Laptop and Desktop PC graphics perform differently despite having the same names, let’s compare Nvidia RTX 2070 performance between a laptop and desktop to see what the differences are in games and applications.
Let’s start with the spec differences between the desktop and laptop 2070. They’re similar for the most part, same GPU core, same amount of CUDA cores and memory. The key difference is the desktop 2070 has faster memory and higher power limit. The desktop option has a higher TDP, which is possible due to cooling being less of a restriction when compared to the smaller space of a laptop. More power equals more heat, so the laptop runs with a lower power limit, and more power will result in more performance with higher clock speed.
For the laptop I’ll be testing with a Clevo chassis, the Metabox Prime-AI here in Australia, Eluktronics THICC-15 in the US, or XMG Apex 15 in Europe. What’s interesting is that this laptop uses the AM4 socket and has a desktop Ryzen 7 3700X processor inside. This has allowed me to do a somewhat fair comparison, as I’ve tested my desktop PC with the same 3700X CPU and Aorus Extreme 2070 from Gigabyte. There is a difference in the memory though, most DDR4-3200 laptop memory runs at CL22, while CL16 is more common for the desktop side. Rather than trying to match these, I’ve left them as is as I feel it best represents the differences you’d see between laptop vs desktop.
Both setups were tested with the same version of Windows and Nvidia drivers, so let’s get into the results. We’ll start out by comparing 15 different games, then look at thermals, power draw and other applications afterwards.
Red Dead Redemption 2 was tested using the games benchmark tool. I’ve got the desktop PC shown by the purple bars, and laptop shown by the red bars below. I’ve also tested all setting presets available, which are arranged on the left with lowest at the bottom to highest up top. As expected, the desktop was faster at all settings, with a 15% higher average frame rate at ultra settings.
Battlefield 5 was tested in campaign mode by running through the same section of the game on both laptops. The desktop 2070 again saw a 15% boost to average FPS at ultra settings, though strangely the laptop was scoring better in 1% low performance at all setting levels.
Shadow of the Tomb Raider was tested with the games benchmark. There’s less of a difference at the lower setting levels as presumably the game is less GPU heavy there, less than a 3% lead from the desktop system, but at max settings it was now 12% faster. Control was tested by walking through the same part of the game on both machines, and as more of a GPU heavy title, the desktop PC was 23% faster than the laptop at max settings, one of the biggest differences out of all titles tested. I’ve also compared results with RTX enabled, and again a similar 21% boost to average FPS at high settings, though the 1% low from the PC is now better than the averages from the laptop. When we enable DLSS though, the desktop was around 18% faster at high settings, and the differences to 1% lows weren’t as big. Fortnite was tested using the replay feature, with the exact same replay file used on both machines. The difference was minor at low settings as again the GPU presumably gets utilized less compared to the processor, but then at max settings the desktop was 23.5% faster than the laptop, the second biggest difference out of all games tested. Ghost Recon Breakpoint was tested with the games benchmark and the difference at max settings was below the average of all games tested at 13% better than the laptop, and this was similar throughout the different setting presets. Borderlands 3 was also tested using the games benchmark tool, and this game saw the biggest difference out of all 15 titles tested at ultra settings, with the desktop offering a 24% boost to average FPS, but the difference was much less at lower settings, where the laptop was actually slightly ahead in average FPS at low, but either way quite close. Apex Legends was tested in the World’s Edge map running through the Drill Site on both machines. This test saw one of the smallest differences out of all games tested, with the desktop 2070 just 4% faster in average FPS at max settings, no major differences between the two here. The Division 2 was tested using the game’s benchmark tool, the desktop 2070 was seeing 19% higher average FPS at max settings, then all the way down at low settings the difference was much smaller. Monster Hunter World was tested running through the main town in the same manner on both machines. Like some of the other games, the two systems were quite close together at low settings with the laptop once more actually slightly ahead, but then at the higher levels the desktop pulls out in front, resulting in a 14.5% higher average frame rate at highest settings. Rainbow Six Siege was tested with the games benchmark tool using Vulkan, and there was an above average improvement with the desktop PC which was 17% higher in average frame rate with the ultra setting preset. Metro Exodus was also tested using the games benchmark, and there was an above average boost from the desktop 2070 at the highest extreme preset, which was able to reach 19% higher average FPS. With the RTX preset enabled though, it was just 6% faster. The Witcher 3 actually saw a bigger difference at the low setting preset, where the desktop was 17% ahead in average FPS, then at the highest ultra setting preset the desktop was 16.5% ahead, so not much different, but most other games were closer together at the lowest settings which wasn’t the case here. Assassin’s Creed Odyssey was tested using the games benchmark, and the difference here was one of the lowest out of all games tested, resulting in a 12% boost to average FPS at ultra high settings. Far Cry New Dawn was also tested with the games benchmark tool, and this game saw the smallest difference out of all games tested, with just a 4% higher average frame rate at ultra settings.
In terms of overall improvement, on average over all of these games, the RTX 2070 in the desktop was performing 16% better in terms of average FPS. As you can see, the performance difference can vary a fair bit between games. Tests that depend more on CPU power tend to be towards the bottom with less of a difference, while GPU heavy games are up the top.
It’s worth noting that when I previously compared the 1660 Ti in a laptop and desktop, I saw a 15.5% difference there, and when we looked at how the 2060 compares in a laptop and desktop, there was a 15.7% difference so on average between these three tiers of graphics options we’re looking at a similar boost going from laptop to desktop.
Outside of gaming I’ve also got some 3DMark results, and the desktop was around 15% higher in the firestrike graphics score, and 18% higher for time spy. Although temperatures obviously aren’t a fair comparison due to the different form factors, the desktop was 6 degrees cooler but this will also vary based on the specific graphics card in use as cooler designs vary. I thought this was interesting as more power typically equals more heat, and although the desktop PC was pulling 50% more power from the wall under a GPU only stress test, it’s still able to run cooler. 50% more power for an on average 16% boost to FPS in games isn’t that great, so it would seem that the laptop parts definitely run more efficiently in terms of wattage per frame.
Despite using so much more power, the clock speed difference wasn’t all that large between them. Laptops generally have shared heatpipes between CPU and GPU too, so with both under load the desktop would have an advantage as thermals aren’t directly tied together by the same cooling solution.
Now for some content creator workloads. I’ve tested DaVinci Resolve with the Puget systems benchmark, and the desktop was scoring almost 12% higher than the laptop. Adobe Premiere was also tested with the Puget systems benchmark, and the difference is less pronounced here with the desktop scoring just 5% higher than the laptop. When looking at raw export times though, Premiere did seem to gain a little more improvement with the desktop, completing nearly 6% faster. The puget systems test previously tests more than only export times, such as live playback and more, but either way not much of difference. SPECviewperf tests out a number of professional 3D applications, and although results would vary based on the particular test, as expected the desktop was ahead, although the difference could vary a fair bit depending on the specific workload.
Prices are difficult to compare, as a full desktop PC can be configured with a number of different options, and there are of course also a vast range of 2070 based laptops on the market.
So then, which should you get? Laptop or desktop graphics? In the end, the question as to whether or not you need a laptop depends on how much you need portability. If you don’t need portability, then build a better performing desktop PC for less money, you’ll have more upgrade options in future too. If you need to take your machine with you, then a laptop is a better choice. In a laptop where we’re mostly limited to 1080p displays anyway, a 2070 is already more than enough for most people, the 1660 Ti or 2060 are good sweet spots for the money.
When comparing laptop and desktop 2070s, In most games the difference wasn’t actually all that large, and in content creator specific tasks it was even smaller. Most laptops don’t have a 3700X desktop processor inside, so while I have aimed to somewhat fairly compare laptop and desktop GPUs, results would also differ with a more average laptop processor. Anyway let me know what you thought of the performance differences between these two down in the comments, would you go for a laptop or desktop PC?