As I’ve moaned about in previous videos, PC Gaming as a hobby wasn’t cheap to get into at the best of times, and with the ever rising prices of RAM and graphics cards, we are clearly not living in the best of times. If we are trying to build an at least somewhat capable PC on a strict budget, even tighter than my recent V4 Xeon builds, is it worth going even further back in time?
Welcome to Esk Computers, I’m Scott and today on the desk of Esk that is what I’m looking to answer here, by trying to get our usual suite of modern-ish games running on a 2013 era V2 Xeon on a cheap X79 motherboard with DDR3, paired with an RX480 8GB card, all for £175. Canny price to performance optimisation or a bridge too far for the aging parts? Let’s find out.
The heart of this system, and the idea for the video, was another AliExpress temptation, in this case a Qiyida X79 motherboard combo with an E5 2450 V2 Xeon and 16GB of 1600MHz DDR3, all of which came in at £27.
I never get over the thrill of getting a processor that at launch cost over a thousand dollars for peanuts, regardless of age. It’s an eight core, sixteen thread part, with a base clock of 2.50 GHz, turboing up to 3.3 GHz. Workable specs even with the age, or at least I hope, we’ll find out soon enough. Checking back on the listing, either it’s changed since I ordered it, or I should have got the 6 core E5 2520 V2 processor. And it’s now £46, so in any case rate buyer beware.
While I was there I also ordered this 600W Aigo AK PSU for £26 – I know, I know, standard advice is never cheap out on the power supply but I’ve bough many of these Aigo models over the years and they’re absolutely fine for older and lower end builds, we’re not powering a 5090 here, and cheaping out is kinda the point of this build.
Storage is tricky, the same NAND shortages that affects memory pricing is also hitting NVME prices, and SATA SSD production is being phased out so that’s driving those prices up too. I’d bought this 250GB Netac NVMe drive for £16 in October, it’s tenner more expensive checking today. Eek.
The CPU is cooled by this Metalfish DH670, coming in at £21, with its 6 heatpipes and 120mm fan claiming it can cool 180W parts. Hmm. We’ll see about that.
The other part we’re perhaps most interested in today is this Gigabyte RX480, an 8GB variant that came in from eBay for £27. This has the potential to be the best price to performance ratio we’ve seen on the channel. It’s a remarkably tidy example externally, it must have been given a good clean before being listed for sale. After testing it was working, and holding a reasonable 71 degrees in stress testing, I thought I’d still bettter check on the delicious thermal pasta, and it was quite dry. Cleaning off some dust on the board and applying fresh paste didn’t actually make any difference at all to the test temperatures, but made me feel better, which is very much its own reward.
If we’re being completely minimalist, we could stop here and run the platform without a case, like I’m doing here to test the components are working, and we’re only in for £117. But I want this to be a proper build, and as part of that I’ve added in this PCIe to USB3 card, as there’s only USB2 ports on the motherboard. I purchased this for £4.20 a few months ago, which allowed me to run the test suite from an external SSD and saved hours of downloading and deleting after testing, so well worth it.
The case was selected by an exacting process of finding the cheapest suitable one on Amazon, meaning one that will fit a MicroATX board, and had case fans included. That landed on this MSI MAG Forge 100R mid tower at £37. It’s a pretty decent case with two ARGB front fans for extra FPS, and another more modest exhaust fan at the rear. It has it’s own RGB controller, useful as the motherboard doesn’t have any ARGB headers. I’m pleased to see it defaults to purple, clearly a case of taste. This fits full sized ATX boards, curiously the MicroATX version was more expensive, but as a result the motherboard does seem a little lost in there to be sure, but hey, at least it keeps your options open for future expansion. Oh, and not that it matters for this build, but there’s no front panel USB C, so maybe not a great fit for more modern boards. The styling of the front panel is a bit too needlessly angular for my taste, but I can’t argue with the value.
In parts accounted for but yet to be added, I realised too late I didn’t have any USB Wifi 6 / Bluetooth 5.4 dongles on hand, so another £3 AliExpress purchase will solve that. In retrospect, I wish I’d bought a 500GB drive at the time, as it stands I’ve ordered a 1TB hard drive on eBay for £13 as 250GB is just too tight for all but the most orderly of digital minimalists. If you’ve been keeping score, that’s us up to £175. Let’s see what a reasonably small budget has bought us.
The build process was reasonably unremarkable, as there’s plenty of room to work in and the case has a good number of cable management points. The only snags came from the compact nature of the motherboard layout – with the cooler installed, it obscures one of the motherboard screw points. Taking the cooler off is easy enough, putting it back on is a pain, but we get there in the end. Or, well, I thought we did, until I noticed this back fan cable somehow getting trapped behind the IO shield. Thankfully for you, my camera battery died sparing you a montage of me stripping that all out and reassembling with the cable routed properly, and we can instead go into the benchmarks.
I’ve kept the RX480 running at the standard power limits and clocks, but applied a modest undervolt in the AMD Adrenalin tool. This doesn’t always quite hit the maximum boost clocks, but upping the power limit to reliably hit them introduces odd frame pacing issues and stutters, and ultimately wound up giving the same or slightly reduced performance, so it seems like Gigabyte’s factory tuning is pretty much getting the most out of the silicon.
The full system power draw from the wall was around 260W when running Steel Nomad, which I think is a reasonable estimate of gaming power draw. At full load, during a Furmark and Prime 95 Stress Test after 10 minutes it’s pulling 390 – 395W from the wall, with the GPU temp levelling out at 74 deg C, so only about 3deg C more than running in open air. The CPU levels out at 65 deg C using the full 95W CPU package power, so I’m sorry for doubting you, Metalfish cooler.
To get the problems out of the way first, Cinebench 2024 would not run, as it requires AVX2 instruction support in the CPU, which this Xeon doesn’t have. I can’t find many games that need this instruction set to run, although it may help them run better, particularly PS3 and Switch emulators, so bear that in mind.
As a replacement test, my usual Davinci Resolve project render finished in 6 minutes 32 seconds. Pretty good results, roughly twice as fast as my Mac Pro 2013 with a similarly aged processor, albeit slower graphics cards, and four times faster than my old Ryzen 5 2500U based laptop. It’s also pretty responsive when editing at 1080p, at least once everything’s loaded in. Sure, my 9800X3D / 7800XT desktop blows it out of the water at 1 minute 33 seconds, but given the vast gulf in price difference, that’s a more than acceptable trade off.
In the synthetic 3DMark benchmarks, Time Spy gives us 4421 points, Fire Strike 11703 points, and Steel Nomad 983 points, so the same or a hair faster than the RX580 SP we looked at recently. The Superposition benchmark gets us 5626 at 1080p High. For context we are looking at about 20ish% of the speed of my RX 7800XT, or about 33% of the results a 2080Ti gets, but I think that’s enough to work with at 1080p. Let’s find out.
Shadow of the Tomb Raider at the usual 1080p High settings gives a nice 69 average fps, so no tweaking needed here.
Counter-Strike 2 gets an entirely acceptable 76 average fps at the 1080p Very High settings, although if you want more then we can get 119 average fps at the “Competitive” 1080p Low Settings, High Shadows, so nothing to complain about on this title.
Let’s give it some problems. In Cyberpunk 2077, we are only getting 35 fps at the usual 1080p Ultra Settings, but dropping to 1080p Low settings puts us back up to a very respectable 61 average fps.
Maximum problems come from Unreal Engine 5 games, where Black Myth Wukong’s benchmark gets a 25 fps average at 1080p High, with No Framegen or resolution scaling, and no raytracing of course. It also recovers, at a high visual cost, to 63 average fps at its self-determined recommended settings, those being Low with 50% TSR resolution scaling.
Marvel Rivals nets us 23 average fps at the usual 1080p Medium settings at Native Res, increasing to 52 average fps at 1080p Low settings, with Performance FSR3 resolution scaling. Not perfect results in either UE5 game, but playable results, and that’s what matters.
Assassin’s Creed Mirage will give you 46 average fps at 1080p High settings, or 57 if you don’t mind Adaptive Resolution Scaling. Lower settings would improve those results, but I think this performance level is fine for single player experiences.
Forza Horizon 5 continues to be dependably scalable, getting 63 average fps at 1080p High settings.
Red Dead Redemption 2 at the 1080p favour quality setting runs at 42 average fps, which I think is fine for this style of game, or if you’d rather, 1080p favour performance settings gives us 75 average fps.
It’s a very similar story in Horizon Zero Dawn, with 1080p Favour Quality settings giving a completely playable 52 ave fps, or dropping slightly to the Original settings gives us 62 average fps.
I’m quite happy with all of that, to be honest. I have tried to come up with a few that might be more CPU bound to make sure there’s no obvious weak spots, but as we’ve determined in videos past at all but the weirdest settings, the GPU is always the limiting factor.
Remember Starfield? The much maligned Bethesda RPG was in part muchly maligned because of its performance, and it’s not amazing here to be fair. The most demanding action I can think of would be causing trouble in a heavilly populated area, so doing that in Akila “City” at 1080p low settings averages 28 fps, or 52 with frame generation enabled. In areas more designed for combat it will do a bit better, but maybe irrelevant given that we all seem to have agreed to move on and never speak of this game again.
I’ve had a request to add Hearts of Iron IV into the test suite, which is hampered by me not having the first idea what I’d doing with these grand strategy games. Framerate’s not the most important thing here, although for the record it’s getting 60 fps at the slowest game speeds and half that at the fastest. It seems what’s most important is the tickrate, roughly speaking the length of time taken for the computer’s turn simulation to complete. Not so easy to measure, but I startd a new game in the 1939 setting and set it running, and it’s taking 25 min 20 seconds to advance the game one month at the slowest game speed, and 1 min 50 seconds at the fourth gamespeed, making this the first time I’ve ever measured something in minutes per month. Is that good? I don’t know, but for sure that’s faster than I can think so it clears that low bar at least.
As it’s doing so well with fairly new games, I’ve not tested the older ones, but these will also be dealt with easily, and I’m confident that the only games it’s not going to be able to play are those that require secure boot for the kernel level anti-cheat systems which this motherboard just doesn’t support.
The £27 RX480 is doing some seriously heavy lifting for this PC, and it really has deserved it’s aging like fine wine reputation. To its credit the old Xeon processor has been able to keep the card fed with the data it needs, and even if it’s not the fastest in 2026, it’s not limiting the GPU so it’s doing exactly what it needs to. I suppose I should point out that the RX480 is no longer getting driver updates, so maybe some future games will have bugs that may never be patched but all the games I’ve thrown at it have been fine.
So, what have we learned here? I’m a bit perplexed to be honest. When planning this video I expected to be saying that the old X79 dog has had its day, and is best left for older games and / or less demanding games only. And, to be clear, that is where it’s going to have the best results, but, look, this twelve year old processor when paired with this ten year old graphics card plays Cyberpunk 2077 and Unreal Engine 5 games.
Sure, it’s not the optimal way to experience these games, and if you have the money to spare then a bigger budget will give higher performance. But for £175, this has give some very impressive chops to it. That £175 cost includes the new case, new PSU, new cooler, and the new SSD, all of which you could reasonably expect to find cheaper in local deals second hand to bring this closer to or below £150.
We’ve put together a PC gaming rig that’s significantly less expensive than even an Xbox Series S, gives halfway decent performance in modern titles and of course can handle the usual Windows productivity tasks with ease.
I’m actually very happy I’ve done this exercise. I’d been getting very gloomy about the future of PC gaming, as there’s no question the mid to high end of it’s getting increasingly expensive, so it’s very good to see that bargains can be had at the budget end with a bit of savvy shopping around, that will provide a solid entry into this world of PC games.
This is not the most powerful PC build I’ve put together for this channel, but it’s by far the one that’s delivering the best price to performance ratio, and surely you have to respect it for that. If you don’t, let us know why in the comments, if you think there’s any shortfalls in spec or vital games that’s not going to run that I’ve not covered, please let me know in the downbelow.
Otherwise, if you enjoyed this videotronic missive then consider subscribing. Until next time, take care of yourself, and each other.