GPU-Z 2.69 Drops Support for RTX 5090 D V2, Pro Blackwell, and China’s Moore Threads S30

TechPowerUp’s GPU-Z has a long history of spoiling hardware announcements before manufacturers are ready, and version 2.69 continues that tradition with a support list that reads like a roadmap leak. The update — released in early 2026 — adds detection for over 18 new GPU variants spanning NVIDIA’s consumer and professional lines, AMD’s budget Radeon tier, Intel’s embedded Arc series, and China’s Moore Threads S30, a domestically designed GPU that most Western users have never encountered. For a lightweight diagnostic utility that hasn’t changed its core UI since roughly 2007, GPU-Z remains one of the most reliable leading indicators of what’s actually shipping (or about to ship) in the GPU market.

The headline additions fall into four distinct categories, and each one tells a different story about where the graphics industry is headed in the second half of 2026. NVIDIA dominates the changelog with consumer, professional, and region-specific variants; AMD fills in budget gaps that have been empty for months; Intel continues its quiet embedded push; and Moore Threads signals that China’s domestic GPU effort hasn’t stalled despite U.S. export restrictions tightening through 2025.

RTX 5090 D V2 Confirms a Second China-Market Revision

The most eyebrow-raising entry in GPU-Z 2.69’s support list is the GeForce RTX 5090 D V2 — a designation that confirms NVIDIA has already revised its China-export variant of the flagship Blackwell consumer GPU. The original RTX 5090 D shipped as a compliance-trimmed version designed to meet U.S. export control thresholds, with reduced CUDA core counts and memory bandwidth caps compared to the unrestricted RTX 5090 sold in other markets. A “V2” revision this quickly after launch typically means one of two things: either NVIDIA adjusted the hardware to comply with updated export rules from the Bureau of Industry and Security, or the company optimized the silicon binning process to improve yields on the restricted SKU.

Having tracked NVIDIA’s China-market GPU variants since the original RTX 4090 D appeared in late 2023, the V2 pattern feels more regulatory than commercial. Washington tightened compute density thresholds twice during 2025, and each adjustment forced NVIDIA to re-spec its China offerings — a rolling compliance headache that GPU-Z’s database quietly documents with each new entry. The RTX 3050 A Mobile also appearing in this update follows the same logic: a region- or OEM-specific variant that exists because global GPU distribution has become a regulatory maze rather than a simple supply chain.

NVIDIA’s Full RTX Pro Blackwell Workstation Stack Revealed

GPU-Z 2.69 adds support for eight RTX Pro Blackwell variants, and the sheer breadth of the lineup — from the RTX Pro Blackwell 5000 down to the RTX Pro Blackwell 500 — confirms that NVIDIA’s professional GPU rebrand is far more extensive than the company has publicly detailed. The naming scheme abandons the old Quadro-era numbering (RTX A4000, A5000) in favor of a tiered approach that mirrors the consumer GeForce stack: 5000 for flagship workstation use, 4500/4000 for mid-range CAD and visualization, 2000/1000 for entry-level professional certification, and 500 for basic multi-display office configurations.

What caught my attention is the inclusion of both SFF and Embedded variants alongside standard desktop and mobile SKUs. The RTX Pro Blackwell 4000 SFF targets small-form-factor workstations — a growing segment driven by studios and engineering firms replacing tower machines with compact nodes — while the RTX Pro Blackwell 5000 Embedded aims at industrial and medical imaging applications where GPU compute runs inside sealed, fanless enclosures. NVIDIA hasn’t published official specs for most of these models, so GPU-Z’s device ID entries represent the first concrete evidence that the full product matrix is locked and silicon is reaching validation stages. For IT departments planning workstation refresh cycles, this list is more useful than anything NVIDIA’s professional sales team has shared publicly.

Moore Threads S30 and the Quiet Expansion of Chinese GPUs

The Moore Threads S30 addition deserves more attention than it’ll get in most English-language coverage. Founded in 2020 by former NVIDIA China chief Zhang Jianzhong, Moore Threads has been building x86-compatible GPUs designed specifically for the Chinese domestic market — targeting government, enterprise, and data center buyers who need alternatives to restricted NVIDIA and AMD hardware. The S30 appears to be a successor or variant of the company’s MTT S80 series, which offered roughly GTX 1060-class performance when it launched in 2022 alongside a custom MUSA architecture that intentionally avoided dependencies on Western IP.

GPU-Z adding Moore Threads support signals that these GPUs are reaching a maturity level where standard Western diagnostic tools need to account for them — either because evaluation units are circulating outside China, or because the installed base has grown large enough that TechPowerUp’s user telemetry flagged the gap. Neither scenario is trivial. Four years ago, the idea that a Chinese-designed GPU would appear in GPU-Z’s database alongside GeForce and Radeon entries would have seemed premature; today it reflects a supply chain bifurcation that’s reshaping the global semiconductor landscape in real time.

AMD and Intel Round Out the Budget and Embedded Tiers

On the AMD side, GPU-Z 2.69 adds support for the Radeon RX 7400, Radeon 820M, Radeon Pro W6600X, Radeon Pro W7800, and Radeon Pro V710. The RX 7400 fills a gap that AMD’s consumer lineup has desperately needed — a true sub-$200 RDNA 3 card for budget builders who can’t justify the RX 7600’s $250+ street price. The Radeon Pro additions suggest AMD continues pushing into the professional visualization market, though the W6600X’s appearance (based on older RDNA 2 architecture) indicates that some workstation OEMs still need driver-certified options for legacy compute workflows.

Intel’s entries — the Arc A380E and Arc A310E — carry the “E” suffix that designates embedded and industrial variants, joining Intel’s quiet push into digital signage, kiosk, and edge computing applications where discrete GPU capability matters but gaming performance doesn’t. These aren’t the Arc GPUs that Intel wants headlines about, but they represent steady revenue from design wins that don’t depend on consumer sentiment or Steam hardware surveys.

Beyond new GPU support, version 2.69 fixes a shader count calculation bug affecting certain AMD and Intel GPUs — a correction that matters for anyone using GPU-Z to verify hardware specifications before purchasing used cards, where misreported shader counts could mask a partially disabled die. Updated device IDs for NVIDIA’s older Quadro P1000, Quadro P2000, and RTX 6000 round out the maintenance work, along

Sources: TechPowerUp GPU-Z, WCCFtech