PC Hardware Audit and Benchmark on Windows

This is a repeatable methodology to inventory, verify and benchmark a Windows PC from BIOS strings to port-by-port testing to thermal stress, and produce a clean acceptance report.

Useful whenever you receive an unknown machine (refurbished purchase, second-hand, fleet handover, customer device) and need to answer one question with evidence: does this hardware do what it claims to do?

---

Overview

Why audit a PC

Concretely, a structured PC audit lets you:

• Detect silent faults before they bite: a battery at 60 % health, a dead USB port, an SSD with degrading SMART, RAM stuck in single-channel.
• Verify advertised performance: does the i5 actually sustain its boost clock, or is it thermally throttling at 1.2 GHz under load?
• Establish a baseline: useful for later comparisons (driver updates, thermal paste renewal, OS reinstall) and for insurance/warranty disputes.
• Document a refurbished/second-hand purchase: proof that the unit was inspected, with date and measurements, before being put in service.

A good report can also be archived as part of an asset inventory alongside license keys and warranty details.

---

When to use this methodology

Scenario	Why it matters
🛒 Refurbished/second-hand purchase	Verify within the return window that the unit matches the listing and is healthy
🏢 Fleet onboarding / handover	Capture the state of the machine when it enters service (lifecycle baseline)
🔧 Pre/post repair	Compare measurements before and after a hardware fix (e.g. thermal paste, SSD swap)
📉 Performance complaint	Replace “feels slow” with numbers: clock under load, SSD latency, battery health
📦 Resale / disposal	Provide a transparent technical sheet to the next owner

---

Workflow

flowchart LR
    A[System info] --> B[Components<br>CPU · RAM · SSD · GPU]
    B --> C[Battery health]
    C --> D[Network<br>Wi-Fi · Ethernet · BT]
    D --> E[Port tests<br>USB · USB-C · HDMI · SD · audio]
    E --> F[Synthetic benchmarks<br>winsat · CDM · Cinebench]
    F --> G[Thermal stress<br>10 min full load]
    G --> H[Verdict<br>strengths · concerns · score]

The order matters: discover before you measure, and measure before you stress.

You want to know what you’re stressing, and you want a clean baseline reading before pushing the machine into thermal headroom.

---

Sections of the report (template)

The example below is structured around eight sections.

Treat them as a checklist: drop or expand sections to fit the machine class (desktop, laptop, mini-PC).

#	Section	What to capture
1	System summary	Vendor, model, family, SKU, form factor (redact serials)
2	CPU	Model, generation, cores/threads, base/boost, cache, TDP, socket
3	RAM	Total, sticks × size, type, speed, channel config
4	Storage	Drive model, bus (NVMe/SATA), capacity, firmware, SMART health, free space
5	GPU	Model, VRAM, driver, current resolution
6	Battery (laptops)	Design vs current full-charge capacity, cycle count, health %
7	Network	Wi-Fi / Ethernet / Bluetooth — chip, link speed, status (redact MAC and IP)
8	OS & firmware	Windows edition + build, BIOS/UEFI version + date, security state (VBS, HVCI, Secure Boot, TPM)

Plus three test sections:

• 🔌 Port matrix: every physical port, with a detection + speed result and a one-line verdict.
• 🧪 Benchmark table: synthetic results (winsat, CrystalDiskMark, Cinebench R23) side by side.
• 🌡️ Thermal stress: sustained-load behaviour (clocks, package temp, SSD temp) over ~10 minutes.

---

Tools used

Tool	What it gives you	License
PowerShell (built-in)	Inventory: hardware, drivers, network, volumes, OS build	Built-in
winsat (built-in)	WEI scores + raw CPU / RAM / disk / GPU throughput	Built-in (admin)
powercfg /batteryreport	Battery health report (HTML): design vs full-charge, cycle count	Built-in
CrystalDiskMark 9	Sequential & random SSD throughput at varied queue depths	Free
Cinebench R23	CPU benchmark + 10-minute stress test in one click	Free
HWiNFO / HWMonitor	Live sensors: per-core clock, package temp, power, SSD temp	Free
Reference USB stick (e.g. Kingston DataTraveler 3.0)	Probe to classify USB ports (USB 3.x ≈ 100 MB/s read vs USB 2.0 ≈ 30 MB/s read)	n/a
External monitor + cable	Validate HDMI / DisplayPort / USB-C DP-Alt	n/a
Wired headphones + microSD card	Validate the audio combo jack and the card reader	n/a

TIP

Read raw, not cached. The single most common mistake when measuring storage or USB throughput is reading data that Windows still has in RAM. The numbers come back as multiple GB/s and mean nothing. Always read with FILE_FLAG_NO_BUFFERING (see the appendix script).

---

What “good” looks like for each section

• Battery (laptop): > 80 % health and a sane cycle count for the unit’s age. Anything below 70 % is replacement territory.
• SSD: SMART Healthy, sequential read ≥ 1 GB/s for NVMe (≥ 500 MB/s for SATA), no reallocated sectors.
• RAM: matched sticks in dual channel (check DeviceLocator: different channels). Single-channel cuts bandwidth roughly in half.
• CPU sustained: 10-minute Cinebench R23 should hold a multi-core score within ~10 % of the database average for that chip. Big shortfall → thermal throttling or power limit.
• Thermals: package temp under sustained load should stay below TjMax (100 °C on most modern Intel/AMD parts). Hitting TjMax = throttling.
• Ports: every physical port detects a device and delivers its rated speed. A USB-A port that detects but only delivers ~30 MB/s on a USB 3.x stick is wired as USB 2.0 (or the controller is degraded).

---

Example: Lenovo ThinkPad L14 Gen 1

Acceptance report for a refurbished unit Inspection date: 2026-06-29 · Hostname: [redacted]

System summary

Field	Value
Vendor	LENOVO
Model	20U10028US
Family	ThinkPad L14 Gen 1
Type	14” Notebook (x64)
SKU	LENOVO_MT_20U1_BU_Think_FM_ThinkPad L14 Gen 1
Serial number	[redacted]
Motherboard S/N	[redacted]
Form factor	Portable

Processor (CPU)

Field	Value
Model	Intel® Core™ i5-10210U
Architecture	Comet Lake (10th gen)
Cores / Threads	4 cores / 8 threads
Base clock	1.60 GHz
Max turbo	4.20 GHz (Intel spec)
Cache L2 / L3	1 MB / 6 MB
Socket	U3E1 (BGA, soldered)
TDP	15 W

RAM

Field	Value
Total installed	16 GB
Configuration	2 × 8 GB (Dual Channel)
Type	DDR4 SODIMM
Speed	3200 MT/s
Module vendor	Samsung
Part number	M471A1K43DB1-CWE
Slots	ChannelA-DIMM0 · ChannelB-DIMM0

Storage (SSD)

Field	Value
Model	Micron MTFDHBA256TDV (2300 series)
Type	NVMe SSD (M.2)
Capacity	256 GB (238.5 GB usable)
Bus	NVMe / PCIe
Firmware	3011P4LN
Health	✅ Healthy
Partitions	3

Volume C: usage

Volume	Total	Free	Used
C: (NTFS)	≈ 236.7 GB	≈ 180.3 GB	≈ 56.4 GB (24 %)

GPU

Field	Value
Model	Intel® UHD Graphics (integrated)
VRAM	1 GB (shared)
Driver	31.0.101.2137
Current resolution	1920 × 1080 (Full HD)

Battery

Field	Value
Part number	5B10W13896
Design capacity	45,730 mWh
Full-charge capacity	44,400 mWh
Cycle count	16
Battery health	✅ 97.1 % (excellent)
Charge at inspection	99 % (on AC)

💡 97 % health with only 16 charge cycles indicates a battery that is effectively new — a strong signal of high-quality refurbishment.

Network and connectivity

Interface	Device	Status
Wi-Fi	Intel® Wi-Fi 6 AX200 160MHz	✅ Connected (144.4 Mbps)
Ethernet	Intel® Ethernet Connection (10) I219-V	✅ Connected (1 Gbps)
Bluetooth	Bluetooth Device (PAN)	⚪ Disconnected

• Wi-Fi 6 (802.11ax) + integrated Bluetooth
• MAC addresses and current IP redacted

Operating system and firmware

Field	Value
OS	Windows 11 Pro
Version	25H2
Build	26200.8737
Install date	2026-06-22
Language / locale	Italian (it-IT)
BIOS/UEFI	LENOVO R17ET45W (v1.28)
BIOS date	2026-04-13
Security	VBS on · HVCI on · App Control for Business (enforced)

---

Port and peripheral tests

Connectivity probe performed with a Kingston DataTraveler 3.0 USB stick (USB 3.0). Uncached read throughput reflects the real port class: ≈100 MB/s = USB 3.x · ≈30–35 MB/s = USB 2.0. Write ~8 MB/s is the stick’s own ceiling, not the port.

#	Port	Type	Detected	Read	Verdict
1	USB-A left	USB-A 3.2 Gen 1	✅ Yes	99.7 MB/s	✅ USB 3.x OK
2	USB-A right (Always-On)	USB-A 3.2 Gen 1	✅ Yes	100 MB/s	✅ USB 3.x OK
3	USB-C (data)	USB-C 3.2 Gen 1	✅ Yes	94.9 MB/s	✅ USB 3.x OK
4	USB-C (charging / PD)	USB-C 3.2 Gen 1	✅ Yes	95.5 MB/s	✅ USB 3.x + charging OK
5	HDMI	HDMI 1.4b	✅ Passed	video OK	✅ External monitor via HDMI: EDID read + image confirmed on screen
6	microSD reader	microSD (SD bus)	✅ Yes	n/a	✅ Reader OK: 2 GB FAT card mounted and Healthy (the first card tried was genuinely defective)
7	Ethernet RJ-45	Gigabit	✅ Passed	1 Gbps	✅ Link up at 1 Gbps, DHCP lease obtained from router
8	3.5 mm audio jack	Combo headset/mic	✅ Passed	OK	✅ Output confirmed by listening; jack detects insertion; external mic recognised as active endpoint

Port test result: 8/8 verified and working ✅: 2× USB-A and 2× USB-C at full USB 3.x speed, microSD reader, audio jack, HDMI (external monitor) and Gigabit Ethernet all confirmed.

---

Benchmarks and tests

Benchmarks run on 2026-06-29 / 2026-06-30: built-in Windows tools (winsat) + CrystalDiskMark 9, Cinebench R23, HWMonitor / HWiNFO.

Test	Tool	Result	Note
CPU multi-thread	winsat	AES256 2950 · SHA1 2054 · Vista-compr. 714 · LZW 253 MB/s	WEI CPU 8.9
CPU single-thread	winsat	AES256 612 · SHA1 376 · LZW 61 MB/s	i5-10210U 4C/8T
CPU Cinebench R23 multi-core	Cinebench R23	3466 pts	✅ Strong for i5-10210U (typical 2500–2900)
CPU Cinebench R23 single-core	Cinebench R23	1054 pts (MP ratio 3.29×)	✅ Within norm
SSD read	winsat	Sequential 2339 MB/s · Random 16K 850 MB/s	WEI Disk 8.9
SSD write	no-admin test (1 GB)	Sequential 891 MB/s	single-thread, realistic value
RAM bandwidth	winsat	19,918 MB/s (~19.9 GB/s)	WEI Mem 8.9 · DDR4-3200 dual channel
GPU Intel UHD	winsat	Desktop/DWM WEI 6.5 · D3D 9.9 · video mem 7581 MB/s	integrated
WEI base score	winsat	6.5	capped by integrated GPU (CPU/RAM/SSD all 8.9)
Thermal stress test	Cinebench R23 10 min + HWiNFO	✅ Stable · CPU ~3.0–3.2 GHz · peak 95 °C	see 🌡️

SSD detail: CrystalDiskMark 9

Pattern	Read	Write
SEQ1M Q8T1	3,333.45 MB/s	1,559.83 MB/s
SEQ1M Q1T1	1,555.10 MB/s	1,266.87 MB/s
RND4K Q32T1	255.27 MB/s	170.14 MB/s
RND4K Q1T1	45.95 MB/s	66.88 MB/s

Peak sequential read 3,333 MB/s at deep queue (Q8): consistent with and exceeding the winsat figure (2,339 MB/s, shallow queue). Sequential write ~1,560 MB/s. Excellent values for an NVMe drive of this class (Micron 2300).

CrystalDiskMark screenshot (SSD result):

Thermal stress test: Cinebench R23

✅ CPU healthy: HWiNFO log, 36 samples under full load:

🌡️ Test conditions (severe): ~33 °C ambient (very warm room), no extra cooling, laptop flat on the desk, no riser. In a cooler room and with a ventilated base the temperatures would be noticeably lower — so the 95 °C below is a good result for the context.

Parameter	Under load	Peak
All-core clock	~3.0–3.2 GHz	3.3 GHz
CPU package temperature	~91 °C	95 °C
Core temperature (avg)	~87 °C	92 °C
SSD temperature	~45 °C	52 °C

Under sustained load the CPU holds ~3.0–3.2 GHz on all cores: a strong result for a 15 W i5-10210U, with temperatures peaking at 95 °C but staying below TjMax (100 °C). The confirmation is Cinebench R23 Multi-Core 3466 pts, fully in line with the part.

📌 CPU conclusion: the hardware works correctly and performs as it should (3,466 pts, ~3.2 GHz sustained).

Cinebench R23 screenshots:

Multi-Core 3466 pts:

Multi-Core 3466 + Single-Core 1054 (MP ratio 3.29×):

---

Acceptance checklist

• Hardware inventory complete
• SSD health verified (Healthy)
• Battery health verified (97.1 %)
• Performance benchmarks: CPU, RAM, SSD, GPU (winsat) + SSD write
• Stress test (Cinebench 10 min): system stable; CPU healthy at ~3.2 GHz / 95 °C → 3466 pts
• USB-A ×2, USB-C ×2, microSD, audio jack verified: OK
• HDMI and Ethernet verified: ✅ OK (external monitor via HDMI · Gigabit Ethernet at 1 Gbps)

---

Final remarks

Overall verdict: excellent refurbished unit, approved. The ThinkPad L14 Gen 1 passes the audit with very good results.

Strengths

• 🔋 Battery as good as new: 97.1 % health with only 16 charge cycles.
• 💾 Excellent NVMe SSD: 3,333 / 1,560 MB/s sequential read/write, Healthy status.
• 🧠 Healthy and performant CPU: Cinebench R23 3,466 pts multi / 1,054 pts single, ~3.2 GHz sustained under load.
• 🧩 Optimal RAM: 16 GB DDR4-3200 in dual channel (~19.9 GB/s).
• 🔌 Ports: 8/8 tested, all OK (USB-A ×2 and USB-C ×2 USB 3.x, microSD reader, audio jack, HDMI with external monitor, Gigabit Ethernet at 1 Gbps).
• 🪟 Software: Windows 11 Pro 25H2, security stack (VBS / HVCI / Secure Boot) active.

Things to watch

• 🌡️ Temperatures under full load: up to 95 °C, but the test was run in harsh conditions (~33 °C ambient, very warm room, no extra cooling, laptop flat on the desk without a riser): in normal conditions, expect lower values. Still under the 100 °C limit; for more headroom, a thermal-paste renewal can be considered.

All tests completed

• 🟢 HDMI (external monitor) and 🟠 Gigabit Ethernet: verified and working. No test deferred.

Judgement

Solid, reliable hardware with good performance for the category (14” business ultrabook). No blocking defect found: purchase approved. ✅

Final score

Per-component rating (0 to 10 scale):

Component	Score
🔋 Battery	9.5
💾 NVMe SSD	9.5
🔌 Ports and connectivity	10
🧩 RAM	9
🧠 CPU	8
🌡️ Thermal behaviour	7.5
🎨 GPU (Intel UHD integrated)	4.5

🟦 Overall score, GPU included: 7.5 / 10 Rated as a general-purpose machine (with the GPU weighted as a primary capability axis, not just a component average): the integrated GPU is the real bottleneck and rules out gaming and accelerated editing / 3D. Still a decent all-rounder, but not a “complete” machine.

🟩 Score for intended use (office and productivity, GPU excluded): 8.9 / 10 (≈ 9/10) For browsing, Office, multitasking and day-to-day work the GPU doesn’t matter: CPU, RAM, SSD, battery and connectivity are all high quality. For that use case, an excellent purchase.

---

Commands used

PowerShell commands (Windows 11). Open PowerShell and paste them. 🔒 = requires PowerShell launched as administrator. This section is updated each time a new test is run.

System information

# General overview (OS, model, BIOS, RAM, network)
systeminfo
 
# Model / vendor / total RAM
Get-CimInstance Win32_ComputerSystem | Format-List Manufacturer,Model,SystemFamily,TotalPhysicalMemory
 
# CPU (cores, threads, clock, cache)
Get-CimInstance Win32_Processor | Format-List Name,NumberOfCores,NumberOfLogicalProcessors,MaxClockSpeed,L2CacheSize,L3CacheSize
 
# RAM modules (slots, speed, dual channel)
Get-CimInstance Win32_PhysicalMemory | Format-Table Manufacturer,PartNumber,Capacity,Speed,DeviceLocator -AutoSize
 
# Physical disks
Get-CimInstance Win32_DiskDrive | Format-List Model,InterfaceType,Size,MediaType,SerialNumber,FirmwareRevision
Get-PhysicalDisk | Format-Table FriendlyName,MediaType,BusType,HealthStatus
 
# GPU
Get-CimInstance Win32_VideoController | Format-List Name,AdapterRAM,DriverVersion,CurrentHorizontalResolution,CurrentVerticalResolution
 
# Motherboard and BIOS/UEFI
Get-CimInstance Win32_BaseBoard | Format-List Manufacturer,Product,SerialNumber,Version
Get-CimInstance Win32_BIOS | Format-List Manufacturer,SMBIOSBIOSVersion,ReleaseDate,SerialNumber
 
# Network adapters
Get-NetAdapter | Format-Table Name,InterfaceDescription,MacAddress,LinkSpeed,Status -AutoSize
 
# Volumes and free space
Get-Volume | Format-Table DriveLetter,FileSystemLabel,FileSystem,HealthStatus,SizeRemaining,Size -AutoSize
 
# Exact Windows version (e.g. 25H2 + build)
Get-ItemProperty 'HKLM:\SOFTWARE\Microsoft\Windows NT\CurrentVersion' |
  Select-Object ProductName,DisplayVersion,CurrentBuildNumber,UBR

Battery (health and charge cycles)

# Official Windows report: Design Capacity, Full Charge Capacity, Cycle Count
powercfg /batteryreport /output "$env:USERPROFILE\Desktop\battery-report.html"
# Open the generated HTML and read the "Installed batteries" section
 
# Full-charge capacity via WMI (mWh)
(Get-CimInstance -Namespace ROOT\WMI -ClassName BatteryFullChargedCapacity).FullChargedCapacity

TPM and Secure Boot

Get-Tpm                  # 🔒 TPM status
Confirm-SecureBootUEFI   # 🔒 returns True/False

USB ports: detection and speed

# USB controllers present
Get-CimInstance Win32_USBController | Select-Object Name
 
# Detection: run BEFORE and AFTER plugging the device, then compare
Get-PnpDevice -PresentOnly -Class USB | Select-Object FriendlyName,InstanceId

Port speed: use the usb-bench.ps1 script (section H). Key rule: read uncached (FILE_FLAG_NO_BUFFERING), otherwise Windows serves the data from RAM and reports fake multi-GB/s numbers. Rule of thumb with a USB 3.0 stick: read ≈100 MB/s = USB 3.x port, ≈30–35 MB/s = USB 2.0 port.

microSD reader

# Card reader controller
Get-PnpDevice -PresentOnly | Where-Object FriendlyName -match 'SD|Card|Reader|Realtek'
# Disk and volume of the inserted card
Get-Disk   | Where-Object BusType -eq 'SD'
Get-Volume | Where-Object DriveType -eq 'Removable'

Combo audio jack

# Audio endpoints: plug headphones in and watch a "2nd output" / "External Mic" appear
Get-PnpDevice -Class AudioEndpoint | Select-Object Status,FriendlyName
 
# Sound test on the default output
(New-Object System.Media.SoundPlayer "C:\Windows\Media\tada.wav").PlaySync()
[console]::beep(880,400); [console]::beep(660,400); [console]::beep(988,600)

Performance benchmarks

WEI / winsat (CPU, RAM, disk, graphics): 🔒 admin required:

# Full assessment: prints raw MB/s for CPU, memory, disk, graphics
winsat formal
 
# Or single sub-tests:
winsat disk      # 🔒 SSD/HDD
winsat mem       # 🔒 memory bandwidth
winsat dwm       # 🔒 desktop graphics
 
# Synthetic WEI scores (1.0–9.9):
Get-CimInstance Win32_WinSAT |
  Format-List CPUScore,MemoryScore,DiskScore,GraphicsScore,D3DScore,WinSPRLevel

SSD write without admin: writes 1 GB, measures sequential write + uncached read (deletion uses [IO.File]::Delete because some environments block Remove-Item on system paths):

$SizeMB = 1024; $path = "$env:TEMP\__disk_bench.bin"
$buf = New-Object byte[] (1MB); (New-Object Random).NextBytes($buf)
$fs=[IO.File]::Create($path); $sw=[Diagnostics.Stopwatch]::StartNew()
for($i=0;$i -lt $SizeMB;$i++){$fs.Write($buf,0,1MB)}; $fs.Flush($true); $sw.Stop(); $fs.Close()
$w=[math]::Round($SizeMB/$sw.Elapsed.TotalSeconds,1)
$align=4096; $NO=0x20000000; $over=New-Object byte[] (1MB+$align)
$h=[Runtime.InteropServices.GCHandle]::Alloc($over,'Pinned')
$b=$h.AddrOfPinnedObject().ToInt64(); $o=[int]((($align-($b%$align))%$align))
$fs=New-Object IO.FileStream($path,[IO.FileMode]::Open,[IO.FileAccess]::Read,[IO.FileShare]::None,1MB,[IO.FileOptions]$NO)
$sw2=[Diagnostics.Stopwatch]::StartNew(); $r=0
while($r -lt $SizeMB*1MB){$n=$fs.Read($over,$o,1MB); if($n -le 0){break}; $r+=$n}
$sw2.Stop(); $fs.Close(); $h.Free()
$rd=[math]::Round($SizeMB/$sw2.Elapsed.TotalSeconds,1)
[System.IO.File]::Delete($path)
"SSD: Write $w MB/s | Read (uncached, queue 1) $rd MB/s"

Note: shallow-queue read under-states NVMe drives; for peak sequential read refer to winsat disk (it uses deep I/O queues).

Sustained CPU (Cinebench R23) + thermal stress

• Cinebench R23 → CPU (Multi Core) button → Start (runs ~10 min: CPU benchmark + thermal stress in one).
• During the test, monitor temperature / power / clock with HWMonitor (CPUID) or HWiNFO (CPU Temperatures and Powers sections).
• Live CPU clock via PowerShell (language-independent, no admin):

$pi = Get-CimInstance Win32_PerfFormattedData_Counters_ProcessorInformation -Filter "Name='_Total'"
"Effective clock ~ $([math]::Round(1.6 * $pi.PercentProcessorPerformance / 100, 2)) GHz (base 1.6)"

• Power source and active plan:

(Get-CimInstance Win32_Battery).BatteryStatus   # 2 = mains (AC), 1 = battery
powercfg /getactivescheme

Full support scripts

usb-bench.ps1 — measures a USB stick’s speed (write + uncached read):

param([int]$SizeMB = 200)
$ErrorActionPreference = 'Stop'
 
$usbDisk = Get-Disk | Where-Object BusType -eq 'USB' | Select-Object -First 1
if (-not $usbDisk) { Write-Host "No USB disk detected." -ForegroundColor Red; return }
$part = Get-Partition -DiskNumber $usbDisk.Number | Where-Object DriveLetter | Select-Object -First 1
if (-not $part) { Write-Host "USB disk has no drive letter." -ForegroundColor Red; return }
$dl = $part.DriveLetter
$path = "${dl}:\__bench_tmp.bin"
 
$buf = New-Object byte[] (1MB); (New-Object Random).NextBytes($buf)
 
# WRITE (with flush to disk)
$fs = [System.IO.File]::Create($path)
$sw = [Diagnostics.Stopwatch]::StartNew()
for ($i=0; $i -lt $SizeMB; $i++) { $fs.Write($buf,0,1MB) }
$fs.Flush($true); $sw.Stop(); $fs.Close()
$writeMBps = [math]::Round($SizeMB / $sw.Elapsed.TotalSeconds, 1)
 
# READ uncached (FILE_FLAG_NO_BUFFERING = 0x20000000)
$align = 4096; $NO_BUF = 0x20000000
$over = New-Object byte[] (1MB + $align)
$h = [Runtime.InteropServices.GCHandle]::Alloc($over, 'Pinned'); $uncached = $true
try {
    $base = $h.AddrOfPinnedObject().ToInt64()
    $off  = [int]((($align - ($base % $align)) % $align))
    $fs = New-Object IO.FileStream($path,[IO.FileMode]::Open,[IO.FileAccess]::Read,[IO.FileShare]::None,1MB,[IO.FileOptions]$NO_BUF)
    $sw2 = [Diagnostics.Stopwatch]::StartNew(); $read = 0
    while ($read -lt $SizeMB*1MB) { $n = $fs.Read($over,$off,1MB); if ($n -le 0){break}; $read += $n }
    $sw2.Stop(); $fs.Close()
    $readMBps = [math]::Round($SizeMB / $sw2.Elapsed.TotalSeconds, 1)
} catch {
    $uncached = $false
    $sw2 = [Diagnostics.Stopwatch]::StartNew()
    $fs = [System.IO.File]::OpenRead($path); $rb = New-Object byte[] (4MB)
    while (($n = $fs.Read($rb,0,$rb.Length)) -gt 0) {}
    $sw2.Stop(); $fs.Close()
    $readMBps = [math]::Round($SizeMB / $sw2.Elapsed.TotalSeconds, 1)
} finally { $h.Free() }
Remove-Item $path -Force
 
Write-Host ("Drive {0}:  Write {1} MB/s  |  Read {2} MB/s" -f $dl,$writeMBps,$readMBps) -ForegroundColor Green

port-test.ps1 — before/after device comparison (USB, disks, monitors, audio, network). Save the file and run it once to create the baseline, then again after plugging the device in:

param([string]$Label = "")
$ErrorActionPreference = 'SilentlyContinue'
$stateFile = Join-Path $PSScriptRoot 'port-baseline.json'
 
function Get-State {
    $usb   = Get-PnpDevice -PresentOnly -Class USB           | Select-Object FriendlyName, InstanceId
    $disk  = Get-Disk                                        | Select-Object Number, FriendlyName, SerialNumber, BusType
    $mon   = Get-PnpDevice -PresentOnly -Class Monitor       | Select-Object FriendlyName, InstanceId
    $audio = Get-PnpDevice -PresentOnly -Class AudioEndpoint | Select-Object FriendlyName, InstanceId
    $net   = Get-NetAdapter | Select-Object Name, Status, LinkSpeed, MacAddress
    [pscustomobject]@{ usb=@($usb); disk=@($disk); mon=@($mon); audio=@($audio); net=@($net) }
}
$new = Get-State
if (-not (Test-Path $stateFile)) {
    $new | ConvertTo-Json -Depth 6 | Set-Content $stateFile -Encoding UTF8
    Write-Host "BASELINE saved. Plug in a device and run again." -ForegroundColor Cyan; return
}
$old = Get-Content $stateFile -Raw | ConvertFrom-Json
Write-Host ("===== PORT TEST: {0} =====" -f $Label) -ForegroundColor Yellow
foreach ($cat in 'usb','mon','audio') {
    $added = @($new.$cat) | Where-Object { $_.InstanceId -and ($_.InstanceId -notin (@($old.$cat).InstanceId)) }
    foreach ($a in $added) { Write-Host ("  [+] {0}: {1}" -f $cat.ToUpper(), $a.FriendlyName) -ForegroundColor Green }
}
$addedDisk = @($new.disk) | Where-Object { $_.SerialNumber -and ($_.SerialNumber -notin (@($old.disk).SerialNumber)) }
foreach ($d in $addedDisk) { Write-Host ("  [+] STORAGE: {0} | Bus={1}" -f $d.FriendlyName,$d.BusType) -ForegroundColor Green }
foreach ($n in @($new.net)) {
    $o = @($old.net) | Where-Object { $_.MacAddress -eq $n.MacAddress } | Select-Object -First 1
    if ($o -and $o.Status -ne $n.Status) { Write-Host ("  [~] RETE: {0} {1}->{2} ({3})" -f $n.Name,$o.Status,$n.Status,$n.LinkSpeed) -ForegroundColor Green }
}
$new | ConvertTo-Json -Depth 6 | Set-Content $stateFile -Encoding UTF8

---

TIP

Reproducibility. Store the resulting .md alongside the battery-report.html, the CrystalDiskMark screenshot and the HWiNFO log in a folder named after the unit (e.g. audit-2026-06-29-thinkpad-l14/). The next time you touch the machine, a 5-minute re-run of the same commands tells you if anything has degraded.