Review Graugear G-M2HP04-F NVMe Cooler Review: Two heatpipes and a fan deliver strong performance

[Admin]

Graugear’s SSD cooler performed very well in our testing. Read on for our thermal benchmarks and detailed information.

Graugear G-M2HP04-F NVMe Cooler Review: Two heatpipes and a fan deliver strong performance : Read more

 

[Marlin1975]

If my hard-drive requires that much maybe I should buy another brand.
I am in the process of building a new computer. Heat/power from the NVME was one of the reasons I choose 1 drive over others.

 

[das_stig]

Prefer Thermalright HR09 or HR10

 

[Alvar "Miles" Udell]

Most users will never encounter a workload this stressful.

Articles like this make me think of the hard drive cooler craze of the early to mid 2000s.

 

[palladin9479]

Yeah .... if your NVME needs this kind of cooling solution then something is wrong.

I suspect this is due to PCIe 5 implementations using the same older fabrication process's that PCIe 3 and 4 drives used to keep costs down. Pumping up the clock rate that high causes the energy usage to spike and therefor needs active cooling.

 

[thestryker]

I kind of love the overkill solutions like this despite them not really being necessary. This seems like a rather well designed cooler which is reflected in the price.

Every once in a while something will make me think about what could have been with BTX. While it was designed to maximize CPU cooling if the memory trace routing issues could have been worked around it would have been a significantly better form factor for today's systems.

Most, if not all, of the SSD heat issues are really due to the M.2 format and where these drives are located due to it. The highest power consumption I think I've seen on one of the SSD reviews here was a bit under 14W peak, but with average around 8W. These aren't exactly high numbers (compared to other desktop PC components), but when located between the CPU and GPU in an area with little airflow it becomes a problem. I would bet that most SSD heatsinks would be able to cool even the hottest drives with some form of direct airflow.

 

[PiersPlowman]

The highest power consumption I think I've seen on one of the SSD reviews here was a bit under 14W peak, but with average around 8W. These aren't exactly high numbers (compared to other desktop PC components)...

These are outrageous numbers for power consumption - worse than a 7,200 RPM HDD - if you disregard spin-up power draw.

 

[thestryker]

These are outrageous numbers for power consumption - worse than a 7,200 RPM HDD - if you disregard spin-up power draw.

While I purposely picked the worst SSD I'd seen so it's bad compared to other SSDs this is a really stupid comparison. First of all unless you set your HDDs to stop spinning when not in use pretty much all client SSDs use less power (most use a lot less) at idle. Also the SSD in question is over 20 times faster than the fastest HDD Tom's has tested (which also averages almost 8W in the copy test). Then there's the peak power consumption where almost every HDD uses more than 14W.

 

[PiersPlowman]

While I purposely picked the worst SSD I'd seen so it's bad compared to other SSDs this is a really stupid comparison.

No it is not. In terms of pure power consumption, the example you provide is comparable.

 

[thestryker]

No it is not. In terms of pure power consumption, the example you provide is comparable.

Oh so you're just ignoring the significant performance difference which makes SSDs way more efficient, got it.

 

[helper800]

Oh so you're just ignoring the significant performance difference which makes SSDs way more efficient, got it.

Exactly. If the SSD does the work 20 times faster at the same wattage as another storage device that means it also uses just 1/20th the power for the same work.

 

[evdjj3j]

I thought it was a slot 1 cooler when I first opened the page.

 

[snemarch]

This is ludicrous 😖

It would be nice if desktop motherboards moved from NVMe toward U.2 or whatever – something that allows a larger form factor and more flexible placement, so we wouldn't need these... contraptions.

(Oh, and perhaps a bit less focus on the BIG NUMBER BETTER that's not attainable in real-life situations, but instead using process improvements to drive down power consumption rather than increasing BIG NUMBER, yeah?)

 

[Alvar "Miles" Udell]

This is ludicrous 😖

It would be nice if desktop motherboards moved from NVMe toward U.2 or whatever – something that allows a larger form factor and more flexible placement, so we wouldn't need these... contraptions.

(Oh, and perhaps a bit less focus on the BIG NUMBER BETTER that's not attainable in real-life situations, but instead using process improvements to drive down power consumption rather than increasing BIG NUMBER, yeah?)

Or if more desktop boards had proper cooling. My X570S board uses decently large aluminum heatsinks that are cooled with case airflow, which isn't the greatest in my case as I have no side fans, and my drives struggle to hit 70°C under typical use.

 

[Albert.Thomas]

Yeah .... if your NVME needs this kind of cooling solution then something is wrong.

Everyone's workloads are different. While the average user won't "need" a device as strong as this, there are definitely folks who do!

 

[Albert.Thomas]

This is ludicrous 😖

It would be nice if desktop motherboards moved from NVMe toward U.2 or whatever – something that allows a larger form factor and more flexible placement, so we wouldn't need these... contraptions.

(Oh, and perhaps a bit less focus on the BIG NUMBER BETTER that's not attainable in real-life situations, but instead using process improvements to drive down power consumption rather than increasing BIG NUMBER, yeah?)

Honestly, I don't understand why internal USB isn't being used for storage. The fastest USB and Thunderbolt drives are much better than anything SATA could offer.

 

[palladin9479]

Honestly, I don't understand why internal USB isn't being used for storage. The fastest USB and Thunderbolt drives are much better than anything SATA could offer.

Short answer
Running OS file systems off USB is inefficient and dangerous suitable only for RO RAMDISK based images.

Long answer
A bit late but it's because of the way USB works vs SATA vs eSATA. USB doesn't support device initiated DMA, this is a good thing because it protects against evil maid attacks. Instead the the device needs to ask the USB host controller and then that controller handles the setting up and transferring of the data from the device into memory so that it's available to other devices. This introduces a non-trivial amount of latency at the start of any access. Furthermore the USB standard is designed for removable peripherals and every device on the bus is treated as temporarily and therefor queued caching is a very dangerous thing. SATA on the other hand supports device initiated DMA, meaning the device can just send it's data directly into memory upon request along with pulling data directly out of memory, while also being a permanent connection and therefor "safe" for caching and other RW optimizations. Then we have eSATA, which is just the SATA protocol that supports the device being removed while the system is powered on, if you check in the BIOS that is what the "removable device" setting is on a SATA port. eSATA didn't live very long because USB was just more convenient and anyone needing professional DAS solutions would use SAS anyway.

 

[snemarch]

Honestly, I don't understand why internal USB isn't being used for storage. The fastest USB and Thunderbolt drives are much better than anything SATA could offer.

I don't know about Thunderbolt – I honestly don't know how it relates to USB with regards to e.g. the driver stack. And USB-C might also have improved the situation.

But my experience with USB-connected storage is that it's... flaky. I don't know if it's because of the physical connection, bugs in OS drivers, hardware on the device side or whatever, but I've had all sorts of bad experiences, ranging from flaky intermittent connection issues, to losing all data (massively corrupted filesystem) on a drive that was mounted for read-only usage (not in read-only *mode*).

Also, Windows *really* doesn't like read errors on USB media. If I get a hard read error on e.g. an old USB flash drive, I'm unable to "safely eject" the drive, and whatever failstate results in Windows not being able to do a clean shutdown – in that situation, I'll have to manually close down applications, use SysInternals sync tool to at least flush outstanding write operations, and then do a hard power-off of the system.

I'd much prefer a connector that's tailor-made for storage – performance- and complexity-wise – over USB 😅