RTX 3080 Laptop vs Desktop Comparison – HALF AS GOOD!!
RTX 3080 laptops are here, sporting massive performance for laptop chips. The performance jump from the last gen is always welcome, but there is one question I’d like to answer. How does the laptop RTX 3080 compare to a desktop 3080? Well, lets test that and find out! But first, if you haven’t already, consider subscribing for more videos like this one every Monday, Wednesday and Friday.
Last generation, NVIDIA used the same GPU cores in both laptop and desktop parts meaning the same number of CUDA cores, the same memory bus, the same physical chip. The only difference, generally, was the clock speed it could run at. Desktop 2080 Super’s were spec’d to run at up to 1.82GHz (as standard, many AIB cards came factory overclocked), whereas the mobile parts peaked at 1.56GHz instead. That means, just from clock speed reduction, you’d get around 15% less performance. When TechSpot/Hardware Unboxed tested 2070 laptops and desktops last year, they found the 2070 they tested with was roughly 25% faster than the laptop part.
So what about this generation? Well, NVIDIA has decided that, despite using the same display name with no added clarification – no “m” or “laptop” actually in the product name that is – they are using entirely different chips between laptop and desktop platforms. How different? Very. The RTX 3080 desktop card is derived from their GA102 design, whereas the RTX 3080 laptop part is derived from their GA104 design, the same as their 3060, 3060ti and 3070. That’s not the worst of it though. The number of cores is substantially different. The desktop card has 8704 cores, the laptop part sports just 6144. That’s 30% less cores. They also use the smaller 256 bit memory bus with slower memory at 14000Mhz instead of 19000MHz.
But that’s it right? Nope. The mobile part ALSO runs significantly slower clock speeds. Even taking the fastest the two are (factory) spec’d for, 1.71GHz on desktop, 1.25GHz on mobile, that’s around 30% slower too. Add that together and you find that the laptop part is about 50% slower – just on core count and clock speed alone. But yet they are being sold under the same name.
So if the desktop 3080 is going to obliterate the laptop part, what desktop card matches the mobile one? Well we can use the clock speed and core counts to get a rough estimate. Multiply the core count, 6144, by the clock speed, 1.25GHz. That works out to 7.68TOP/s. For reference, the 3080 desktop runs 14.88TOP/s. The closest match available right now is the RTX 3060Ti which runs 8.12TOP/s, a hair over the laptop 3080. So, now we know what to test, lets run it!
I’m running all these benchmarks at 4K as it helps eliminate the CPU as much as possible, although I am using the same ‘simulated’ Ryzen 5800X – a 5900X with 4 cores disabled – as it is about as close as I can get in performance to the 5900HX that’s in this Zephyrus Duo laptop.
Starting with COD Modern Warfare, the 3080 laptop is sitting at 63FPS average, whereas the desktop 3080 is pushing 91FPS. That’s around 30% slower, and a tad slower than the 3060Ti which gets 70FPS average. To make it clear, 63FPS average at 4K ultra settings is nothing to scoff at, and the fact that is using a laptop GPU is nothing short of incredible in this form factor, but it’s not even remotely close to the desktop card’s performance for which is shares its name.
In Cyberpunk, even with ray tracing disabled, this is an immensely difficult game to run. Even the desktop 3080 only gets 38FPS average, and both the mobile 3080 and desktop 3060Ti get around 25FPS average. That’s around 34% slower than the desktop 3080.
In Fortnite the laptop card runs about 10% slower than the 3060Ti, and 35% slower than the desktop 3080. It still gets almost bang on 60FPS average at 4K on epic settings, but with the desktop card getting 91FPS average it’s hard to argue they should share the same model number!
And finally for this video, Watchdogs Legion. Much like Cyberpunk, this is an incredibly difficult game to run on ultra settings, let alone at 4K, so squeezing 35 FPS out of the laptop part, slightly beating the desktop 3060Ti mind you, is incredibly impressive. Nonetheless, the desktop 3080 offers 46% more performance than the mobile chip – and at these sorts of frame rates, the playing difference between 35FPS and 51FPS is massive.
So, the mobile part just about matches a 3060Ti, or in the real world is about 30% slower than the desktop card – or to put it the other way, in my testing on average the desktop card is 48% faster than the laptop one. And don’t get me started on the fact this laptop GPU is the 115W (and up to 130W with Dynamic boost) TDP model, not the lower power 90W or 80W spec versions which would perform ENTIRELY differently.
NVIDIA: you’ve made an amazing GPU here. This laptop can legitimately play games at 4K AND ultra settings and it’s not a thermo-nuclear disaster. It has all the fancy added features like Ray Tracing, DLSS, NVIDIA Broadcast, NVENC encoder, Reflex, all that stuff. But, please, please. Don’t mislead people like this. Call it RTX 3080 Mobile. RTX 3080m. Something. ANYTHING. It’s not the same chip, you can’t call it the same. And it offers HALF the performance of the desktop part. HALF. Come on! For every tech nerd like me that knows to look for CUDA cores, clock speeds and TDPs, there are thousands of people who don’t even know what a CUDA core is, let alone it’s relationship to the thing they are going to spend thousands of pounds on’s performance in games. They read RTX 3080 and have seen the reviews of the desktop cards and not put it together that they aren’t even remotely the same. So, please, fix this. Love, everyone.
So, there you have it. You’ve heard my clearly impassioned thoughts, but I would love to hear yours in the comments below. What do you think of the mobile part, the performance difference, and the name scheme… woes. Let me know in the comments below!
