They say you can’t go home again. They’re full of crap. I’m rebuilding my previous daily driver system, and seeing if 2018’s best consumer processor, the i9 9900K can still perform in recent games with the 1080Ti.
Welcome to Esk Computers, I’m Scott and today on the desk of Esk, let’s build something. It’s been a while.
Okay, I say this is a rebuild of my last rig, it’s more like a very close cousin. For the bulk of its life my previous machine used a GIGABYTE Z390 AORUS PRO motherboard along with a i7 9700K processor. After a few years I decided to put it in a new case and upgrade to the i9 processor for a slight and frankly not economically justifiable bump in multi-core performance, which went okay for a few months until a stick of RAM died. I took that excuse to buy faster RAM back when it was cheap, remember that, but it soon killed one of those sticks in the same slot. At which point I stopped throwing good money after bad and pulled the trigger on the upgrade to the AM5 Ryzen 79800X3D platform I’ve using for the past year or so. The 9700K cost £400 back in February 2019 and I sold it for £140 to part fund the £220 purchase of the 9900K in April 2023.
Some of the parts of that build have been sold on, others scavenged in various builds, but I never got round to selling the 9900K, and one of the old ADATA XPG Spectrix D41 3000MHz RAM sticks has been knocking around too. Two were used in a pre-this channel build that’s long since sold. I managed to find one solitary replacement stick on eBay for £30, so that’s back to 16GB. The four stick, 32GB pack originally cost me £200 back in February 2019.
After the plan to put dual 1080Ti’s in my server build failed, I’ve had one of those knocking around spare, and we’re now getting close enough to a full parts list that my palms were getting itchy enough to pull the trigger. This 1080Ti cost about £160 in December 2025.
I’d hoped to find a cheap z370 motherboard as a same class replacement for the Gigabyte board, but there’s not so many of them kicking about on eBay for the prices to come down. I’ve settled on this MSI Z390 PLUS board. Oh, look, it says SLI on it. I suppose you now know what next week’s video is about. It was £63, more than I’d wanted to spend on this part of the project but that’s computing in 2026 for you.
It will need a decent power supply, and I only had broken Gigabyte UD PSUs knocking about which are of little use, but if you need a paperweight, let me know. I’ve picked up this 1000W EVGA GQ unit for £53.
For a case, I’ve sold the Corsair Carbide 175 case it used to live in, but I still have this X79 build in a MSI MAG Forge 100R case. I’ve never been happy with the microATX board in an ATX case aesthetic, so let’s use that and we’ll find a downsized home for this down the line. This case cost £36 in November 2025, and came with the three fans, two of them ARGB. It’s a solid if unremarkable case to build in, and good value for money. For the CPU, I’ll use the Thermalright Frozen Warframe 240mm cooler we recently swapped out of my daily driver. It cost £38 back in March 2025.
Storage is an issue as much as RAM these days. I’m using this Samsung PM9A1 1TB nvme that I had been using as an external drive and while I really still need that for shuffling stuff around various projects for the channel, eBay’s asking for around £40 for used 500GB drives, which i can’t stomach. This cost me £45 back in April 2025.
Ideally I’d rather have a larger nvme drive and not rely on old school spinning rust, but this is what the neo-dark ages of the AI-pocalypse has reduced us to. If you wondered why I was bothering to do that short video the other week on shucking a hard drive, it was both to have a drive for this, and to buy some time to complete the build. It’s a 3TB drive from the dark ages of 2010 so I’m side-eying its reliability until I can prove otherwise. My records don’t go that far back, but from reviews I’d guess it was £200 and bought sometime in 2011.
If all goes to plan, I’ll have been intercutting footage of the build as we’ve gone along rather than have you staring at my gormless visage for minutes at a time, as it’s been a mostly straightforward build for once and Windows 11 installs with no issues. Only slight problem is that the PSU is a bit too long to fully coexist with the hard drive brackets in their proper mounting points, but this is fine. Also, the case comes with an addressable RGB controller and fans, but the motherboard only has the older 12V header. It’s tidy enough on the business end, although it needs a bit of a cable management session round the back. However, I’ve got a video to get out, so that’s going on the to-do list until the testing is done.
Given how the parts for this have been sourced it’s not quite so easy to tell you what this build has actually cost me. I’d say if you exclude the outlandish for 2026 hard drive price it’d be about £700-£750. On a quick eBay search I’d say this would have cost about £650 if buying all the parts today. Let’s find out if it’s worth that by running through the same tests we recently put the 1080Ti and my 9800X3D through.
I trust you will take my word that this build isn’t going to have any problem with web browsing, office tasks, photo and graphic work, and all the day to day computing work. The i9 9900K is an 8 core, 16 thread part that boosts up to 5Ghz on one core, and 4.7GHz on all cores. It has a TDP of 95 W, getting up to 120W while boosting. That’s plenty of horsepower to bring to bear, way more than today’s general computing needs even at eight years old.
To prove it, I set it rendering last week’s video in DaVinci Resolve, where it’s taking 5 minutes and 37 seconds to render out at its 1080p YouTube preset. That’s about 75% slower than my 9800X3D system’s 3 minutes 13 seconds.
I gave Cinebench 2024 a spin next, and we’re getting results of 80 single core, and 736 multicore. That’s still fairly good in 2026, although as you’d expect a good bit slower than the 9800X3D’s 135 single and 1294 multi core results. The GPU test gets 5570 points.
Let’s get into some gaming tests, and all of these results are at 1080p resolution. This is roughly chronological, using the results of the recent 1080Ti vs modern gaming video. Let’s see if the slower processor makes a meaningful difference to the results. I trust we’ve covered the 1080Ti well enough previously that I don’t need to tell you much about this old beast of a card, and I’ll point you in the direction of the linked videos that have been popping up if you want my takes on the legendary GPU.
Kicking off with 2018’s Shadow of the Tomb Raider benchmark at its Highest Settings, where we’re getting 133 average fps. The 9800X3D system gets 131 average fps, that’s technically 2% slower but very much margin of error, run to run variance stuff.
Here’s the Red Dead Redemption 2 benchmark with everything cranked to Highest settings, where we’re getting 66 average fps. The 9800X3D system gets 64 average fps, that’s technically 3% slower, but again, margin of error stuff.
Final Fantasy XV next, at its Highest settings, where we’re getting 105 average fps, with 1% lows of 75 and 0.1% lows of 55 fps. The 9800X3D system gets 105 average fps, that’s 1% faster. Say it with me now, margin of error, yes that’s right have a cookie.
This is Control at Ultra settings, no raytracing obviously, where we’re getting 69 average fps, with slightly disappointing 1% lows of 36 and 0.1% lows of 13 fps. The 9800X3D system gets 66 average fps, that’s 5% slower. I’ll stop saying margin of error now, I think you get it.
Old testing stalwart Cyberpunk 2077 next, here at Ultra Settings. The usual run-around and cop bothering gets us 80 average fps, with 1% lows of 56 and 0.1% lows of 44 fps. The 9800X3D system gets 83 average fps, that’s 4% faster.
Gor Blimey guvnor, apples and pears and all that, it’s Watch Dogs: Legion at Ultra settings, where we’re getting 73 average fps, with 1% lows of 61 and 0.1% lows of 54 fps. The 9800X3D system gets the very same 73 average fps.
Bothering some machines in Horizon Zero Dawn at Ultimate quality now, and we’re getting 104 average fps, with 1% lows of 79 and 0.1% lows of 47. The 9800X3D system gets 91 average fps, that’s 14% slower, but looking back I’ve taken results from a different area of the map. Both processors are letting the GPU run full tilt, so there should be no functional difference.
This of course is Forza Horizon 5, a well optimised game but Ultra settings can be a bit of a stretch. We’re getting 83 average fps, with my records saying the 9800X3D system gets 109 average fps, that’s 24% faster. I can’t think why that would be the case, and I think it’s more likely I’ve mislabelled that 9800X3D result. At High settings the 9900K system gets a suspiciously similar 113 fps result. In any case, both run Forza well.
A Plague Tale: Requiem next, at Ultra settings, where we’re getting 66 average fps, with 1% lows of 30 and 0.1% lows of 6 fps. The 9800X3D system gets 62 average fps, that’s 6% slower.
Getting into the harder to run stuff now, starting with the much maligned Starfield, here at High settings with all the upscaling off, where we’re getting 41 average fps, with 1% lows of 28 and 0.1% lows of 26. The 9800X3D system gets 49 average fps, that’s 16% faster.
This is Immortals of Aveum, at High settings with FSR disabled, where we’re getting 51 average fps, with 1% lows of 26 and 0.1% lows of 8 fps. The 9800X3D system gets 56 average fps, that’s 9% faster.
Next is Avatar: Frontiers of Pandora, at High settings, with no upscaling or framegen, where we’re getting 42 average fps, with 1% lows of 31 and 0.1% lows of 27 fps. The 9800X3D system gets 48 average fps, that’s 13% faster.
The Unreal Engine 5 boogeyman pops back up with the Black Myth: Wukong benchmark where at High settings, without any RT or upscaling, we’re getting 41 average fps. The 9800X3D system gets 38 average fps, that’s 8% slower.
This is S.T.A.L.K.E.R. 2 at High setting and native resolution, where we’re getting 41 average fps, with 1% lows of 26 and 0.1% lows of 15 fps. The 9800X3D system gets 42 average fps, that’s 2% faster.
Next up is Senua’s Saga: Hellblade II at Medium settings with Balanced FSR, where we’re getting 68 average fps, with 1% lows of 55 and 0.1% lows of 44 fps. The 9800X3D system gets 76 average fps, that’s 11% faster.
Silent Hill f recommends its Balanced preset, where we’re getting 52 average fps, with 1% lows of 41 and 0.1% lows of 20 fps. The 9800X3D system gets 58 average fps, that’s 10% faster.
Metal Gear Solid Delta: Snake Eater next, where at High settings with balanced FSR, we’re getting 50 average fps, with 1% lows of 36 and 0.1% lows of 28 fps. The 9800X3D system gets 41 average fps, that’s 22% slower. I think that’s due to getting these figures at an earlier section of the game. It felt exactly the same to play.
Next, Resident Evil Requiem, where I’m using its “Prioritise quality” settings. We’re getting 89 average fps, with 1% lows of 36 and 0.1% lows of 33 fps. The 9800X3D system gets 96 average fps, that’s 7% faster.
This is Death Stranding 2 with its suggested Medium settings, with quality FSR upscaling. I turned the recommended dynamic resolution scaling off, and that’s giving us 64 average fps, with 1% lows of 53 and 0.1% lows of 49. The 9800X3D system gets 86 average fps, that’s 26% faster.
We’ll round off with a few tests I don’t have a 9800X3D/1080Ti result to compare with, but nonetheless felt were worth including starting off with the usual synthetic benchmarks where Fire Strike gets 25,159 points, Time Spy gets 10,612 points, Steel Nomad scores 2,333 points, and Unigine’s Superposition at 1080p High gets 16,204 points.
This is Total War: Warhammer III’s Mirrors of Madness benchmark at Ultra settings, where we’re getting 68 average fps.
Trying to stretch the CPU a bit with Civilisation VI’s Gathering Storm AI Benchmark at Highest Settings, where we’re getting an average turn time of 35 seconds. The 9800X3D with my 7800XT gets 27 seconds here, about 30% faster.
The other usual Unreal Engine 5 test next with the Marvel Rivals benchmark at Medium settings, which gives us 62 average fps.
And how could we not test out Counter-Strike 2. The FPS Heaven Benchmark at 1080p Very High Settings gives us 239 average fps, and dropping to Competitive settings of Low but with All Dynamic Shadows on gives us 412 average fps.
So, there we go, a good selection of games tested there, and I think we can at least say that it’s still getting the most out of the 1080Ti, as there’s not been a huge gulf in performance between this build and the more modern, now much more expensive AM5 platform. Actually more often that not it had a slight lead, probably due to this being a nice clean fresh install of Windows, but more likely just being the margin of error in the results.
I think there’s maybe a just about statistically significant edge to the 900X3D results on the more recent games, so arguably the faster processor gives it a slight edge in games that are relying on modern tricks like upscaling or otherwise just have heavier engines with more physics simulations and so on. But certainly not enough of an edge to make it worth upgrading the entire platform, it’s still handling the 1080Ti with headroom to spare, so there’s something to investigate further
Given how much RAM and Storage prices have ballooned in the recent years, all my ideas of cost to performance ratios have gone completely out the window so I’m going to have to leave the judgment on whether it’s good value or not up to you.
Starting from scratch, I don’t think this platform would be my first choice for a budget build in 2026, but hey, it’s a really weird market right now. You certainly could have built a more modern and more powerful system a year ago for the £650 ballpark I’m estimating this costs today. For example, the small form factor AM5 build with a Ryzen 5 7500F, a RX 9060 XT, and 32GB of DDR5 was about £800 in parts around a year ago, but is about 50% more expensive now, on a good day. If you can get this or something like this 9900K build at a reasonable price it’s still an exceedingly capable machine for general computing, content creation and gaming, so there’s a lot to like here.
I hope you’ve enjoyed this look at the system, as there’s a few other questions I want to answer over the coming weeks with this platform, like how far we can push it, if the cost difference between this generations i9 and i5 processors was justified, and if it can properly handle faster graphics cards. I’ll sprinkle in some other topics for variety, but we’re be returning to the 9900K a couple of times over the coming month or so.
I hope you are looking forward to that but in the meantime, if you have any questions or want further details please leave a comment down below, and if you enjoyed this videotronic missive then consider subscribing. Until next time, take care of yourself, and each other.