Run your most demanding scientific models on NVIDIA® Tesla® GPU Accelerators. Based on the NVIDIA Kepler™ Architecture, Tesla GPUs are designed to deliver faster, more efficient compute performance.

Find a Distributor

One Stop Systems offers a wide variety of products that can be integrated with NVIDIA GPUs. Our status as an NVIDIA Partner Network (NPN) Preferred OEM, acknowledges OSS’s achievement in accelerated computing, including technical expertise; customer service; and the ability to design, implement, and maintain best-in class accelerated computing solutions from NVIDIA. Call 877-438-2724 for more information.



The NVIDIA CUDA® parallel computing platform is enabled on GeForce®,Quadro®, and Tesla® products. Whereas GeForce and Quadro are designed for consumer graphics and professional visualization, respectively, the Tesla product family is designed ground-up for parallel computing and programming and offers exclusive High performance computing features.


Tesla products are designed with exclusive features to maximize performance for supercomputing professionals:

Full double precision floating point performance

  • 1.43 TFlops on the Tesla K40 GPU
  • Higher double precision than consumer products

Faster PCIe communication

  • The only NVIDIA product with two DMA engines for bi-directional PCIe communication

Higher performance on technical applications with large data sets

  • Larger on-board memory (12 GB on the Tesla K40 GPU)

Faster communication with InfiniBand using NVIDIA GPUDirect™

  • Special Linux patch, InfiniBand driver, and CUDA driver

Higher performance CUDA driver for Windows OS

  • TCC driver reduces CUDA kernel overhead and enables Windows Remote Desktop and Windows Services

Learn more about exclusive software for Tesla products

Features Tesla K40 Tesla K20X Tesla K20 Tesla K10
Number and Type of GPU 1 Kepler GK110B 1 Kepler GK110 2 Kepler GK104s
Peak double precision floating
point performance
1.43 Tflops 1.31 Tflops 1.17 Tflops 0.19 Tflops
Peak single precision floating
point performance
4.29 Tflops 3.95 Tflops 3.52 Tflops 4.58 Tflops
Memory bandwidth (ECC off) 288 GB/sec 250 GB/sec 208 GB/sec 320 GB/sec
Memory size (GDDR5) 12 GB 6 GB 5 GB 8 GB
CUDA cores 2880 2688 2496 2 x 1536

* Note: Tesla K10 specifications are shown as aggregate of two GPUs. With ECC on, 6.25% of the GPU memory is used for ECC bits. For example, 6 GB total memory yields 5.25 GB of user available memory with ECC on.

One Stop Systems warrants this product to be free of defects in material and workmanship for the warranty period purchased by the customer (1 to 5 years) from OSS beginning on the date of delivery to the original purchaser from OSS . During this period, One Stop Systems will, at its option, repair or replace this product at no additional charge to the purchaser, except as set forth in this warranty agreement. OSS will, at its option, repair or replace this product at no additional charge to the purchaser, if the defect is related to the OSS manufactured product. OSS is not liable for any defects in material or workmanship of any peripherals, products or parts, which OSS does not design or manufacture (third party products). OSS will honor the original manufacturer’s warranty on these products. One Stop Systems will analyze the defective component and the customer will be charged only in the following instances:

  • No problem found: $125(U.S dollars)
  • Damage Due to Negligence or Mishandling: Parts and labor at $75 per hour with a $100 minimum charge (U.S. dollars). Receipt of damaged goods voids the One Stop Systems warranty.

Repair parts and replacement products will be furnished on an exchange basis and will be either new or reconditioned. All replacement parts and products shall become the property of OSS, if such parts or products are provided under this warranty agreement. In the event a defect is not related to the OSS manufactured product or is not covered by the third party product, OSS shall repair or replace the defective parts at the purchaser’s cost and deliver the defective parts to the purchaser. This limited warranty shall not apply if the product has been misused, carelessly handled, defaced, modified or altered, or if unauthorized repairs have been attempted by others. The above warranty is the only warranty authorized by OSS and is in lieu of any implied warranties, including implied warranty of merchantability and fitness for a particular purpose. In no event will One Stop Systems be liable for any such damage as lost business, lost profits, lost savings, downtime or delay, labor, repair or material cost, injury to person or property or any similar or dissimilar consequential loss or damage incurred by the purchaser, even if OSS has been advised of the possibility of such losses or damages.

Depending on the warranty service level purchased, warranty service is obtained by advanced replacement when accompanied by a signed Warranty Service Agreement form or by delivering the product to an OSS authorized repair facility with all parts and accessories as originally shipped, along with the proof of purchase and a Returned Merchandise Authorization (RMA) number. The RMA number is obtained in advance from OSS Support or Customer Service Department and is valid for 30 days. The RMA number must be clearly marked on the exterior of the original shipping container or equivalent. Purchaser will be responsible and liable for any missing or damaged parts. Purchaser agrees to pay for shipping charges one way, and to either insure the product or assume the liability for loss or damage during transit.

Ship to: One Stop Systems



2235 Enterprise Street, Suite 110

Escondido, CA 92029

Support: or 877-438-2724


The GPUltima is 14.3 petaflops of computational clusters in a rack that contains up to eight nodes with each node containing up to 16 accelerators and one or two dual-socket servers. Customers can build up to the full rack, one node at a time, depending on their application requirements. Multiple nodes provide different performance capabilities, dependent on the number accelerators employed. All NVIDIA accelerators exchange data through Infiniband 100Gb transfers using GPU Direct RDMA. All GPUs are connected to a root complex through up to four 128Gb PCI Express connections. Root complexes are connected to the GPUs through Infiniband and each other and the outside world through Ethernet.

Learn More

Compute Accelerators

Our compute accelerators support from one to sixteen double-wide PCIe cards and can be cabled up to four host computers through PCIe x16 Gen3 connections each operating at 128Gb/s. The all-steel construction chassis house power supplies, fans, and a system monitor that monitors the fans, temperature sensors and power voltages. Front panel LEDs signal minor, major or critical alarms. The compute accelerators are transparent and do not require software except for the drivers required by the PCIe add-in cards. Compute accelerators are the best appliance for applications that require a large amount of compute power.

Learn More