T-Force Cardea Liquid M.2 PCIe NVMe SSD Review

When it comes to modern NVMe drives, heat actually becomes a real concern. Often times M.2 drives are buried under hot graphics cards, which tends to exacerbate the heat problem. To combat this growing concern for NVMe temperatures, and to stop the drives from thermal throttling, companies like Team Group are taking things to the next level. Their latest NVMe SSD release, the T-Force Cardea Liquid, features a liquid-cooled heat sink. How can such a small device be liquid cooled? Follow along as we investigate this new drive and run it through a gambit of benchmark tests.

Specifications and Features

The T-Force Cardea Liquid NVMe drive certainly garners attention with the claim of liquid cooling on an NVMe drive, but in reality, it might not be what you expect. The liquid cooling aspect of this drive is a closed cavity that is indeed filled with colored liquid. This is not something you will be plumbing to in your custom water cooling loop, rather, it’s a self-contained cooling system. That’s right, it’s a passive water cooling solution.

The heat sink is composed of an aluminum base, plastic top, and a sliding aluminum screw cover. Considering that most NVMe drives come with no integrated cooling system, we have high hopes that although it may be passive, it will be highly effective at cooling the drive. Team Group claims that this cooling method will drop the drive temperature by 10ºC.

The liquid serves two major functions. Firstly, it acts as a heat dissipation device for the high-performance drive it’s paired with. Second, and likely most impact to water coolers, is that the water color can be changed easily. This means you can change the water in this drive to match the coolant you are running in your loop.

The heat sink may be adding bling to your system, but the drive beneath is all business. Running the Phison PS5012-E12 controller and 3D TLC NAND flash memory, it’s a foregone conclusion that this drive will be fast by modern standards. For the 1TB size, Team Group touts a sequential read speed of 3,400 MB/s in CrstyalDiskMark.

Team Group T-Force Cardea Liquid M.2 PCIe NVMe SSD Specifications
Capacity256GB, 512GB, 1TB
InterfacePCIe 3.0 x4 with NVMe 1.3
ControllerPhison PS5012-E12
Flash3D TLC NAND
Form FactorM.2 2280
Dimensions (LxWxH)83.9(L) x 24.3(W) x 14.1(H) mm
Weight42g
Sequential Read/Write Speeds (1TB)Read:  3,400 MB/s
Write: 3,000 MB/s
4 KB Random Read / Write (1TB)Read: 450 k IOPS
Write: 400 k IOPS
Warranty5-year limited warranty (SSD)
1-year limited warranty (Heat Sink)
Endurance / MTBF2,000,000 hours
Pricing1 TB – $169.99
512 GB – $102.99
256 GB – Pricing N/A

Packaging/Product Tour

The T-Force Cardea Liquid ships in a predominantly black box with a stylized picture of the drive and eye-catching artwork on the front. The back of the package contains some details about the drive’s performance and about the sliding mechanism. While it’s certainly not the most elaborate packaging we can imagine, it doesn’t leave us wanting more.

Looking at the drive itself, we are instantly greeted with a pleasant teal/aqua color. The block comes pre-filled and includes a bottle of the same blue liquid and a micro funnel to aid in filling. To round things out, the package also contains an owner’s manual and a case badge sticker.

On the underside of the drive, we find the product label. The heat sink and NVMe drive are sealed together with an anti-tamper label which poses a dilemma. Many modern high-end motherboards come with integrated M.2 heat sinks. If users want to use the integrated motherboard heat sink, they cannot do so without voiding the warranty.

 

At the back end of the drive, we have a fill port and a sliding metal cover. The metal cover simply slides about 10mm to hide the fill port and also the mounting screw. Filling the drive or changing the liquid is as easy as simply unscrewing the plug and using the included funnel. Users who want to color match the liquid in the drive can do so with just a few minutes.

While we don’t like the fact that the warranty is voided in the process, removing the heat sink was exceptionally easy — It’s held in place with four plastic tabs. The heat sink makes decent contact with the controller. In the image below, you can see the oil impression left from the thermal pad contact.

To install it in our test system, we needed to remove the OEM M.2 heat sink, which was well worth it. The drive simply looks stunning in our test system.

Once installed, the drive is noticeably taller than the PCIe lane. While it may be taller than the metal shrouded lane, it’s approximately the same height as the PCH cooler. Thus, there shouldn’t be any conflict with the height.

In a world dominated by RGB lighting, we were surprised that the Cardea Liquid didn’t come packed with addressable RGB LEDs. Perhaps in the next revision of this drive, they will add an RGB element…

Software

Team Group offers a nice piece of free software that applies to the Cardea Liquid NVMe SSD. To measure drive temps and overall health, Team Group has a nice utility. Aptly named the SSD Toolbox, this utility has includes all the built-in features you need to check the status of the drive. We tested Version 5.13 with Windows 10.

SSD Toolbox

Coolant Options

While you can fill the T-Force Cardea Liquid with just about any liquid you want, Team Group recently introduced a new line of liquid coolants. Available in a test kit of 5 different colors, the new CK5 coolant is designed to meet the needs of high-demand cooling applications.

  • Environmentally friendly anti-corrosion material
  • Natural biological decomposition
  • Low conductivity
CK5 Coolant

Test System

To test the drive, we will install it in our system as a secondary drive and run benchmarks programs on an empty drive. Under normal conditions, many users would install their OS on the drive. However, for our tests, we wanted to isolate the hard drive to maximize the performance and reliability of scores.

Below are the tests we run with a brief description.

  • Crystal Disk Mark Version 5.5.0 – Run at Default Settings
  • AS SSD Version 2.0.7316.34247 – Run at Default Settings
  • ATTO Version 4.01.0f1 – Run at Default Settings except for the QD Set to 10
  • Anvil Storage Utility Benchmark Version 1.1.0 – Default Settings
CPUAMD RYZEN 9 3900X
CoolerNZXT Kraken X62 280mm AIO
MotherboardASRock X570 TAICHI AM4
Graphics CardPowerColor RED DEVIL Radeon RX 580
SSDADATA SU800 256GB (OS)
Power SupplySeasonic 1200W Platinum PRIME
Operating SystemWindows 10 x64 v1909

 

Benchmark Results

Throughout all of our testing, the Cardea Liquid never exceeded temperatures of 45ºC. We did notice that once the drive heated up, it took a while for the temps to drop back down to ambient. It’s clear that the heat sink acts as a heat reservoir with no active cooling.  That being said, once we took off the heat sink, our drive quickly reached 60ºC when running CrystalDiskMark. The downside to passive coolers is that it can take some time for the temps to recover back to ambient, but as our tests show, the drive runs much cooler with the heat sink.

With a naked drive and no cooling at all, there was no thermal throttling observed, even running the benchmarks back-to-back to simulate a worst-case scenario. All benchmark results and stress testing with done in an open-air test bench with no fans or direct airflow.

CrystalDiskMark

Crystal Disk Mark – Reads

Unfortunately, we were not able to achieve the rated Crystal Disk Mark speed for read performance. We tried multiple different versions of the benchmark software but never attained it. Although the drive did not reach the expected sequential read score of 3,400 MB/s, the drive did prove to be quite fast compared to our other test drives in the lineup.

Crystal Disk Mark – Writes

When it came down to the read performance, our results landed exactly where we expected them to for the Q32T1 test. While the multi-queue tests are right where we expected them to be, the single thread, single queue sequential test showed exceptional performance compared to our other test drives.

AS SSD

AS SSD – Reads

There were no big surprises here. As we saw in the Crystal Disk Mark benchmark tests, this drive is quite fast and easily scores over 3000 MB/s in the read performance.

AS SSD – Writes

At this point, we know that the drive is blazing fast and a top competitor compared to the others we tested. However, our result in the 4K-64Thrd test looked exceptionally fast. We achieved consistent results with multiple tests, but it does appear to be an outlier.

ATTO

ATTO – Reads

Comparing to other drives our scores in ATTO were a little lower at higher I/O sizes. However, in the lower I/O size testes the Cardea Liquid outperformed the competition.

ATTO – Writes

Our results for the write test in ATTO showed incredible performance. As with the read performance, our drive showed particularly high scores at the lower I/O levels.

Anvil Storage Utility

Lastly, we will take a quick look at the drive performance in Anvil’s Storage Utility. The tool can monitor and test read and write speeds on hard drives and also produces an output performance score for comparison. It’s a great utility that also provides further information such as partition and volume information.

Anvil Storage Utilities

Conclusion

Over the past few years we’ve seen Team Group release several innovative products, so it didn’t come as much of a surprise that they were the first company to create a liquid-cooled M.2 drive. The T-Force Cardea Liquid is a welcome addition to an NVMe market. With a stylish and effective cooling solution, it’s got a leg up on the competition as many drives are still being sold with no built-in cooling option.

Team Group bundled the Phison PS5012-E12 controller and 3D TLC NAND flash memory, so we had high expectations for the performance. While we didn’t actually achieve one of the rated speeds for this drive, we did observe some very impressive benchmark results overall. Our sample produced faster read times than everything else we compared in CrstalDiskMark. The multi-queue tests were right where we expected them to be, but in the single thread, single queue sequential tests, Team Group crushed the competition.

The Cardea Liquid is surprisingly competitive with a price tag of $169.99 for the 1TB model. Looking at other 1TB PCIe Gen3. X4, NVMe 1.3 drives rated for 3000+ MB/s max sequential read, we find the majority are selling between $130 and $200. Considering that most of those drives come with no heat sink, or just a simplistic aluminum cooler, we feel the Cardea Liquid has the advantage here. With exceptional performance numbers and a stunning heat sink, we are happy to recommend the T-Force Cardea Liquid for your next build.

Overclockers_clear_approvedClick the stamp for an explanation of what this means.

David Miller – mllrkllr88

Related Reading:

Loading new replies...

Avatar of petteyg359
petteyg359

Likes Popcorn

6,518 messages 13 likes

Sure it looks cool, but how long does it actually run before it overheats because the silly bling on it is actually insulating? I expect the cheapest of metal sinks would perform better than a plastic box with a puddle in it.

Reply Like

Avatar of EarthDog
EarthDog

Gulper Nozzle Co-Owner

77,325 messages 3,984 likes

Sure it looks cool, but how long does it actually run before it overheats because the silly bling on it is actually insulating? I expect the cheapest of metal sinks would perform better than a plastic box with a puddle in it.

I don't believe our reviewer tried to run it hard for an hour, but I believe he had mentioned an extended run time and no throttling?? Maybe I am confusing reviews?

Reply Like

Avatar of Robert17
Robert17

Premium Member

3,698 messages 160 likes

You're not missing anything ED. It appears from the review that the fluid enhances the dissipation of heat. Water has a specific heat of 1. There are few better liquids for that purpose. I can see how it would match a metal strip easily. My main concern would be a leak. If they designed it well and the owner doesn't damage the unit it should be capable past the time it's replaced with DNA drives. Or not depending on the R/W cycles.

Reply Like

Avatar of Blaylock
Blaylock

"That Backfired" Senior Member

8,053 messages 981 likes

According to the review the length of time for the water temps to drop back to ambient were excessive compared to an aluminum heatsink. My question is how many NVMe drives are stressed for hours at a time without a break? Who uses NVMe drives as file servers?

Just my opinion but, if I were to manufacture a water cooled "heatsink" for an M.2 drive it would be a waterblock with a cold plate, not a heatsink. One that could be added into a custom loop.

Reply Like

Avatar of petteyg359
petteyg359

Likes Popcorn

6,518 messages 13 likes

According to the review the length of time for the water temps to drop back to ambient were excessive compared to an aluminum heatsink.

That was my point. The water is going to eventually reach the drive's maximum temperature because the plastic is insulating, and then this would perform worse than a bare drive, let alone one with a metal sink.

My question is how many NVMe drives are stressed for hours at a time without a break? Who uses NVMe drives as file servers?

Who uses NVMe drives for anything but that? If you're not needing to repetitively store and load terabytes of data where the GB/s transfer rate is useful, what is the benefit of having an NVMe drive at all?

I think 3 million files for a total of 264GB is plenty for a 'real world' test :D ...snip... This was a pretty serious torture test

I have to strongly disagree. If you're paying for an NVMe drive, you're paying for GBps transfer speeds. Your benchmark, with STR well under 200 MBps, is hardly a test for a cheap SATA drive, let alone NVMe. Try again with a single hundred-gigabyte file (random or a tarball of a bunch or music or something psuedo-random) copied a few times.

Reply Like

click to expand...
Avatar of EarthDog
EarthDog

Gulper Nozzle Co-Owner

77,325 messages 3,984 likes

Who uses NVMe drives for anything but that? If you're not needing to repetitively store and load terabytes of data and actually need the GB/s transfer rate, what is the benefit of having an NVMe drive at all?

Me...and many many others USE it, but NEED (the speeds) I/we don't. I'd go as far to say that an OS and game storage is more common than using these as loading TB/s of data and requiring GB/s transfer for extended periods of time. Surely some users beat on these and for that, our testing isn't good enough... but we can't please all use models for every part. We just try to look at typical use scenarios for the consumer.

That out of the way, we are looking at updating our storage testing methods to include more significant temperature testing. :)

You should see an update from the article author in this thread on this specific drive soon!

Reply Like

click to expand...
Avatar of mllrkllr88
mllrkllr88

Member

121 messages 0 likes

To answer your question I thought I would setup a little test. As in the main review, the drive is tested in an open air test bench with no active airflow pointed at it. There is an AIO close by but it does not blow directly on the drive, so it's close to a worst case scenario.

I decided to create a large file and copy it from drive to drive. I started with some random benchmark software files I had on my USB drive, and copied them many times. Here is a screenshot of the file creation process, as you can see the dive is getting a good workout even before the main event takes place. I copied files back-to-back and the drive never got much of a break doing the initial file build up.
File Creation.jpg

Here is a screenshot of the finished file. I think 3 million files for a total of 264GB is plenty for a 'real world' test :D
Transfer File.jpg

Here is a screenshot of the final file copy and benchmark result. This was a pretty serious torture test but I think the heat sink worked out well. With decent airflow in a case, I believe the max temperature would be considerably less. With the system running idle and no workload, it took 25 minutes to drop the temp back down from 52c to 25c.
Bench Result.jpg

Would you like to see this kind of testing implemented in future NVMe reviews?

Reply Like

click to expand...
Avatar of caddi daddi
caddi daddi

Godzilla to ant hills

11,186 messages 41 likes

"My question is how many NVMe drives are stressed for hours at a time without a break? Who uses NVMe drives as file servers? "
Out here in the wild nvme drives can REALLY help in video editing rigs.

Reply Like