TL;DR
Play games on Amazon Luna for your next adventure, included with Prime
- A rotating selection of games, no download needed
- Play on TV, laptop or phone
- Fast, free delivery for your gear, too
An XDA Developers writer saw a 2TB XPG GAMMIX S70 BLADE drop from about 6,000 MB/s to 1,000–3,000 MB/s after 19 seconds of copying a 235GB file. The reported slowdown occurred as the drive’s dynamic SLC cache filled; the exact performance and timing will vary by SSD, free space and workload.
An XDA Developers writer recorded a sharp write-speed drop while copying a 235GB file on a 2TB XPG GAMMIX S70 BLADE SSD: throughput started around 6,000 MB/s and fell to roughly 1,000–3,000 MB/s after 19 seconds. The report links the change to the drive filling its fast SLC write cache, showing why a headline sequential-write specification may not describe performance throughout a long transfer.
The drive in the test is a PCIe Gen4 NVMe SSD rated at 6,800 MB/s write speed. The writer said it used TLC NAND and a DRAM cache, and had about 24% of its 2TB capacity free. During a sustained copy of a media-server file between partitions, the transfer began at about 6,000 MB/s before dropping. The report says the drop became apparent after 19 seconds and after roughly 117GB had been written to the fast cache.
The article attributes the slowdown to the dynamic SLC cache, a portion of NAND that the drive temporarily operates in single-bit mode to handle writes quickly. Once that cache fills, incoming data is written largely at the drive’s native TLC or QLC pace while the controller folds cached data into its normal storage. The report describes throughput after the drop as fluctuating between 1,000 and 3,000 MB/s, rather than remaining at the advertised peak.
The article also describes a way to observe the behavior with HD Tune Pro’s File Benchmark, using a sufficiently large test file and a Random or Mixed data pattern. It warns that a Zero pattern can produce misleading results and advises against using the program’s main write benchmark, which could erase existing data. The test writes data to the SSD and consumes some endurance, so the writer recommends running it sparingly.
Storage decoded · Sustained write performance
Why Did My SSD Hit a Speed Cliff After 19 Seconds?
One 235GB copy exposed the gap between a headline peak and a drive’s long-write pace. Here’s what the XDA Developers test observed—and what it can tell you.
A fast start. Then a sudden change in pace.
The reported transfer began near the drive’s peak, then settled into a lower, fluctuating range after its fast write area filled.
Conceptual graphic based on the reported speed ranges; it is not a raw benchmark trace. Exact speed and timing vary with drive state and workload.
The SLC cache is a fast lane with a changing size.
Temporary single-bit operation helps consumer TLC or QLC drives accept data quickly. Once that space fills, the native NAND pace becomes more visible.
Fast temporary writes
SLC stores one bit per cell. A drive can temporarily use part of its NAND this way to handle incoming writes quickly.
Denser native storage
TLC stores three bits per cell and QLC stores four. Greater density supports capacity and cost, with lower sustained write pace after cache exhaustion.
Free space affects the buffer
A dynamic SLC cache can borrow available NAND. A fuller drive may have less room for this fast area. It is separate from dedicated DRAM used for data mapping.
What happens when the fast area fills?
The controller keeps moving data as the workload shifts from cache-assisted writes to the drive’s native NAND behavior.
Data arrives
Incoming file data is accepted at high speed while the fast cache has room.
Cache fills
The temporary SLC area reaches capacity; the reported test saw a drop around 117GB.
Native pace shows
Writes proceed largely at TLC or QLC speed while the controller folds cached data into normal storage.
“This fast lane isn’t actually a fixed size.”
XDA Developers writerPeak speed can need an asterisk.
Long writes reveal moreCache behavior is part of the design.
Not, by itself, a faultWho is most likely to notice?
The impact depends on the kind of work you do and how long each uninterrupted write lasts.
Large projects, backups, datasets
Long media transfers and other large writes can outlast a fast cache, making sustained throughput more relevant than a brief peak.
Gaming is often read-heavy
Typical gameplay primarily reads from storage. Installs and copies write data, but they do not always behave like one uninterrupted transfer of this size.
Compare long-write reviews
Check sustained-write tests for the exact model and capacity. Product-page maximums alone do not show how long peak speed lasts.
One drive, one workload, useful limits.
The test illustrates cache behavior on a particular drive state. It does not establish a universal threshold or prove a defect.
What the report can show
An XDA Developers writer recorded a sharp drop during a sustained 235GB transfer on one 2TB XPG GAMMIX S70 BLADE, with about 24% of capacity free.
The article links the change to a dynamic SLC cache and describes post-drop speeds fluctuating around 1,000–3,000 MB/s.
What remains uncertain
- Not a universal timer: 19 seconds is not a fixed limit for all SSDs or every use of this model.
- Not a guaranteed cache size: The reported ~117GB behavior depends on drive capacity and state.
- Not a broad comparison: No controlled multi-drive or repeated-trial results were provided.
- Conditions can vary: Temperature, data pattern, system conditions, free space, and workload can affect results.
How to inspect sustained writes on your own SSD.
Use a large enough file to pass the cache, and treat a write benchmark as a real workload on the drive.
Choose a file test
The writer recommends HD Tune Pro’s File Benchmark for observing a throughput graph.
Use Random or Mixed
Choose a sufficiently large test file and a Random or Mixed data pattern; zeros can give misleading results.
Watch for the transition
Look for a sustained fast section followed by a lower-speed region as the cache fills.
Run sparingly
Make sure the partition has room and compare independent long-write reviews for purchase decisions.
Avoid the program’s main write benchmark, which could erase existing data. Do not repeat a write-heavy test unnecessarily.
Quick answers for SSD owners.
Use the reported numbers as one example of cache behavior, not as a pass-or-fail rule.
Why did the speed drop?
The report attributes it to the dynamic SLC write cache filling. Afterward, writes largely proceed at native NAND pace while cached data is processed.
Does 19 seconds mean my drive is faulty?
No. A speed drop alone does not establish a fault; the article describes cache exhaustion as expected behavior. Timing varies by drive and workload.
Can a fuller SSD sustain less?
It may have less free NAND available for a dynamic cache. The source gives no universal formula for how much free space changes performance.
How should I compare drives?
Look for independent sustained-write tests on the same model and capacity, alongside the advertised maximum speed.
Why Long Writes Miss Peak Speeds
The result matters most to people who regularly write large files in one sustained operation, such as media projects, backups or large data sets. A drive can reach high speeds for a shorter burst yet deliver lower throughput once its fast cache is exhausted. Product-page figures alone do not show how long the drive can sustain peak performance, and the XDA test illustrates that difference on one particular model and workload.
It is not evidence that the SSD is defective. The article says the drop is a consequence of how common TLC and QLC consumer drives balance cost and write performance. Buyers comparing drives for sustained work may need to look beyond the advertised maximum and check reviews that measure long writes. For gaming, the report says the finding is less likely to affect typical use because games primarily read data; a game installation also does not necessarily involve one uninterrupted write of the same size.
2TB PCIe Gen4 NVMe SSD sustained write speed TLC DRAM
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
How the Cache Creates a Speed Cliff
NAND flash stores data in cells, and the number of bits held in each cell affects the number of voltage states the drive must distinguish. SLC stores one bit per cell, while TLC stores three and QLC stores four. SLC can accept writes quickly, but consumer SSDs commonly use denser TLC or QLC NAND to provide more capacity at lower cost.
To speed up writes, a drive can temporarily use part of its NAND in SLC mode. This is separate from a dedicated DRAM cache, which helps manage the drive’s data mapping. The source describes the tested drive’s cache as dynamic: its available size depends on free NAND capacity. As a result, a fuller drive may have less room for this fast write area, while an emptier drive may be able to allocate more. The reported 117GB cache behavior is specific to this drive’s capacity and tested state, not a universal SSD threshold.
““All it took was 19 seconds for that speed to dip hard.””
— XDA Developers writer
As an affiliate, we earn on qualifying purchases.
What the Single Drive Test Cannot Show
The reported figures come from one SSD and one transfer test. The source does not provide a controlled comparison with other drive models, a full set of repeated trials, or results at different levels of drive capacity. It is not clear how much the timing or post-cache speeds would change with different system conditions, data patterns, temperatures or workloads.
The result also should not be treated as a fixed 19-second limit or a guaranteed 117GB cache size for this model in every situation. The article reports those observations with about 24% of the drive free, but does not establish a universal threshold. Buyers should check independent sustained-write tests for the specific model and capacity they are considering.
USB 3.2 Gen 2×2 NVMe SSD enclosure for sustained file transfers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
How Buyers Can Check Sustained Writes
Readers who want to test their own drive can use a large-file benchmark and watch the throughput graph for a sustained high-speed section followed by a drop. The XDA writer recommends a Random or Mixed data pattern in HD Tune Pro’s File Benchmark and cautions against using zeros, which may be compressed or handled specially. The test should be large enough to exceed the drive’s cache, and the selected partition needs sufficient free space.
Because this is a write-heavy test, it should not be repeated just to reproduce a graph. For purchasing decisions, the next useful step is to compare long-duration write tests for the same model and capacity, rather than relying only on a peak specification. The source does not report a manufacturer response or a follow-up test across additional SSDs.
HD Tune Pro SSD file benchmark software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
Why did the SSD’s write speed drop?
The XDA report attributes the drop to the drive filling its dynamic SLC write cache. After that fast area is used, writes largely proceed at the speed of the drive’s native TLC NAND while the controller processes cached data.
Does a drop after 19 seconds mean my SSD is faulty?
Not by itself. The 19-second result came from one drive and one sustained transfer. The article describes cache exhaustion as an expected behavior, though the timing and speeds can differ among SSDs and workloads.
Does a fuller SSD have slower sustained writes?
It may have a smaller dynamic SLC cache. The source says the cache can borrow from available NAND capacity, so less free space may mean less room for the fast write area. It does not provide a universal formula for the effect.
Will this affect gaming?
The report says the effect is less relevant to typical gaming because games mostly involve reading data from the SSD. A large installation or file copy can involve writes, but it may not take the form of one uninterrupted transfer that exhausts the cache.
How can I test my SSD’s sustained write speed?
The writer used HD Tune Pro’s File Benchmark with a large test file and a Random or Mixed data pattern. A write test uses drive endurance, so run it carefully, avoid the program’s potentially destructive main write benchmark, and do not repeat the test unnecessarily.
Source: XDA Developers
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
