Precision 7530 & Precision 7730 owner's thread

New Thunderbolt drivers and ASMedia drivers this morning.
"Improves power management stability" - possibly addressing the USB stuttering issue. I'll try it out in a bit...

https://www.dell.com/support/home/us/en/04/drivers/driversdetails?driverid=886c9
https://www.dell.com/support/home/us/en/04/drivers/driversdetails?driverid=jcdn0
[Edit] - Nevermind, only the ASMedia driver is new, the Thunderbolt driver is a few weeks old...
[Edit 2] - OK, these are both not new today, the ASMedia driver is from 9/7. The page here was not updated until today so my auto checker did not notice. I installed the ASMedia driver and checked with C states turned on and still got stuttering, so no fix. However, I do not have the issue described by @Regular_Ragnor a few pages back.


Also, I've received word that my case regarding the TB18DC power issue has been escalated and I "might" be getting a call from the engineering department to work on it in a few days...

@brazzmonkey

TB16 power - Even with a 240W adapter attached, it can only deliver 130W to the laptop, this is a limitation of the Thunderbolt connection. The TB18DC uses both Thunderbolt ports solely to handle this issue (at least that would be the case if it worked as claimed). Otherwise it's basically exactly the same dock. It has no additional capabilities.
The Pioneer BD-RE drive that I mentioned is USB powered.
An M.2 SATA SSD would absolutely be faster than an HDD.

 

Hey guys—I was twiddling with ThrottleStop, and realised that even at idle, with nothing in the background and no open apps, the CPU stays in the C0 for far too long—my C0% readout fluctuates between 5% to 12% on idle. For a notebook with Optimus enabled and the dGPU sleeping, and with a nearly 100 Wh battery, I would expect something like 12-15 hours of runtime of light use (i.e. browsing the web and scrolling through PDFs). However, the battery runs down extremely quickly: I doubt I get anything better than 4.5 hours on a full charge, and I have to plug in halfway through the day. I don't use the dGPU at all, because I hardly have time for games or video rendering.

Has anyone else experienced this? I suspect a driver that is waking up the CPU too often. Resting a finger on the touchpad causes the C0% to jump up to 15-16%.

I've seen other screenshots of ThrottleStop on other machines (notably, in the ThrottleStop thread itself) where the C0% was around 0.5%. This is frankly unacceptable, and it's another one of my motivations to move to Linux.

 

I see 1.1 to 2.9% C0% on idle on my 7730 with i9 and no dGPU with Dock + 2 external displays attached. Current BIOS and drivers installed.

 

Hello @Ionising_Radiation please could you do a set of cinebench runs to see the performance of your Xeon and the impact of throttling ? I hesitate between a Xeon and a 8950hk.

I'm also looking for the same kind fo benchmark for the new ZBook 15 G5 using these processors. I did heard they did set a very good cooling system on it (but we have yet to see...).

 

@SvenC
out of curiosity, is the Toshiba XG5 SSD you mentioned in your signature the one provided by Dell as a class 40 SSD?

 

Correct.

CrystalDiskMark 5.1.2 values on Win10:
(Seq Q32 / 4K Q32 / Seq / 4K)
Read: 2736 / 139 / 1607 / 19
Write: 352 / 107 / 349 / 44

 

It depends on which type of RAID you're going to use. I'll use two 1 TB drives as an example.

RAID 0 (striping) gives you a 2 TB logical drive. Data is distributed evenly over both drives, where even numbered bytes are written to the first drive and odd numbered bytes are written to the second drive. In case of a single drive failure, you will have 100% data loss, because 50% of the data is gone and the remaining 50% is useless. With NVME drives, the speed boost in a RAID 0 setup is only marginal and not noticeable in real-life. You will only notice the extra performance in a benchmark. The risk of 100% data loss, in my opinion, far outweighs the marginal speed boost gained from RAID 0.

RAID 1 (mirroring) gives you a 1 TB drive that is 100% protected against a single drive failure. The second drive contains an exact copy of the first drive, but the RAID set appears as a single drive. If you plan to backup regularly, RAID 1 is just a waste of valuable drive space.

There's also RAID 5, which requires at least 3 drives and uses approximately 33% of each drive to store checksums of the data on the other 2 drives. The result is a single logical drive that has a size of approximately 66% of the sum of all 3 drives. In the event of a single drive failure, the data on the remaining 2 drives, combined with the stored checksums, can be used to reconstruct the lost data from the failed drive. RAID 5 gives you a large logical drive with protection against data loss, but no speed boost. Again, if you plan to backup regularly, RAID 5 is a waste of valuable drive space.

In other words, RAID on NVME drives makes no sense unless you need a single large volume (using RAID 0 or 5) or protection against data loss (using RAID 1 or 5).

 

I'm pleased to report that I just got off the phone with a Dell technician in the resolution expert center at corporate headquarters, regarding the dock power issue, and I have some information on both dock issues to present:

Power issue -
The support team successfully replicated the issue for the first time, today. He had me describe the issue in detail (in particular the behavior of the LED light on the dock cable) to make sure that my issue lined up with what they saw. They had the lead engineer behind the TB18DC dock come and check it out, and their test setup is being packaged up and sent over to the engineering team so that they can try and figure something out. Apparently they have been trying to replicate this in their lab for some time without success, but they have first-run hardware there (pre-dating what was shipped out in the first batch of retail systems), so there is some speculation that something has changed between the original engineering samples and when it went to mass production that has caused this problem. Anyway, as an engineer myself, I know that it is hard to fix a problem that you cannot replicate. They have now successfully replicated the problem. Hopefully this leads to a fix.

USB keyboard/mouse stuttering -
While I had the guy on the phone, I brought up the USB keyboard and mouse stuttering issue. They are also aware of this issue already and have made some progress on it. They believe that the issue is with the Intel Management Engine driver. He was already aware of the fact that you can resolve the issue by disabling C states on the CPU as we have discovered, but he presented a better way to resolve the issue, which is to go to Device Manager, find the Intel Management Engine Interface under "System devices", go to Power Management there and uncheck the box for "Allow the computer to turn off this device to save power". I am currently running in this mode (C states on, power saving on Intel ME off) and at least I have had no stuttering while I typed this. We'll see if it lasts. Anyway, they're working on getting a driver update prepared and pushed out (coordinating with Intel, I guess) which will fix this problem without the need to make the power management change in device manager.

I noticed that Dell posted an Intel Management Engine driver update on 9/6. I haven't installed this yet (I usually ignore the Intel ME updates and just use whatever pops down through Windows Update). Looking over the case notes, the technician could not tell me if this one contained the I/O stuttering fix, but he could tell me that the engineering case is still open so the engineering team doesn't think that they are "done" even if the fix was included. He said that he'd look into this further and get back to me.

[Edit] I checked on the Intel ME update that was posted last week, it has a date from Intel of 4/11 so I'm kind of doubtful that it includes a fix. I did go ahead and install it, though, the one I had previously installed was dated 1/16.

It was nice to talk to an expert. The guy I spoke with last week was eager to help but didn't know the system and dock well at all. This guy was clearly very familiar with the system. In the end, I'm now reasonably assured that they are taking these issues seriously and working towards resolution.


[Edit 2]
USB stutter still happening. Turning C states back off...

 

@SvenC
well I'm no expert but those numbers seem good, especially the read rates. I'd expected much less from a so-called "class 40" SSD.
Writes are a bit on the low side, but this should be good enough for an OS disk. I guess I wouldn't see any real-life significant improvement by upgrading to a Samsung SSD...
According to Toshiba website this SSD support SED (Dell's doesn't mention that), so that's a good match for me (not that I know how exactly this should be configured... BIOS?).

@Michiko
My point exactly. I don't think that RAID 0 would bring any noticeable performance in daily use.
As for RAID 1 and 5, I'd rather rely on regular backups.
In RAID, "I" and "D" stand for "Inexpensive Disks". Not really what NVME SSD are at the moment...

 

I performed a Cinebench loop a while ago, forgot to post it here. I really want to do a thorough benchmark, a la NotebookCheck, but I cannot find the time at all. Anyway, here you go: