Low 4K and 4K-64Thrd Scores in AS SSD Benchmark — Causes and Step-by-Step Checks

Sequential numbers look fine, but the 4K rows are in single figures and 4K-64Thrd barely improves on them — so the total score, which weights those rows heavily, collapses. This pattern has a short list of usual causes, and most of them are settings rather than the SSD. This guide explains what the two 4K tests measure, why IDE mode and misalignment hit them first, and the order in which to check everything else.

Before you start: back up important files on the drive you are troubleshooting. Most checks below are read-only, but the fixes that touch BIOS settings, partitions or drivers should never be attempted without a current backup.

What the 4K rows actually measure

The 4K test measures random reads and writes of tiny 4 KB blocks, one operation at a time. The 4K-64Thrd test measures the same small-block pattern spread across 64 threads, queuing many operations at once — the workload where a drive’s controller, its command queuing and the host interface all have to work together. AS SSD Benchmark runs six synthetic tests plus three copy tests without using the operating system cache for the synthetic set, so these rows reflect the drive path, not memory speed.

Two expectations to set before troubleshooting. First, single-thread 4K numbers are always dramatically smaller than sequential numbers, on every drive ever made — small random operations simply cannot stream. A “low-looking” 4K read next to a huge sequential read is physics, not a fault. Second, the gap between 4K and 4K-64Thrd is itself diagnostic: on a healthy AHCI setup the threaded test should pull well ahead on reads because queued operations can be served efficiently. When 4K-64Thrd refuses to rise above plain 4K, something is preventing effective queuing — and that something is the target of this guide.

Cause 1: IDE mode (pciide) — the queue killer

Native Command Queuing is an AHCI feature. In IDE mode there is no NCQ for the drive to exploit, so the 64-thread test loses its main advantage and 4K-64Thrd scores sag exactly as described above. Check the header of the AS SSD window first: pciide – BAD confirms this cause in one glance. The fix is a controller-mode change in BIOS/UEFI with Windows prepared beforehand — a job with a genuine boot-failure risk if rushed. Follow the pciide – BAD guide in full, backup included, and use the mode check to see every drive’s state before you start.

If the header already shows storahci or msahci, queuing is available and the cause lies elsewhere — do not change BIOS settings that are already correct. Note the driver word down and move on; our storahci explainer covers why either Microsoft driver name is fine.

Cause 2: misalignment and write-side pain

A misaligned partition forces extra work on small writes in particular, so suspect alignment when 4K writes and 4K-64Thrd writes disappoint while reads look healthier. The header reports alignment next to the driver: BAD there is your confirmation. The remedies — leave it, realign with a partition tool at your own risk, or reinstall/re-clone cleanly — are compared honestly in the alignment BAD guide, and the alignment check shows how to record the offset before deciding. As with every partition job: back up first, and never promise yourself a lossless outcome from an in-place tool.

While you are looking at writes, check free space. A nearly full SSD has little room for its internal housekeeping, and small random writes are the first casualty. Freeing real space and letting the drive idle before a retest is a boring fix that regularly works.

Cause 3: CPU, power and background competition

Sixty-four threads of small operations need the host awake and responsive. A laptop on battery with an aggressive power plan, a CPU being held back by power limits, or a system quietly running updates, sync and antivirus scans will post 4K numbers that say more about the moment than the drive. For a diagnostic run: mains power connected, a performance-oriented power plan, a fresh reboot, background disk work paused, and no other drive utilities open. Then run the test twice and compare — consistency between runs is itself a health sign.

Thermal behaviour belongs here too. Sustained testing warms an NVMe drive, and a throttling drive fades as the run progresses. AS SSD Benchmark reports no temperatures, so if you suspect heat, watch it with the drive maker’s tool or a hardware monitor during a run, and think about airflow and heatsinking only once you have seen the temperature actually climb. Guessing at heat without a reading wastes money.

A clean retest protocol

• Reboot, connect mains power, and pause updates, sync and scans.
• Confirm the header: AHCI driver shown, alignment OK — or note which one is not.
• Leave the system idle for a few minutes so internal drive housekeeping can settle.
• Run the full test at the default 1 GB size, twice, and save or screenshot both results (AS SSD Benchmark can export results and save screenshots from its File menu).
• Compare patterns, not just totals.
• 4K-64Thrd now well above 4K on reads means queuing is working. Flat 4K-64Thrd still points to mode or driver. Weak writes only points to alignment or free space. Everything weak points back to the general low-score checklist, including port, cable and enclosure checks.

Keep those saved runs. A dated series of consistent results is how you spot real decline later, instead of arguing with a single number remembered from months ago. You can always get a fresh, verified copy of the tool from our download page if your current one has an uncertain history.

When low 4K is the drive telling the truth

If mode, alignment, space, power and background are all clean and 4K results remain far below the drive’s class across repeated quiet runs, accept the possibility that the drive is worn or failing. Very small, heavily used drives and drives that have lived nearly full for years can genuinely lose random-write pace. Back up while the drive still cooperates, and if replacement is due, choose on the strength of real 4K behaviour and capacity for your workload rather than headline sequential figures — that is the principle behind our NVMe picks. Low 4K is rarely a mystery once the header, the layout and the test conditions have each been given a fair hearing.