我們接受加密貨幣付款 — 讓您持有的幣種也能買到真正的硬體設備。以加密貨幣付款,直接購入真正的硬體設備。 全球訂單滿 US$399,即享免費保險運送。簡約外箱,不印任何品牌標誌,從德國出貨。訂單滿 US$399,即享免費保險運送。 結帳一開始,您的匯率即鎖定 30 分鐘。匯率鎖定 30 分鐘。 每件商品皆原廠密封、序號經過核對,並享有 24 個月保固。原廠密封,保固 24 個月。

追蹤清單 登入

Lenovo

Lenovo ThinkStation PGX 30KL0003US AI edge system 128GB 4TB

SKU CH-H3MS-QNYGK MPN 30KL0003US

Compact ARM mini PC with 20-core CPU, 128GB LPDDR5X unified memory, 4TB PCIe Gen4 SSD, Wi-Fi 7, ConnectX-7 Smart NIC, and NVIDIA DGX OS in a 240W chassis

  • TypeMini PC
  • CPU Type20 core Arm, 10 Cortex-X925 + 10 Cortex-A725 Arm
  • Installed Memory Size128GB
  • Memory TypeLPDDR5x, coherent unified system memory
  • Memory FeaturesTensor Performance: 1 PFLOP Memory Interface: 256-bit Memory
  • SSD (Installed)4TB SSD M.2 2242 PCIe Gen4 TLC Opal
  • 20-core Arm CPU delivers 1 PFLOP tensor performance
  • 128GB LPDDR5X unified memory provides 273 GB/s bandwidth
  • 4TB PCIe Gen4 SSD offers high-speed local storage
  • Wi-Fi 7 and ConnectX-7 Smart NIC enable fast networking
  • Compact 1.98" x 5.91" x 5.91" design fits edge deployments

Compact ARM AI workstation

The Lenovo ThinkStation PGX 30KL0003US is a mini PC built around a 20-core ARM processor with 10 Cortex-X925 and 10 Cortex-A725 cores. It ships with 128GB of coherent unified LPDDR5X memory delivering 273 GB/s bandwidth and a 4TB M.2 2242 PCIe Gen4 TLC Opal SSD. The system runs NVIDIA DGX OS and draws 240 Watts.

Replaces discrete GPU servers for edge inference

This unit replaces rack-mounted GPU servers when the workload fits within 128GB of unified memory and tensor performance of 1 PFLOP. The jump is worth it when you need Wi-Fi 7, Bluetooth 5.3, and a ConnectX-7 Smart NIC in a 1.98 x 5.91 x 5.91 inch chassis weighing 2.64 lbs instead of a full-depth server. HDMI 2.1a supports 4K at 60Hz for local display.

Check power, mounting, and OS compatibility

Verify that a 240W power supply meets your site requirements and that the mini PC dimensions fit your mounting location. Ensure your workflow runs on ARM architecture under NVIDIA DGX OS. Confirm that four USB-C 20Gbps ports plus two USB-C to USB-A dongles cover your peripheral needs and that the single RJ45 Ethernet port satisfies wired networking demands.

商品特色

  • 20-core Arm CPU delivers 1 PFLOP tensor performance
  • 128GB LPDDR5X unified memory provides 273 GB/s bandwidth
  • 4TB PCIe Gen4 SSD offers high-speed local storage
  • Wi-Fi 7 and ConnectX-7 Smart NIC enable fast networking
  • Compact 1.98" x 5.91" x 5.91" design fits edge deployments

商品規格

BrandLenovo
SeriesThinkStation PGX
Model30KL0003US
TypeMini PC
CPU Type20 core Arm, 10 Cortex-X925 + 10 Cortex-A725 Arm
Installed Memory Size128GB
Memory TypeLPDDR5x, coherent unified system memory
Memory FeaturesTensor Performance: 1 PFLOP
Memory Interface: 256-bit
Memory Bandwidth: 273 GB/s
SSD (Installed)4TB SSD M.2 2242 PCIe Gen4 TLC Opal
WLAN802.11be Wireless LAN
WiFi GenerationWi-Fi 7
BluetoothBluetooth 5.3
Other ports4 x USB-C (USB 20Gbps)
HDMI 2.1a (supports resolution up to 4K@60Hz)
ConnectX-7 Smart NIC
Ethernet (RJ45)
Operating SystemNVIDIA DGX OS
Power Characteristics240 Watts
Dimensions1.98" x 5.91" x 5.91"
Weight2.64 lbs.
Package ContentsLenovo ThinkStation PGX
240W power supply unit & power cord
Quick User Guide
2x USB-C to USB-A dongle
出貨重量13.15 kg
包裝尺寸584 × 508 × 330 mm

商品問答

What clearance does the 1.98 inch high mini PC need for the 240W power supply and HDMI 2.1a cable?

The chassis measures 1.98 by 5.91 by 5.91 inches and weighs 2.64 lbs, so allow at least 2 inches above the unit for the 240W brick and cable bend radius behind the HDMI 2.1a port.

Which step is often missed when installing the NVIDIA DGX OS on the 128GB LPDDR5X unified memory system?

After connecting the 240W power supply and the 4TB M.2 2242 PCIe Gen4 drive, the first boot must complete the DGX OS initialisation before attaching any USB-C 20Gbps peripherals or the ConnectX-7 Smart NIC.

When should the 4TB M.2 2242 PCIe Gen4 TLC Opal SSD be replaced in this 240W AI edge system?

Replace the 4TB M.2 2242 drive when SMART health drops below 80 percent or sustained write throughput falls under the 273 GB/s memory bandwidth ceiling during Tensor workloads.

同類商品

顧客也比較過

同一分類,同一套結帳流程 — 支援八種加密貨幣,匯率鎖定 30 分鐘。