Anyone using Quadro m500m successfully for f@h ?

[PFM]

Have been considering a Lenovo P series laptop that has a nVidia Quadro m500m mobile gpu. I have been told it has its own cooling fan.
But before jumping for it just wanted to check if anyone has any experience with this gpu - good or bad.
I did check the gpus.txt list and I found the m600m but not m500m, so not sure if its not supported or just not listed.

[bruce]

In general, laptops have rather limited GPUs. Some of them are capable of completing a FAH WU before it expires and some are not. I suspect that the M500M is in the latter category but I have not found enough information about it to confirm that as a fact.

Each GPU type has a binary identification which can be found if you run "lspci" or any of the GPU utilities, including the one delivered with the NVidia drivers.

Also, you'll need to install the proprietary drivers from NVidia if there's any hope of being able to use this GPU but I'm not optimistic.

[JimboPalmer]

OK I googled Quadro m500m and it has a GM108 chip with 384 shader cores

The M in GM stands for Maxwell, so it is current and should run all Modern WUs if the drivers support OpenCL (the video card supports OpenCL just fine, with the right drivers)

384 shaders is enough so that all WUs should finish before the deadline if the laptop runs 24/7.

[PFM]

Currently I am running v7.4.4 on an Intel P9600 which is an old core2duo and without interruption it finishes WUs well before expiration. Would the m500m gpu be even slower than this old cpu ?

If the new laptop had a latest generation i7 or even i5 would it be much better to just run the cpu client rather than m500m gpu ?

[7im]

Most GPUs currently outperform all CPUs, at this time.

Since this product was just announced, not even sure it's available yet. Can't find any reviews on it yet, so that's why it's not in the Grey List yet.

However, the specs on the m500m look similar to the existing m600m, so the fah performance would be in the same neighborhood.

[bruce]

In general, GPUs are faster than CPUs and the complexity of the assignments reflect that expectation. CPU projects have to be more limited so that they can run on the most basic of CPU although they'll do much better on an i5/i7. Some CPU projects can even run on servers with many more cores than an i7 while other projects need to be limited to 8 CPUs.

The same is true for low-end vs. high-end GPUs. The GM108 is most likely sufficient to fold but we'd need the GPU identification codes I mentioned earlier.
You can also run lspci inside of FAHCore by entering "fahcore --lspci" in a text window.

Laptops generally have a limited capability to dissipate heat and as you add more CPUs and add work to a GPU, you might be exceeding the heat output that your hardware can dissipate -- or it might mange it reasonably well. That has not been tested.

[7im]

I didn't look very hard, but I didn't even see the new model available for sale yet on the laptop makers we site. Tough to know any real performance info on something no one has seen yet.

[artoar_11]

If this information is correct: http://www.notebookcheck.net/NVIDIA-Qua ... 132.0.html

The Quadro M500M is the successor to the Quadro K620M which is also based on the Maxwell GM108 but lists a higher memory bandwidth of 16 vs. 14.4 GB/s. Therefore, it is also similar to the consumer GeForce 930M.
Date of Announcement: 15.12.2015

[bruce]

We can't do anything until somebody provides the hex codes that identify the GPU from LSPCI.
GPU Mfg: 0x10de
GPU ID: 0x????

Apparently the answer to the original question is "not yet"

So I'll lock this thread. When the value from LSPCI is available, start a new topic.