Who needs a PS5?

First a couple of pictures…

  • Shadow of the Tomb Raider
  • Cyberpunk 2077

Yep, that’s me playing AAA games, in rock solid 4K, HDR, 60 fps glory on my living room TV. And that’s all with no console or PC attached. Hm…

The secret is a combination of several pieces:

Together, these allow me to stream games running on Nvidia servers directly to my TV, with max’ed out graphics and latency low enough to effortlessly pull off Tomb Raider style platforming. Wow!

The available games come from my Steam, Epic Games, and Ubisoft libraries, so I’m not paying for games, even if the cost of the Shield Pro + RTX 3080 subscription was non-trivial. (Note that it’s still less than a PS5 over two years of service, and who knows, by 2023 I might actually be able to buy a PS5 in Ottawa. 😛 )

Some caveats…

  • My internet service is shared cable running on Rogers infrastructure, so if too many of my neighbours are watching Netflix, I imagine my experience will degrade. I haven’t seen this yet.
  • There is a long standing and irritating bug in the GeForce NOW service that causes you to need to log in to Steam every time you start a new game. My understanding is this is something to do with where the servers are running, versus some geofencing that Steam does. Regardless, until they find a workaround, I recommend adding on a wireless keyboard to your setup, since typing a 20 character, upper/lower/numeric/specials password with a game controller sucks.
  • Speaking of add on hardware, you will also need to get a game controller, since none is provided with the Shield TV Pro. I’m using an Xbox One wireless controller I already had over Bluetooth and it’s working just fine.

Overall, I’m extremely pleased with this setup. I’ve been missing the ability to play games in my living room since my son got my PS4. This is a great replacement.

120Hz FTW!

I have been using a 144Hz monitor for gaming for a while now, but I recently replaced the monitor I was using with the M1 Mac Mini with a 120Hz, 1ms G2G one.

Wow! With that monitor, Mac OS is as smooth as an iPad Pro. Ok, I get it. The ProMotion display on the iPad Pro is 120Hz, so of course it’s that smooth. The thing is, I’m not used to seeing that while working on a desktop. The difference is *immediately* obvious.

Problems on the M1 frontier

I hadn’t gotten around to writing about my newest Mac, even though I’ve had it for a while now. Unfortunately, now that I am finding the time, it’s because of an unexpected problem with it. Oh well, that’s life.

Backstory

A few months before the new M1 Macs were announced, my mom’s iMac (20-inch, Early 2009) started to fail for her. This machine had already had its HD replaced once and, from the symptoms I was seeing, it looked like it was on the verge of happening again.

Since I had stopped using my iMac to host GCW, and I had a gaming PC with enough horsepower that I could use it as a backup to make music, I wasn’t really doing anything mission critical on the Mac, so decided to wipe it and give it to my mom as an upgrade. That machine was incredibly badass for its day, and is still running fine. Between the processor upgrade (Core2Duo to i7), the Fusion Drive, the top (at the time) Mac graphics chip (780M), and the increased system RAM (4Gig to 24Gig), the new machine behaves easily >10x faster than what she had. Woot!

Of course, my secret plan in all this was to replace my iMac with a new M1 Mac when they became available. 😉

What I got

What I ended up getting was this…

Mac mini – M1, 16Gig, 8-Core GPU

Of course, this wasn’t my first Mac mini, and I had lots of peripherals lying around, so “downgrading” from the iMac wasn’t a big deal, but I have to say I was not prepared for how much faster this new machine is.

I can honestly, say that this is the most responsive Macintosh I have ever worked on, and that includes my (almost) current gen, max spec, work MBP, which cost four times as much.

Starting up an application? Instant. Starting an Intel-based application via Rosetta 2? Almost as fast (after the first startup). Gaming performance? Everything is faster, even comparing against the dedicated 780M in the iMac, and that includes games written for the Intel architecture.

Wow! To say I’m pleased with the performance would be a massive understatement.

And now the bad news.

Like any first generation product, there have been growing pains — Blutooth dropouts, USB issues, etc. Generally, these have been minor and new releases of macOS 11 are bringing improvements.

At this point though, there are two issues which are pretty serious:

Lack of support for M1
Because of Rosetta 2, most applications you run work fine on M1. However, the closer you get to things that are at the edges, like printer drivers and music software, the more likely you will be to see problems. I’m still nowhere close to having all of my music software running on the new machine, but at least my mainstay, Reason, works as does Logic Pro (of course).

The display flickering bug
This is a reasonably wide-spread issue that causes certain combinations of content being displayed to flicker in brightness and have faint vertical lines. The problem is quite bad, when it happens, but I usually only see it when watching videos with dark backgrounds.

Here is an example:

Notice how the whole screen flickers when the lightning flashes

It’s hard to get a good demonstration of the problem, because it usually only occurs when the display is in flux, but at one point I did manage to capture a video of a static case. Here’s a close up:

At this point, there’s no fix for the problem. Anecdotally, there’s some evidence that this is a software bug, which hopefully means it will eventually get fixed, but who knows.

Also, in case you think it’s just me, here are some links from around the net:

I have a case open with Apple, of course, but they weren’t able to provide any useful suggestions, and fell back to the old standbys:

I did try running in safe mode (with no effect), but the other two are particularly sad because neither of those are possible on an M1. Argh.

I guess we’ll see. Hopefully, this gets fixed at some point. I’ll update this post when something changes.

Mini PCs

As you can tell from my Raspberry Pi addiction, I’ve always been fascinated by “mini” PCs (i.e. very inexpensive computers in tiny packages). My latest foray into this world was triggered by the hope that I could set up my living room TV as a Zoom/Webex/Skype/Hangouts space for family get togethers.

I tried using a Pi for this, but it was a struggle, with a mix of failures from missing features to inability to recognize the camera to performance problems, etc. I think if I was only using one service, I probably could have fought to make it work, but it felt like it was going to be too hard to get all of the services going.

This led me to wonder whether I could find a cheap “actual” PC to do the job and coincidentally (or at least I hope it was a coincidence), Amazon decided to point me at a sale on these…

  • Terryza Mini PC

In typical Amazon fashion, this is called

Mini PC, Windows 10 Pro Mini Computer Intel Gemini Lake J4125 Processor(up to 2.7GHz) Desktop Computer, 6GB RAM 120GB ROM, 4K@60fps, Dual Wi-Fi, Gigabit Ethernet, Bluetooth 4.2

On sale, I was able to pick one up for less than 200 $CDN. This is a surprisingly powerful little beast at that price point. It’s about 13cm on a side, making it only a bit larger than a Pi in an Argon ONE case. It’s quiet, it drives my 4K TV fine, it even has an internal slot to add a SATA drive that I filled with a 1TB SSD I had lying around.

Now, I will say it’s not fast. It just barely has enough horsepower to display 4K video, but that did work for the YouTube and Netflix examples I tried.

As to the reason why I bought the box, since it’s running Windows, all of the standard teleconferencing services worked OOTB using a Logitech C920 video camera. I was even able to do a video call through my home NextCloud instance, which surprised me.

Overall I’m quite happy.

Pi, home theatre, and Harmony remotes

Just for grins, I took my Raspberry Pi in the Argon ONE M.2 case upstairs and plugged it into the living room home theatre. I have a small wireless keyboard/trackpad device (that I got with an old QNAP NAS, of all things), which I could use to talk to it. This worked, but the range was terrible — basically, I had to be within 50cm of the Pi for it to work. I have ordered a “real” Bluetooth wireless keyboard/trackpad from Logitech. I’ll let you know how well it works once it shows up.

I also have a Harmony remote, that I’ve been using with the living room gear, and I have to say this has been the best universal remote that I’ve ever had. For grins, I added the Pi as a “Computer” in the app for the remote, and was surprised to find out that the remote wanted to pair with the Pi over Bluetooth. After doing this I am now able to move the cursor and type text on the Pi from the remote app on my iPhone. This despite the fact that the Pi isn’t running Mac OS or Windows. Very cool!

Using the remote, I was able (from my couch) to run Kodi and VLC (both of which worked well) and was able to do some light web browsing. Typing text into the remote app isn’t a good enough experience to want to do anything serious there, but the Logitech keyboard should fix that.

Looks like “argonPi” has a new home.

It works!

Yes, my little temperature sensor app works with the real Sense Hat hardware. Here’s a picture.

  • Temperature sense app in action

Note: that picture is taken in low light so the shutter speed is long enough to catch all of the pixels. You don’t notice the LED scanning with the naked eye, but the camera really shows it (see picture below).

When that picture was taken, it was still in a case with a fan, but it was hard to see the display and the fan noise bothered me. I took it out of the case and added a right angle extender to the GPIO connector on the Pi, so the display stands up, which keeps everything cool enough that no fan is required. I then added four risers and the lid off another old Pi case with some rubber feet, which keeps the board off the ground (and also increases the airflow).

  • The new look — it’s showing “25”, but the scan lines weren’t all lit when I took the picture

It actually reads a couple of degrees too high, but the accuracy of the readings isn’t all that important as long as it tracks relative temperature values. I tested it by taking it out into my garage (in January), where the temperature dropped, the display turned red, and email was sent. Yay!

(See part 1 for more details about the app.)

Python ?

Wow. I have to say I never thought I was going to end up writing Python programs for fun and (no) profit. That is until I got a Raspberry Pi Sense Hat, at least.

The specific project I’m working on is a remote temperature monitor for an unoccupied site that I’m responsible for, to allow us to react if the furnace goes out there before the pipes burst.

Given the Python library for the Sense Hat, it was basically trivial to build a program that would watch for the temperature to go outside a given range, and send email to me if it did. It took a couple of hours longer than I expected it would to write, but my excuse is that it really was my first Python program. 🙂 (Also, it seems there are quirks around getting accurate temperature readings from the SH, so I’m still tweaking the program, but it’s basically working.)

Given the 8×8 display on the Hat, I even managed to provide a continuous temperature display. 8×8 is not enough pixels to make something look good, so you end up with something like this (as drawn by the Sense Hat Emulator developer tool):

Ugly but readable. For the inquisitive, here’s the “McQ’s First Python Program” version of a function to display the digits:

###########################################################
# Display two digits on the sense hat.
 

# Digit patterns
digits0_9 = [
    [2, 9, 11, 17, 19, 25, 27, 33, 35, 42],  # 0
    [2, 9, 10, 18, 26, 34, 41, 42, 43],      # 1
    [2, 9, 11, 19, 26, 33, 41, 42, 43],      # 2
    [1, 2, 11, 18, 27, 35, 41, 42],          # 3
    [3, 10, 11, 17, 19, 25, 26, 27, 35, 43], # 4
    [1, 2, 3, 9, 17, 18, 27, 35, 41, 42],    # 5
    [2, 3, 9, 17, 18, 25, 27, 33, 35, 42],   # 6
    [1, 2, 3, 9, 11, 19, 26, 34, 42],        # 7
    [2, 9, 11, 18, 25, 27, 33, 35, 42],      # 8
    [2, 9, 11, 17, 19, 26, 27, 35, 43]       # 9
]
     
def display_two_digits (a_number, color): 

    global digits0_9
    black = (0, 0, 0)
     
    if a_number < 0:
        negative = True
        a_number = abs(a_number)
    else:
        negative = False
         
    first_digit = int(int(a_number / 10) % 10)
    second_digit = int(a_number % 10)
 
    # set pixels for the two digits
    pixels = [black for i in range(64)]
    digit_glyph = digits0_9[first_digit]
    for i in range(0, len(digit_glyph)):
        pixels[digit_glyph[i]] = color
    digit_glyph = digits0_9[second_digit]
    for i in range(0, len(digit_glyph)):
        pixels[digit_glyph[i]+4] = color
     
    # set pixels for a minus sign for negatives
    if negative:
        pixels[56] = color
        pixels[57] = color
        pixels[58] = color
     
    # set bottom right pixel if number is more than 2 digits
    if a_number > 99:
        pixels[63] = color
     
    # display the result
    sense.set_pixels(pixels)

Of course, you pass in a color for the digits to show, so when the temperature is outside the range, you can show it in red. 😉

Anyway, feel free to point out all the noob Python programming mistakes. I’m always happy to improve.

Site news (again)

So, I realized my last post was probably a bit confusing. The reason I was seeing issues with my older posts was that I had moved them to a new WordPress install because, as it happens, I have once again moved GCW to a new machine. This time, it’s another Raspberry Pi, but this new install has a few nice things going for it:

  1. It’s an 8Gig model. I don’t actually believe I need that much space to run a web server most of the time, but I am running quite a few web apps on it now, and depending on what was going on, I was seeing some peaks on the 4Gig one that were pushing into swap.
  2. It’s a brand new, fresh install on SSD. The old one was still running off a MicroSD card, with the website content stored on a separate SSD. Although this gave me lots of storage for the website, it meant that the install was just that tiny bit more complex than a standard one, so I was forever fiddling with it when updating, etc.
  3. It’s even smaller! I love the size of the finished result. Here’s a picture to give you the idea. Note the size of the ethernet jack on the side for scale.

This time around, I’ve only got 500Gig of storage for the whole install including the website, but I’m no where near hitting that limit yet, so I’m very happy.

Test post

Most of my older posts seem to have extended entries for the titles. I’m not sure what’s going on, but I suspect it’s a side-effect of one too many export/imports. Something to figure out eventually.

Update: I managed to turn it off by hiding post abstracts. I’m not sure why this suddenly started happening however. Hm…