RTX 3080 Laptop 80W TDP vs 115W TDP Comparison
If you are looking to buy a new gaming laptop, you are likely sifting through countless options that all look the same but can offer significantly different levels of performance. Even if you know you want the RTX 3080 Laptop GPU, the highest end option NVIDIA currently offers, you then have to choose which TDP or thermal design power variant you want. And all of that is on top of the other factors like the CPU, RAM, storage, display, battery life, keyboard, trackpad, I/O and so, so much more. It’s a minefield.
In the hopes of explaining the differences between the TDP variants, I’ve got these two machines from Asus. One is the Zephyrus G15 featuring an RTX 3080 Laptop GPU with an 80W TDP, and the other is a STRIX SCAR 15 which also uses an RTX 3080 Laptop GPU, but with a 115W TDP.
Before we go any further, it’s a good idea to give a quick rundown of what TDP actually is so you can understand why it can be so important. Like I said, TDP is thermal design power, and is basically how much power in Watts the GPU is designed to run at based on the cooling solution in the laptop. The higher the TDP the more power it can draw, and the more power the higher the chip can boost and run faster. It’s also important to note that the TDP rating isn’t exactly how much power the GPU can draw, so this Zephyrus G15’s “80W” 3080L peaked at 96W during my benchmarks. In part that’s thanks to dynamic boost which for this machine offers up to 20W more power (that’s often taken from the CPU’s power budget and given to the GPU).
So, when comparing the TDP variants, it’s probably more accurate to say this is a 95W GPU versus a 130W GPU. So, is this SCAR 15 over 35% faster than the Zephyrus G15? In short, no, but as always it’s not quite that simple. First you need to know that despite the CPUs in these machines being technically different, this is the 5900HS and the SCAR has the 5900HX, both are 8 core 16 thread Zen 3 Ryzen CPUs and in my testing both ran at the same TDP, and the same performance across the board.
But, just to be sure, since both of these machines came with the stunning 1440p 165Hz panel anyway, I’ll be running these tests at 1440p. That helps remove the CPU bottleneck considerably, making the results more dependent on the GPU’s speed rather than the balance of CPU and GPU.
Right, enough talk, more numbers. Lets start with CSGO on max settings at 1440p. The SCAR 15’s 115W chip ran at 198FPS average, which is pretty respectable. It hit 102FPS in the 1% lows which means with the 165Hz panel you’d be having a pretty good time on this. By comparison the 80W version ran a bit slower at 178FPS average, and 94FPS in the 1% lows. That makes the 115W version around 11% faster.
In Cyberpunk the 80W chip ran at 42FPS average, and the 115W hit 45FPS. That’s only 7% faster and only 3FPS average different, meaning you could reasonably account for some of that in run to run variation. In my testing Cyberpunk can be more CPU dependent even at 1440p, and I should add that the SCAR was tested with a higher ambient temperature meaning it had less thermal headroom to boost quite as hard.
Watchdogs also has them close, 52FPS for the Zephyrus and 54FPS for the SCAR, although the extra 4FPS the SCAR holds in the 1% lows would go a long way to making it feel like a better playing experience even if the average isn’t much higher.
Fortnite shows a bigger difference, with the 80W chip hitting 79FPS average, whereas the 115W hit just over 93FPS. That’s nearly 20% faster and is a sizable difference you might actually notice in games. Even the 1% lows are better at over 53FPS rather than 45FPS.
Finally in Microsoft Flight the 115W offers a respectable (for this game) 32FPS average and 18FPS in the 1% lows, which is up from the 80W’s 26FPS average and 12FPS in the 1% lows. That makes the 115W version 24% faster on average, but that’s still nowhere near the 36% the power figures would suggest.
Why is that? Well, there is one very good reason. The law of diminishing returns. Performance out of these chips isn’t linear with the amount of power you put in, it’s a curve. If we plot the performance from 0, 95W and 130W you’ll see that it isn’t a straight line, it plateaus and you could assume that with 160W it’d perform better, but not that much better. If you compare the 3080 Laptop chip to the closest desktop card in terms of core count, which is an RTX 3070, that desktop card would run at 220W, but wouldn’t provide 2x more performance than the 95W TDP 3080 Laptop chip.
The other reason the performance gap might not be so big is the CPU, even testing at 1440p the CPU still has to work hard to keep up with the GPU and in certain titles that’s evident. Being a laptop with a shared cooling solution, it’s a fine balance between offering more power to the graphics core while taking some from the CPU, and vice versa. There is a limited thermal budget, and a limited supply budget, making it both a difficult balancing act, and a much more closed system than a desktop where the CPU and GPU can draw as much power as they like and neither has a significant impact on the other’s thermals or ability to boost.
So, if you want an RTX 3080L equipped machine, the 115W variant will offer more performance, but not by much. Some games benefit more, but some don’t care which, and as I noted in the review of this SCAR even some well cooled 3060 and 3070 machines can keep up with both of these in some titles. To throw my own opinion in here, I’d personally go with a higher TDP 3060 or 3070 machine. You save a good amount not getting the 3080 and in some cases still get the same final performance.
