I’m just about to buy a new laptop for coding and work, and I have the following two options:
I can summarize the main differences as below:
CPU Intel i7-11800H Ryzen 7-5800H (24M-L3 Intel vs 16M for AMD)
GPU RTX 3050 (4G) RTX 3060 (6G)
Memory 16GB (2x8) MAX 16GB (2x8) Expandable to 32
Display 16" (2560x1600) 15.6" (1920x1080)
Ports Thunderbolt 4 Not supported
Brick 230W (lighter) 300W (Heavy)
Design Gamery Look More professional
Weight 5.07 lbs 5.30 lbs
Battery 7.2 Hrs 9.2 Hrs (80Wh battery for both)
Price $1815 $1705 (international shipping for the Pro)
I’m very confused which one to buy after I sold my laptop (a previous version of the same model). Any advice or recommendations? Or should I pass on both of them and consider something else given the current shortages and availability issues?
What kind of work are you planning to do with this machine? There seem to be tradeoffs in both cases; at first glance, I’d probably lean towards the first mainly because of the higher resolution screen and the thunderbolt capability.
The GPU in the first is a bit weaker, but unless you’re planning on doing a lot of GPU-intensive work, that probably doesn’t really matter, and even if you are, these are both relatively weak GPUs (from a computational perspective, they’d be fine for games and such) that are not well-suited for a lot of intense numerical work.
Is there some reason you’re attached to the Lenovo Legion? There are probably better options available, even from within Lenovo’s lineup.
Is it possible to run Julia code on an eGPU via Thunderbolt?
If your going to code directly on your computer screen (without an external display), I would say take the bigger screen and the higher resolution
Having the option to upgrade to 32 GB of RAM might be useful or even necessary depending of what your doing on your machine (VMs, huge DataFrames and Matrices)
Sure. The headaches might come in making sure the eGPU is properly setup with drivers and all that, but
CUDA.jl and the like have made interfacing with Julia quite simple. It pulls in its own CUDA installation, etc., so as long as you can get the gpu to work properly in the first place you should have little to no problem getting GPU-accelerated Julia code to work.
My workload is not so heavy, normal office work besides Julia simulations and other languages. Yes both devices have tradeoffs, that’s why I’m confused. The legion lineup sells very well in my country, I used my previous machine for two years and sold it at approximately the same price I bought it. I can afford the more premium business models, but it will cost three times that price to get the same specs and then the law of diminishing returns applies.
I don’t game but nowadays GPUs became more an more important. Some normal tasks like converting a video in handbrake can benefit a lot from the GPU, not to mention new trends in programming. And I love that 16:10 aspect ratio, my external monitor is a Dell ultrasharp with that ratio. It’s nice for coding and documents work and I found it surprisingly useful in watching videos as the title goes into that black bar above.
That makes sense. They’re good machines, so you should be safe either way, but I wanted to make sure that there wasn’t some kind of other restriction in play.
It looks like the screen is the deciding factor: higher-resolution screens are hugely beneficial, but only if you plan to use the built-in screen regularly. If you’re planning to use an external monitor, it’s not as big a deal.
The processors are virtually identical (see benchmarks here).
The GPU in the second machine is slightly more powerful, and the ram expansion possibility is nice. If you don’t care about the built-in screen or thunderbolt, the second is a slightly better bet. But thunderbolt capability could also be important if you’re planning on using the machine mostly in a docked set-up.
That’s very interesting, the benchmarks suggest they are too close in performance, even in power performance they have identical marks. I’m now more confused, why the first one has low battery times. I sold the previous machine mainly because of short battery life and a huge power brick at the same time. It had an Intel i7 9750H by the way.
The battery estimates are not all that different. Might just be down to the physical limits of the battery itself (i.e. the second just has a battery with a larger capacity).
The processors are also not the only major user of power. The difference in GPUs might also be to blame
Yes, other hardware than processor can hugely affect battery life. The higher pixel density and brighter display can make a big difference. They have the same 80Wh battery though.
I wll be using the laptop built in display mostly, and I feel 16G of memory will be ok, I rarely use vms on my laptop, I prefer the desktop for that. The shorter battery life of the first option is what worries me more.
I’m leaning towards Apple, I find the OS more polished than Ubuntu, but otherwise very similar. I like the build quality and the M1.
What I don’t like is a lack of a full keyboard, even on the 16" intel mac pro. I find that I’m constantly using home and end, number keys for any intensive arithmetic or data entry.
My HP I have now feels light on build quality, I think Lenovo are solid, not sure who else.
Latest MacBooks are really impressive with the M1 chip. I thought about that too, but unfortunately, many Engineering softwares are still unsupported by Mac. Imagine a simple IDE like CodeBlocks, the latest supported version is 2013. You cannot easily find a windows machine that rivals the performance of an M1 Mac, but software is still a big issue.
Could dual boot, but I see your point.
If this is for coding and work, I think the Legion 5 Pro with the Intel processor is going to give you a better experience. Since you are here, the AVX-512 is something you might take advantage of from Julia as LLVM support for that instruction set matures. The Ryzen 7-5800H does not support AVX-512 as far as I know.
I also noticed some differences in cache sizes between the two processors. Can that have any noticable peformance difference in certain problems including BLAS, Linear Algebra, etc.?
||64K (per core)
||80K (per core)
||512K (per core)
||1250K (per core)
I also say take the bigger screen, but not so much for the physical size or resolution but for the 16:10 aspect ratio. The extra vertical space is a surprisingly large upgrade over 16:9 screens for pretty much every application except gaming and watching movies. Even better: several laptop manufacturers now also have an even taller 3:2 aspect ratio as an option, e.g. Microsoft’s Surface line.
The discussion with comments from the pros in this other thread might also be interesting.
For me, the main benefit of having a laptop is to be able to carry it around easily and use it anywhere (small and light, touch screen, long battery life) and yet get decent performance (multiple cores, large RAM and SSD disk).
I have no connection to Dell, but I’m very happy with my home setup: XPS13 + large external monitor + thunderbolt hub. If one day I’ll be able to do some serious computing, then I might consider buying an eGPU or a standalone desktop workstation.