I have had a request on a previous X99 build to push it’s limits with a faster graphics card, to see what’s the most performance it can give. Meanwhile, I want to see how much performance the T1000 left on the table in our tiny little hot box experiment last week.
Welcome to Esk Computers, I’m Scott and today on the desk of Esk, let’s satisfy both those curiosities.
The first test subject today is the Machinist X99 motherboard, paired with two 8GB sticks of DDR4 2133MHz and an Intel Xeon E5 2680 V4, a 14-Core, 28-Thread part that clocks at 2.5 GHz with a turbo up to 3.3 GHz. This is normally paired with a GTX1660 6GB, which I’m not going to fully retest today but we have some results on file from the previous video to compare where appropriate.
Also on the operating table, this 8GB, single slot, half height Nvidia T1000. I’m hoping that the T1000 will get us closer to the 1660’s numbers when it’s not thermally throttling, or being held back by a processor bottleneck with the Ryzen 5 2400G as it was in last week’s video.
In order to get the most out of the Xeon processor, I’ll then be transplanting the fastest graphics card I have access to, AMD’s 7800XT, and seeing what it gives me in comparison to my system with a 9800X3D processor to see if it’s leaving any performance on the table.
The 7800XT’s girthier than this case can stomach, so we’ll be doing this testbench style.
Just to be abundantly clear, I’m not saying it makes any sense to directly compare the T1000 to the 7800XT for reasons that are surely obvious just by looking at the size difference in the cards. It’s just the best way I can think of to present that data without wasting your time.
Perhaps you can think of it as a reasonable budget build with the T1000 figures, compared with the maximum performance you could expect with a future GPU upgrade. Yes, that makes sense. I knew we’d find a point to this video somewhere. Onwards, for science!
In 3DMark’s Fire Strike, the T1000 gets 8164 points, which is up 11% from its 7333 when paired with the 2400G. Meanwhile the Xeon and 7800XT combo gets 24225, which is a 44% drop from the 43054 points the 7800XT gets when paired with the 9800X3D processor. The 1660 with the Xeon gets 12512 points, so there seems to be headroom to pair a more powerful card with it.
Even better results for the T1000 in Time Spy, it’s 3486 points being a 19% improvement on the 2400G, and the 7800XT and Xeon paring’ s 15334 points only being a 20% drop from the 9800X3D result. The 1660’s 5878 point result is again a good ways below the performance ceiling.
Superposition at 1080p High Settings sits between those results, with the T1000 improving 14% to 4531 points, and the 7800XT showing a 38% drop with 14739 points. The 1660 came in at 7695 points.
The Shadow of the Tomb Raider benchmark at 1080p sees the T1000 get 53 average fps, an 18% gain, and the 7800XT loses 49% of it’s performance with 133 fps. The 1660 and Xeon combo scored 88fps.
Counter-Strike 2 with the FPS Heaven Benchmark at 1080p Very High Settings sees the T1000 improve 25% to 85 fps, compared to the 2400G pairing’s 68 fps. The 9800XT result is 161 fps, a 56% drop from the 9800X3D’s 368 fps. The 1660 comes in at 95 fps.
Dropping to competitive settings, and the T1000 oddly gets 143 fps, barely any difference to the 2400G result of 141 fps. Meanwhile, the 7800XT scores 185 fps, a solid result but a whopping 77% drop from the 815 fps it gets with the 9800X3D.
A trip to Night City in 1080p Ultra Settings sees benchmark results of 31 fps with the T1000, a 19% improvement, and the 7800XT get 80 fps, a 53% drop. The 1660 scored 53 fps.
Re-running the benchmark at 720P Low Settings sees the T1000 get 83 fps, a 24% improvement. Not much is gained by dropping in the 7800XT, where it scores 90 fps. That’s a 69% drop from the 289 fps it can get with the 9800X3D.
Some actual gameplay with a run around Kubuki in 720p Low settings with high crowd density next, and the T1000 gets 49 fps, that’s 31% better than the 2400G result. The 7800XT gets only 47 fps, which is a a 59% drop off from the 9800X3D result of 116 fps, and lower than the T1000 result. We’ll see the same with Baldur’s Gate 3 – I think these settings do not agree with the Xeon processor.
Assassin’s Creed Mirage’s benchmark sees the T1000 score 60 fps, a marginal improvement over the 57 fps the 2400G gets, although I think the 1080p High setting leaves adaptive upscaling on, so those numbers aren’t exactly reflective of the visual quality it can sustain. The 7800XT gets 128 fps, a 30% drop from the 9800X3D’s 182 fps.
Unreal Engine 5 now, as Black Myth Wukong’s benchmark at 1080p High, with no Framegen or upscaling sees the T1000 score 15 fps, technically 15% better than the 2400G’s 13 fps. The 7800XT gets 75 fps, only dropping off 10% from the 9800X3D. The 1660 can get 25 fps in this test.
Dropping to 720p Low, with Framegen, and 66% FSR upscaling lets the T1000 report 101 fps, a 22% improvement over the 2400G figure. While it’s not a sensible setting for the other combos, for the record, the 7800XT gets 218 fps, a 43% drop off from the 383 the 9800XT scores.
Marvel Rivals at 1080p Native resolution, Medium settings is a demanding test, and the T1000 gets 28 fps, that’s again 22% better than the 2400G result. The 7800XT gets 86 fps, a 26% drop from the 9800X3D paring. The 1660 gets 35 fps on this test.
If we add Performance FSR and frame generation to the mix, at Low settings the T1000 gets 80 fps, a 23% gain on the 2400G, while the 7800XT gets 145 fps, a 61% drop from the 370fps it gets with the 9800X3D.
Forza Horizon 5’s benchmark at 1080p Medium now, where the T1000 gets 67 fps, a 24% improvement, and the 7800XT scores 135 fps, a 53% drop off.
Horizon Zero Dawn in 1080p Favour Performance sees the T1000 score a nice 69 points, up 21% from the 2400G result, and the 7800XT get 156 fps, a 28% drop from the 7800XT’s 218 fps.
The Civilization VI Gathering Storm AI Benchmark at 1080p Highest settings is more of a CPU test than of graphics, so the Xeon’s scoring pretty much the same average turn times of 53 and 51 seconds with the T1000 and 7800XT respectively. The 2400G gets a 61 second time, and the 9800X3D gets 24 seconds.
Total War: Warhammer III’s Mirrors of Madness benchmark at 1080p Ultra sees the T1000 get 19.4 fps. 21% better than the 2400G. The 7800XT gets 43.9 fps, a 61% drop from the 113.3 fps it gets with the 9800X3D.
Dropping to 720p Low settings paints a similar picture, the T1000 gaining 30% over the 2400G baseline to 108 fps with the Xeon, and the 7800XT’s 190 fps result representing a 76% drop from the 517.7 result with the 9800X3D.
Lastly today, Baldur’s Gate 3, with the usual run around Lower City on the T1000 getting 34.6 fps, a 24% improvement on the 2400G result. The 7800XT gets 54.7 fps, a 58% drop from the 166.7 fps the 9800X3D achieves.
Dropping to 720p Low settings will see the T1000 give 65.3 fps, a 55% improvement on the 2400G pairing’s 42 fps. Like Cyberpunk, The 7800XT bafflingly only gets 61.3 fps, less than the T1000 as well as being a 55% drop on the 9800X3D’s 218.5. It’s barely getting above 30% GPU utilisation, so there’s either some weird bug or the relatively low clock speed and single thread performance of the Xeon is holding it back, perhaps.
So, to return to our mostly unrelated questions that prompted this, the T1000 was able to run a few degrees cooler by blasting a fan at it, roughly 79 deg C rather than 81 deg C in the restricted case used last week, and from a seperate round of testing with and without that fan, it’s accounting for between 1 and 6% improvement across the test runs. We can, loosely speaking, seperate that out from the difference in the CPU results, which tells us that there’s more often than not a double digit percentage improvement in going to the faster processor than the 2400G, so the bottleneck was more CPU than thermal.
In most of these games, there’s an around 15-20% difference in running the T1000 with this Xeon, which is certainly significant. I’d expected better of the 2400G to be honest, I thought it would be about as well as the Xeon processor but that’s not really the case, so the Xeon remains a good budget option.
In terms of the pairing with the 7800XT, as expected that’s a significant bottleneck, although comparing with the 1660 results shows there’s room to get a faster GPU that can be adequately serviced by the Xeon E5 2680 V4. Hard to pin it down exactly with the test data I have to hand, but it should be more than enough for a 1080Ti, but maybe not quite grunty enough for a 2080Ti. Other contenders in that bracket would be the RX 5700 XT, the 3060 & 3060 Ti, the 2070, maybe the RX 6650 XT and 4060.
The surprising thing to me from this is that the T1000, even in a case with plenty of space to circulate air, just seems to want to live in the borderline unacceptably toasty range under load. Maybe that shouldn’t be so surprising given the diminutive heatsink and fan. Perhaps I owe an apology to the “hot box” case used last week, there’s no helping the card. Or is there? I have an idea, let’s see how it pans out next week.
Until then, 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.