Jon Jacobi reported that a 2TB Samsung P9 USB4 SSD maintained its transfer rate while receiving a 900GB file immediately after a separate 450GB write. He also described the portable drive as remaining almost cool to the touch during extended writes. Those observations concern one tested sample and have not been independently confirmed.
The large-file transfers were the main distinction Jacobi identified, although he did not describe the P9 as the fastest drive across his tests. His findings covered synthetic benchmarks, folder transfers and individual large files, with results varying by workload. He described capacities reaching 8TB, but the performance observations apply specifically to the 2TB unit Samsung supplied for testing.
Reported design and warranty details
Jacobi described the P9 as a 40Gbps USB4 external SSD measuring approximately 3.5 inches long, 2.6 inches wide and 0.3 inches thick, with a weight of 3.3 ounces. These specifications have not been independently verified. The dimensions and weight were part of his description of the enclosure used during testing.
He described the enclosure as lacking the large heatsink, cooling fins or fan found on some other external drives in his testing experience. His observation of limited noticeable heat therefore concerned a relatively small enclosure without those visible cooling features. No measured surface temperature accompanied that observation.
Jacobi also described a limited five-year warranty. The limited designation matters: the stated duration alone does not establish the coverage conditions or any applicable endurance restrictions. His comments do not support describing the warranty as unconditional or assigning a specific write allowance to the drive.
He said Samsung had not supplied a terabytes-written, or TBW, endurance rating. No numerical endurance limit is established by that statement. He also described the P9 as lacking an IP rating; his observations do not establish certified protection against water or dust, and they do not support a water-resistance claim.
Large-file transfers were the main reported strength
The most specific sustained-write observation involved two consecutive transfers. Jacobi said he first wrote a 450GB file and then immediately copied a 900GB file to the drive. During the second transfer, he reported no significant decline in transfer rate. He did not provide a numerical minimum speed for that sequence in the accompanying text.
That limits the conclusion to the behavior he described under those conditions. It does not establish an exact sustained throughput figure, a guaranteed rate across the drive’s entire capacity or equivalent results for the 1TB, 4TB and 8TB versions. No separate results for those capacities were provided.
The 450GB transfer did not produce the fastest result in his comparison using either FastCopy or Windows Explorer, Jacobi said. His findings therefore distinguish sustained behavior during consecutive large writes from finishing an individual transfer ahead of every competing drive. The reported consistency was a separate observation from the drive’s position in any single benchmark.
Jacobi described the 48GB transfer results as generally good, while noting comparatively slower folder transfers. He also reported varied results in CrystalDiskMark’s 4K tests. These descriptions do not establish one uniform performance advantage across small files, folders and large sequential transfers, and no exact benchmark speeds can be assigned from them.
His comparison included the OWC 1M2, LaCie Rugged SSD4 and TerraMaster D1 SSD Plus. Those references establish the products he considered alongside the P9, but the available descriptions do not support a numerical performance table or a precise speed difference between the drives.
The thermal observation was similarly qualitative. Jacobi said the P9 remained almost cool to the touch during long writes. He had asked Samsung about the energy demands of its NAND memory and controller but said he had not received an explanation at that point. The reported behavior therefore does not establish how Samsung achieved it or quantify the drive’s power consumption.
Cited prices ranged from $340 to $2,880
Jacobi cited the following prices from Samsung’s website. The amounts have not been independently confirmed, and no publication date was supplied for the price observations. They should be treated as an undated price snapshot.
| Capacity | Cited price |
|---|---|
| 1TB | $340 |
| 2TB | $650 |
| 4TB | $1,440 |
| 8TB | $2,880 |
The 2TB model used for testing carried a cited price of $650. The $2,880 figure applied to the 8TB version, whose performance was not established by the 2TB sample’s results. The capacity and price figures therefore should not be read as evidence that every version delivers identical sustained-write behavior.
The cited amounts also do not establish a broader change in SSD prices or support a prediction about future costs. Without a dated, independently confirmed retail listing, they cannot establish a present purchase price or stock position for any of the four capacities.
Test conditions limit how broadly the results apply
Jacobi described a test system running 64-bit Windows 11 on a Z790 motherboard with an Intel Core i5-12400 processor. The stated configuration included two Kingston Fury 32GB DDR5-4800 memory modules, giving it 64GB of system memory. These are his stated test conditions and have not been independently verified.
He listed integrated 20Gbps USB and Thunderbolt 4 connections on the system’s rear panel, along with Intel integrated graphics. The 48GB transfer tests used an ImDisk RAM disk allocated 58GB of the available memory. The 450GB file was transferred from a 2TB Samsung 990 Pro that also held the operating system.
Jacobi said each test used a freshly NTFS-formatted drive on which TRIM had been performed. That preparation describes the condition of the drive during the reported measurements. It does not establish how the same unit would perform after extended use or with substantially less free space.
He cautioned that external SSD performance can decline as a drive fills, including when less NAND remains available for secondary caching. He also noted that capacity can affect results because versions may differ in the number of memory chips available for parallel reads and writes and in the space available for caching.
His stated limitations also included the possibility of component changes by vendors. The sustained-write and thermal observations remain specific to the supplied 2TB sample, its configuration and the workloads he described.
