Inspired by a helpful comment on a previous video where we took a look at this ageing but surprisingly performant Xeon X79 E5-2450 V2 system with a RX480 graphics card, I thought it might be interesting to push it a bit further and see if we can hit its limits.
Welcome to Esk Computers, I’m Scott and today on the desk of Esk I’m going to compare the X79 system to the fastest system I have, my daily driver, named Goliath. This has a beastly Ryzen 7 9800X3D and 7800XT graphics card – not the fastest card in the world, but the fastest that I own. If anyone would like to send me a 5090, I’d welcome it. Just for science, of course. I’ll run through both my usual test suite, and a few more CPU pushing scenarios that should show where the differences lie.
Given the nature of this comparison, it was never going to be completely fair. For example, I’m sure the 64GB of 6000Mhz DDR5 is going to give a further boost compared with the 16GB of 1600Mhz DDR3, and the NVME drive on my rig is faster, and so on, but I think it’s still worthwhile to compare them as holistic platforms.
I’m sure I’ll get round to covering my PC setup at some point, but for the purposes of this exercise I’ll just say that I’ve not spent any time tuning the Precision Boost Overdrive settings or fully tweaking the setup to wring every FPS from it, as I’m prioritising stability over peak performance. However, it’s obviously still significantly faster than the Xeon from over a decade prior.
So, let’s slap that 7800XT in the X79 build, which is the appropriate technical term, and see how it’s holding up. Not for nothing did I name this build Useful Barometer. It’s certainly filling out the case better. And just to compare, we’ll see how the RX480 performs in several of the scenarios.
There are a few games that gave playable results under all the scenarios, such as Shadow of the Tomb Raider, but the results are still telling. In the built in benchmark at 1080p High settings, the Xeon/480 combo gives 69 average fps, compared to 262fps on the 9800X3D/7800XT system. The Xeon paired with the 7800XT gives only a modest improvement to 87fps, a 26% improvement on the RX480 figures but a massive 67% drop from the X3D system.
Counter-Strike 2 saw an overnight update that broke the FPS Heaven benchmark on the Dust map after a couple of seconds, so we’ll have to take these Xeon/7800XT results as a best estimate from me eyeballing it at 120fps on Very High Settings and 160fps on Competitive settings, a seemingly a massive drop off of 80% of the 815fps the X3D system delivers at Competitive settings, or 67% of the 368fps at Very High. Comparing to the RX480, at Very High it’s a 58% improvement over its 76fps result, but just 34% better with 119fps at Competitive settings.
Using the built in Cyberpunk 2077 benchmark at 1080p Ultra settings is a similar story. From a base of 34 fps with the RX480, there’s an 83% improvement to 64fps by moving to the 7800XT. Still, that’s a huge 62% drop from the 168fps the 7800XT can give when paired with the 9800X3D.
Sticking with Cyberpunk, here I’ve turned off the resolution scaling that the Ultra setting adds by default and taken a quick run around Kubuki with High crowd density, and causing a bit of a ruckus to see how it holds up. On the X3D system this gives 125fps, and drops 68% to 40fps in the Xeon system.
Black Myth Wukong’s Benchmark at 1080p High Settings next, no framegen or upscaling, and clearly this is even more GPU limited than cyberpunk. The 9800X3D system gives 83 fps, with only a 17% drop to 69fps when moved to the Xeon system. The RX480 paired with the Xeon only gave 25fps at these settings, the massive 176% increase by moving to the 7800XT showing the relative weakness of the older graphics architecture in Unreal Engine 5.
The other Unreal Engine 5 game tested around these parts is Marvel Rivals, with a similar theme but less dramatic numbers. This struggled its way to 23 fps on the Xeon/RX480 system at 1080p Native resolution, Medium settings, and we saw “mere” 130% gain for the 7800XT at 53fps. The 9800X3D can squeeze 116fps out of the 7800XT, so a sizable 54% drop by going to the Xeon system.
Assassin’s Creed Mirage benchmark at 1080p High Setting next, and a familiar story, with a 57% performance drop going from 182fps to 79fps as we move the 7800XT from the X3D to Xeon systems. That’s giving a 72% improvement on the 46fps that the RX480 gave.
Very similar percentages in Horizon Zero Dawn at the 1080p Favour Quality Setting, a 57% drop from 212fps to 91fps when moving the 7800XT, which means it’s a 75% improvement on the 52fps the RX480 can deliver.
Total War: Warhammer III is new to me, I think it came in a recent Humble Choice selection, and seemingly the Ultra settings give a nice large unit size that gives the CPU a workout. Using the inbuilt Mirrors of Madness map benchmark the X3D system gives 114fps, and this falls all the way off to 27fps on the Xeon system, a 76% drop.
Civilization VI should also care far more about processor performance than graphics settings, but I thought it worth examining. The Gathering Storm Benchmark 1080p simulates a late-stage game, and here the X3D system gives an average turn speed of 24 seconds, while the Xeon takes 237% longer at 57 seconds. Big numbers, although I’m sure the game’s perfectly playable on the Xeon.
Rounding off the testing, I’m working my way through Baldur’s Gate 3, so although I hear the Lower City in Act 3 is quite CPU intensive, I’m not that far into it yet. At the grave risk of spoilers, I’ve found a save online and used that to take a run through the area while trying to ignore anything plot related. The X3D system gives me 179fps at 1080p Ultra settings, dropping 77% to 41fps when paired with the Xeon.
So, what have we proved here? Well, arguably what was already obvious, that although the Xeon can get very playable results in a lot of these games, when pushed to its limits it’s not going to be able to keep faster modern graphics cards fed with the data to make the most of them. The 7800XT was frequently down at 50% utilisation during the testing, so while that X79 system has plenty of life left in it, it’s no longer in its prime, and indeed is closer to collecting its free bus pass.
Of course, with more realistic graphics card pairings and more CPU friendly settings it can still perform well. Looking back over test data for other cards I think there’s a decent chance it won’t bottleneck cards as powerful as the 1080Ti, which I happen to have here, so let me know if you want me to test that theory in the comments below.
Otherwise, if you enjoyed this videotronic missive please consider subscribing and until next time, take care of yourself, and each other.