HOT ! Various 1TB NVMe with coveted E12 Controller $135 aprox retail

Status
Not open for further replies.

Nenu

[H]ardened
Joined
Apr 28, 2007
Messages
19,347
A lot of people are reporting the drives are "30-50% slower" but I dispute this finding. The E12S has the same specs as the E12, the NAND is the same or superior (96L), but there is less DRAM (which would only impact heavy workloads/benchmarks). I suspect there might be a change in the SLC cache design that might result in poor sequential performance in benchmarks but I have no evidence/proof of this yet. As an example, compare the SN750 to the SX8200 Pro: the SN750's small, static SLC cache runs out rapidly and drops to 1500 MB/s almost immediately while the SX8200 Pro's large, dynamic SLC cache remains twice as fast for quite a while. This does not mean the SN750 is inferior whatsoever, just a design decision, so my theory so far is that the new drives might forego SLC for some reason. Still investigating.
If mine is a 512MB cache drive I would dispute it for general use as well.
I also didnt notice a slowdown copying my games onto this drive, which I would have expected.

It is approximately twice as fast as my old 840 Pro for game load times, it surprised me.
Windows load time is around 33% faster, from when the OS load starts to when the desktop appears.
The 840 Pro had been defragged about a month earlier which made it a lot snappier, there was little wrong and it benchmarked well.
I have the CrystalDiskMark figures if you are interested.

My issue is what I would repurpose this drive as once I have finished with it.

Perhaps I should copy a large file say 50GB and let you know the time windows reports it has finished?
I realise Windows also caches the copy so there will be a small error but with such a large file it should only be a few % out.
(I am on Windows 7)
Let me know.
...
Although thinking about it, a smaller file size copy time will let us differentiate between the 512MB and 1GB SLC cache sizes better, barring the windows cache error.
Can you suggest a tool to do the copy that will show when the transfer rate drops?
 
Last edited:

Maxx

[H]ard|Gawd
Joined
Mar 31, 2003
Messages
1,644
If mine is a 512MB cache drive I would dispute it for general use as well.
I also didnt notice a slowdown copying my games onto this drive, which I would have expected.

It is approximately twice as fast as my old 840 Pro for game load times, it surprised me.
Windows load time is around 33% faster, from when the OS load starts to when the desktop appears.
The 840 Pro had been defragged about a month earlier which made it a lot snappier, there was little wrong and it benchmarked well.
I have the CrystalDiskMark figures if you are interested.

My issue is what I would repurpose this drive as once I have finished with it.

Perhaps I should copy a large file say 50GB and let you know the time windows reports it has finished?
I realise Windows also caches the copy so there will be a small error but with such a large file it should only be a few % out.
(I am on Windows 7)
Let me know.
The Intel 660p only has 256MB of DRAM even on its 2TB SKU. It makes no difference on that drive. Of course, that's a four-channel controller with QLC, but the point stands: you're only going to exceed 512MB of DRAM cache if you're really hitting the drive hard. The reason this concerns me, though, is that one area the E12 has over the SM2262/EN drives is such workloads (being a "budget 970 EVO" of sorts); by halving the RAM, that pushes the drives instead firmly in the "desktop performance" (everyday NVMe) category instead. Which is fine since it's priced and has been priced that way for a while now, but nevertheless worth noting.

Since people are calling it slower it must be the SLC cache design. Honestly, it's possible the flash is 96-layer (not 64-layer) which if anything would be a bit faster. The easiest way to test SLC would be a prolonged/sustained, sequential write, at speed. That last part is crucial, a file copy would have to come from another NVMe drive that can read fast enough.

Benchmarks are more squirrelly. If you see my recent Reddit post on the Addlink S22 review, the reviewer clearly overestimates the ability of the drive to cache, likely because he is not benchmarking properly. The old tool HDTach might work (worth a shot). HDTune only works if it uses specific settings but also worth checking. The normal E12 drives have only a 30GB dynamic SLC cache, quite small by consumer standards, so if there's a noticeable drop it must be static-only (which has to be in the overprovisioned/reserved space, and thus would be 12GB at most and probably smaller). This would actually give the drive more consistent performance, which would have two benefits - one, less hard on the controller (if it has changed at all) and two, less demanding on the DRAM.
 

Nenu

[H]ardened
Joined
Apr 28, 2007
Messages
19,347
Benchmarks are more squirrelly. If you see my recent Reddit post on the Addlink S22 review, the reviewer clearly overestimates the ability of the drive to cache, likely because he is not benchmarking properly. The old tool HDTach might work (worth a shot). HDTune only works if it uses specific settings but also worth checking. The normal E12 drives have only a 30GB dynamic SLC cache, quite small by consumer standards, so if there's a noticeable drop it must be static-only (which has to be in the overprovisioned/reserved space, and thus would be 12GB at most and probably smaller). This would actually give the drive more consistent performance, which would have two benefits - one, less hard on the controller (if it has changed at all) and two, less demanding on the DRAM.
Perhaps copy to/from the same drive?

I have version 3.040 of HDTach which at best uses 32MB zones and seems to max at 2GB/s, is this useful?
I dont have HDtune Pro, is the free version useful?
 

Maxx

[H]ard|Gawd
Joined
Mar 31, 2003
Messages
1,644
A copy to/from would technically work since NVMe is full duplex, although I'm not sure how accurate it would be. Ideally you'd have one giant file/ISO to test rather than lots of small files. You're right, HDTach is limited, although it may still be useful as seen here; it would just take longer for the drop to occur. I'd have to extrapolate SLC cache size from the results which is a bit tricky but not impossible. Not sure on the limitations of HDTune's free version actually.
 

AnIgnorantPerson

Limp Gawd
Joined
Jan 10, 2019
Messages
207
Looks like 4TB versions using the Phison E12 are slowly starting to appear: https://www.amazon.com/dp/B07ZQSDQDB/
bought the 1TB premium but what are prospects of newer and bigger ones? i recall 2TB was supposed to come but am out of loop. can you give me low down on controller and nand and whats expected?

thanks!
Got my 512GB Inland in the mail yesterday. I should be kicking myself for waiting this long to get one. I've never seen Windows load up so fast.

Question, does Windows 10 1903 handle these drives better than past Windows version? I know in the Windows 7 days that SSD drives needed some tweaking to make sure they lasted a long time.
i think you mean turning off virtual memory. you still have to do that in win 10. i burned up like 3 TB in 1-2 months because of that. i could look at HDSentinal to be exact later
 

Maxx

[H]ard|Gawd
Joined
Mar 31, 2003
Messages
1,644
Looks like 4TB versions using the Phison E12 are slowly starting to appear: https://www.amazon.com/dp/B07ZQSDQDB/
Yes, and you can tell by the design that it's the "new" E12S layout with four NAND packages per side. It's possible they moved to 512Gb/die, 96L TLC, but they could still use 256Gb/die 64L TLC but it would have a higher performance hit than even the current 2TB (~10%). Further the DRAM shown in the picture is https://www.nanya.com/en/Product/4115/NT5CC128M16JR-EK which is only 256MB per side. I think it's more likely it'll be 1GB per side for 2GB total (1/2 normal ratio) based on the E12S designs I've seen. So in total, a move towards making these drives cheaper but not necessarily better.
 
  • Like
Reactions: Luckz
like this

Snowknight26

Supreme [H]ardness
Joined
May 8, 2005
Messages
4,256
Yes, and you can tell by the design that it's the "new" E12S layout with four NAND packages per side. It's possible they moved to 512Gb/die, 96L TLC, but they could still use 256Gb/die 64L TLC but it would have a higher performance hit than even the current 2TB (~10%). Further the DRAM shown in the picture is https://www.nanya.com/en/Product/4115/NT5CC128M16JR-EK which is only 256MB per side. I think it's more likely it'll be 1GB per side for 2GB total (1/2 normal ratio) based on the E12S designs I've seen. So in total, a move towards making these drives cheaper but not necessarily better.
The irony is that the E12 supports up to 8TB..
 

Maxx

[H]ard|Gawd
Joined
Mar 31, 2003
Messages
1,644
The irony is that the E12 supports up to 8TB..
Technically. Phison only lists it up to 2TB in the M.2 factor; it requires U.2 for 8TB. The E12 reference design has a total of four NAND packages which at 16DQ (sixteen dies per package) is only 2TB with 256Gb/die TLC. It would be possible to hit 4TB with 512Gb/die NAND, which is possible with both 64L and 96L - the E16 (basically an E12) uses the latter at 2TB to avoid oversaturating the controller (32 CE), SN750 & 970 EVO drives use the former at 2TB to stay single-sided while the 970 EVO Plus uses the latter as well for the same reason. None of them have a 4TB SKU, the E12/E16 would still oversaturate with 512Gb/die NAND. So you'd have to jump to QLC (768Gb/die or 1.33Tb/die for 64L/96L Toshiba, 1Tb/die for 64L/96L Intel) despite Phison listing just "3D TLC" on their spec page. Basically, the controller was not meant for more than 2TB.

Further difficulty exists with the DRAM. Maintaining 1GB:1TB is costly after a certain point, specifically the SN750 & 970 EVO/EVO Plus drives need a 16Gb (2GB) module to stay single-sided which is not cost-effective. Likewise, a 4TB E12 drive would need two such modules. This is why I believe they've halved the DRAM as 8Gb/1GB is much easier. So 4TB+ here is also difficult with the stock E12 layout.

tl;dr they likely cut down the DRAM and increased the number of NAND packages to save money and allow higher capacities on the cheap, plus likely redid the E12 in a smaller process package ("E12S") for the same reason; the E12 is, after all, 28nm. Nevertheless I question its performance at more than 2TB even if it does have 96L & 512Gb flash which it might not, although the NAND code does point towards 96L anyway (likely chosen to unify production with their 4.0 drives, which again do not use 512Gb except for the 2TB SKUs). I would be surprised to see more than 4TB either way.
 
Last edited:
  • Like
Reactions: Luckz
like this

Lateralus

More [H]uman than Human
Joined
Aug 7, 2004
Messages
15,951
Thanks for all of the awesome info on these, Maxx. Let me know as you continue to research if you can verify a list of single-sided 1TB E12S part numbers (I'm particularly interested in the Sabrent Rocket, addlink S70, or Corsair MP510 but actually getting a single-sided one in the event that two versions exist might prove to be a challenge -- mixed stock and retailers might be slow to update product listings, if they do at all).

Also - are you aware of any single-sided 1TB SM2262EN drives in existence? TIA.
 

1911Shootist

[H]ard|Gawd
Joined
Nov 15, 2012
Messages
1,782
Picked up one of these for $99 at the local MC for use in my Razer Blade Stealth ('17) ultrabook. It fits, but the laptop was meant for a single sided drive, so I'm flexing the drives PCB a bit. So far it's working great, but I'm wondering if I shouldn't swap it for a single sided drive...
 

AnIgnorantPerson

Limp Gawd
Joined
Jan 10, 2019
Messages
207
Picked up one of these for $99 at the local MC for use in my Razer Blade Stealth ('17) ultrabook. It fits, but the laptop was meant for a single sided drive, so I'm flexing the drives PCB a bit. So far it's working great, but I'm wondering if I shouldn't swap it for a single sided drive...
yea thats playing with fire in my opinion. Think about the pressure it can cause when something rests against it like sandwiched in a backpack or if your resting your hand on that spot.

Is the chance it could transfer any static or electricity and short that way? I honestly dont know but i am always paranoid about that with those things touching
 

1911Shootist

[H]ard|Gawd
Joined
Nov 15, 2012
Messages
1,782
yea thats playing with fire in my opinion. Think about the pressure it can cause when something rests against it like sandwiched in a backpack or if your resting your hand on that spot.

Is the chance it could transfer any static or electricity and short that way? I honestly dont know but i am always paranoid about that with those things touching
Yeah, I think I should swap it out for single sided... Now the question is Inland Pro ($95 - Gen3 x2), or Samsung 970 Evo ($170 - Gen3 x4)...
 

Lateralus

More [H]uman than Human
Joined
Aug 7, 2004
Messages
15,951
New versions of the Sabrent Rocket 1TB are single sided. This is Gen3 x4.
I have one.
Did you get yours from Amazon? I can't remember if you are in the US or not. I just worry about getting an older, double-sided version but I want that drive!
 

Shonk

n00b
Joined
Sep 30, 2008
Messages
5
here's the spec's (it has some custom firmware on it btw..)

Summary
Controller : PS5012-E12
Nand = Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
DRAM Size,MB : 512


Code:
v0.2a
OS: 10.0 build 18362
Drive: 3(NVME)
Scsi    : 1
Model   : Sabrent                           
Fw      : RKT303.1
Size    : 976762 MB
LBA Size: 512
Read_System_Info_5008 error: 4294967295
Firmware lock supported [02 01]
Drive unlocked
P/N     : 511-190916140
Bank00: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank01: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank02: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank03: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank04: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank05: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank06: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank07: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank08: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank09: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank10: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank11: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank12: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank13: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank14: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Bank15: 0x2c,0xc4,0x18,0x32,0xa2,0x0,0x0,0x0 - Micron 96L(B27A) TLC 512Gb/CE 512Gb/die
Controller    : PS5012-E12
CPU Clk       : 666
Flash CE      : 16
Flash Channel : 8
Interleave    : 2
Flash CE Mask : [++++++++ ++++++++ -------- --------]
FlashR Clk,MT : 666
FlashW Clk,MT : 666
Block per CE  : 944
Bit Per Cell  : 3(TLC)
DRAM Size,MB  : 512
DRAM Clock,MHz: 1600
DRAM Type     : DDR3
PMIC Type     : PS6102

Defects Early Read Prog Erase
Bank00:   14    0    0    0
Bank01:   13    0    0    0
Bank02:   11    0    0    0
Bank03:   14    0    0    0
Bank04:   10    0    0    0
Bank05:   12    0    0    0
Bank06:   14    0    0    0
Bank07:   16    0    0    0
Bank08:   13    0    0    0
Bank09:   10    0    0    0
Bank10:   12    0    0    0
Bank11:   14    0    0    0
Bank12:   13    0    0    0
Bank13:   21    0    0    0
Bank14:   15    0    0    0
Bank15:   13    0    0    0
Total :  215    0    0    0

Defects Early Read Prog Erase
Ce00Pl0:    5    0    0    0
Ce00Pl1:    2    0    0    0
Ce00Pl2:    3    0    0    0
Ce00Pl3:    4    0    0    0
Ce01Pl0:    5    0    0    0
Ce01Pl1:    2    0    0    0
Ce01Pl2:    3    0    0    0
Ce01Pl3:    3    0    0    0
Ce02Pl0:    4    0    0    0
Ce02Pl1:    3    0    0    0
Ce02Pl2:    2    0    0    0
Ce02Pl3:    2    0    0    0
Ce03Pl0:    5    0    0    0
Ce03Pl1:    3    0    0    0
Ce03Pl2:    4    0    0    0
Ce03Pl3:    2    0    0    0
Ce04Pl0:    4    0    0    0
Ce04Pl1:    2    0    0    0
Ce04Pl2:    2    0    0    0
Ce04Pl3:    2    0    0    0
Ce05Pl0:    4    0    0    0
Ce05Pl1:    3    0    0    0
Ce05Pl2:    3    0    0    0
Ce05Pl3:    2    0    0    0
Ce06Pl0:    4    0    0    0
Ce06Pl1:    3    0    0    0
Ce06Pl2:    3    0    0    0
Ce06Pl3:    4    0    0    0
Ce07Pl0:    6    0    0    0
Ce07Pl1:    3    0    0    0
Ce07Pl2:    2    0    0    0
Ce07Pl3:    5    0    0    0
Ce08Pl0:    5    0    0    0
Ce08Pl1:    2    0    0    0
Ce08Pl2:    3    0    0    0
Ce08Pl3:    3    0    0    0
Ce09Pl0:    4    0    0    0
Ce09Pl1:    2    0    0    0
Ce09Pl2:    2    0    0    0
Ce09Pl3:    2    0    0    0
Ce10Pl0:    4    0    0    0
Ce10Pl1:    3    0    0    0
Ce10Pl2:    3    0    0    0
Ce10Pl3:    2    0    0    0
Ce11Pl0:    5    0    0    0
Ce11Pl1:    3    0    0    0
Ce11Pl2:    3    0    0    0
Ce11Pl3:    3    0    0    0
Ce12Pl0:    5    0    0    0
Ce12Pl1:    3    0    0    0
Ce12Pl2:    2    0    0    0
Ce12Pl3:    3    0    0    0
Ce13Pl0:    4    0    0    0
Ce13Pl1:   10    0    0    0
Ce13Pl2:    3    0    0    0
Ce13Pl3:    4    0    0    0
Ce14Pl0:    4    0    0    0
Ce14Pl1:    2    0    0    0
Ce14Pl2:    6    0    0    0
Ce14Pl3:    3    0    0    0
Ce15Pl0:    5    0    0    0
Ce15Pl1:    2    0    0    0
Ce15Pl2:    3    0    0    0
Ce15Pl3:    3    0    0    0
Total  :  215    0    0    0
Code:
-------- NVME SMART --------
0 Critical Warning           : 0
1 Composite Temperature           : 26
2 Available Spare           : 100
3 Available Spare Threshold       : 5
4 Percentage Used           : 0
5 Data Units Read           : 0
6 Data Units Written           : 0
7 Host Read Commands           : 12
8 Host Write Commands           : 0
9 Controller Busy Time           : 0
10 Power Cycles               : 1
11 Power On Hours           : 0
12 Unsafe Shutdowns           : 1
13 Media and Data Integrity Errors   : 0
14 Number of Error Information Log Entries: 2
15 Warning Composite Temperature Time   : 0
16 Critical Composite Temperature Time   : 0
-------- Temperature --------
Max Temp All Time   : 59
Min Temp All Time   : 27
Max Temp Last Power : 42
Min Temp Last Power : 26
Temp Linkup         : 26
Temperature         : 26
Internal Temperature: 37
Flash temp. per die : 19 20 19 18 19 19 18 18 20 20 19 18 19 19 18 18
Code:
-------- SYSTEM STATUS LOG --------
disk init fail               : 0
disk hw status               : 0
write protect               : 0
ftl err path               : 0
hardware initial error           : 0
fw code update count           : 0
security state               : 33
gpio                   : 0
power cycle count           : 0
abnormal power cycle count       : 1
fw internal power cycle count       : 0
power on time               : 821(0h)
flash ip reset count           : 1
host e3d err count           : 0
flash e3d err count           : 0
ddr ecc err count           : 0
dbuf ecc err count           : 0
gc table trigger count           : 0
d1 gc data trigger count       : 0
d2 d3 gc data trigger count       : 0
dynamic d1 gc data trigger count   : 0
d1 gc block rate of data       : 100
d2 d3 gc block rate of data       : 100
dynamic d1 gc block rate of data   : 100
vendor aes set key status       : 0
axi err slave               : 0
axi err zone               : 0
d1 wear leveling check count       : 64
d1 wear leveling trigger count       : 128
d1 wear leveling block rate       : 100
d2 d3 wear leveling check count       : 32
d2 d3 wear leveling trigger count   : 64
d2 d3 wear leveling block rate       : 100
vuc protect mode           : 2
vuc protect state           : 3
Code:
-------- FLASH STATUS LOG --------
max erase count d1           : 0
max erase count d2 d3           : 0
average erase count d1           : 0
average erase count d2 d3       : 0
min erase count d1           : 0
min erase count d2 d3           : 0
total flash erase count d1       : 0
total flash erase count d2 d3       : 0
total flash program count d1       : 0
total flash program count d2 d3       : 116
total flash read count           : 30
total flash write count           : 3712
read flash unc retry ok count d1   : 0
read flash unc retry ok count d2 d3   : 0
read flash unc retry fail count d1   : 0
read flash unc retry fail count d2 d3   : 0
raid ecc recovery ok count d1       : 0
raid ecc recovery ok count d2 d3   : 0
raid ecc recovery fail count d1       : 0
raid ecc recovery fail count d2 d3   : 0
logical good block count d1       : 0
logical good block count d2 d3       : 226
total early bad physical block count   : 215
total later bad physical block count   : 0
total read fail block count d1       : 0
total read fail block count d2 d3   : 0
total program fail block count d1   : 0
total program fail block count d2 d3   : 0
total erase fail block count d1       : 0
total erase fail block count d2 d3   : 0
raid ecc entry               : 0
read disturb count           : 0
flash max pecycle           : 2000
revoked vb number           : 20
acceptable revoked vb number       : 20
Code:
-------- DRIVE INFO --------
maximum pca value           : 250051158
vuc buffer base               : 1196859392
vuc buffer size               : 131072
host read count               : 12
host write count           : 0
background read count           : 0
 
Last edited:

Maxx

[H]ard|Gawd
Joined
Mar 31, 2003
Messages
1,644
Thanks, Shonk. I suspected 96L NAND and it seems like confirmation there. I also figured they might use 512Gb/die, although I didn't expect it at lower capacities. Also I figured the 1/2 DRAM ratio which you confirm. It states Micron TLC which is a bit odd as physically it looks like Toshiba, but I'll confirm that later unless you want to take a look. Reason being I saw "G63" for the flash code in one picture and that's 96-layer generation Toshiba (BiCS4). I have seen B27A before (over a year ago!) but didn't ever see it coupled with a Phison controller...but it is 512Gb/die.
 
Last edited:
  • Like
Reactions: Luckz
like this

Maxx

[H]ard|Gawd
Joined
Mar 31, 2003
Messages
1,644
Thanks for all of the awesome info on these, Maxx. Let me know as you continue to research if you can verify a list of single-sided 1TB E12S part numbers (I'm particularly interested in the Sabrent Rocket, addlink S70, or Corsair MP510 but actually getting a single-sided one in the event that two versions exist might prove to be a challenge -- mixed stock and retailers might be slow to update product listings, if they do at all).

Also - are you aware of any single-sided 1TB SM2262EN drives in existence? TIA.
It's my suspicion that most or all E12 drives will make these changes moving forward, just a matter of when. It makes sense as it fills in the gap before we get true blue 4.0 drives in 2H 2020. It's cheaper and just as fast. However if you follow me or my sub on Reddit, I will post information/confirmations as I get them.

There are no single-sided SM2262EN drives at ANY capacity, same as with the SM2262 drives. Except the Intel 760p! This could change of course but if you're looking for that I'd suggest the Kingston A2000 (SM2263) instead. Excellent drive if you can find it at MSRP. My suspicion is SMI will wait for their new controllers next year, but who knows? I've already seen them moving to 96L flash so maybe a layout change to match the E12S is in the works. (none of my sources could confirm the E12S, even now)
 
Last edited:

Lateralus

More [H]uman than Human
Joined
Aug 7, 2004
Messages
15,951
Thanks, friend.

One of you UK gents feel like shipping me a drive? I'd love to upgrade this Precision with a 1TB Rocket. I'm not really feeling like playing the return game with Amazon US, as painless as it might be. Not sure how long to wait until the old double sided stock is flushed from inventory.
 

Maxx

[H]ard|Gawd
Joined
Mar 31, 2003
Messages
1,644
added the smart info and flash log
Much appreciated. It does seem to be accurate, definitely some changes were made. I wouldn't consider it inferior though. Well, less DRAM, but the SLC cache might have changed to compensate.
 

Shonk

n00b
Joined
Sep 30, 2008
Messages
5
There's some photo's of it here
btw in older photo's the sabrent sticker was clearly just plastic
its a thin strip of copper now that has been printed ontop of
see picture7


Ram module image3
Kingston D2516ECMDXGJD
4Gb 96 ball FBGA DDR3L 7.5x13.5x1.2
256Mx16 1866 Mbps 1.35V* 0°C ~ +95°C


https://postimg.cc/gallery/2i8ixxyec/
 
Last edited:

Lateralus

More [H]uman than Human
Joined
Aug 7, 2004
Messages
15,951
That's the one I need! Guess I'll wait. Or maybe I can get my Irish buddy to post me one.
 

Maxx

[H]ard|Gawd
Joined
Mar 31, 2003
Messages
1,644
There's some photo's of it here
btw in older photo's the sabrent sticker was clearly just plastic
its a thin strip of copper now that has been printed ontop of
Probably an aluminum heatspreader. I'd keep an eye on the temps there (your above readings show 59C max lifetime so far). Phison previously had a design like that with the E7, Corsair MP500, which was prone to overheating. Check it after CDM or some such.
 

Shonk

n00b
Joined
Sep 30, 2008
Messages
5
Its copper take a good look close up at image7 its copper colour
its so thin that it wouldnt make a huge diff anyway but atleast it will transfer through to my heatsink
that i have fitted on it

It hasnt gone over 27 since i owned it (havnt done any nand reads or writes to it still unpartitioned)
when it reached 59 that was at the factory
 

Nenu

[H]ardened
Joined
Apr 28, 2007
Messages
19,347
Its copper take a good look close up at image7 its copper colour
its so thin that it wouldnt make a huge diff anyway but atleast it will transfer through to my heatsink
that i have fitted on it

It hasnt gone over 27 since i owned it (havnt done any nand reads or writes to it still unpartitioned)
when it reached 59 that was at the factory
Thats too cold for NAND, it can reduce its lifespan.
See post 1317.

Its a tricky drive to cool properly because the controller and NAND are linked with the plate, most likely to keep the NAND warm enough.
You really want to cool only the controller unless in a very hot environment, assuming the controller does get too hot a lot of the time.
 
Last edited:

1911Shootist

[H]ard|Gawd
Joined
Nov 15, 2012
Messages
1,782
I wound up switching to the 512GB version to get a single sided drive. Pics on MicroCenter show they are single sided, but out of the ~20 my local MC had, only one was single sided. Glad I could go in person and find one.

I'll keep my eyes peeled for SS 1TB versions and maybe hand this one down to my dad later (he's still on a ~8 year old 120GB Kingston 2.5" SSD boot drive).
 

Shonk

n00b
Joined
Sep 30, 2008
Messages
5
Thats too cold for NAND, it can reduce its lifespan.
See post 1317.

Its a tricky drive to cool properly because the controller and NAND are linked with the plate, most likely to keep the NAND warm enough.
You really want to cool only the controller unless in a very hot environment, assuming the controller does get too hot a lot of the time.

They will be warm dont worry about that

The Rocket is sandwiched between an i9 9900KS (upto 230w Heat output) and a Sapphire RX Vega 64 Nitro + with a 1682 Core and a 50% powerlimit (Around 400W heat output)

and the Toshiba XG3's are sitting right infront + Below of the Vega 64
 

Lateralus

More [H]uman than Human
Joined
Aug 7, 2004
Messages
15,951
I wound up switching to the 512GB version to get a single sided drive. Pics on MicroCenter show they are single sided, but out of the ~20 my local MC had, only one was single sided. Glad I could go in person and find one.

I'll keep my eyes peeled for SS 1TB versions and maybe hand this one down to my dad later (he's still on a ~8 year old 120GB Kingston 2.5" SSD boot drive).
If you find any of the SS 1TB versions, please let me know if you'd be willing to ship one to me. My buddy in Ireland was willing to post me one from Amazon UK but after the £ to $ conversion and shipping, it was going to be within spitting distance of a Samsung 970 EVO. I'm trying to get performance on a budget here! :smuggrin:
 

radeon962

[H]ard|Gawd
Joined
Mar 13, 2008
Messages
1,238
Tried the 512gb version in my Intel Nuc8i3bek and the drive is not recognized regardless of what I did. Latest BIOS, reseated the drive, etc. have a Crucial MX500 m2 ordered so I’ll try that as NVMe drive seem to be hit or miss with the NUC.
 

53oval

n00b
Joined
Jun 21, 2008
Messages
26
I installed a Inland 2TB in my new HP Pavillion laptop with the HM370 chipset. I was worried that the chipset might be gimped compared to a Desktop chipset but I cannot complain about the numbers I get.
 

Attachments

  • Like
Reactions: xx0xx
like this

Monstieur

Limp Gawd
Joined
Jun 10, 2011
Messages
263
My Inland Premium 2TB drives came with ECFM22.4 out of the box. I don't want to risk accidentally downgrading them to ECFM12.3.
 

tangoseal

[H]F Junkie
Joined
Dec 18, 2010
Messages
8,834
I havent been on this thread in a hot minute. Did they change the drive or something? Different silicon? Controller etc.?

Sorry way too many pages to read.

I have the original model that is blazing fast. I was gonna get another but not if it was neutered. Sorry if I'm not making sense
Just a few replies I did read, I may have mis- read/understood.
 
Last edited:

Maxx

[H]ard|Gawd
Joined
Mar 31, 2003
Messages
1,644
Same controller, just smaller. 1/2 the DRAM. Newer NAND (96-layer). Now single-sided at 1TB. New 4TB SKU. SLC cache changes unknown.
 
  • Like
Reactions: Luckz
like this

Dark12

[H]ard|Gawd
Joined
Dec 13, 2003
Messages
1,683
Same controller, just smaller. 1/2 the DRAM. Newer NAND (96-layer). Now single-sided at 1TB. New 4TB SKU. SLC cache changes unknown.
Hmm I just picked up the 1TB Saturday and it was double sided. I guess I got the old batch.
 

xx0xx

Gawd
Joined
Oct 20, 2005
Messages
657
The pressure finally got to me. I gave in and got the 512GB version and an external enclosure.
I won't get absolutely full speed but it's going to become my iMac boot drive and is going to have much better 4K QD1 I/O performance.

My current 256GB SATA-based SSD (MX100) struggles a bit on writes, especially on small files. I feel like it's noticeable in the OS now.
Letting it idle and garbage collect helped, but I figure why not upgrade both protocol and capacity, especially at this price.

The MX100 is still in great condition though, and it has lasted me over 5 years now. It will become a drive for a USFF system probably.

Edit: never fails, every single time I order Amazon shipments it rains. I should become a meteorologist
 
Last edited:
Status
Not open for further replies.
Top