Alienware 18 Area-51 AA18250: RAID0 volume under Intel VMD freezes for 60 s
(Intel erratum ARL004, RST 20.2.13.1038 A06)
System
- Alienware 18 Area-51 AA18250, Intel Core Ultra 9 275HX, NVIDIA RTX 5090, BIOS 2.9.0 (current), Windows 11 Pro 26H2 (build 26300).
- Storage: three WD_BLACK SN8100 2 TB M.2 2280 PCIe Gen5 NVMe SSDs (TLC, firmware 830ZRR0A), installed by me in the three M.2 slots. One RAID0 volume (128 KB stripe) created with the Intel Rapid Storage Technology utility built into the BIOS, under the Intel VMD controller (PCI\VEN_8086&DEV_AD0B&SUBSYS_0CCD1028), BIOS setting SATA/NVMe Operation = "RAID On". The OS runs on this volume.
- Driver: Intel Rapid Storage Technology Driver 20.2.13.1038, A06 — Dell package 78WWJ, the latest Dell-validated RST driver for this model. Its driver files (iaStorVD.sys, iaStorVD.inf, iaStorVD.cat, RstMwService.exe, RstMwEventLogMsg.dll) are byte-for-byte identical (SHA-256) to the 20.2.13.1038 package that Windows Update had already installed on this system on 10 July 2026, which the Microsoft Update Catalog lists as published by Intel Corporation ("Intel Corporation SCSIAdapter Driver Update (20.2.13.1038)"); when the Dell installer ran later, Windows reported the package as already imported. Dell's INF contains no Dell- or OEM-specific entries. The same holds for the previous version: Dell's 20.2.6.1025 package and Intel's own SetupRST 20.2.6.1025.3 contain the same iaStorVD.sys code and an identical INF, differing only in the WHQL signature. In other words, Dell ships Intel's driver unmodified.
Problem
Since the first day of use (20 March 2026, before any tuning of BIOS, registry or power settings) the RAID0 volume periodically freezes for about 60 seconds: every I/O stops, then all outstanding requests complete in the same second. Windows logs bursts of Microsoft-Windows-Ntfs/Operational event 147 ("IO operation took more than 30000 ms"; latencies mostly 30–65 s, median about 55 s, maximum about 94 s) and sometimes iaStorVD event 129 ("Reset to device \Device\RaidPort0"). Frequency: about 0.2–0.4 stalls per hour of active use, plus a stall within 1–3 minutes after most boots (35 of the first 59 boots). My log holds more than 7,000 event-147 records. In most stalls (109 of 183 analysed) the second stuck request starts within one second of the first, on a different RAID member, and in 162 of them all requests are released within 2 seconds: one missed completion freezes the whole volume. Any I/O type is affected (paging reads, writes, flushes).
What has been ruled out (no effect on the stalls unless stated otherwise)
- Drives: SMART of all three SSDs clean (no media errors, spare 100%, 0% used, error counters static); extended self-test clean; firmware 830ZRR0A is the latest per SanDisk Dashboard; stalls start on each of the three drives about equally often (189, 162 and 167 of 518 first-stuck-request samples), so no single drive is at fault; one drive removed and reseated after a full power drain; chkdsk C: /f clean.
- BIOS: Load Defaults; BIOS 2.5.0, 2.7.0 and 2.9.0 (updated and downgraded with Dell installers); Adaptive Optimization, Adaptive C-States for discrete graphics and Block Sleep on and off; SMART Reporting off and on (no errors at POST); Bluetooth, LAN and Wake on LAN off; hard reset (AC off, power held 30 s); F12 diagnostics, all tests passed.
- PCIe link: SSD link speed Gen4 and Gen5; ASPM and L1 substates disabled and enabled on the SSD root ports; link speed downgrade off and on.
- RST driver versions: 20.2.4.1019, 20.2.6.1025 and 20.2.13.1038 (Dell A06) all behave the same.
- Driver and storage registry settings: iaStorVD Parameters\Device NvmeApstEnabled=0, PcieRtd3Enabled=0, UnitAdaptiveRtd3Disabled=1, PowerSavingsOff=1, StorageD3Enable=0; StorPort EnableIdlePowerManagement=0 on the VMD controller; Disk\TimeOutValue lowered to 5 s (a port reset every 2–6 s during a stall, but the stall still lasted about 60 s); GPU TdrDelay/TdrDdiDelay=10; right now they are returned to default values. Legacy INTx instead of MSI is not possible: the VMD controller has no interrupt pin.
- Windows power: Modern Standby disabled (PlatformAoAcOverride=0; 7 stalls in the next 26 hours with no sleep transitions); Fast Startup off; disk idle timeout changed. Stalls occur during sleep, seconds after wake, mid-session and even before logon, so sleep/resume is not the trigger.
- Windows software and services: Windows Search off; Dell SupportAssist, Dell TechHub Instrumentation and Killer network services off; monitoring-tool drive polling off; background sync reduced; scheduled Optimize Drives/TRIM never coincides with a stall; the storage filter stack contains only Microsoft filters plus iaStorVD; no service starts or stops in the 15 s before any of 75 checked stall onsets.
- NVIDIA: Control Panel power management mode "Prefer maximum performance" and default.
- SanDisk Dashboard: Game Mode on and off.
Most of these were tested separately and some in combination. The stalls are in the logs from the first day of use, before any of these settings were changed.
Root cause
The behaviour matches Intel erratum ARL004 "MSI From VMD-Owned Device May Pass Memory Write" (Intel Core Ultra Processors Series 2 Specification Update, doc 834774). In RAID 0/1 mode the VMD controller may signal an MSI before the NVMe device's completion-queue write reaches memory. The driver then reads stale completion data, and the I/O hangs until the driver's 60-second timeout. Intel's documented handling is a dummy register read of the MSI initiator in the VMD interrupt handler, i.e. a change inside iaStorVD.sys. The mainline Linux kernel now applies exactly this workaround to the same controller (8086:AD0B): commit 2c88c2dc1852 "PCI: vmd: Flush DMA writes before handling MSI on Arrow Lake", which cites ARL004 and is marked for stable backports (https://github.com/torvalds/linux/commit/2c88c2dc18528489b403fe95741f5731e292f7ba). Ubuntu bug 2166325 lists other Arrow Lake laptops with the same stalls (MSI Vector 17 HX AI, MSI Raider 16 HX, Acer Predator PHN16-73, Dell Pro Max 16 Plus). As far as I can tell, the Windows RST driver shipped by Dell (and by Intel) does not contain this workaround.
Intel case
Intel case 06987682 is open. Intel support first replied that OEMs may customise drivers and directed me to Dell. After I showed that the installed driver is Dell's unmodified package, they asked for the Dell case number and Dell's findings "to correlate and facilitate further escalation if required". Intel case is still under review.
See my next post reply below.. (description limit sorry)
Dell case
I opened Dell Support Request with this information. It was closed with this reason: "the system appears to have a storage configuration that is not part of the original supported factory configuration. Due to this, the reported issue falls outside the scope of Dell technical support." A closed case cannot be replied to.
<To protect you, your private information was removed from public view. All private data was saved to your private Case. DELL-Admin>
Why I believe the closure reason does not apply
1. RAID 0 on three NVMe drives is a configuration Dell sells for this model. The Dell store configurator for the AA18250 offers "12TB M.2 2280 RAID 0 (3 x 4TB) TLC NVMe M.2 PCIe Gen5 SSD" as well as 2 x 2TB and 2 x 4TB RAID 0 options (https://www.dell.com/en-uk/shop/gaming-and-games/alienware-18-area-51-gaming-laptop/spd/alienware-area-51-aa18250-gaming-laptop/aa18250_cto2). My array has the same layout: three M.2 2280 TLC PCIe Gen5 NVMe SSDs in RAID0 under Intel VMD.
2. The defect does not depend on the SSD model. ARL004 is in the VMD controller of the Core Ultra 9 275HX supplied in this system, and the Linux workaround applies to the controller (8086:AD0B) whatever drives are attached; Ubuntu bug 2166325 reports the same stalls on Arrow Lake laptops from several vendors, including a Dell Pro Max 16 Plus. The workaround belongs in the Intel RST VMD driver that Dell delivers and validates for this model (package 78WWJ lists the AA18250 as a supported system). As shown above, no single drive is at fault. I therefore expect the same behaviour with Dell-supplied SSDs in RAID0, but I cannot test this myself.
3. The drives themselves are not exotic. Dell's own driver page for the AA18250 carries firmware for SanDisk/Western Digital OEM SSDs (package Y0KG7, "Sandisk SN8000s SED Solid State Drive Firmware Update", 21 Oct 2026, lists the Alienware 18 Area-51 AA18250 and 16 Area-51 AA16250 as compatible systems), which suggests Dell itself uses SanDisk/Western Digital drives in this model family. The WD_BLACK SN8100 is the same manufacturer's retail PCIe Gen5 TLC drive in the same M.2 2280 form factor.
4. RAID is a documented feature that customers can use on this model. "RAID On" is a standard option of the SATA/NVMe Operation setting in the AA18250 BIOS, and the RAID volume is created with the Intel RST utility built into Dell's BIOS. Dell publishes general guidance for Intel RAID 0/1 on Dell computers: Knowledge Base article 000132392, "How to Configure Intel Based RAID on a Dell Computer" (https://www.dell.com/support/kbdoc/en-us/000132392/how-to-configure-raid-on-a-dell-pc). Replacing the M.2 SSDs is a customer-serviceable procedure in the Owner's Manual (Storage section).
What I am asking
- DELL-Cares: please pass case to the engineering / L2 team. The Service Tag is available by private message. I can provide the System and NTFS/Operational event logs and the Intel SSU report on request but you can also find them inside my Dell case.
- If it helps, the issue should be reproducible on an AA18250 with three Dell-supplied SSDs in RAID0 under VMD, Windows 11 and RST 20.2.13.1038 A06: reboot about 10 times and check Microsoft-Windows-Ntfs/Operational for event 147 in the first 3 minutes after each boot.
- Open an engineering request to Intel for an RST VMD driver build with the ARL004 workaround for DEV_AD0B, released through Dell's driver page for the AA18250 and AA16250. I am available to test a pre-release build.
- Intel asked for the Dell case number and Dell's findings so it can consider escalating case 06987682. Any written answer from Dell would let me pass that on.
<to protect your privacy, all private information was removed from this public post. dell-admin>
Those SSD are all very-fast natively, so RAID-0 is not required.
RAID-1 is replaced with a daily drive-imaging backup.
No Intel-RST and No RAIDs.
(edited)
Dell Rockstar
Microsoft Windows and Apple iOS Developer (Retired) - Like many of you, I can appreciate a good game-engine. - I answer questions here, but I'm not a Dell employee. - Consider giving posts you like a "thumbs-up" - Posting models-numbers and software versions speeds trouble-shooting. - Click "Mark as Accepted Answer" on any post that answers your question best.
Thank you, Tesla1856. Running the three SSDs as separate drives would indeed avoid the problem, and for a personal setup that is a reasonable choice. In my case the OS lives on the RAID0 volume, so it would mean reinstalling Windows, but that is not the point of this thread.
The point is that this is a defect in a driver Dell publishes for this model (package 78WWJ, RST 20.2.13.1038 A06), in a configuration Dell sells for this model (the configurator offers 3 x 4 TB RAID 0). The root cause is a documented erratum in the Core Ultra 9 275HX's VMD controller (ARL004), and the documented workaround is a small change in the VMD driver, which Linux already carries for this very controller. Avoiding RAID on my machine leaves every other owner who bought or configured RAID 0/1 with the same 60-second freezes, so I am asking Dell to raise it with Intel.
On the elevenforum link: that 2023 post is about the legacy RST management application and Optane, not about the VMD RAID driver on Core Ultra systems. The driver on this laptop is the one Dell publishes for this Windows 11 model and Intel publishes on Windows Update ("Intel Corporation SCSIAdapter Driver Update (20.2.13.1038)", April 2026), so this is a supported configuration.
The RAID volume is GPT and NTFS; SMART and self-tests on all three drives are clean.
DELL-Cares, I would be grateful if you could pass case to the L2/engineering team. Service Tag by private message.
<to protect your privacy, all private information was removed from this public post. dell-admin>
Update, 11 November 2026: Intel's written answer (<To protect you, your private information was removed from public view. All private data was saved to your private Case. DELL-Admin>)
I asked Intel Customer Support to put three questions to the RST / VMD engineering team. Their answers:
Does iaStorVD 20.2.13 (or 20.4) implement the ARL004 workaround for DEV_AD0B in RAID 0 mode?
"No. Current production releases of iaStorVD.sys (including 20.2.13.1038 and the 20.4.0.x baseline) do not incorporate the non-posted dummy register read in the Windows StorPort interrupt service routine (ISR / DPC) path for DEV_AD0B."
Is there any supported mitigation on Windows short of leaving RAID mode?
"Currently, there are no known Windows-level driver parameters or registry toggles that can mitigate ARL004 while remaining on Intel VMD RAID."
Is a driver build with the workaround planned?
Intel does not comment on future releases, and adds: "Dell is responsible for testing and validating drivers on their hardware, we only provide generic drivers ..."
What this means
1. 20.2.13.1038 is the driver in Dell package 78WWJ (A06), the latest RST package Dell publishes for the AA18250. Intel confirms that this version lacks the ARL004 workaround, and that Intel's own newer 20.4.0.x release lacks it too.
2. The missing workaround is not SSD-specific. The erratum is in the processor's VMD controller, and the workaround belongs in the driver. Dell sells the AA18250 with factory RAID 0 (2 x 2 TB, 2 x 4 TB and 3 x 4 TB) on the same controller and the same driver. I cannot test Dell-supplied SSDs myself.
3. Intel knows of no Windows driver or registry setting that mitigates it under VMD RAID. Short of leaving RAID mode, which means reinstalling Windows and giving up RAID 0, the fix has to come in the driver.
4. Intel refers driver validation to Dell, and Dell support closed my case as out of scope. As a customer I cannot take this further.
Request to DELL-Cares / moderators
Please forward this thread and case to the Alienware engineering team and ask them to: (1) request from Intel, through Dell's OEM channel, an RST VMD driver build with the ARL004 workaround for DEV_AD0B; (2) validate and publish it for the AA18250 (and the AA16250, if affected). I can send the Service Tag and Intel's full reply by private message.
Other AA18250 owners with RAID 0 or RAID 1 (BIOS: SATA/NVMe Operation = "RAID On"), especially factory RAID 0: in Event Viewer open Applications and Services Logs > Microsoft > Windows > Ntfs > Operational and look for warning event 147 (an IO operation that took more than 30,000 ms). If you find several of them logged in the same second, please reply with your RAID level and whether the SSDs were factory-installed, and click "I have this problem too".
<to protect your privacy, all private information was removed from this public post. dell-admin>
@DELL-Daniel V, could you please look at this thread? Intel has confirmed in writing that RST driver 20.2.13.1038, which is the version Dell ships in package 78WWJ for the AA18250, lacks the workaround for processor erratum ARL004 (the answers are quoted in my post above). Dell sells this model with factory RAID 0 on the same controller and the same driver. Could you notify the platform team liaison for the Alienware 18 Area-51? I can send the Service Tag and Intel's full reply by private message. I was not able to send you one from my account, so please message me first.
Ayni
6 Posts
•
22 Points
4
1
Posted November 6th, 2026 17:45
Dell case
I opened Dell Support Request with this information. It was closed with this reason: "the system appears to have a storage configuration that is not part of the original supported factory configuration. Due to this, the reported issue falls outside the scope of Dell technical support." A closed case cannot be replied to.
<To protect you, your private information was removed from public view. All private data was saved to your private Case. DELL-Admin>
Why I believe the closure reason does not apply
1. RAID 0 on three NVMe drives is a configuration Dell sells for this model. The Dell store configurator for the AA18250 offers "12TB M.2 2280 RAID 0 (3 x 4TB) TLC NVMe M.2 PCIe Gen5 SSD" as well as 2 x 2TB and 2 x 4TB RAID 0 options (https://www.dell.com/en-uk/shop/gaming-and-games/alienware-18-area-51-gaming-laptop/spd/alienware-area-51-aa18250-gaming-laptop/aa18250_cto2). My array has the same layout: three M.2 2280 TLC PCIe Gen5 NVMe SSDs in RAID0 under Intel VMD.
2. The defect does not depend on the SSD model. ARL004 is in the VMD controller of the Core Ultra 9 275HX supplied in this system, and the Linux workaround applies to the controller (8086:AD0B) whatever drives are attached; Ubuntu bug 2166325 reports the same stalls on Arrow Lake laptops from several vendors, including a Dell Pro Max 16 Plus. The workaround belongs in the Intel RST VMD driver that Dell delivers and validates for this model (package 78WWJ lists the AA18250 as a supported system). As shown above, no single drive is at fault. I therefore expect the same behaviour with Dell-supplied SSDs in RAID0, but I cannot test this myself.
3. The drives themselves are not exotic. Dell's own driver page for the AA18250 carries firmware for SanDisk/Western Digital OEM SSDs (package Y0KG7, "Sandisk SN8000s SED Solid State Drive Firmware Update", 21 Oct 2026, lists the Alienware 18 Area-51 AA18250 and 16 Area-51 AA16250 as compatible systems), which suggests Dell itself uses SanDisk/Western Digital drives in this model family. The WD_BLACK SN8100 is the same manufacturer's retail PCIe Gen5 TLC drive in the same M.2 2280 form factor.
4. RAID is a documented feature that customers can use on this model. "RAID On" is a standard option of the SATA/NVMe Operation setting in the AA18250 BIOS, and the RAID volume is created with the Intel RST utility built into Dell's BIOS. Dell publishes general guidance for Intel RAID 0/1 on Dell computers: Knowledge Base article 000132392, "How to Configure Intel Based RAID on a Dell Computer" (https://www.dell.com/support/kbdoc/en-us/000132392/how-to-configure-raid-on-a-dell-pc). Replacing the M.2 SSDs is a customer-serviceable procedure in the Owner's Manual (Storage section).
What I am asking
- DELL-Cares: please pass case to the engineering / L2 team. The Service Tag is available by private message. I can provide the System and NTFS/Operational event logs and the Intel SSU report on request but you can also find them inside my Dell case.
- If it helps, the issue should be reproducible on an AA18250 with three Dell-supplied SSDs in RAID0 under VMD, Windows 11 and RST 20.2.13.1038 A06: reboot about 10 times and check Microsoft-Windows-Ntfs/Operational for event 147 in the first 3 minutes after each boot.
- Open an engineering request to Intel for an RST VMD driver build with the ARL004 workaround for DEV_AD0B, released through Dell's driver page for the AA18250 and AA16250. I am available to test a pre-release build.
- Intel asked for the Dell case number and Dell's findings so it can consider escalating case 06987682. Any written answer from Dell would let me pass that on.
<to protect your privacy, all private information was removed from this public post. dell-admin>
(edited)