12-02-2019 08:55 AM
My workflow is normally Solidworks, ANSYS, Matlab, and some other engineering software. I prioritized single core processing effeciency, and the GPU, so I ordered the following:
So, a few questions:
i.) It was hard for me to say no the the 8-core i9 as a gear nerd, I know the benefits of more cores. Would be interested in hearing feedback on if any performance differences between my setup and the i9 would even be noticable other than on a benchmark test?
ii.) I wanted the 16GB VRAM in the GPU, so that was a factor in why I went with the GTX 5000. I am a bit ignorant to how hungry software are on resources, and where we are in technology today if 16GB VRAM is even utilized and beneficial? Are we a bit ahead of time with this one or is it worth having? Would my engineering software benefit from and utilize the additional VRAM? I know work load of course is a factor but please answer as best as you can.
iii.) I was worried about the thermals with the i9 and even having a RTX 4000 paired with it. As we know the RTX 5000 is not offered with the i9. Am I ignorant in saying the setup I went with will have lower thermals and a more reliable and consistent performnce curve than any of the i9 iterations?
iv.) Overall thoughts on my setup with my given workflow compared to a a different spec'd P53? Would you have purchased differently? If so what would you have gone with?
Thank you in advance any input/experience is appreciated.
12-02-2019 09:54 AM - edited 12-02-2019 09:57 AM
Fellow ME here. I work in the AEC industry and my work loads benefit the most from a few fast CPU cores. Autocad, for example, was optimal on up to 8 cores, and the user consensus was that you could get by with a decent gaming GPU (as opposed to an expensive workstation version). As I remember (and I haven't touched ANSYS and Matlab in more than 10 years), what you're using is very heavily bound by the CPU. Puget Systems has some Solidworks benchmarks that may highlight this.
I got the Core i7 9850H with the Quadro T2000 and 32 GB RAM (16 GB from the factory, bought a matching stick and installed it myself afterwards) and two NVME SSDs (but my projects are on an external HDD, so these don't help as much). I'm satisfied with my decision. Back in August, when I ordered the laptop, there weren't any options for custom configurations. I would have liked the Xeon paired with the T2000, but it doesn't seem like this is an option even now. The 8 core i9 seemed very appealing, but I had to remind myself time and again that:
- Turbo is no substitute for base clock, as much as the marketing hype says otherwise
- I don't need as many cores (I upgraded from a 2c/4t Core i7 with a 2.8 GHz base clock and a 3.2 GHz turbo, and the actual speed in single threaded loads is not that different)
- The price was better (paid for it myself), and the difference went into accessories (dock, cooling pad, RAM, additional USB hub, new carrying case).
I'm not sure how much of a benefit the RTX5000 is for your work load, but you can see how much you use it in Task Manager. Worst case scenario, there are external TB3 graphics cards that you can use, and they will give you more power than anything Lenovo could have added into the P53 chassis. If you look at how the heat pipes are laid out, the one that services the CPU is also connected to the GPU, so simultaneous CPU+GPU loads will always be a problem with this design.
12-02-2019 10:08 AM
I appreciate the feed back. I meant to include note on how I also favored the Xeon due to its higher base clock speed versus the i9. I think for any longer workloads such as simulations that a higher base clock speed would be superior to turbo spurts, especially when simulations heat the system up over extended times.
My goal was to have something that 4 years down the road I could still use it if I wanted to, so I am hoping it lasts and that i is not outdated too quickly by whatever Intel puts out over the next few years.
12-02-2019 12:37 PM - edited 12-02-2019 12:38 PM
My experience has been that the higher Turbo speed is more important for single-thread work than the base clock speed. My testing was limited, as I don't usually have a Xeon here to use in a comparison. I used the program SuperPi to start calculating 32k digits of Pi over and over using a single thread, so it did a long-running, sustained test. In real life, however, other things might start and stop affecting other threads' availability to contribute to the speed. I have no real information but my gut reaction would be to expect the Xeon models to use more power and run hotter. Both the Xeon and the I9 are 45w TDP CPUs and after reading the posts about throttling, I would expect the CPU will use all the power available. The Xeon models come with an RTX 5000 GPU vs. RTX 4000 with the I9. I would expect that the RTX5000 might run hotter. Again, this is a gut reaction with no real data.