HP ZBook Studio G7 Aims To Attract Linux Developers, Data Scientists

[phoronix]

Phoronix: HP ZBook Studio G7 Aims To Attract Linux Developers, Data Scientists

The HP ZBook Studio G7 aims to attract Linux developers and data scientists by not only offering a powerful hardware combination and by pre-loading Ubuntu 20.04 LTS but in also shipping a variety of tools and other software packages pre-configured for a modern developer and data scientist workload. We have been testing the HP ZBook Studio G7 for the better part of two months for this Linux-loaded mobile workstation and in this article is a look at this new HP device along with plenty of benchmarks, including Windows vs. Linux performance tests and more.

http://www.phoronix.com/vr.php?view=29953

[CochainComplex]

Nice. If it only had an AMD CPU. I really would like to know how the state of mobil dGPU of amd.

[sn99]

Nice. If it only had an AMD CPU. I really would like to know how the state of mobil dGPU of amd.

Yep, rn unless it is a light book with no dedicated GPU, AMD seems to be just a far far better choice especially considering a data scientist or developer will profit more from more cores than higher clocked cores.

[Guest]

Nice. If it only had an AMD CPU. I really would like to know how the state of mobil dGPU of amd.

For CPU, sure Zen 3 makes sense, although their shoddy Linux support is a real showstopper. For dGPU, well NVIDIA or Intel makes better sense. Way better Linux support, compute libraries and APIs.

OpenCL on NVIDIA is problematic so if you want that, AMD or Intel is better. But if you want hardware accelerated CUDA, NVIDIA is the only choice.

[Leprechaunius]

No numeric keyboard and tiny adjacent up/down keys? Only two USB ports? How on Earth could that attract developers? I would not take that one even if they gave it for free.

[tildearrow]

Two browsers?! Could I at least choose which one(s) to keep?

[pracedru]

I have had so many issues with HP laptops that i would rather buy another one, even if it is slower.

[Paradigm Shifter]

Also worth noting is that while the link to HP.com takes you to the base spec, the "Data Science" configuration clocks in at $8,100 ($8,750 if you want that 2TB SSD) although to be fair that is with a Quadro card, which basically adds between $1,000 and $4,000 (depending on card) to the asking price over an equivalent consumer GPU. Yes, I know it's on sale right now, so it's more like $5,200. Their configuration page, though, is troublesome. No individual selection of CPU, RAM, GPU - you have to read through all the details of each selection to check what you're actually getting.

That said, while I might use this if one was given to me by work (ha, I should be so lucky!) and I loved my first HP laptop (an nx7000, one of the first Centrino laptops) I have since sworn off HP completely due to several poor experiences.

[stormcrow]

Really kinda of skeptical about a laptop class computer aimed at data scientists. Back when I was still running HPC, the heavy work was done on clusters while the visualization could be done on basic desktop class hardware with few bells and whistles. Development only needed a dumb terminal. What do they expect people to burn a hole in their desk waiting for their latest data queries to be done local on a single CPU/GPU, especially when laptops, regardless of who makes them, are thermally limited even with fans that sound like a 737 taking off?

Data scientists are going to be connected to a cluster at a lab somewhere, or a cloud computing facility, or they'll have a boat anchor desktop tower under their desk chewing their heavy work.

Linux devs are more likely to want something like this, sans the Nvidia card. I've seen a handful lugging around a back breaker workstation because they don't have room at home/school for a desktop. After lugging my massive Dell gaming laptop around through airports a few times I decided to just drop the money for a regular desktop with an ultrabook class laptop that weighs half as much (and 3x the battery life) for whenever mobile is a thing again.

[CochainComplex]

Really kinda of skeptical about a laptop class computer aimed at data scientists. Back when I was still running HPC, the heavy work was done on clusters while the visualization could be done on basic desktop class hardware with few bells and whistles. Development only needed a dumb terminal. What do they expect people to burn a hole in their desk waiting for their latest data queries to be done local on a single CPU/GPU, especially when laptops, regardless of who makes them, are thermally limited even with fans that sound like a 737 taking off?

Data scientists are going to be connected to a cluster at a lab somewhere, or a cloud computing facility, or they'll have a boat anchor desktop tower under their desk chewing their heavy work.

Linux devs are more likely to want something like this, sans the Nvidia card. I've seen a handful lugging around a back breaker workstation because they don't have room at home/school for a desktop. After lugging my massive Dell gaming laptop around through airports a few times I decided to just drop the money for a regular desktop with an ultrabook class laptop that weighs half as much (and 3x the battery life) for whenever mobile is a thing again.

i wouldn't consider my self a data scientist but working as a physicist at particle accelerator institute - I have to work on measurement data. Some of the analysis is rather computational heavy. I do have a quite powerfull laptop at least it was 2019 (xeon 2286). Why? because I have to repeat this analysis at other institutes too. During traveling. Indeed I use the sheer power of our HPC but sometimes you can not have access to it or if you have access downloading and uploading data might be slow due to bad internet connection. Or the que priority is not in your favor because other project have been temporarily given higher priority. Besides I want to check/develop my code locally - because of the mentioned circumstances. Compiling over submitting it as batchjob is not very comfortable nor fast. And abusing the submitnodes as compiler workers is not always liked. i prefer to build the singularity image locally.
Since we are in a Pandamic I'm glad that my "desktop performance equal notebook" is portable so that I can easily work at home or in the institute without having to rely on a stable vpn/remote solution*.

*During the first lockdown in germany the DSLAM in the neighborhood was fried so that it took them almost 3 weeks to "make my Internet great again".