 0.025v) Core Voltage Offset: 0x662 0x50 (should result in 80 aka 0.080v) Core Voltage Offset: 0x662 0x0 (default 0.000v) PROCESSOR GRAPHICS VOLTAGE OFFSET GT Voltage Offset Negative: 0x86D 0x1 GT Voltage Offset: 0x86B 0x28 (should result in 40 aka 0.040v) GT Voltage Offset: 0x86B 0x14 (should result in 20 aka 0.020v) GT Voltage Offset: 0x86B 0x0 (default 0.000v) PROCESSOR GRAPHICS VOLTAGE OFFSET FOR HD MEDIA? HAD NO EFFECT CAN NOT FIND REGION FOR EQUIVALENT XTU GRAPHICS MEDIA OFFSET GT Voltage Offset Negative: 0x876 0x1 GT Voltage Offset: 0x874 0x14 (should result in 20 aka 0.020v) GTU Voltage Offset: 0x874 0x0 (default 0.000v) Uncore Voltage Offset Negative: 0x865 0x1 (was 0) Uncore Voltage Offset: 0x863 0x14 (should result in 20 aka 0.020v) (was 0) Note: Standard Long Term Power Limit Override, limits, and time windows did not work (0x4C5, 0x4C1, 0x4C6, 0x4C7, 0x4C8). Possible a different setting area (SLPC?) or another setting had to be enabled?. However most other addresses had 0x0, or non-0x0 variables that did not seem to make sense if converted to decimal.
Really, unless you completely understand the process, the values, and how to read/interpret your own IFR you shouldn’t even consider tinkering with this stuff. But because some people do and/or will tinker with this stuff anyway, hopefully having my data as a reference will help point you in the direction you’re hoping to go.
Keep in mind that if something goes wrong (and it may, even if you do everything correctly), you’ll have a dead machine. If you can’t afford that, you probably shouldn’t be tinkering with this stuff.
I’m quite happy with the machine overall.
Sure, the display is basic, and disassembly was a royal pain. Sticky trackpad and 10/100 ethernet are things I’ll get over. USB-C is one of those “would have been nice to have in case I care about it some day” aspects, but for most of the current Inspirons it only shows up in models with a dedicated GPU so given the choice between not-having-it or having-it-but-paying-more I’ll take the former.
Once the panel is off, maintenance/access is great. In this age of glueing and soldering everything together so that broken/old devices go to landfills (hello Apple), Dell definitely deserves credit for letting you swap out the innards.
It runs Linux (Ubuntu) very well. Heck, I had macOS running on it so we’ll put that in the plus column too.
At idle it’s also the quietest laptop I’ve owned. I was able to do some light web browsing and have a YouTube video going while Chrome downloaded installed all without the fans making a sound. The fans are annoying when they do spin up, but fortunately that’s rare under light usage.
For now, that concludes my review (or observations/experiences anyway). As I did with the ASUS, I’ll make updates if I come across anything applicable as time goes on.
If you’ve got an Inspiron 5570 (or the AMD variant 5575), feel free to share your own experiences below!