Question Thermal throttling in Silicon Power P34A80 NVMe SSD and strange SMART status.

[atonement]

I am using the SSD in my Acer Nitro 5, i5 8th gen. laptop which only has a PCIex2 link. Thermal throttling has been an issue right from the beginning but it seems it's getting worse. Anything (mostly file transfers to and from ext. HDDs but even over MTP over a USB 2.0 connection can cause the SSD to heat up) that takes more than a few minutes is enough to cause temps. to rise quickly to 80-85°C. Temps. on more than a few occasions don't go back to normal unless I put the laptop to Sleep. What's worse is nowadays whenever this happens CrystalDiskInfo shows the SSD health as "Bad". There's a bunch of issues in the SMART info. But after temps. fall back to normal all the SMART info also gets reverted back to the original values. Is this a cause for concern? Currently on FW version 12.1. Silicon Power hasn't releases a FW update AFAICT. Will updating to 12.3 help in thermals? Investing in an inexpensive copper heatsink from Aliexpress is worth considering? I don't know what thickness of the heatsink can be accommodated.
There's still 2+ years of the warranty period still left.
Thanks.

 

[Justinus]

A lot of NVME drives using silicon motion or phison controllers, especially a couple years old ones, run really hot. While it's reporting health as bad while it's throttling, I'm not seeing any of the visible smart data actually being reported as bad. My guess is it's solely the temperature, unless you've got a screenshot showing non-temperature related smart data reading bad as well.

It doesn't really hurt the drive, all modern nvme drives have smart throttling profiles which is why they throttle so low relative to other hot-running chips like cpus and GPUs. It may eventually shorten the lifespan simply due to increased heat, though.



If this is a correct photo of your laptop (not sure if it's the right version of the nitro 5) it looks like Acer went the route of the awfully ineffective heatspreader that acts more like a heat cover and traps heat.

I would open it up and see if the thermal pad is positioned correctly on the controller, making good contact, and in good shape. You might consider replacing it with a new, high quality pad for a few bucks on Amazon.

You could also inspect the clearance to the bottom cover, maybe there are a few MM to use a thermal adhesive pad to stick on a few copper shims (or ultra low profile heatsinks per NewMaxx) to increase thermal mass. It wouldn't solve the problem but it would give you more heatsinking mass to absorb heat before reaching throttle point.

Unfortunately it doesn't look like it was designed to work without the heat cover, otherwise I'd suggest removing the cover entirely and re-testing.


Regarding the firmware I am confused - you say they haven't released a new firmware as far as you can tell, but then follow it with asking if updating from 12.1 to 12.3 could help? A few nvme drives in my memory have received firmware updates to improve thermal performance/throttling profiles. There's no "magic fix" but it's definitely possible for it to be somewhat improved if a firmware update was specifically designed to.

Degradation in thermal performance is typically due to physical reasons though, like above where I say you should check the heat cover and thermal pad.

 

[NewMaxx]

You can get low profile heatsinks (2mm, 4mm in this case), and as Justinus shows above you can find internal videos/pics.

 

[solidsnake1298]

I know that NVMe SSD controllers tend to throttle at a lot lower temps than a CPU or GPU would. It is my understanding that this is done to keep the temps below a point where it would be hot to the point of silicon degradation and to prevent it from throttling hard. If it did throttle in the same way as CPUs and GPUs it would significantly impact performance and performance consistency. By throttling early and less severely the SSD doesn't have to throttle nearly as hard and there is usually no perceivable change in performance.

Most SSD controllers start lightly throttling in the ~75*C range. In terms of silicon health, 77*C is nothing to worry about.

 

[atonement]

A lot of NVME drives using silicon motion or phison controllers, especially a couple years old ones, run really hot. While it's reporting health as bad while it's throttling, I'm not seeing any of the visible smart data actually being reported as bad. My guess is it's solely the temperature, unless you've got a screenshot showing non-temperature related smart data reading bad as well.

It doesn't really hurt the drive, all modern nvme drives have smart throttling profiles which is why they throttle so low relative to other hot-running chips like cpus and GPUs. It may eventually shorten the lifespan simply due to increased heat, though.

View attachment 47232

If this is a correct photo of your laptop (not sure if it's the right version of the nitro 5) it looks like Acer went the route of the awfully ineffective heatspreader that acts more like a heat cover and traps heat.

I would open it up and see if the thermal pad is positioned correctly on the controller, making good contact, and in good shape. You might consider replacing it with a new, high quality pad for a few bucks on Amazon.

You could also inspect the clearance to the bottom cover, maybe there are a few MM to use a thermal adhesive pad to stick on a few copper shims (or ultra low profile heatsinks per NewMaxx) to increase thermal mass. It wouldn't solve the problem but it would give you more heatsinking mass to absorb heat before reaching throttle point.

Unfortunately it doesn't look like it was designed to work without the heat cover, otherwise I'd suggest removing the cover entirely and re-testing.


Regarding the firmware I am confused - you say they haven't released a new firmware as far as you can tell, but then follow it with asking if updating from 12.1 to 12.3 could help? A few nvme drives in my memory have received firmware updates to improve thermal performance/throttling profiles. There's no "magic fix" but it's definitely possible for it to be somewhat improved if a firmware update was specifically designed to.

Degradation in thermal performance is typically due to physical reasons though, like above where I say you should check the heat cover and thermal pad.

There's no heatspreader etc for the SSD slot in the laptop. The exact model no. is AN515-31.
Regarding the FW, Silicon Power AFAICT didn't release any upgrade. But I there's a generic firmware available. I had come across it the Reddit thread by NewMaxx.

 

[atonement]

You can get low profile heatsinks (2mm, 4mm in this case), and as Justinus shows above you can find internal videos/pics.

Lot's of items on Amazon USA can't be shipped to India and even otherwise shipping charges and import duties are prohibitive. So I was looking to buy from Aliexpress. There also 2mm,4mm options are available for Copper heatsinks. I guess I need to try one.

 

[NewMaxx]

Lot's of items on Amazon USA can't be shipped to India and even otherwise shipping charges and import duties are prohibitive. So I was looking to buy from Aliexpress. There also 2mm,4mm options are available for Copper heatsinks. I guess I need to try one.

Right, I was just giving an example. Yeah, you may have to measure.

 

[atonement]

I know that NVMe SSD controllers tend to throttle at a lot lower temps than a CPU or GPU would. It is my understanding that this is done to keep the temps below a point where it would be hot to the point of silicon degradation and to prevent it from throttling hard. If it did throttle in the same way as CPUs and GPUs it would significantly impact performance and performance consistency. By throttling early and less severely the SSD doesn't have to throttle nearly as hard and there is usually no perceivable change in performance.

Most SSD controllers start lightly throttling in the ~75*C range. In terms of silicon health, 77*C is nothing to worry about.

When throttling sets in performance absolutely craters. In the low tens of MB/s. Occasionally it falls to as low as a few MB/s. It's a 512GB drive limited to x2 and I usually keep nearly a third of the drive free most of the time. While running Prime95 once temps. were hovering at 87°C. Ambient temps. rarely cross early thirties, mostly in the mid-twenties. CPU and GPU run pretty cool even when not undervolted and/or underclocked. It's not because of strenous workloads transferring hundreds of MB/s that can cause throttling but even sustained transfers over USB 2.0 (15-30GB) are sufficient to cause perf. to nosedive.

 

[solidsnake1298]

Interesting. Then it sounds like you could benefit from a low profile heat sink, as suggested.

A secondary thought: How is your laptop situated? Is it on a hard surface (table, counter top, etc.)? Or is it on a soft surface that may restrict airflow underneath the laptop?

 

[atonement]

Interesting. Then it sounds like you could benefit from a low profile heat sink, as suggested.

A secondary thought: How is your laptop situated? Is it on a hard surface (table, counter top, etc.)? Or is it on a soft surface that may restrict airflow underneath the laptop?

On a cooling pad which is on top of a book for further elevation. I use my laptop as a desktop replacement of sorts. It never leaves the desk. Other than the SSD temps. aren't a concern at all. If I had a desktop I could probably confirm whether the SSD or inadequate cooling in the laptop was at fault.

 

[solidsnake1298]

Woooooow. So it is on a cooling pad and still thermal throttles? Heat sink all the way. That feels like a major oversight on Acer's part.

 

[atonement]

Update: Yesterday I flashed the v12.3 firmware and so far it seems thermal throttling is vastly reduced. In fact I transferred nearly 100GB data to an external HDD and temps. didn't cross 70°C even once. Maybe it was a case of fake temps. earlier. Although in the Task Manager SSD usage was also shown as 100% when throttling occurred.

 

[atonement]

I think the SSD is dead. I ran the 64GB test in CrystalDiskMark and a few minutes later I saw the "No Bootable Media" message in the BIOS. In the BIOS itself it doesn't show the SSD. Only the original HDD. Is there any Phison utilty to revive the drive or something? Any help will be much appreciated. Thanks.
Edit: I booted from pen drive hosting Linux Mint and it didn't see the SSD as well.

 

[atonement]

WTH?! I had installed Windows back on the HDD. Switched on just a little while ago and everything was as it was earlier. Suffice to say I was shocked. It's still on FW 12.3. And its throttling again though not as badly as before.

 

[atonement]

@NewMaxx Is there any (safe) way to downgrade to the original firmware? As of now the SSD has done the vanishing act at least 4 times and afterwards during boot Windows reoprts what seems to be several disk errors and apparently corrects them(SMART info still shows no glaring issues). Besides several BSODs (WHEA:Uncorrectable error) while playing lightweight games, benchmarking in AS SSD (10GB file size) transferring several files totalling 14GB,each no more than 1GB. I think I was wrong about thermal throttling being improved by the 12.3 fw. More likely incorrectly much lower temps. are being reported and when the critical temp. is reached the SSD shutsdown.

 

[NewMaxx]

There were WHEA errors with NVMe drives after X570 came out, actually, although not as serious as that. It's rare to get disk errors on SSDs actually...that sounds like a hardware issue, although cannot be sure it's the drive alone without trying a different one.

 

[atonement]

There were WHEA errors with NVMe drives after X570 came out, actually, although not as serious as that. It's rare to get disk errors on SSDs actually...that sounds like a hardware issue, although cannot be sure it's the drive alone without trying a different one.

Is there anyway to safely downgrade to the original firmware? Any file bigger than a few GBs is sufficient to cause the SSD to suddenly vanish. I can't install any game or a VM for this reason. I really need to reinstall Windows but that doesn't seem to be doable.

 

[NewMaxx]

Yeah that sounds like a serious issue...although I've heard "it's not compatible" from many support people on laptops. So that's unfortunate. I'm not aware of people downgrading firmware on that controller, it's not something that's done terribly often, not to say it's impossible but I wouldn't do it without asking support simply to maintain RMA consistency.

 

[atonement]

Yeah that sounds like a serious issue...although I've heard "it's not compatible" from many support people on laptops. So that's unfortunate. I'm not aware of people downgrading firmware on that controller, it's not something that's done terribly often, not to say it's impossible but I wouldn't do it without asking support simply to maintain RMA consistency.

The damn SSD is crashing at every possible chance nowadays. I came across this SSD firmware zip in Silicon Power website but there's no detail at all https://www.silicon-power.com/web/firmware. Any idea about it?
Edit: I booted into Linux Mint and ran the "nvme id-ctrl" command which didn't list the controller! I wonder if mine was a Silicon Motion unit instead of Phison which I had assumed al this time. Original firmware was ECMF v12.1.
I got in contact with the customer care but unsurprisingly no response yet.
This is the last time I buy from such lousy manufacturers and no experimenting with any unofficial firmwares etc.

 

[atonement]

Yeah that sounds like a serious issue...although I've heard "it's not compatible" from many support people on laptops. So that's unfortunate. I'm not aware of people downgrading firmware on that controller, it's not something that's done terribly often, not to say it's impossible but I wouldn't do it without asking support simply to maintain RMA consistency.

Customer care responded and provided a firmware "SP NVME M.2 FW_ECFM12.6 (E12+Bics3)" This is a newer firmware and E12 + Toshiba Bics3 sounds right? While filing the complaint I had explicitly chosen SP P34A80.

 

[NewMaxx]

Customer care responded and provided a firmware "SP NVME M.2 FW_ECFM12.6 (E12+Bics3)" This is a newer firmware and E12 + Toshiba Bics3 sounds right? While filing the complaint I had explicitly chosen SP P34A80.

Original E12s used BiCS3, that is correct, if your drive only had two NAND packages on the top side then it's the old layout.

 

[atonement]

I finally tried the new firmware today a few weeks ahead of schedule, as the system was crashing quite frequently. While the flashing process went all right it seems the SSD was to far gone. Just before flashing CrystalDiskInfo was showing a yellow caution about the Spare area parameter. Only 10GB writes. Windows installation crashed just after the first boot and the SSD had vanished again. Now will have to try my luck at the service center. Even if the replacement in an inferior one as reported elsewhere it can't be any worse.

 

[atonement]

Received the replacement SSD a few days ago and so far the thermals are vastly better. It looks like the original SSD was a bad one as far thermal throttling was concerned. Firmware version is VB431D61. Now I am kinda glad that I gave a try with the unofficial firmware otherwise I would have been stuck with a defective SSD. Don't think I will need to buy a heatsink anymore.
Edit: No thermal throttling whatsoever after over nearly 3 weeks of fairly heavy usage.