How Is the AS SSD Benchmark Score Calculated? (Formula + Worked Example)
You run AS SSD Benchmark, wait a few minutes, and get a number: 1247. Is that good? Where does it even come from?
Unlike most benchmark tools, AS SSD's author made the scoring formula public. That means you can take any drive's raw test results, plug them into the formula yourself, and arrive at the exact same score the tool reports. In this guide, we'll do exactly that — using a real, independently published test of our own OSCOO MD100 portable SSD as the worked example, so you can verify every step.
The Three AS SSD Formulas
AS SSD produces three scores: a read score, a write score, and a total score. Each one is a weighted sum of the raw speeds (in MB/s) from the three main tests:
- Seq — sequential read/write of a large file
- 4K — random reads/writes of 4 KB blocks, single thread
- 4K-64Thrd — the same 4K random test, spread across 64 threads
The formulas:
Read score = (Seq Read × 0.1) + 4K Read + 4K-64Thrd Read
Write score = (Seq Write × 0.1) + 4K Write + 4K-64Thrd Write
Total score = (Seq Write × 0.15) + (Seq Read × 0.1) + (4K Read × 2) + 4K Write + (4K-64Thrd Read × 1.5) + 4K-64Thrd Write
Two things jump out immediately:
- Random performance carries the heaviest multipliers. 4K Read gets 2× and 4K-64Thrd Read gets 1.5×, while sequential speed — the big number on the box — is cut down to 0.1–0.15×.
- But don't confuse multipliers with contributions. Sequential speeds are 10–60× larger than 4K speeds to begin with, so even after the ×0.1 haircut they still claim a substantial share of the final score. We'll quantify exactly how much below.
Note that access time, while measured and displayed, is not part of the score formula at all.
Worked Example: Verifying the Formula with Real Test Data
In July 2026, the independent testing site ssd-tester.com published a full benchmark review of the OSCOO MD100 512GB. Here are the raw AS SSD results from that test:
| Test | Read | Write |
| Seq | 1,906.74 MB/s | 1,695.93 MB/s |
| 4K | 30.59 MB/s | 78.23 MB/s |
| 4K-64Thrd | 248.33 MB/s | 290.26 MB/s |
| Score | 470 | 538 |
| Total score | 1247 |
Let's verify all three scores by hand.
Read score:
(1,906.74 × 0.1) + 30.59 + 248.33 = 190.67 + 30.59 + 248.33 = 469.6 → 470 ✓
Write score:
(1,695.93 × 0.1) + 78.23 + 290.26 = 169.59 + 78.23 + 290.26 = 538.1 → 538 ✓
Total score:
(1,695.93 × 0.15) + (1,906.74 × 0.1) + (30.59 × 2) + 78.23 + (248.33 × 1.5) + 290.26 = 254.39 + 190.67 + 61.18 + 78.23 + 372.50 + 290.26 = 1,247.2 → 1247 ✓
Every score matches the published result to the point. The formula isn't an approximation — it's exactly how the tool computes the number on your screen.
Weights vs. Actual Contribution — They're Not the Same Thing
The multipliers tell you the formula's intent: reward random 4K performance, downweight the sequential marketing number. Small random reads are what make a system feel fast — booting, launching apps, loading game assets — so 4K gets 2× and the marketing number gets 0.1×.
But the intent and the arithmetic diverge, because sequential speeds are vastly larger raw numbers. Here's what each test actually contributes to the MD100's scores:
| Component | Share of Read score | Share of Write score | Share of Total score |
| Seq (× 0.1–0.15) | 40.6% | 31.5% | 35.7% |
| 4K (× 1–2) | 6.5% | 14.5% | 11.2% |
| 4K-64Thrd (× 1–1.5) | 52.9% | 53.9% | 53.1% |
Three takeaways:
- The 64-thread test is the real backbone of the score — over half of it, on every axis.
- 4K single-thread has the highest multiplier but the smallest actual influence. Ironically, it's the test that best predicts everyday responsiveness (real desktop use runs at low queue depths), yet it moves the score the least.
- Sequential still matters more than the ×0.1 suggests. The formula dates from the SATA era, when ~500 MB/s sequential speeds kept that share small. On a modern 20 Gbps drive pushing ~2,000 MB/s, sequential quietly claims over a third of the total — the score has drifted toward rewarding interface bandwidth over time.
None of this makes the score useless — it makes it comparative. Two drives tested the same way can be ranked fairly against each other; just don't read the total as a pure measure of how fast the drive will feel day to day.
How to Read Your Own Score
A few practical rules of thumb when interpreting results:
Compare within the same interface class. A SATA SSD physically cannot exceed ~550 MB/s sequential, so its score ceiling is far lower than an NVMe or USB 3.2 Gen 2x2 drive. Comparing scores across interfaces mostly tells you about the interface.
Check the status line before trusting anything. In the top-left of the AS SSD window, you should see your storage driver reported as OK and the partition alignment as OK. A "BAD" flag (misaligned partition, or legacy IDE driver mode) will crater your 4K results and your score — and it's fixable.
External drives take a penalty on 4K. USB adds protocol overhead and latency that hits small random operations hardest. This is normal. The MD100's results above are strong for a portable drive; an internal NVMe drive on the same controller class would post higher 4K numbers.
A single run is a snapshot. Background tasks, thermal state, and how full the drive is all affect results. If a number looks off, close everything and run it again.
What the Score Doesn't Tell You
AS SSD's synthetic tests run against a relatively small test file, which means they mostly measure the drive's fast cache region. They don't capture sustained write behavior — what happens when you write 50–100 GB continuously and the pSLC cache fills up. For that, you need a long-duration test like HD Tune Pro's full-file write. (In the MD100 review linked above, ssd-tester.com ran exactly this test alongside AS SSD, which is what a thorough review should do.)
It also doesn't replace CrystalDiskMark, which uses different queue/thread configurations and typically reports higher peak sequential numbers. If you've ever wondered why the two tools disagree, we've covered that in detail in our guide to CrystalDiskMark results vs. real-world speeds.
The Drive Behind the Numbers
The worked example above comes from ssd-tester.com's independent review of the OSCOO MD100 512GB, a USB 3.2 Gen 2x2 (20 Gbps) portable SSD rated at 2,100 MB/s read and 1,900 MB/s write. At the time of testing, it held the top performance score in its capacity class on the site — a result you can now verify line by line yourself.
