I’ll admit it: I tend to upgrade more than I should. I spend money on parts for minor gains and then re-run my benchmarks, hoping to have bumped my scores up a little bit. I have a problem.
However, that’s only on my “main” machine that sits in my office. I’ve always had a lower powered machine in the living room for couch co-op games or games that are best suited to a gamepad. Sometimes this machine has been WOEFULLY out of date, so slow that only streaming was practicable. I recently gave away the machine I had been using, so it was time to buy another. Then, I had a thought: what if I could build a machine that was capable of AAA gaming for the price of a PS4 (standard, not Pro)?
The Test
In my locality, a PS4, tax included, comes out to roughly $324. That’s not a lot of money for a gaming PC, but I figured it’d be worth a shot. Off to the Internets! And cue disappointment. You see, I had just a few rules in this challenge. First, total price (shipping and everything) had to be $324 or less. Second, quad core minimum. Third, discrete graphics. Finally, a minimum target of 1080P.
On the Intel side, the best brand new system I could put together was:
- Core i3 9100F – $81
- MSI H310M motherboard – $59
- 2x Patriot DDR4 (total of 8GB) – $36
- DIYPC mATX tower – $24
- Raidmax 500W PS – $33
- SanDisk 256G SSD – $25 (nice sale!)
This brought me to a with-tax total of $280 or so. Leaving me a bit over $40 for a graphics card. This is where the bottom started to fall out. For $40 or so, I could get a Geforce GT 710 w/2GB of DDR3. Fail.
Well, what about AMD, you ask? Oh, you didn’t ask? I could have sworn…
- Ryzen 3 1200 – $65
- ASRock A320M – $55
- Rest of the parts are the same
That took me to a post-tax $258. That’s better than Intel, but only left me around $66 for a GPU. This got me a lightning-fast…R7 240 w/2GB of DDR3. Seems hopeless, doesn’t it? The rest of the system is probably alright, if a bit weak. But those GPU’s aren’t going to be able to run anything appreciably well.
One Man’s Trash
Finally, I turned to my old friend eBay. Perhaps there was something on the used market that would work? The very first machine I happened to notice was a used, barebones (no CPU/RAM/GPU) HP z420. It is a workstation-class machine, with an LGA2011 socket for Xeons. It could only go up to Ivy Bridge, but it covered the case, motherboard, power supply and even an optical drive for $115. Still, that was about where I’d be with the other builds. Let’s see what else we could find to round that out.
This is where it started to get interesting. I found a Xeon 1620 v2 (Ivy Bridge), a quad core, 3.7Ghz chip, for only $28 shipped. It got even cheaper with RAM. For 16GB of registered DDR3 at 1866Mhz, I was able to slide in at $33 shipped. I was even able to get a Samsung 512GB SSD (supposedly lightly used) at $25 shipped (auction, not Buy it Now). I was now at $201 shipped, leaving $123 for a GPU.
If I were buying a NEW GPU, that would probably only get me an RX 550 or possibly a 560. Used, however, I was able to score an RX 580 with 8GB of RAM for $120 shipped. Three dollars under budget, but will it game?
To give it a baseline, I am comparing it to my office machine. This consists of:
- Core i7 9700K
- GeForce RTX 2080
- 16GB of DDR4
- 256GB m2 SSD
- Other odds and ends
Results
In short, they’re very impressive.
Test | 9700K (Office) | 1620 v2 (Living Room) |
3DMark Firestrike (DirectX) | 21925 | 11284 |
3DMark TimeSpy (DirectX) | 10408 | 4206 |
Unigine Heaven (1080P, Ultra w/Max Tessellation, 8xAA) | 3510 | 1340 |
Unigine Superposition (1080P Extreme) | 6829 | 2482 |
GTA V (1080P, Very High) Pass 4 | 95fps Avg, 10.47ms Avg frame time | 35fps Avg, 24.67ms Avg frame time |
35 frames per second in GTA V may make this seem like a failure. I admit, it’s somewhat low, but that’s with EVERYTHING maxed out. Dropping the presets from Very High to High give me an average frame rate of 58. That’s still way nicer looking than XBone/PS4 at a higher frame rate, to boot.
Beyond the numbers above, which admittedly only feature one actual semi-current game, I usually go based on how games “feel”. I turn on all the eye candy and see if it feels sluggish. Do I notice frame drops? Is there pop-in?
While playing numerous games in my library on both systems, I can only point to one instance where the living room machine feels a bit slower than it should (7 Days to Die, a Unity game). Otherwise, it performed like a champ throughout a day of gaming. Games tested include:
- DOOM (2016)
- Wolfenstein: New Order
- Project Cars
- GTA V
- Elite Dangerous
- 7 Days to Die
- Civilization VI
- Stellaris (note: slows down late game on ANY system)
- World in Conflict
- FEAR
- Xonotic
Conclusions
I’d call this a success. Better performance than a PS4 or Xbox One for the same price? Check. Better value than new PC hardware for the same price? Check. Near flawless 1080P experience (with eye candy)? Yep.
Now, admittedly, my machine in my office games at 1440P, not 1080P. But if I were a struggling gamer on a budget (and believe me, I have been), I’d be tempted to look at the performance per dollar between these two builds. Sure, the office machine is double or more the performance in synthetic benchmarks, but breaking it down (using the TimeSpy demo and GTA V for reference):
Machine | Dollar per TimeSpy Unit | Dollar(s) per GTA V Frame |
Living Room ($321) | $0.076 | $9.17 |
Office ($2000 est) | $0.192 | $21.05 |
That’s a stark budgetary difference at well over double the value for the Living Room machine. It may not be a screamer and it’s certainly not the most impressive looking machine, but dollar for dollar, I’m pretty happy with this experiment.
Should we have gone another way? Have your own tales of budgetary gaming goodness? Let us know!