• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Sabrent Rocket 4 Plus 2 TB + PS5 Heatsink

W1zzard

Administrator
Staff member
Joined
May 14, 2004
Messages
28,873 (3.74/day)
Processor Ryzen 7 5700X
Memory 48 GB
Video Card(s) RTX 4080
Storage 2x HDD RAID 1, 3x M.2 NVMe
Display(s) 30" 2560x1600 + 19" 1280x1024
Software Windows 10 64-bit
The Sabrent Rocket 4 Plus uses the famous Phison E18 controller. In our real-life testing, it achieves a top score, almost matching the Samsung 980 Pro and WD Black SN850 at better pricing. We also tested the Sabrent PlayStation 5 heatsink, which is custom-shaped to fit the opening in the PS5 for maximized heat transfer.

Show full review
 
Wasn't this released a while back? Why is it still so damn expensive :wtf:
 
Wasn't this released a while back? Why is it still so damn expensive :wtf:
It has a good price, but you can buy better drives like the 980 PRO or the SN850.
 
SN850 pricing in Australia is an utter joke, would be the last drive I would buy same as the 980 EVO (yes EVO, this is no longer a Pro model).

If Hynix Gold P31 came in 2TB I would buy one immediately.
 
Great review as always...

SN850 pricing in Australia is an utter joke, would be the last drive I would buy same as the 980 EVO (yes EVO, this is no longer a Pro model).

If Hynix Gold P31 came in 2TB I would buy one immediately.
Amozon Sells the Hynix gold 2TB for $244.99 just a heads up
 
Making it into the top 4 in performance with a heatsink designed for consoles, niiice.


I was expecting a budget 'just good enough' since the PS5 was involved


SN850 pricing in Australia is an utter joke, would be the last drive I would buy same as the 980 EVO (yes EVO, this is no longer a Pro model).

If Hynix Gold P31 came in 2TB I would buy one immediately.

*looks around*
[whispering] The secret is to use a credit card [/whispering]
 
Still hideously priced.
 
When did they introduce the 2TB? Damn will check it out. Thanks for the heads up.
September :)
 
So instead of the NVMe premium coming down for all drives, we're back to the same old 'cash cow' issue of "the newest shiniest 'must have' 2TB NVMe4 drive costs roughly the same as a 4TB MX500 SATA which is a whole 1 second slower for real-world game load times but can fit twice as many increasingly bloated games on..." Decisions, decisions...
 
So instead of the NVMe premium coming down for all drives, we're back to the same old 'cash cow' issue of "the newest shiniest 'must have' 2TB NVMe4 drive costs roughly the same as a 4TB MX500 SATA which is a whole 1 second slower for real-world game load times but can fit twice as many increasingly bloated games on..." Decisions, decisions...
It's great to have that option, is it not?

NVME
SATA SSD
SATA Mech


For me, i want those fast sustained write speeds. After suffering Sata and NVME drives with sustained writes under 50MB/s, i learned my lesson about going the cheap route
 
Does this drive support formatting to 4096B LBA size? Some of the latest NVMe SSDs allow this, and it's most easily done with command line programs (nvme-cli) on Linux from a separate drive or live distribution. In theory performance should be slightly higher than with 512B size. This can be checked with standard SMART-enabled utilities like smartmontools. It would be illuminating if at some point a 512B/4096B comparison could be professionally done by TPU on supported NVMe SSDs:

Code:
PS C:\Users\ssb> smartctl -c /dev/sda
smartctl 7.2 2020-12-30 r5155 [x86_64-w64-mingw32-w10-b22000] (sf-7.2-1)
Copyright (C) 2002-20, Bruce Allen, Christian Franke, www.smartmontools.org

=== START OF INFORMATION SECTION ===
Firmware Updates (0x14):            2 Slots, no Reset required
Optional Admin Commands (0x0017):   Security Format Frmw_DL Self_Test
Optional NVM Commands (0x00df):     Comp Wr_Unc DS_Mngmt Wr_Zero Sav/Sel_Feat Timestmp Verify
Log Page Attributes (0x1e):         Cmd_Eff_Lg Ext_Get_Lg Telmtry_Lg Pers_Ev_Lg
Maximum Data Transfer Size:         128 Pages
Warning  Comp. Temp. Threshold:     84 Celsius
Critical Comp. Temp. Threshold:     88 Celsius
Namespace 1 Features (0x02):        NA_Fields

Supported Power States
St Op     Max   Active     Idle   RL RT WL WT  Ent_Lat  Ex_Lat
0 +     9.00W    9.00W       -    0  0  0  0        0       0

Supported LBA Sizes (NSID 0x1)
Id Fmt  Data  Metadt  Rel_Perf
0 -     512       0         2
1 +    4096       0         1

See the "Supported LBA Sizes" section. Right now my SN850 is set to 4096B format.
 
Looks like a Yes
1638911603823.png



Although i thought 4K cluster size was old news by now? Isn't it the default?
 
Although i thought 4K cluster size was old news by now? Isn't it the default?

It seems that client SSDs come by default with 512B LBA size for compatibility purposes, and even the one you showed a screenshot of above has a default setting of 512B (see where the + sign is in the "Fmt" column).

There are some limitations with a 4096B native LBA size: notably, it isn't possible to clone partitions or restore images made from 512B units to 4096B ones, which would be a major inconvenience for many people wanting to transfer their old system effortlessly. Also, some storage-related software (partitioning tools, etc) may refuse to work with 4096B storage; certain VMware products won't work with it either:

4096 byte (4K) sector drives emerged as a way for hard drives to support higher capacity by creating better utilization of space with optimized data placement. Like hard disk drives, SSDs can be formatted to different sector sizes and since many operating systems provide optimal performance with a sector size of 4096, this is the default for many of the latest devices. Some systems and applications are not yet able to take advantage of 4Kn (4K native) drives, and these may require reformatting for compatibility.

Intel suggests using a 4096B size for optimal performance for some of its datacenter SSDs:

These Intel® Solid State Drives support a 512 byte and 4096 byte (4K) physical sector size. Most operating systems provide optimal system performance when using the larger physical sector size. [...]

This blogpost from a WD SN850 user suggests that performance is slightly higher with a 4096 LBA setting, at least on Linux:

Another source suggesting better performance with 4096B LBA:

Many NVMe SSDs support both 512-byte sectors and 4096-byte sectors. They often ship with 512-byte sectors, which are less performant than 4096-byte sectors. Some also support metadata for T10/DIF CRC to try to improve reliability, although this is unnecessary with ZFS.

NVMe drives should be formatted to use 4096-byte sectors without metadata prior to being given to ZFS for best performance unless they indicate that 512-byte sectors are as performant as 4096-byte sectors, although this is unlikely. Lower numbers in the Rel_Perf of Supported LBA Sizes from smartctl -a /dev/$device_namespace (for example smartctl -a /dev/nvme1n1) indicate higher performance low level formats, with 0 being the best. The current formatting will be marked by a plus sign under the format Fmt.

You may format a drive using nvme format /dev/nvme1n1 -l $ID. The $ID corresponds to the Id field value from the Supported LBA Sizes SMART information.
 
Last edited:
Is this a linux issue? Windows uses 4K on the OS partition by default, and has a long time (Since Vista SP1)

No, this is not a Linux-specific issue. It's a different setting from the file system's cluster size.
The file system can still be using 4 kB clusters while the drive uses a 512-bytes LBA size.

Most modern client storage uses a 4 kB physical block size, but with a 512-bytes logical block size for compatibility purposes as mentioned earlier. This mode of operation is called "512 emulation" or "512e". Some large-capacity or enterprise-class storage comes equipped with both a 4 kB size for physical and logical blocks. This mode of operation is called "4K native" or "4Kn".

Some easy to digest related information here:

On Windows you can check the drive's logical block size under System Information. Here are my WD SN850 and Crucial MX500. The Crucial has a single partition with the standard NTFS cluster size of 4 kB, yet it has a Bytes/sector size of 512 bytes (which cannot be changed, unlike on the SN850 or the Sabrent Rocket 4 Plus of this review). This property is independent of the partitions made:

systeminfo-lba.png
 
Everything i'm reading basically says, thats hardware.

The hardware shows 512 to be compatible, even when its 4K

As long as you use 4K, you get the proper performance - but if you use 512, it's emulated and has to load the whole 4K cluster to regardless

Those posts you're showing are about esoteric file systems that needed to be changed to the 4K sizing, since they defaulted to lower values. Note how most are many years old, and usually not about NTFS?

1638939119516.png

1638939153027.png



While outside my knowledge, this could also be one of the differences between MBR formatting and GPT formatting
 
Everything i'm reading basically says, thats hardware.
The hardware shows 512 to be compatible, even when its 4K

Yes, it can be considered a hardware setting that is configurable on many (but not all) NVMe SSDs.

With the 512 bytes setting, a 1 TB SSD may present itself to the operating system as having 1,953,125,000 logical sectors
With the 4096 bytes setting, the same SSD may instead present itself to the OS as having 244,140,625 logical sectors

In both cases the SSD may still internally use 4096 bytes physical sectors, but in the 512 bytes logical sectors case (512e mode), there is a "translation process" in firmware that maps them to 4096 bytes sectors. This translation presumably decreases performance in a few ways, which is how setting the SSD to 4096 bytes sectors may improve it slightly.

Again, this is not directly related to NTFS cluster size.

As long as you use 4K, you get the proper performance - but if you use 512, it's emulated and has to load the whole 4K cluster to regardless

Yes, that's basically it.

Those posts you're showing are about esoteric file systems that needed to be changed to the 4K sizing, since they defaulted to lower values. Note how most are many years old, and usually not about NTFS?

1638939119516.png

Some of the posts I linked suggested that the 4K sizing may be incompatible with some applications or environment (e.g. VMware vSphere), others that the 4 KB setting (i.e. 4096 bytes per sector) has higher performance and should be used if available, which is what I wanted to highlight (i.e. higher performance but potential compatibility issues).

On my WD SN850 I set 4096 bytes per sector before installing Windows, and the command which gave the about output (fsutil fsinfo ntfsinfo x: ) reports this on it:

bpyQgJ4.png


This is instead what it reports on a Crucial MX500 where the "Bytes Per Sector" setting (i.e. logical sector size) cannot be changed from its default 512 bytes size. I also formatted the NTFS partition to 64 kB cluster size, ("Bytes Per Cluster") which shows that the "Bytes Per Sector" property is independent of cluster size:

5bX4SCf.png


While outside my knowledge, this could also be one of the differences between MBR formatting and GPT formatting

No, this is not directly related to MBR/GPT. GPT formatting can still use 512 bytes per sector.
Both the SSDs installed in my system are GPT-formatted, but only the NVMe one has been set (by me) to 4096 bytes per sector.
 
How did you set it, for what file system

I feel like you may be over-riding that compatibility setting by changing that, and not much else
 
How did you set it, for what file system

I set it from a Linux installation on a separate SSD, using the "nvme-cli" tools as described in the links below. Changing this option deletes the contents of the affected SSD and permanently changes the logical sector size presented by the SSD, regardless of the filesystems used afterwards (whether from Windows, Linux or other OS).


Some more general details on the above tools are available on the official NVM Express organization website:

I feel like you may be over-riding that compatibility setting by changing that, and not much else

Yes, it's entirely possible that in the end only compatibility is affected in practice, although this might depend on the specific SSD. I saw small differences on Windows, but they might have been within error margins. Some people have claimed larger differences; corporate websites generally claim that the 4096 byte logical block size performs better, but I have not seen actual performance numbers for this.

It is certainly not something that has been tested in SSD reviews.
 
Last edited:
You can also use the Sabrent sector size tool.


That seems convenient to use for Windows-only environments, and also quite in-topic with the review here. I wonder if it will work also on NVMe SSDs from other vendors which support changing the logical sector size. (EDIT: I couldn't get the program to start at all on my configuration, so I guess not)

The page above does not mention anything about possible performance changes from this setting, only issues pertaining to "data cloning scenarios".
 
Last edited:
You can also use the Sabrent sector size tool.

Ooh, thats convenient
Sigh i'm going to need to format a drive and test this out now


Okay weirdly, that only detects one of the four drives in my system.


The more i read on this, it basically comes down to that setting only being relevant to OS and file systems that arent 4K sector aware
As long as the partitions are aligned correctly (formatted with an updated tool, basically) this should not be an issue whatsoever


Maybe one of you should start a new thread on this, showing before and after results on your drives?
 
Last edited:
@Mussel I made a test about a week ago when I first got it, which showed slight differences between the two settings, but with Windows 11 having consistency issues with NVMe write performance as everybody later found, it could have been due mainly because of that. I didn't test it on Linux with fast filesystems like EXT4, however (which is what the user who claimed 10% faster performance with the SN850 in 4Kn mode was using).

wsrxebg.png
r0QOw07.png
 
@Mussel I made a test about a week ago when I first got it, which showed slight differences between the two settings, but with Windows 11 having consistency issues with NVMe write performance as everybody later found, it could have been due mainly because of that. I didn't test it on Linux with fast filesystems like EXT4, however (which is what the user who claimed 10% faster performance with the SN850 in 4Kn mode was using).

wsrxebg.png
r0QOw07.png
Interesting. A bit better on randoms, good. How did you convert it? If under linux, could it be done on a live image like Parted Magic, since I don’t have (and don’t need) a linux install on another drive?
 
Back
Top