OVERVIEW
When nVidia released the Desktop RTX 50 series to the market, it was not received well for many reasons. It was expensive, often above the MSRP, almost unavailable in any quantity, and does not have significant gains from the previous generation of nVidia RTX GPUs. There are still more down the pipeline that are yet to be released but shouldn't be the saving grace either.
But what about the Laptop GPUs? Let's list some that were released over the last month or two:
RTX 5090 Mobile: 10496 CUDA Cores, 24GB RAM, 256-bit Bus.
RTX 5080 Mobile: 8192 CUDA Cores, 16GB RAM, 256-bit Bus.
RTX 5070 Ti Mobile: 5888 CUDA Cores, 12GB RAM, 192-bit Bus
RTX 5070 Mobile: 4608 CUDA Cores, 8GB RAM, 128-bit bus
There is one in particular that seems like a good proposition when I looked at it, not just in terms of performance but also balancing out power and heat. And that's the 5070 Ti.
EXAMPLES OF NOTEBOOK / LAPTOPS WITH RTX 5070 Ti MOBILE GPU:
But before I delve down a little, just let you all know I'd certainly would love to review one of these great examples that carry it. I'll have to spend time to use and give my unbiased reviews about it. But if any company would like to give me one to try, I would do it. Here's a couple of examples. I'm sure there will be more down the pike but I'd love to get my hands on the Chuwi GameBook, or the MX-16, lighter examples that use this GPU.
CHUWI GameBook: Launching Soon.
 |
| Chuwi Gamebook: Ryzen 9 9955HX, 5070 Ti (140W), 32GB RAM Price and yet to be released. 2.3 Kilos which is light for a 16" |
AftershockPC MX-16: Intel Ultra 9 275X, 5070 Ti
 |
| Aftershock MX-16: Intel Ultra 9 275X, 5070 Ti, 32GB RAM, and any kind of SSD available now 2699SGD (About 2090 USD) at base config. 2.3 Kilos like the Gamebook. |
MSI Vector A18X A9W (Image from notebookcheck):
 |
| From notebookcheck's review also with a Ryzen 9 9955HX. The Heaviest and most costly one of the three: at 3.5kg plus a 1.327 400W Power Brick 3200 USD (About 4150SGD) at their configuration |
I'd certainly would love to review one of these great examples in a more unbiased manner. So if anyone would oblige to give me one, do let me know.
RTX 5070 Ti (Mobile) is in a nutshell, a refined Desktop RTX 4070
On paper, they are actually very similar to each other. Same CUDA core count, Same VRAM count. The only differences are power profiles, clock speeds and using the faster GDDR7 RAM which dramatically boosts the overall memory bandwidth of the mobile GPU. Below is a table I did to compare other similar GPUs and last generation GPUs including a few desktop ones.
 |
| Left side, the 5888 Core Candidates plus the desktop 5070. Right side: the 4608 Core Candidates. |
12GB is the new minimum for any GPU, even in Laptops.
From various benchmarks, the combination of using less VRAM on top of a narrower memory bus really slows down the RTX 5070 Laptop significantly and presents as a memory bottleneck, even in just Full HD 1080p resolution. Even though the amount of memory bandwidth looks decent enough on paper and the amount CUDA cores differs by 20%, you'll find the 8GB VRAM gets fully saturated much more frequently than having 12GB.
Still the mobile RTX 5070 Ti can prove to be capable enough for gaming at 1440p and to some extent 4K on most modern titles. That's the benefit of having more VRAM available before it is fully used or can be allocated for more demanding graphical features.
General Performance:
The RTX 5070 Ti Laptop in general is:
- On par, Trades blows, or slightly faster than the:
RTX 4070 Desktop, RTX 5060 Ti Desktop and 4080/4090 Mobile
- Faster than the:
Desktop RTX 3090 and below
Laptop 5070 by a Significant amount.
RTX 5070 Ti Laptop is a true successor to the RTX 4070 laptop.
According to various benchmarks, the RTX 5070 Ti (Mobile) can be as high as almost 40-45% faster than even just the non Ti Version of the Laptop GPU. As expected, with lesser CUDA cores, the Laptop 5070 Ti performs in between the Desktop 5070 and 5060 Ti, and in between the RTX 4080 and 4090 Laptop GPUs, filling in an actually considerable gap in between these two GPUs, and also performs slightly better than a desktop RTX 4070 if we compare the GPUs with the same amount of CUDA cores. It's helped because of a 192-bit bus and 12GB VRAM. 12GB is just about the minimum that you need these days but it shows that the RTX 5070 Ti (Laptop) GPU can do it.
The non-Ti version of the 5070 Mobile GPU is only marginally faster, again depending on wattage, than the previous RTX 4070.
RTX 5070 Ti carries the best balance
From Notebookcheck's various reviews of the 5090 and 5080, they all mostly share a common quirk, high power consumption and heat, especially if the GPUs are tuned to run on Full tilt. Razer and Asus do get around this with optimised power figures which also reduce heat and power, to the point a 250W external power brick can be made with their 5090 laptops. But most of them need a 400W PSU. Chuwi shrinks down to a 250W brick, and likewise XMG Pro 16 which notebookcheck also reviewed, does use a 250W one too.
So it wouldn't be a surprise if a 250W will be standard in a laptop that uses the 5070 Ti GPU which is great if you need to transport it around often. But even if you're more than likely to use it in a Desktop Replacement role, using less power will also mean less money toward your power bill compared to models to use the more powerful GPUs.
That's the reason I wish to review. How powerful and how much weight you can conserve before your laptop really becomes a desktop replacement because 2.3Kg laptops that use full wattage GPUs seem to be the sweet spot here. There are some good examples of more powerful GPUs like the Razer Blade, but really be prepared to pay a premium for these.
Interesting quirk about the RTX 5070 Ti Laptop's hidden potential.
I've only just seen a video of Dave2D that he thinks there is actually wasted potential hidden in the RTX 5070 Ti laptop. Most laptops GPUs will cap out to its best performance after it reaches a certain maximum wattage as seen below:
The AD107 based 4050 to 4070s do tap out after about 105 Watts, and the GB206 based RTX 5070 mobile starts to tap out at around 115W.
However, for the RTX 5070 Ti, that seems to be not the case as it's those higher end boards. The #05 to the #03 boards do have potential for power. According to his graph, the RTX 5070 Ti hasn't really capped and plateau-ed out yet even after reaching the maximum 145W that NVidia allows.
That is a clear indicator of some performance and TDP still left on the table that the GPU can still provide, waiting to be unleashed.
There's a reason why they limit the amount power available for the RTX 5070 Ti, understandable in a business perspective and marking perspective. Because if unlocked, can trade blows or beat the higher ranked AND more expensive 5080 and 5090 mobile GPUs, which is true and proven if the latter two are tuned with a lower TDP as shown in some gaming comparisons off notebookcheck which you can see below. It's already fast enough for what the market is aimed at, so leaving at 140-145W makes sense.
 |
| Assassin's Creed Shadows: The RTX 5070 Ti is massively better than the 5070 in 4K. Almost doubling the 5070 and 4070. |
 |
| Indiana Jones: the RTX 5070 Ti can trade blows with a lower TDP 5080 and even a 5090 as shown by a shorter grey meter. Moreover, 5070 Ti's average is not that far away either than theirs. |
 |
| Pretty much same story with Alan Wake 2. Higher resolutions is helped by higher VRAM and higher memory bus width. And it's weird that even the 5070 loses to the 4070. |
 |
| Cyberpunk: The 5070 only has 70% of the performance of the 5070 TI. The 5070 Ti bridges the gap quite nicely between the 5070 to 5080, closer to the latter. |
You can always check the rest of the numbers at notebookcheck. So you can kinda imagine how much closer the 5070Ti can be to the 5080 if the limits were removed. They did an article and sorta VS showdown between the 5070 and 5070 Ti Laptop GPUs which you can see here:
https://www.notebookcheck.net/GeForce-RTX-5070-Ti-Laptop-vs-RTX-5070-Laptop-The-successors-to-the-RTX-4070-Laptop-in-review.1006203.0.html
I won't be surprised if someone can tap on its hidden potential with a new vBIOS down the line though, in which case I'd be very excited to see what it can really really do. Moreover, bridging the gap that is 15-20% away from the RTX 5080-5090 Laptop GPUs can be also done with regular overclocking and undervolting. Of course that won't get as much gains as having a vBIOS that uncaps the TDP of the GPU.
THE GPU IS STILL NOT PERFECT THOUGH.
Personally I'd hold off and wait for a while long if I was the buyer. Because some reviewers are facing some performance issues in games like Doom: The Dark Ages and some AAA titles. To me, it's very likely driver related or the game engines are abit finicky with the RTX 50 series. Till these bugs are ironed out, my advise is to hold off buying them if you're not sure on updating drivers and stuff. Should have a good out of the box experience. So I hope laptop manufacturers are on top of all these issues that have surfaced recently.
