CAD laptop that wont break the back - or bank, looking for reccomendations

After looking for a long time, I have resorted to asking people for choosing a University laptop

Looking for a laptop with decent build quality and a good GPU(either discrete or strix halo), It’ll be used for CAD(mainly Siemens NX), Matlab and some simulation workloads.
Actualy doing any intensive workload will be pretty rare and so I dont really mind bad thermals and noise.

Specs;

  • Preferably 14" and less than 2kg(5lb) , 16 is fine too
  • minimum requirements for performance:
    • 13th Gen Intel® Core™ i7-13700H
    • Nvidia RTX A1000
      I dont really need the “Quadro” class GPU, as the software is certified for intel/amd/nvidia’s non-studio drivers as well, so only the raw performance really matters
  • Decent Build as I’m planning on using this for a while
  • Upgradable or at least 32gb of RAM
  • Available in EU, for less than 2000€
  • nice to have proper linux support
  • nice to have Dual NVME slots(basicaly impossible)

I’m also willing to go 2nd hand if it means saving money

Here are some laptops that almost fit what im looking for:

HP Zbook Ultra G1: This is perfect but it is too expensive
Asus ROG Zephyrus G14 (2025) GA403WM-QS049X: I dont want an asus laptop, also the price
Framework 16: Too large, too heavy

Yes, CAD has softened, their GPU requirements [*to a point]
Doing dually nVME, will likely push laptop sizing, towards larger side [15.6 min]

https://www.newegg.com/p/1TS-000E-17XA0?Item=9SIA24GK1P0329
https://www.newegg.com/p/1TS-000E-1CVY2?Item=9SIAZ2GKGW4333

In case I wasn’t clear, the P16 gen 2 is basically the opposite of what i want, its a proper workstation and at 2.95 kg (6.5 lbs) i don’t think It’ll work for me

I’m prioritising the weight and size over almost everything considering I’ll need to carry this on my back (almost) every day, and as noted, dual nvme isn’t a must

I was considering the p14 gen 5 with RTX 500 Ada but the GPU is not enough

tbh I should probably be looking at 14" gaming laptops from the 3000 generation but i dont know where to look

I think the ASUS Zephyrus G14 is a great option but not easy to upgrade. Acer Nitro V is cheaper and works well. MSI Stealth 14 has good features. Lenovo Legion 5 is strong but heavier. All are about 14” and under 2kg. Pick what fits your budget and needs.

What do you need the GPU for? If it’s for cad then a workstation gpu will be much faster in NX, (but not for example in inventor). The raw performance is not indicative in performance. But if you only make small models it would be fine to have a slower gpu.

In NX, amd gpu’s are actually faster than geforce gpu’s compared to raw performance. I think it is because nvidia specifically only targets quadro’s with some optimisations.

A desktop RTX 4090 is slower than integrated amd 780M or slowest quadro. As for cost effectiveness, a RTX 1000 ada / 2000 ada or the 7700S in the framework 16 would be a give good performance to cost ratio. the RTX ada a500 would also be much faster than the Geforce options you have.

The data is interesting, can you share where can i find more of those benchmarks you posted?

You’re right, I should’ve further specified the workload,
NX/SW/Inventor/Catia will run off of basically any system, even for complex models without bottlenecking the GPU, and everything on the benchmark posted is way above “good enough”

The issue is running simulations and analysis, which does actually use the GPU(look at ansys/Marc benchmarks), and i dont know what I’ll need to run in the foreseeable future.

Also, I was under the impression that nvidia’s studio(workstation) drivers could be installed in consumer cards as well, does that not work?

It’s also worth noting that the specs posted was the uni’s reccomended specs so i dont want to go below that

Aware that P16 isn’t what you’re chasing down
But was mentioning dual nVME, is more likely in this size category

this is from notebookcheck, nvidia studio drivers do the same thing as normal geforce drivers, just slightly more “stable”. they don’t give any different drivers for applications than a geforce driver.

you mentioned those results being fine as well, but it can drop down quite a bit more if you go into geforce gpu’s.

are you sure simulation runs on the gpu? Lots of FEA stuff still doesn’t run on a gpu, but all cpu. this does depend on what simulation you are running.
For simulations getting zen5 cpu would be the best bet, because they have accelerated avx512. and even still, for school you are not running big analysis, so it will still be fast enough. an asus p16 even though it is asus, would be nice, small package with a 4060, but it is also expensive…

I would go for an amd hx 370 with a geforce 4060, and then in NX you can force the use of the integrated gpu as it is faster in that application.

Is there any reason for the proart over the zephyrus g16? the price and specs are similar.
Also, the benchmarks make strix-halo seem like a sweet deal, so I might also go for the Asus z13, which has a form-factor i really like, while still being pretty affordable and light

Most commertial packages support GPU acceleration for the simulations now, its somewhat new so there might not be many benchmarks for it

You didn’t mention battery life, so might I suggest a portable ITX build?

I don’t envy your position, OEMs hate their customers and you rarely have the perfect option available.

If you haven’t gotten it yet, I do wonder how limiting integrated graphics tied to slower LPDDR RAM is like, you can turn graphics down to lower the strain on VRAM in games, but I assume CAD is far more demanding on VRAM bandwidth (doing blender sculpting lately and notice stutters and pausing when overflowing VRAM), Strix Halo does look really nice for performance, but the lack of dedicated VRAM might be a huge issue, I suspect a dGPU might be a better option due to bandwidth issues, but if you run out of dVRAM then a large amount of shared might be better.

I hope you found a solution that isn’t giving you headaches anyway.

Having seen NX be used in enterprise environments with some truly potato grade hardware, those minimum requirements seem like they’ve been given to your university by Siemens themselves or whatever company the university uses for their software management.

I would suggest a larger screen like @GoldenAngel1997 suggested, because the side bars of NX take up quite a bit of the model space on the screen. Having done 3 years of Matlab work on a dual-core 2015 Macbook pro, you can use basically anything for programming work. I’ve upgraded my laptop to a Framework 16, 7840HS, 64GB ram and the iGPU. It’s more than enough performance for ‘normal’ cad use, and is actually more graphically powerful than my official work workstation, that has an Nvidia 3000 series Quadro card and an 11th gen intel processor (6-core).

The main thing I would recommend is 32GB ram, but beyond that there aren’t too many things to worry about. For university work, you won’t be running anything too computationally intensive for a reason, they don’t want people to have to spend lots of money on new laptops when they start. In terms of simulations, if you’re using NX you’ll likely do FEA within NX itself. Even for work, it’s all done on CPU, and doesn’t take that long on modern hardware.

CFD is the only thing that is run on GPUs regularly, and that is done on the cloud so massive simulations can be run overnight or over a weekend. I wouldn’t worry about the GPU compute performance for now, but if it is something that you may eventually use, it’s going to be on university cluster hardware where you will remote in. GPU simulation is best used on very large datasets, where speedups are counted in hours or days, rather than minutes. You won’t (and shouldn’t) be expected to do anything this large at university.

I would also agree with @nutral, with getting a cpu that runs avx512, looking at AMD Zen4+. It helps, and is slightly faster for specific tasks. I would steer away from Intel at the moment, because CAD packages and simulation isn’t playing too nicely with multiple core architectures on one chip without a fair amount of manual tweaking and scheduling. My honest recommendation at the moment would be a Framework 13, with the 7640U or 7840U. You can go for the newer zen5 based chips, but I wouldn’t go above 8 cores on a laptop, I have a desktop for larger tasks as a preference.

[EDIT] I forgot to mention the most important thing with whatever you decide on getting. 250£/$/Eur of peripherals will make more of a difference to you than 5000£/$/Eur of compute hardware will. Make sure you have a few hundred left over for a good mouse (I’m partial to a Logitech MX master), and if you want to be fancy but ready for business level CAD, get a spacemouse too.

1 Like

(avoid both types of mice for linux support, though! 3dconnexion haven’t released a linux driver since 2014, and the last time I tried, macros and reassignments were impossible on MX Master mice)

1 Like

That is true. Unfortunately, there are no enterprise CAD packages that have front end linux support anymore, so if you do get a spacemouse it’s limited to CAD use only.

Bar Onshape, since it’s browser-based (and my tool of choice), but otherwise, I absolutely concur!

What’s the point? Barely any of the portability and much heavier.

I plan to get a half decent monitor for at-home use, and I have used NX enough on my 14" laptop that i can handle it.

Yeah, especially with newer laptops having soldered memory, im considering either 32 or 64gb if the budget allows

I’ve beenn eyeing the mx master 3s ever since it came out,

I wish i can keep using Onshape but nope :frowning:

Thanks everyone for the suggestions btw!(i still dont get to actually buy it tho)