We take cryptocurrency — so the coins you hold can buy real hardware.Pay in crypto, spend it on real hardware. Free insured shipping worldwide over $399. Plain, unbranded boxes, sent from Germany.Free insured shipping over $399. Your rate is locked for 30 minutes once checkout opens.Rate locked 30 minutes. Every item sealed, serial-checked and covered by a 24-month warranty.Sealed, 24-month warranty.

Saved items Sign in

AMD

AMD J0H11A FirePro S9150 AI accelerator card 16GB GDDR5 512-bit 235W

SKU CH-8G0C-N1AYS MPN J0H11A

Passively cooled AI accelerator with 2816 stream processors, 16GB GDDR5 on a 512-bit bus and 5.07 TFLOPS single-precision performance

  • Graphics processorOpenGL 4.3, OpenCL 1.2
  • Memory capacity16 GB
  • Interface typePCI Express x16
  • Shader units2,816 stream processors (44 compute units)
  • Type of memoryGDDR5
  • Chipset makerAMD
  • 2816 stream processors deliver 5.07 TFLOPS single-precision compute
  • 16GB GDDR5 memory on 512-bit bus provides 320 GB/s bandwidth
  • Passive heatsink enables silent operation in 20 CFM server airflow
  • 53 TFLOPS double-precision with ECC support for accuracy-critical workloads
  • PCIe 3.0 x16 interface at 235W fits dense server deployments

Server AI accelerator with 16 GB GDDR5

The AMD FirePro S9150 is a compute-focused add-in card built for servers that need high-throughput parallel processing. It delivers 5.07 TFLOPS peak single-precision and 2.53 TFLOPS peak double-precision performance through 2,816 stream processors grouped into 44 compute units. The 16 GB GDDR5 frame buffer on a 512-bit interface provides up to 320 GB/s memory bandwidth for large datasets. This card has no display outputs and cannot drive a monitor directly.

PCIe 3.0 x16 slot and 20 CFM airflow required

Installation requires a PCI Express 3.0 x16 slot and a chassis capable of supplying at least 20 CFM of directed airflow across the passive heatsink. The board draws up to 235 W from the slot and auxiliary power connectors, so the platform power budget must accommodate that load. Because cooling is entirely passive, the card frees up internal space that active coolers would occupy but depends on the system fans to maintain stable temperatures. Buyers who need a card with built-in fans or external display connectors should look elsewhere.

Passive heatsink with SMBus thermal reporting

A passive heatsink eliminates fan noise and moving-part wear, relying on the server's forced-air design to dissipate up to 235 W. SMBus temperature reporting at boot lets the management controller verify thermal status before workloads start. AMD PowerTune technology adjusts clocks to stay within the power envelope during sustained compute tasks. The card supports ECC memory externally and OpenCL 1.2, DirectX 11.1 and OpenGL 4.3 APIs for compatible software stacks.

How it compares

Measured against everything we stock in GPUs for AI & Data Centers, at today’s prices. Recalculated every night.

Price $479.99
Cheaper than about 90% of the 36 comparable items (median $5,369.99).
Memory capacity 16 GB
Lower than about 99% of the 12 comparable items (median 79 GB).

New to this? Buying server hardware with crypto — what to check, what it costs to pay, and what is never asked for.

Specifications

Product
Brand nameAMD
Product lineFirepro
Model nameJ0H11A, S9150, 797639-001
Memory
Type of memoryGDDR5
Graphics
Graphics processorOpenGL 4.3, OpenCL 1.2
Memory capacity16 GB
Shader units2,816 stream processors (44 compute units)
Chipset makerAMD
Memory bus width512-bit
DirectX support11.1
OpenGL support4.3
Ports & slots
Interface typePCI Express x16
More details
Floating point formats5.07 TFLOPS peak single precision floating point, 2.53 TFLOPS peak double precision floating point

Questions about this item

What is the 235 W power limit under sustained compute loads?

The card has a maximum power consumption of 235 W; the system must supply this continuously through the PCIe 3.0 x16 slot and any auxiliary connectors.

How much airflow does the passive heatsink need for all-day operation?

A minimum of 20 CFM of directed airflow is required to keep the passive cooler at a stable temperature during continuous use.

Does the card need a host GPU or display outputs to function in a server?

No front-panel video outputs are present; the accelerator works headless in a PCIe 3.0 x16 slot with sufficient airflow and 235 W power.

More in GPUs for AI & Data Centers

Similar items