Thank you, Duracell

I’ve mentioned before on this blog that I’m a hobbyist photographer. Back in December I had a photoshoot with a then-10 year-old girl at a local park under Christmas lights. Knowing I’d be using my flash, I bought an extra set of AA batteries – specifically the Duracell Optimum AA 4-pack.

Didn’t need the batteries during that shoot, so I just held onto them. Fast forward to a couple Fridays ago. I had another night photoshoot with a local graduate. And during the shoot the batteries in my flash kept overheating. So I opted to try to switch them out for the spares, the aforementioned Duracell batteries. But when my wife and I went to open the packet, we discovered the batteries had corroded. An unopened, unused set of batteries.

So a couple nights later, I wrote in to Duracell through the online form. (Unfortunately I didn’t copy off what I sent.) And I attached a few pictures of the batteries and the packaging.

I’ve always believed that what sets a company apart isn’t whether they managed to distribute defective items, since that’s an inescapable reality, but what they do when a customer reports having one. In this instance, Duracell sent me a replacement set of batteries, which I received tonight.

An 8-pack replacement set of AA Duracell Optimum batteries.

Responding to a knife with a firearm

LA Times Letters to the Editor: “Why do police officers keep shooting people armed with knives?

Oh this is a good one…

To the editor: I read with interest Steve Lopez’s column on another police shooting of a man armed with a knife. These shootings first caught my eye in 1979.

At the time, I was in law school when two Los Angeles Police Department officers shot Eula Love multiple times. The 39-year-old Love was holding a steak knife and arguing about why her gas was turned off over a $22 bill.

After reading of these incidents over 40 years, I have picked up a couple of themes. First, officers across the country are permitted to use combat fire, where they empty their guns at the suspect irrespective of opposing risks.

Second, police do not resort to alternative weapons, such as bean bags or even .22-caliber short bullets, often enough. Each LAPD squad car is loaded with special technology; is a 9-millimeter semiautomatic handgun the only option?

If we agree a solution is needed, and we do, my first suggestion is a requirement to hire only college graduates with a degree in criminology.

Kevin H. Park, Westlake Village

It’s interesting that he’d call out “.22-caliber short bullets”, which I presume he means the .22LR, as if it’s somehow less lethal than a 9mm pistol. Out of the right firearm, it’s actually MORE lethal.

But let’s get to the base question: why do police officers use their firearms against a threat armed with a knife? For much the same reason they will shoot someone carrying a firearm if that person refuses to follow orders. When someone is armed with a deadly weapon, on which I hope we can all agree a knife qualifies, the only logistically-sound way to neutralize that threat is with deadly force. And the best deadly force for neutralizing a threat while still keeping distance from that threat is… a firearm.

How villains use knives in movies is far different from what happens in reality. In reality, a person armed with a knife can close distance and kill or seriously injure a target very quickly. Even in the hands of someone who isn’t all that skilled, a blade weapon can pose a significant threat.

This video is older, but it still shows the reality of how quickly a situation can change when you don’t know the person is armed with a knife. But even if you know they have one and the knife is in plain view, it’s still possible to get surprised and ambushed.

And that is why police officers respond with deadly force.

Police shootings are still very much in the spotlight after several high-profile shootings in recent years. And while it has rightly caused people to question the use of deadly force by police, it has also revealed the ignorance of the general public with these type of circumstances. Since many believe a person armed with a weapon that is not a firearm, or isn’t armed at all, cannot be a deadly threat to a police officer or civilian. And such is evident in the above letter to the editor, and the continual emphasis on “unarmed” with regard to Michael Brown and Trayvon Martin.

A tip for street photographers

I’ve been a photographer for over a year and a half. Hobbyist only. Not really doing anything professional yet.

And I love shooting pictures of strangers, since it’s always different, always new. Indeed a lot of the pictures I’ve taken have been of strangers out in public, with their permission. And when I do that, I tend to also ask if they want the pictures and hand over a business card if they do. But with one little trick.

On the back of the business card I write a 4-digit number. Not just some random 4-digit number, but a number off one of the picture files from the camera. This way when they write back into me, I can easily tell who is writing in if I happen to photograph multiple people when I’m out. Not everyone contacts me. And I didn’t always do this, either.

But it proved fruitful today.

Back on June 14, I took a couple pictures of two women out biking, and gave them a business card with a corresponding 4-digit number on the back. If I don’t hear from someone for at least a week, I presume I’m not going to. So I wrote off ever hearing from them. But one of the two wrote to me today to request the pictures, noting the 4-digit number on the back of the card while also noting they were out biking. My card got buried in her bag, and she found it when cleaning it out. Without the number, I would not have been able to locate the pictures without asking a few more questions.

So moral of the story: if you take pictures of someone in public, don’t just hand them a business card. Write some additional information on the back so you can locate the pictures afterward, such as the 4-digit file number sequence from your camera.

They’re adapting

This one is a little different from the previous ones I’ve received, but still the same pattern. Interestingly this time they’re demanding Ethereum instead of Bitcoin. I actually used to mine Ethereum several years ago, the money from which helped fund the purchase of my house and the move into it.

Hey, some time ago your computer was infected with my private software, RAT (Remote Administration Tool).

My software gave me access to all your accounts, contacts and it was possible to spy on you over your webcam.

That’s why I know that at the time of infection your password was: [REDACTED]

Sometimes I was spying on you and then once I was shocked seeing you started to MASTRUBATE, so I recorded you with the software called: Bandicam, Google it if you want.

I can share the video of you with all your friends, contacts, post it on social networks and everywhere else.

You can stop me, send 500$ with the cryptocurrency Ethereum (ETH).

It’s easy to buy Ethereum (ETH), for example here: , , , , or Google another exchanger.

My Ethereum (ETH) wallet is: 0x4562B3eEa33b3eb4Ed2e08719A05421e06E452f4

Yes that’s how the wallet looks like, copy and paste it.

After receiving the payment, I will remove everything and you never hear from me again.

You got 2 days time!

Next time update your browser before browsing the web, so you won’t get infected again!

Update 2020-07-16: I received this exact same e-mail again this morning. With the same ETH wallet. Sadly it does appear that one person has paid this fake “ransom”, so spread the word that these type of e-mails are not what they claim to be. They don’t have a video of you “enjoying yourself”, and the scammers blast these e-mails out to thousands of e-mail addresses in an automated fashion hoping someone will pay up.

A more intelligent radiator box

Build Log:

Last I left this series, I mentioned a next step was making the radiator box smarter. The current iteration of the radiator box now includes intelligent control and monitoring of the fans and monitoring of the pump, liquid flow rate, and temperatures. I also removed the DDC pump and figured out how to get the D5 back into the system while also upgrading the power supply. Sounds like a lot went into this, doesn’t it?

First, here’s the current bill of materials (not including fittings, tubing, and wiring):

And obviously you’ll need fittings, tubing, and wire. I used terminal blocks for helping with wiring all this up, so I’d recommend either that or bus bars. It just made things a hell of a lot easier.

Corsair Commander Pro

I mentioned a couple ways to approach this. Initially I went with the Corsair iCUE Commander Pro since it has temperature sensing (standard 10kΩ resistance sensor). The NZXT Grid+ doesn’t have that. Monitoring coolant temperature allows you to set up curves based on that coolant temperature.

I have two temperature sensors: one before the radiators and one after the radiators and before the reservoir. With the iCUE software, I could set up fan curves based on the coolant temperature. The rear fans – Nanoxia Deep Silence 120mm – would run at a near-constant, quiet RPM.

But there was one difficulty that I’d previously mentioned: it appeared to need all three voltages from the SATA power connector.

Or so I thought.

It will function without the 3.3V line. That voltage is likely required only if you’re planning to use LEDs.

Connecting to USB

Connecting this to my system, however, proved interested.

The USB connector on the Commander Pro is a 9-pin motherboard header connector, so I needed an adapter to turn that into two (2) Type A USB plugs, which could then be plugged into a USB hub to give one Type A plug. Which is where this comes in, the modDIY CAB138:

This turned the 9-pin header from the Commander Pro into two USB Type-A plugs. Then I used a USB hub to combine them. Probably could’ve gone with the version that gives a single Type-A plug, but I at least knew before I tried it that this would work.

Proof of concept

Before actually putting this into the radiator box, I needed a proof of concept to ensure it would still operate powered by a separate power supply than the main system. I connected the radiator fans to the first three (3) fan ports, and the rear fan trio to the fourth (4th) fan port. The two temperature sensors I plugged into the first two temperature sensor ports. I left the pump powered by the radiator box power supply. I used a separate power supply to power the Commander Pro.

And it worked beautifully.

But then it is a USB device, and I believe the USB standard requires devices to be able to work with a power supply separate from that of the system it’s plugged into if the entire device isn’t powered directly off the USB port – e.g. keyboard, mouse, and some external storage devices.

Back to the D5

I couldn’t get the PWM DDC to consistently run at full speed, or anything higher than minimum speed, for very long. So I decided to stop trying. The power supply I was using may have been part of the issue, but it was difficult to say for sure. And with having snipped off the plug to screw it into a terminal block, I couldn’t just plug it into another power supply to test.

So back to the D5, which is not PWM.

The vibration isolation mounts I had previously are 10x10mm M4 (from AlphaCool). And I ordered 13x13mm M4 vibration isolators instead. Couldn’t find the original mounts, and didn’t want to wait long for replacements.

Wiring everything up

While waiting for the new vibration isolation mounts to arrive, I drained the loop and tore down the box with the intent of wiring everything up. I did not cut the SATA plug off the Commander Pro – which turned out to be very fortunate.

I retained the terminal block, but cleaned up the wiring. The Commander Pro meant I could remove the voltage regulator I was using to lock the fans at 7V. In its place went another 4-circuit terminal block for wiring up a SATA connector to power the Commander Pro, with a step-down regulator connecting the 12V line to the 5V line.

Along with controlling the fans and watching the pump speed and coolant temperature, I also incorporated a flow monitor: Koolance INS-FM17N and the corresponding frequency adapter ADT-FM03, which converts the flow meter’s signal into an RPM signal. The Commander Pro, D5 pump, and flow meter all mean more power is needed for this as well.

I’d thought the 12V/5A power supply I bought for this was running borderline – though a Kill-A-Watt tells me otherwise – but then the DDC pump also wasn’t running at full speed or anywhere close. But to be on the safe side, I opted to upgrade the power supply to an 8.5A model, which also provided two (2) 12V outputs. This presented a slight challenge to the wiring.

With the previous iteration, the vandal-resistant switch – rated at 5A – was inline with the 12V circuit. And even then that wasn’t a great idea: you should always have headroom with electrical components, such as by not using a 5A rated switch on a circuit that could have 5A current flowing through it. Instead I should have always had it connected on the AC hot line.

And the prospect of current exceeding the switch’s rating left me with no choice this time. The switch is rated at 250VAC/5A, well above the rated current for the power supply, meaning it’s suitable to be used as a DC or AC switch up to the rated current.

Out with the Commander Pro

Not long after I completed the upgrades to the box, the Commander Pro started having… issues. Fan ports were dying off one by one. Port 2 died first, then port 1.

Doing some research, a bunch of forum posts I found said the Commander Pro has issues with fan splitters. Three-way splitters notably, which is how I had the fans connected: 12 fans in four (4) groups of three (3). Nothing in the manual or documentation says to not use splitters, or to not use three-way splitters. Indeed the manual implies they will work: max current per fan channel is 1A, 4.5A combined across all six (6) channels.

I have Cougar CF-V12HB fans on the radiators. The sticker on the back says 0.2A at 12V. The cross-flow fans are Nanoxia Deep Silence 120mm 1300RPM fans, which are rated at 0.16A. So well under 1A per channel – .48A for the Nanoxia group, and .6A for each group of the Cougar fans – and only about 2.3A combined for the fan channels. So what gives?

So out with the Commander Pro. The Grid+ V3 isn’t an option here as well since it supports at most 5W (0.4A) per channel, meaning only 1 or 2 fans. Plus it doesn’t have temperature monitoring.

So back to the drawing board. Along with filing a warranty replacement claim with Corsair.

Aquacomputer aquaero

This was about the only way I could see to get everything working optimally. My choice ultimately was the aquaero 6 LT or the aquaero 5 LT, with or without a Power Adjust 3.

The aquaero 5 and 6 both have only 4 fan channels. The aquaero 5 LT supports up to 1.65A per fan channel, while the aquaero 6 supports up to 2.5A. Either would make a good option for the radiator box. For the aquaero 5, I could split the nine (9) radiator fans between two fan channel. Fan controllers in the aquasuite can be set up for multiple outputs.

But having the fans split would mean a more complicated setup. Using the PowerAdust 3 would also mean a more complicated setup since I’d have to figure out how to run power to it. With the aquaero 6 LT, on the other hand, I could have all radiator fans on one channel using one of the NZXT Grid fan hubs I already have.

So ultimately I went with the aquaero 6 LT. Here’s how I have everything connected:

  • Fan 1: Radiator fans
  • Fan 2: Rear fans
  • Fan 3: RPM sensor from pump
  • Flow 1: Flow meter (calibration = 200 in aquasuite)
  • Temp 1: Radiator out to reservoir
  • Temp 2: Radiator in from system
  • Temp 3: Internal air temperature sensor

Conflicting signals

One… issue with some fan hubs and splitter cables: they don’t isolate the RPM signal to just one connection. The NZXT Grid was one, and the ModMyToys splitters also don’t isolate the fan signal to just one fan. That signal isolation is necessary or the competing signals will confuse whatever is reading it, meaning you won’t get an accurate RPM signal.

The PowerAdust 3 provides for that signal isolation without the need to modify it, meaning it’s the ideal out-of-the-box solution. Modifying one of the Grid hubs, however, was pretty simple: clip off the RPM signal pin from all but one of the plugs.

Rebuilding the radiator box

Rewiring the radiator box wasn’t difficult, though there was quite a bit that needed to be undone. I also took this as a chance to do what I should’ve done: move the power plugs for the fans to the other end of the radiator so they aren’t crowded against the rear fans, which has led to fans being inadvertently disconnected. Plus they reach the Grid while still being able to be tucked out of the way.

The aquaero LT is an aquaero XT without the LCD display. On the back of the device is a pin array that would connect to the screen and control panel – not sold separately from what I can find. The standoffs that are preinstalled allow clearance for those pins, but it complicates mounting it, since I can’t just put double-sided tape on the back of it.

The aquaero is designed to fit into a 5¼ drive bay, so a 5¼ drive bay cover from an old black chassis works out for this quite nicely. Just needed to drill holes for the screws. If you don’t have any of these lying around, you can buy them online for dirt cheap, or just use a strip of aluminum – which I nearly did till I realized this as an option.

I bought a sacrificial LP4 extension cable with a right-angle connector for powering this from the terminal block. The rest of the wiring was pretty straightforward. The pump and flow meter are powered from one of the 12V circuits, and the aquaero is powered from the other with the 12V to 5V regulator providing the 5V line.

Note: I put the aquaero on the radiator box wall before I started in on the power cable. Made it much easier to get near-exact lengths.

All fans are plugged into the fan control ports on the aquaero, with the pump RPM sensor plugged into Fan 3. The flow meter is plugged into the Flow 1 port. The flow rate that is displayed by aquasuite is 10x the actual approximate flow rate, though, as I can’t enter a calibration value high enough to compensate. I’d need to enter 2000 (1000 is the max allowed), so I have it set at 200.

For the time being, I have the USB cable for the aquaero plugged into the USB adapter and hub shown earlier. For the final USB connection, I want something that could be panel mounted so I don’t have a cable dangling out of it. I’ll figure that out later. That would be really the only change to this box that I would make aside from, perhaps, swapping out fans.


Since the time I finished the radiator box and published this article, I upgraded my system from the i7-5820k to the Ryzen 7 3700X with an X470 mainboard. The mainboard and CPU were the only components to be swapped out. So with the D5 pump back in the box and the fan curves configured, how is the performance?

For reference, the CPU is overclocked to 4.3GHz and the GPU is overclocked to +125 on the core, +500 on the memory through MSI Afterburner.

Running OCCT on “CPU: OCCT”, Test mode: “Small Data Set” with Furmark in the background at 1080p windowed, it took about 30 minutes for the coolant to reach equilibrium with the fan curves I had set, topping out a little north of 32.6°C (about 90.7°F) at the radiator inlet, a little under 31.5°C (about 88.7°F) at the radiator outlet (coolant going back to the reservoir), with the radiator fans at about 836rpm. (Note: the time showed on the OCCT test is after I had to restart it after an error).

Click on the image to see it full-size. Word of caution: it’s a 4K screenshot.

As you can tell, the fans are well under 1000RPM, the general threshold for “silent” for even already-quiet fans, so there was no fan noise and virtually no pump noise as well. Any noise from this setup is coming from the fans in the NZXT H440, which I’ve been meaning to replace for a while and just haven’t gotten around to.

I added in the PowerPanel image to show how much power my entire system (including my dual 4K televisions) is drawing from the wall under load. It peaks at a little over 500W.

References   [ + ]

1. I had to special-request the 9x120mm side panel by placing an order for an 18×18 panel and adding in the order comment that I wanted the 9x120mm panel.
2. Appears to have been discontinued, 1500mm PWM is available
3. The AlphaCool clear acrylic housing has since been discontinued
4. As of this writing, I continue to support Singularity Computers via their Patreon

More of the same

Slightly different from previous e-mails I’ve received, but otherwise still more of the same.


I’m a hacker who hacked your operating system a few months ago.

This means that I have full access to your account:
At the time of hacking your account([REDACTED]) had this password: [REDACTED]

You can say: this is my, but old password!
Or: I can change my password at any time!

Of course! You will be right,
but the fact is that when you change the password, my malicious code every time saved a new one!

I’ve been watching you for a few months now.
But the fact is that you were infected with malware through an adult site that you visited.

If you are not familiar with this, I will explain.
Trojan Virus gives me full access and control over a computer or other device.
This means that I can see everything on your screen, turn on the camera and microphone, but you do not know about it.

I also have access to all your contacts and all your correspondence from e-mail and messangers.

Why your antivirus did not detect my malware?
Answer: My malware uses the driver, I update its signatures every 4 hours so that your antivirus is silent.

I made a video showing how you masturbate in the left half of the screen, and in the right half you see the video that you watched.
With one click of the mouse, I can send this video to all your emails and contacts on social networks. I can also post access to all your e-mail correspondence and messengers that you use.

If you want to prevent this, transfer the amount of $950 to my bitcoin address (if you do not know how to do this, write to Google: “Buy Bitcoin”).

My bitcoin address (BTC Wallet) is: 1PNzJwB1CuVnKqKJQnu31E5ckiz9VxTcND

After receiving the payment, I will delete the video and you will never hear me again.
I give you 48 hours to pay.
I have a notice reading this letter, and the timer will work when you see this letter.

Filing a complaint somewhere does not make sense because this email cannot be tracked like my bitcoin address.
I do not make any mistakes.

If I find that you have shared this message with someone else, the video will be immediately distributed.

Clever encoding

Another scam message came to my inbox today, and it used a rather clever encoding. It looks like plaintext, but it isn’t. It uses high-order 32-bit Unicode codepoints (encoded as UTF-8) to construct the text. The codepoints specifically are part of the Mathematical Alphanumeric Symbols group.

So it looks like plaintext. You can read it like plaintext. But it isn’t plaintext. And the codepoints being higher than the UTF-16 boundary means that WordPress’s editor screws up trying to handle it. Nodepad++ had no issues with it though. I’ve posted a screenshot below, and attached the redacted source text from the e-mail.

Click (or tap) the image to see it full size.

* * * * *

And the plaintext (the Bitcoin address was the only part NOT encoded in those high-order Unicode characters):

I am aware, REDACTED, is your password.

I require your full attention for the coming Twenty-four hours, or I may make sure you that you live out of shame for the rest of your lifetime.

Hello there, you do not know me. Yet I know everything concerning you. Your personal facebook contact list, mobile phone contacts and all the virtual activity in your computer from past 175 days.

Which includes, your self pleasure video footage, which brings me to the primary reason why I ‘m crafting this particular e-mail to you.

Well the previous time you went to see the porn webpages, my malware was activated inside your pc which ended up shooting a eye-catching video footage of your masturbation act simply by activating your webcam.
(you got a exceptionally unusual taste by the way lol)

I have got the complete recording. If you feel I am playing around, simply reply proof and I will be forwarding the particular recording randomly to 9 people you’re friends with.

It might be your friends, co workers, boss, mother and father (I don’t know! My system will randomly pick the contacts).

Will you be able to look into anyone’s eyes again after it? I doubt it…

But, it doesn’t need to be that way.

I want to make you a one time, no negotiable offer.

Buy $ 2000 in bitcoin and send them on the below address:

[CASE SENSITIVE so copy & paste it, and remove * from it]

(If you do not understand how, look online how to buy bitcoin. Do not waste my precious time)

If you send this ‘donation’ (why don’t we call it that?). After that, I will go away and under no circumstance contact you again. I will erase everything I’ve got concerning you. You may very well carry on living your normal day to day lifestyle with no fear.

You’ve 24 hours in order to do so. Your time starts off as quickly you check out this e mail. I have an one of a kind code that will alert me once you see this mail therefore don’t try to act smart.

So yeah, that’s definitely something new. I knew that Unicode can go out to 32-bit codepoints, but never saw anything like that used in this fashion. But it’s just another scam e-mail of the same variety I’ve posted here before.

IKEA BESTÅ wall-mount (or free-stand) rack cabinet

Looking online there are plenty of options for a wall-mount 19″ rack. Open frame. Enclosed cabinet with doors and accessible sides. And there are plenty of options on eBay as well that go for much less than what you’ll find most anywhere else.

But they all suffer from the same flaw: they stand out. Hang one on your wall or use it free-standing and it’s quite obvious that it’s a network cabinet.

My wife and I recently bought a house. And after a lot of planning and research for how to wire up the network in my home, I had a couple ideas in mind. The central network hardware in my home is a MikroTik CRS317 10GbE switch (buy it at Amazon or EuroDK) and my ISP’s router, both connected to an uninterruptible power supply. I have a separate wireless access point (Tenda AC1900), but it is located elsewhere in the house at the moment, and I’ll probably buy a second one to have full coverage in my home.

I also have a few sets of rack rails left over from some previous projects I’ve since taken down, so I initially planned on just building an open rack. But given that where I put that rack is likely to be in plain sight, that would’ve been… unsightly. And most any prefab option – whether it’s open frame or enclosed – again would’ve stood out wherever I hung it in the house.

But I didn’t go full DIY either. Kind of… in between.

IKEA’s BESTÅ system includes several cabinet frame options, two of which have measurements almost perfect for making a 19″ rack. They’re 600mm (23-5/8″) wide by 640mm (25-1/4″) tall, and 200mm (7-7/8″) or 400mm (15-3/4″) deep. Perfect for light-weight equipment, such as switches, routers, and patch panels. And they can either be used free-standing or mounted on the wall with the corresponding suspension rail.

The shallower cabinet works for patch panels and PDUs and other very shallow equipment – e.g the Bitscope BR04A. But the deeper option is great for lightweight switches that are 12″ or shorter, such as the aforementioned MikroTik switch.

Adhering to standards

The cabinet has an internal width of 22-3/16″ (~563mm to 564mm). The proper spacing for the mounting holes is governed by the EIA-310 standard. Since the rails I’m using are similar to what I used in a previous attempt to build an open frame rack cabinet from 2x4s, I already know to space the rails to about 19-3/16″ or 19-1/4″ (~488mm).

Meaning 1-1/2″ on each side for mounting the rails. Can you get any more perfect?

Additional materials and tools

Along with the cabinet and 12U rails, you’ll need these materials:

  • 2×2 boards (1.5″ x 1.5″ nominal), one for each rail cut to the length of the rails
  • 1/4″-20 lag bolts, 1″ length, at least 2 per rail
  • #4×1.5″ wood screws – such as Everbilt #808631 – or M3x1.5″ wood screws, at least 2 per rail

Building the cabinet

Assemble the cabinet frame as normal, including the back. If you’re going to be hanging this on the wall, add the suspension rail hardware to the cabinet as well.

Then it’s just a matter of attaching the rail to the 2x2s and attaching that to the inside of the cabinet. And it’s on that which I have suggestions from my experience.

The cabinet frame comes with two (2) screws, about 50mm long, for joining two of these together, and they go through the shelf peg holes on the inside of the frame. That they go straight through means they are perfect for holding a 2×2 to the inside of the frame. But to know where, you need to put the screws through two of those holes going out, that way you have a full pass-through for pushing back in.

The holes I selected were these two (click on the image to view it full-size):

If I were having more equipment in this than just two switches and a patch panel or two, I’d have probably 4 screws each securing the 2x2s to the sides just to be safe.

I mounted the rails to the 2x2s using 1/4″ lag bolts through the #2 and #11 rack units. This all went together for me in less than an hour.


Yes, the cabling is a mess, currently, as I haven’t consolidated all connections here yet. The blue cables are OM4 optical fiber, and there is also Cat6 cable in there, too. That is a GbE switch as well, giving me the option of using various devices off RJ45 instead of wireless – e.g. my work laptop.

This looks better than most pre-fab options, in my opinion, and was less expensive than most pre-fab options. I could’ve used wood that blended in better to the outside, such as oak 2x2s instead of the poplar, or lined them with veneer that matches the cabinet. But since this was going to be out of the way and mostly out of sight, I wasn’t concerned.

So this became the central networking cabinet for my home, where I moved my 10GbE switch and ISP router. With how far out the front the cables stick, unfortunately I wasn’t able to take advantage of one of the doors available for the cabinet. The switch, router, and uninterruptible power supply all fit into this without issue. Now I just need to finish consolidating connections.

Back to AMD

Build Log:

Mira started life back in late 2016 as my first foray back to Intel in… 17 years. Here’s a rough timeline:

  • 1998: 486 DX2 (used Gateway machine)
  • 1999: Cyrix 200MHz
  • Later in 1999: AMD K6-II 333MHz
  • 2001 (I think): AMD K6-II 500Mhz
  • 2003: AMD Athlon 800MHz (cartridge)
  • 2004: Athlon XP 2500+
  • 2007: Athlon X2 4200+
  • 2013: FX-8350

Then in 2016, rather than waiting for the first generation of Ryzen processors, I decided to jump to the Intel X99 platform with the i7-5820K. Now in 2020, it’s back to AMD with the 3rd generation Ryzen.

Old specs

For immediate reference, here are the prior specifications:

  • Processor: i7-5820k (6 core, 12 thread), OC’d to 4.3GHz
  • Mainboard: ASUS Sabertooth X99
  • Memory: 16GB EVGA DDR4-3200

The rest of the system is remaining the same. I’m not upgrading the storage or graphics card since… I just don’t feel the need to do so right now.

Why the switch?

For some reason the mainboard decided to give up the ghost. It wouldn’t detect the RAM, continually illuminating the DRAM_LED, and plugging in the USB port for the TUF Detective app on my phone wouldn’t connect. So the board is, in effect, dead. Might try for a resurrection later. We’ll see.

And for some reason, listings for it are several hundred dollars, used:

Seriously? Over 400 USD for a refurbished desktop mainboard? That’s more than I paid for mine when I bought it. New. Yeah, no thanks.

New specifications

So how far up the Ryzen tiers did I go?

I initially considered a B450 mainboard, and reserved one with my local Micro Center – specifically the ASRock B450 PRO4. But after doing a little more research into the board, I discovered that I’d need a little more headroom than what the B450 could provide.

Along with the GTX 1070, I also have a Mellanox 10GbE card, Samsung 950 PRO NVMe SSD, plus four (4) 1TB HDDs in RAID-0 that I use for additional storage. With the B450 board I was considering, that’s basically running right up against the limits of what that can handle.

The X470 board, on the other hand, has plenty of room with some space left over. I can use all of the SATA ports plus both M.2 slots with none of it sharing bandwidth. Plus there is plenty of room for the graphics card and 10GbE card since the X470 board has a PCI-E 2.0×16 slot, leaving the other PCI-E 3.0×16 slot open for a second graphics card or something else.

Overall it was the better way to go for my setup, and it’ll be the way to go when I upgrade Amethyst later this year. Most, though, would likely be happy with a B450 mainboard. If you have just a graphics card and M.2 SSD, go with the B450 chipset.

Note on ASUS X470 and B450 mainboards: if the box has a red dot near the UPC, it’s BIOS has already been updated to support the 3xxx-series Ryzen processors. Learned that when I bought mine at Micro Center.

Water block

Thankfully most waterblocks are made to be cross compatible. They have coldplates large enough for most desktop processors. So the only consideration is the mounting hardware. With most blocks, the manufacturers make available the mounting hardware separate from the block. This allows you to switch platforms without having to buy a whole new block.

So for my Watercool Heatkiller IV, I just needed to buy the AM4 mounting hardware – part no. 14072 – which was thankfully available through Performance PCs.

Unfortunately, though, I couldn’t keep the same block alignment I had with the X99 processor, since AMD’s mount isn’t square. This also meant I needed to replace the tubing. I didn’t have any clear tubing left over, and I didn’t want to buy a 10-foot box of it when I only need about 18″. But I did still have some UV Blue tubing left over from the maintenance on Desert Sapphire last year, so I opted to give my system a little more color.

Initial Benchmarks

The last benchmarks I have recorded are from two (2) years ago. My 5820k was running at 4.1GHz at that time, and I had about the same overclock on the GTX 1070 as I still have now.

  • Unigine Valley (Extreme HD): 4326
  • Unigine Heaven (1080p, everything maxed): 2612
  • 3DMark Firestrike: 16936, Graphics: 20101
  • 3DMark Sky Diver: 40476, Graphics: 66493

Here are the benchmarks for the Ryzen 7 3700X with a boost to only 3.8GHz on the stock Wraith cooler:

  • Unigine Heaven (1080p, everything maxed): 2523
  • Unigine Valley (Extreme HD): 4254
  • 3DMark Firestrike: 17660, Graphics: 19686
  • 3DMark Sky Diver: 45845, Graphics: 64486

Okay that’s a little interesting. Let’s see how this fares with an overclock.


High core count CPUs are not great overclockers. That’s just the way of it. Though periodically you do get lucky – the full GHz overclock I was able to pull off on my 5820k without touching the voltage. (I do wonder how high I would’ve been able to go before having to touch the voltage.) But with this Ryzen 7 chip, without touching the voltage, I could take it only to 4.1GHz.

And as you can expect, the benchmarks weren’t all that impressive compared to stock.

  • Unigine Heaven (1080p, everything maxed): 2514
  • Unigine Valley (Extreme HD): 4199
  • 3DMark Firestrike: 18196, Graphics: 20028
  • 3DMark Sky Diver: 47098, Graphics: 64749

A little more impressive on the Firestrike overall score, with the graphics score on par with the X99 scores. And the Sky Diver score saw a bit more distance from the X99 score as well, though the graphics score there is a little… interesting.

Temperatures were also very impressive. Now I have an overkill and somewhat unusual water cooling setup. But that setup also meant it was nearly as easy swapping out the X99 for the X470 as if it were on air. Under load with OCCT Small Data Set, the CPU maxed out and held at 60C, reaching toward 61C but not quite getting there.

The 5820k would max out in the low to mid-80s. But I also discovered the IHS had a bow to it similar to what JayzTwoCents discovered about one of his processors that he lapped before delidding:

The risk of that defect is higher with soldered CPUs, since solder solidifies when it cools, giving no play should it be applied too thick.

Had the processor not had that defect in the IHS, it likely would’ve been able to run much cooler than it was. I planned to lap my 5820k about the time the X99 board decided to give up the ghost.

Precision Boost Overdrive

And then there’s gem from BuildZoid:

Summarizing the changes relevant to my system:

  • Enable DOCP – already had it enabled
  • Precision Boost Overdrive: Manual
    • PPT Limit: 300
    • TDC Limit: 230
    • EDC Limit: 230
    • Precision Boost Overdrive Scalar: 2X

And did I see any improvement? Nope.

Setting the multiplier back to “Auto” and enabling these settings in all places where you can configure the Precision Boost Overdrive (of which there are THREE), the processor was able to automatically boost to 4.1GHz across all cores. But it also drew significantly more power and raised the core temperature by 20C. So a lot more heat, but no more performance.


Next step: Asus AI Suite.

Asus AI Suite III

An article on Asus’s Edge Up mentioned the AI Suite as a means of also tuning the CPU. And since that tweaked the voltage along with the multiplier, it was able to pull the CPU up to 4.3GHz. How did scores compare?

  • 3DMark Fire Strike: 18052, Graphics: 19874
  • 3DMark Sky Diver: 47201, Graphics: 64180

And… not much of a difference to benchmarks. This was enough to convince me to just… stick with the 41x multiplier. Especially since the AI Suite also pushed the CPU temperatures up well over 70C due to the added voltage. So not much of a performance gain, but a jump in temperatures and power consumption.

No thanks.


So to summarize, I went from X99 and 6 cores/12 threads to X470 and 8 cores/16 threads. The new processor runs on less power and produces less heat, yet performs about the same or better.

Glad to see AMD is back. And I’m glad to be going back to AMD.

Update 2020-04-19:

I decided to revisit overclocking the CPU, and was able to bump it up to 4.3GHz. The key was disabling Precision Boost Overdrive in the BIOS. So if you’re going to manually overclock your processor, that should be step 1: disable Precision Boost Overdrive in the BIOS. Tried to get to 4.4GHz, but it crashed almost immediately in Cinebench R20.

And the 4.3GHz manual overclock took the 3DMark scores a little higher (stock, overclock):

  • 3DMark Firestrike: 18277 (17660, 18196), Graphics: 19989 (19686, 20028)
  • 3DMark Sky Diver: 48009 (45845, 47201), Graphics: 64821 (64486, 64749)

So it was good seeing those scores actually improve with a better overclock.

Still the same scam…

This one came into my junk folder earlier today:

Your password ιs [REDACTED]. Ι kηοw a lοt mοre thηgs abοuτ you τhαn thατ.


Ι plαced a malwαre oη τhe pοrn websιte αηd guess what, yοu visiτed τhιs web site το hαve fun (you κηow whαt Ι meaη). While you were wατchιηg τhe video, yοur web browser αcτed as an RDP (Remοτe Desκτoρ) αnd a κeylοgger, whιch ρrοided me access το yοur disρlαy screeη aηd webcam. Right αfter τhat, my sofτwαre gαthered all yοur coηtacts frοm yοur Messenger, Fαcebοok αccοunτ, aηd email αccouηt.

Whαt exactly dιd Ι do?

I mαde α sρlιt-screeη video. The first part recorded the νideο yοu were vιewing (you’e gοt αn exceptiοηαl tαsτe haha), αηd τhe ηexτ pαrt recοrded your webcam (Yep! τ’s you \ dοιηg nasty τhings!).

Whαt shοuld you dο?

Well, I believe, $2000 is α faιr price fοr our liττle secreτ. You’ll mακe τhe ρaymenτ νιa βitcoin tο the below αddress (ιf you dοn’τ kηοw τhιs, seαrch “hοw τo buy Βιτcοiη” iη Google).

βiτcoin Address:bc1qg97suky3qqt97mztfzca0xcn0g5u75k9fera3v
(Ιt is cAsE sensιτιve, sο copy aηd paste iτ)


You haνe 24 hours to maκe τhe ρaymeηt. (I hαve α unique pixel wiτhιη τhis email message, and rιghτ nοw I κηοw thaτ yοu hαve read thιs emaιl). Ιf Ι dοη’t geτ τhe ρaymenτ, Ι will seηd your νιdeo τo all of your coηtacτs, iηcludiηg relatiνes, coworkers, αηd sο fοrτh. Nonetheless, if Ι dο get ρaid, Ι wιll erase τhe vιdeo ιmmediaτely. If you wαnτ evideηce, reply wιth “Yes!” aηd I wιll send yοur vιdeo recοrding tο yοur five frιends. Thιs is a nοη-ηegοtιable offer, sο don’t waste my time and yοurs by replying το τhis emαιl.


Update: In the time since I posted this, I’ve received quite a few more from different senders with different names attached, all with the exact same text otherwise:

The sheer number that I’ve received, including four (4) alone within a 24-hour period, along with the fact they all use the same text, is evidence enough it’s entirely vapid.