AgilityAI GPU Servers


AgilityAI GPU Servers

Our AgilityAI GPU server selection offers best in class design and performance. We can work with you and determine which one is best for your use case. SXM2 with NV-Link based systems usually offer more performance and more I/O options. PCIe based systems can support both NVIDIA GeForce and NVIDIA Tesla based GPUs and are available with either AMD EPYC or Intel Xeon Scalable based CPU platforms. IBM’s Power 9 AC922 offers a unique GPU offering as they are the only CPU manufacturer that incorporates the high bandwidth NV-LINK technology integrated in to the CPU improving performance between CPU, GPU and system memory. HGX2, a system similar to NVIDIA DGX2, will soon be available and offers the highest performance available utilizing 16x Tesla V100 GPUs in one platform. Let us help you determine which solution is best.

AgilityAI GPU Enterprise Servers

 

IBM Power9 AC922

2U 2-socket IBM Power9 Four/Six SXM2 GPU platform

  • Two IBM Power9 CPUs, up to 22 cores per processor with integrated NV-Link 2.0
  • Two 2.5” SATA/SSD boot drives
  • 16x DDR4 DIMM slots; 8x memory channels per CPU socket; 256GB, 512GB or 1TB
  • Up to six SXM2 Tesla V100 32GB GPUs
  • 2x PCIe Gen4 x16 FHFL slots, CAPI 2.0; 2x PCIe Gen4 x8 slots, CAPI
  • Burst buffer, 1.6TB NVMe PCIe adapter
  • IBM PowerAI Enterprise DL frameworks

RG104SX-XS2-D

1U 2-socket Intel Xeon Scalable Four SXM2 GPU platform

  • Two Intel Xeon Scalable Processors, up to 40 cores
  • Drive bays: up to two m.2 120GB or 240GB, mirrored
  • 24x DDR4 DIMM slots support RDIMM and LRDIMM; up to 1.5TB
  • Up to four SXM2 Tesla V100 32GB GPUs
  • 2x PCIe Gen3 slots, x16
  • Dual 10GbE RJ45 or SFP+ network ports plus 1x dedicated port for IPMI
  • Redundant, high-efficiency power supplies

RG404P-XS4-D

4U 4-socket Intel Xeon Scalable Four PCIe GPU platform

  • Four Intel Xeon Scalable Processors, up to 112 cores
  • Front drive bays: up to 32x 2.5” SAS/SATA; up to 4x NVMe
  • 48x DDR4 DIMM slots support RDIMM and LRDIMM; up to 6TB; 12x NVDIMMS
  • Up to four double wide GPUs or up to eight FPGAs
  • 12x PCIe Gen3 slots; 6x x16 slots or 2x x16 with 10x x8 slots
  • Dual 10GbE RJ45 or SFP+ network ports plus 1x dedicated port for IPMI
  • Redundant, high-efficiency power supplies

AgilityAI GPU Value Servers

 

RG104SX-XS2-S

1U 2-socket Intel Xeon Scalable Four SXM2 GPU platform

 

 

  • Two Intel Xeon Scalable Processors, up to 56 cores
  • Front drive bays: up to 2x 2.5” SSD/SATA; one internal m.2 NVMe
  • 12x DDR4 DIMM slots support RDIMM and LRDIMM; up to 1.5TB
  • Up to four SXM2 Tesla V100 32GB GPUs
  • 4x PCIe Gen3 slots, x16
  • Dual 10GbE RJ45 network ports plus 1x dedicated port for IPMI
  • Redundant, high-efficiency power supplies

RG408SX-XS2-S

4U 2-socket Intel Xeon Scalable Eight SXM2 GPU platform

  • Two Intel Xeon Scalable Processors, up to 56 cores
  • Front drive bays: up to 16x 2.5” SSD/SATA; two internal m.2 NVMe
  • 24x DDR4 DIMM slots support RDIMM and LRDIMM; up to 3TB
  • Up to eight SXM2 Tesla V100 32GB GPUs
  • 6x PCIe Gen3 slots, x16
  • Dual 10GbE RJ45 network ports plus 1x dedicated port for IPMI
  • Redundant, high-efficiency power supplies

RG408PS-XS2-S

4U 2-socket Intel Xeon Scalable Eight PCIe GPU platform, single root hub

  • Two Intel Xeon Scalable CPUs, up to 205w
  • Eight PCIe Gen3 x16 GPUs; V100; GeForce RTX2080; single root hub architecture
  • 24x 2.5” SATA/SAS
  • One m.2 boot drive; NVMe
  • 24x DDR4 DIMM slots support RDIMM and LRDIMM
  • Network Daughter card options include 2x 10GbE
  • 11 FH, FL PCIe Gen3 x16 slots; 1x PCIe Gen3 x8
  • Redundant, high-efficiency power supplies

RG408P-XS2-S

4U 2-socket Intel Xeon Scalable Eight PCIe GPU platform, dual root hub

  • Two Intel Xeon Scalable CPUs, up to 205w
  • Eight PCIe Gen3 x16 GPUs; Tesla V100, GeForce RTX2080
  • 24x 2.5” SATA/SAS
  • One m.2 boot drive; NVMe
  • 24x DDR4 DIMM slots support RDIMM and LRDIMM
  • Network Daughter card options include 2x 10GbE
  • 8 FH, FL PCIe Gen3 x16 slots; 2x PCIe Gen3 x8
  • Redundant, high-efficiency power supplies

RG208P-AE1-G

2U 1-Socket AMD EPYC Eight PCIe GPU platform

  • One AMD EPYC CPU, up to 32 CPU cores
  • Eight PCIe Gen3 x16 GPUs; V100, GeForce RTX2080
  • 8x 2.5” SATA drive bays
  • Two m.2 boot drives; NVMe
  • 8x DDR4 DIMM slots support RDIMM and LRDIMM; up to 1TB
  • Network Daughter card includes 2x 10GbE SFP+
  • 8 FH, FL PCIe Gen3 x16 slots; 2x PCIe Gen3 x16
  • Redundant, high-efficiency power supplies

RG1016SX-XS2-S

HGX2 16x SXM3 GPU platform, 2-socket Intel Xeon Scalable

  • Two Intel Xeon Scalable CPUs, up to 205w
  • 16 NVIDIA Tesla V100 32GB SXM3 GPUs
  • 16x 2.5″ NVMe drive bays
  • 24x DDR4 DIMM slots support RDIMM and LRDIMM
  • 16x PCIe Gen3 x16 slots
  • Network Daughter card includes 2x 10GbE
  • 6x 3000w Titanium rated redundant power supplies