• 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.

Acer MA200 1 TB

W1zzard

Administrator
Staff member
Joined
May 14, 2004
Messages
28,803 (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 Acer MA200 is a highly compact solid-state-drive using the M.2 2230 form factor. This makes it a perfect fit to upgrade the storage on game consoles like the Steam Deck and ROG Ally. Unlike many competitors which use slower QLC, the MA200 is built with TLC NAND flash, which helps boost performance.

Show full review
 
I think the graph gives me a wrong impression. it ends around 4000mb/s. It should end vertically, the same as most other graphs around 8000 mb/s for pcie 4.0 nvme-SSDs

This graph has shown me to pay special attention first to the division units for the x and y axis. Lessons learnt, only pay attention to the total numbers.

write-over-time.png


First NVME i found which is older wiht a graph

write-over-time.png
 
Call me an imbecile...but that looks a lot like the wifi adapters I used to upgrade in laptops....so was the m.2 slot derived from wifi laptop slots/tech? or are they actually m.2 slots? and could one install an SSD there. I'm assuming not, but idk if: A) the chip that checks the slot is just reading what that hardware is capable of, or B) its got a set priority and expectations from the hardware slotted. Thanks!
 
This graph has shown me to pay special attention first to the division units for the x and y axis. Lessons learnt, only pay attention to the total numbers.
Not sure I understand the issue?
 
I think the graph gives me a wrong impression. it ends around 4000mb/s. It should end vertically, the same as most other graphs around 8000 mb/s for pcie 4.0 nvme-SSDs

This graph has shown me to pay special attention first to the division units for the x and y axis. Lessons learnt, only pay attention to the total numbers.

First NVME i found which is older wiht a graph
-------------------------------------------------------------------------------------------------------------------------------------------------------
write-over-time.png

--------------------------------------------------------------------------------------------------------------------------------------------------
write-over-time (1).png

----------------------------------------------------------------------------------------------------------------------------------------------------
Most of the latest NVMe SSDs (Consumers Client SSD) operate in TLC mode when using maximum capacity. The raw capacity of SLC mode is 333GB for 1TB (1024GB). When this capacity is exceeded, it operates in MLC mode, where performance degrades. MLC mode has a raw capacity of about 666GB. When the storage data exceeds 666GB, it switches to TLC mode, where performance degrades dramatically. The raw capacity of TLC SSD is 999GB for 1TB. If you set excessive overprovisioning of about 70%, it operates in SLC mode and has very good performance. TLC mode is not bad if you want to write in read-only mode. However, there should be no write operations and the data must be locked. In some cases, the MLC section is small and it switches directly from SLC mode to TLC. The graph of the 990 Pro is like that, and the MA200 has at least a fairly long MLC mode section.
 
Call me an imbecile...but that looks a lot like the wifi adapters I used to upgrade in laptops....so was the m.2 slot derived from wifi laptop slots/tech? or are they actually m.2 slots? and could one install an SSD there. I'm assuming not, but idk if: A) the chip that checks the slot is just reading what that hardware is capable of, or B) its got a set priority and expectations from the hardware slotted. Thanks!
The M.2 slot is quite universal, it can carry PCIe, SATA, USB and CNVi signals (CNVi is Intel's proprietary stuff for wireless adapters). Its function is partially defined by its "key". For example, the M-key means that the key is at the position "M" on the slot, and the device must have a cutout in its connector at the same position. M-key implies a 4-lane PCIe connection, used in nearly all modern SSDs.
 
Not sure I understand the issue?

I need to pay more attention on the numbers on the y-Axis.

I estimated wrongly the y-axis could be max always around 7000 or 8000mb/s for pcie 4.0. That was a wrong assumption.
the y-axis seems to be just scaled for that particular drive. I'll pay more attention to this in the next drive reviews.

Thanks for always providing us with such drive reviews.

Call me an imbecile...but that looks a lot like the wifi adapters I used to upgrade in laptops....so was the m.2 slot derived from wifi laptop slots/tech? or are they actually m.2 slots? and could one install an SSD there. I'm assuming not, but idk if: A) the chip that checks the slot is just reading what that hardware is capable of, or B) its got a set priority and expectations from the hardware slotted. Thanks!

What post #7 said

You may check Wikipedia. It has a good explanation with pictures about those "keys"
Keys are mechanical notches which prevent insertions. you need the same mechanical dimensions and cut outs which they call "keys".
 
the y-axis seems to be just scaled for that particular drive.
yeah, thats the norm, otherwise you lose too much resolution on the y axis. the bar charts below are scaled in the way you expect with the best setting the scale range
 
I estimated wrongly the y-axis could be max always around 7000 or 8000mb/s for pcie 4.0. That was a wrong assumption.
the y-axis seems to be just scaled for that particular drive. I'll pay more attention to this in the next drive reviews.
Apart from that, you probably noticed that the highest speed in SLC cache test is not the highest speed achievable, because this test writes non-queued data (Q1). At Q8 or so, it would reach around 4.6 GB/s.
 
so was the m.2 slot derived from wifi laptop slots/tech?
M.2 form factor was developed as successor for mSATA and mPCIe of mobile devices needing slim space consumption.
Hence M.2 "card/device" parallel to "host board".
And resulting hogging of motherboard real estate, instead of rising out vertically like expansion cards...

-------------------------------------------------------------------------------------------------------------------------------------------------------
View attachment 398806
--------------------------------------------------------------------------------------------------------------------------------------------------
View attachment 398807
----------------------------------------------------------------------------------------------------------------------------------------------------
Most of the latest NVMe SSDs (Consumers Client SSD) operate in TLC mode when using maximum capacity. The raw capacity of SLC mode is 333GB for 1TB (1024GB). When this capacity is exceeded, it operates in MLC mode, where performance degrades. MLC mode has a raw capacity of about 666GB. When the storage data exceeds 666GB, it switches to TLC mode, where performance degrades dramatically. The raw capacity of TLC SSD is 999GB for 1TB. If you set excessive overprovisioning of about 70%, it operates in SLC mode and has very good performance. TLC mode is not bad if you want to write in read-only mode. However, there should be no write operations and the data must be locked. In some cases, the MLC section is small and it switches directly from SLC mode to TLC. The graph of the 990 Pro is like that, and the MA200 has at least a fairly long MLC mode section.
There's unlikely to be "MLC" mid step.
That would only increase write amplification even further and wear NAND out even faster.
(while adding another layer of complexity to controller's job)

And what you call as TLC mode is drive in "folding" state:
All NAND is full and controller has to read out SLC cached data and erase those cells to have room for writing cached data again in TLC mode, along with new received data.

What happens in between is more variable and depends on controller/firmware.
But that's where genuine empty state TLC writing happens.
If all space isn't treated as SLC cache, in which case drive goes more or less directly from SLC caching to folding.

Here's article in which they do write test with SLC caching completely disabled and drive writing all time in TLC mode:


yeah, thats the norm, otherwise you lose too much resolution on the y axis. the bar charts below are scaled in the way you expect with the best setting the scale range
SATA drive graph would kind of lack accuracy in scale of PCIe drive...
 
I really appreciate these. I would also like to see the dramless sata drives again as alot of the dram sata drives on the market are either too expensive or gone.

The dramless ones range from being insanely bad, to defective controllers that caused permanent damage with certain configs (maxio), and being diamonds in the rough offering a suedo cache (smi).
 
I really appreciate these. I would also like to see the dramless sata drives again as alot of the dram sata drives on the market are either too expensive or gone.

The dramless ones range from being insanely bad, to defective controllers that caused permanent damage with certain configs (maxio), and being diamonds in the rough offering a suedo cache (smi).
WD SA 510 is dramless.

It performs ok (at least the 4tb model does) but write amplification is pretty bad. Fair amount of reports of early deaths (on the lower capacity units). However if you not fussed about losing data, it does have a 5 year warranty.
 
Last edited:
I really appreciate these. I would also like to see the dramless sata drives again as alot of the dram sata drives on the market are either too expensive or gone.

The dramless ones range from being insanely bad, to defective controllers that caused permanent damage with certain configs (maxio), and being diamonds in the rough offering a suedo cache (smi).
I REALLY despise dramless SATA SSDs as main OS drives. Most of them slow down to a crawl after a year or two of use, and the only way to speed’em up is to secure-erase them (I use Parted Magic). Then they’ll go back to their original speed.
Or at least that’s what I found with many customers who buy the cheapest, crappiest SSDs (usually Kingston A400s or no-name ones from Amazon)…
A permanent fix I do is to clone those drives to MX500s or 870 Evos, which don’t suffer from those problems.

Are there any 2230 SSDs with a DRAM cache though?
Most likely zero, as there isn’t enough real estate on a 22x30mm PCB! Unless double sided, but I haven’t seen any.
 
Back
Top