Dell Precision 5510 Owner's Lounge

Bokeh, thanks for sharing these! I'm trying to make sense of how your lid open/lid closed results compare.

Was the specific test used to create this photo repeated with the lid closed?
(I don't think it was, but I may be misinterpreting something)

What were the parameters of the benchmark software that you ran to get the system running at 130W? And where are you measuring 130W from? I thought the CPU and the GPU each had a TDP of 45W, so the two total 90W when both fully loaded. Where are the other 40W being used? I imagine the motherboard and peripherals use some additional power, but not 40W! Are you measuring the power draw at the AC outlet the power brick is plugged into? I suppose 130W would make sense there.

 

Did you run these same tests with the lid open?

In terms of CPU usage CAD is heavily single threaded and max performance will be achieved by a single thread running at the maximum clock speed. A second thread will likely be used to a comparatively insignificant degree for the non-computational user interface operations. It is rather unique in this regard. Mult-threading is of little benefit to CAD because of the linear hierarchy of dependencies in any part or assembly. When a given feature is dependent on the one before it, the two features cannot be cranked out simultaneously on two separate threads.

A more representative benchmark for CAD would be:
1. 15 seconds of single threaded CPU computations at max clock speed with low GPU utilization.
2. 60 seconds of low CPU usage and heavy GPU usage (heavy for a notebook GPU)
3. Repeat starting at #1

What do you mean by "System is sitting straight up"?

At first, I understood that to mean that closed system is in some sort of stand with the keyboard perpendicular to the ground. (If this is the case which edge of the laptop on top?

If the system is positioned as I described in my interpretation of "sitting straight up", it's rubber feet aren't pressed against any surface.

If the laptop is sitting feet down on a flat surface, then the feet could have a significant impact on preventing hot air from the output vents entering the cool air intake, as air flows in the gap between the laptop bottom and the surface, and the rear foot creates a barrier preventing intake of air from behind the laptop where it's likely to be warmer.

If the laptop is not sitting on a flat surface it's a completely different situation, and the foot would have a much much smaller effect as air is not forced to run through a narrow channel along the bottom of the notebook. If the notebook is positioned standing up with the hinge edge at the top, the foot is doing almost nothing to prevent intake of warmer air. Hot air rises, and the hot air output vents are higher on the laptop than the cool air intake. The only possibility of hot air moving downward to the cool air intake would be if the intake created a large amount of suction which it does not. But assuming there was enough suction to pull down some rising hot air, the lip created by the foot could make a tiny difference by directing airflow a bit.

The bottom rubber foot can be seen because the laptop bottom is aluminum, a good heat conductor. The foot is made of rubber, an insulator, and sits on top of the aluminum bottom between it and the camera lens. As it's an insulator, the surface temperature of the rubber is lower than that of the aluminum beneath it and surrounding it, so it shows up clearly because it's surface is uniformly cooler than its surroundings.

The temperature on either side of the foot doesn't appear substantially different. It doesn't look like the foot is breaking up the natural circular temperature gradient it sits on top of. HOWEVER, if you were to do this test with the laptop sitting on a flat surface, you very well may observe the foot breaking up the natural temperature gradient. So the fact that I don't see this leading me to believe that you did have the laptop positioned with the keyboard perpendicular to the ground as I described.

Anyway, thanks for the thermal imagery! I'd love to see too Apple's to Apples comparisons of the same benchmark with the lid open vs closed if you get a chance.

 

130W from the wall as measured by a Kill A Watt power meter at the wall.

 

System was running on a flat surface. Flipped the system up to get a thermal image of the bottom of the system.

 

For some sorts of software development, the Mac OS is objectively better than other options -- it has a more mature desktop environment than Linux, but it has real Unix-like underpinnings whereas it's a real pain to do some kinds of development on Windows. There's also a bandwagon effect: if the rest of your team/company is all tied to tools which are either not Windows-compatible, or outright Mac specific, then even if the general sort of development you do (e.g. Java stuff) is perfectly happy to run on Windows, running the rest of your outfit's toolchain may not be possible.

I don't find it an appealing desktop, and am happy to deal with the clunky desktop environment for the cleaner infrastructure and better performance of actual Linux, but on my personal machine -- where I have too many Windows-specific things (and where I need the better battery life and more predictable sleep and dock/undock behavior) I literally have to run a Linux VM to do my job away from my main machine, because our deployment and testing tools all assume Linux or Mac (and while we're moving to docker for a lot of it, docker support on windows is still sketchy at best.)

Software development is also a very broad area; there's a lot of it where any old ultrabook is fine, up through big Java or C/C++ projects where the 16GB limit on the MacBook Pro is crippling.

 

Correct. But you need a driver either way, there's no RAID-mode-specific problem, which is what was suggested by the post I was responding to.


Sent from my iPhone using Tapatalk

 

I've just arranged with Dell support for another mainboard replacement. I fell into the trap described here because I wanted to see if disabling Thunderbolt would speed up the boot process. The BIOS described in that link wouldn't work for me, possibly because I have put a password on the BIOS.

I would take the view that Dell would be liable for fixing this type of bricking problem even after the warranty has expired as it is a defect built into the machine.

John

 

John, I'm not sure that you posted the correct link. In the link you posted, the XPS 9550 users (note not precision users like us) were able to regain normal functionality by enabling thunderbolt again. Is this not a possibility for you? What has required you to swap motherboards because firmware updates/downgrades couldn't fix it?

On a related note, I'm curious, is the precision motherboard different from the XPS outside of the different Nvidia graphics card? Outside of the GPU and tangential circuitry, is the Precision motherboard otherwise identical to the XPS 9550 motherboard?

 

No, that's the currect link. It describes my problem including the corrupted display when one gets (with difficulty) into the BIOS using F2. Otherwise the computer just sits with a blank display and refuses to boot. I can't do the BIOS reset as I have a password on the BIOS which requires clicking on the right part of the corrupted display (and there's no sign of any cursor) and Dell don't admit to having a key combination for doing a blind reset.

AFAIK the XPS and the 5510 are extremely similar other than the nominally different GPU. The BIOS versions share the same bugs (eg BIOS 13 having the battery charge problem).

John

 

You could reset the BIOS by removing the CMOS battery. It is under the motherboard on side opposite to the power input.