NVIDIA A10 24GB Specs, Benchmarks & Pricing

• Release Date: April 12, 2021
• MSRP: $2,800 USD^†
• GPU Architecture: Ampere
• Hardware-Accelerated GEMM Operations:
  FP64 FP32 TF32 BF16 FP16 FP8 FP4 INT8 INT4
• CUDA Compute Capability : 8.6

† Estimated MSRP of $2,800 USD for the NVIDIA A10 based on launch-time market pricing. NVIDIA does not publish official MSRP for data center GPUs. Estimate from NextPlatform article (April 2021) discussing the A10 launch. Corroborated by Newegg listing showing current market pricing around $3,299.

Specifications for NVIDIA A10

References

• https://www.nvidia.com/en-us/data-center/products/a10-gpu/
• https://www.nvidia.com/content/dam/en-zz/Solutions/Data-Center/a10/pdf/a10-datasheet.pdf
• https://www.nvidia.com/en-us/data-center/tensor-cores/
• https://www.nextplatform.com/2021/04/15/nvidia-rounds-out-ampere-lineup-with-two-new-accelerators/
• https://www.newegg.com/p/N82E16814133882
• https://www.techpowerup.com/gpu-specs/a10-pcie.c3793
• https://lenovopress.lenovo.com/lp1816.pdf
• https://www.dell.com/en-us/shop/nvidia-ampere-a10-pcie-150w-24gb-passive-single-wide-full-height-gpu/apd/490-bgkh/graphic-video-cards
• https://www.serversupply.com/GPU/GDDR6/24GB/DELL/490-BGKF_396993.htm
• https://www.itcreations.com/product/150448
• https://www.hpe.com/us/en/product-catalog/options/pip.specifications.nvidia-accelerators.1014670677.html
• https://www.scan.co.uk/products/pny-nvidia-a10-cec-24gb-gddr6-passive-graphics-card-8192-cuda-312-tflops-sp-125-tflops-hp
• https://www.pny.com/nvidia-a10
• https://www.exxactcorp.com/Supermicro-GPU-NVTA10-E3770657
• https://smicro.eu/nvidia-a10-24gb-gddr6-pcie-4-0-passive-cooling-gpu-nvta10-900-2g133-0020-100-1
• https://www.serversupply.com/GPU/GDDR6/APPLICATION%20ACCELERATOR/NVIDIA/900-2G133-0020-000_395888.htm
• https://www.cisco.com/c/dam/en/us/products/collateral/servers-unified-computing/ucs-c-series-rack-servers/nvidia-a10-gpu.pdf
• https://www.advantech.com/en-us/products/8d6a8886-b57e-493f-87c8-6f1e02d0608a/sky-tesl-a10-24p/mod_d9268799-585b-46b7-91d5-d49d126718b9
• https://forums.developer.nvidia.com/t/what-is-the-fp64-rpeak-value-for-a5000/196056

Notes

1. fp16TFLOPS of 250 is FP16 Tensor Core throughput with 2:4 structured sparsity, as NVIDIA's A10 page (nvidia.com/en-us/data-center/products/a10-gpu/) publishes it: 125 dense and 250 with sparsity.
2. int8TOPS: NOTE: INT8 Tensor Core
3. Estimated MSRP of $2,800 USD for the NVIDIA A10 based on launch-time market pricing. NVIDIA does not publish official MSRP for data center GPUs. Estimate from NextPlatform article (April 2021) discussing the A10 launch. Corroborated by Newegg listing showing current market pricing around $3,299.
4. NVIDIA part number 900-2G133-0020-000 sourced from NVIDIA official documentation and multiple vendor listings including serversupply.com
5. NVIDIA part number 900-2G133-0020-100 sourced from smicro.eu and multiple vendor listings for standard A10 variant
6. NVIDIA part number 900-2G133-2720-100 sourced from Lenovo ThinkSystem Product Guide (lenovopress.lenovo.com/lp1816.pdf) as the NVIDIA part number corresponding to Lenovo feature code BQZS
7. Board ID PG133 and SKU 215 sourced from TechPowerUp GPU Database (techpowerup.com/gpu-specs/a10-pcie.c3793)
8. Dell part numbers 490-BGKH, 490-BGLB, and 490-BGKF sourced from Dell official store and vendor listings (dell.com, serversupply.com, itcreations.com)
9. HPE part number R9W59C sourced from HPE official product catalog (hpe.com/us/en/product-catalog)
10. Lenovo part number 4X67A71311 (feature code BQZS) sourced from Lenovo Press Product Guide (lenovopress.lenovo.com/lp1816.pdf)
11. PNY SKU TCSA10M-PB sourced from SCAN UK and PNY official website (scan.co.uk, pny.com)
12. PNY SKU NVA10TCGPU-KIT sourced from vendor listings at deploydepot.ca and sabrepc.com
13. Supermicro part number GPU-NVTA10 sourced from Exxact and Supermicro retailers (exxactcorp.com, smicro.eu)
14. Cisco part number UCSC-GPU-A10 sourced from Cisco official datasheet (cisco.com) and vendor listings. Compatible with UCS C220/C225/C240/C245 M6 servers
15. Advantech part number SKY-TESL-A10-24P sourced from Advantech official website (advantech.com) and datasheet listings
16. Phase 5 audit (2026-09-02): gpuArchitecture normalized from lowercase 'ampere' to canonical 'Ampere' capitalization per the canonical table in .claude/agents/gpu-researcher.md.
17. Phase 5 audit (2026-09-02): FP64 removed from supportedHardwareOperations. The A10 uses the GA102 die (confirmed in the summary above), not GA100. Per an NVIDIA staff reply on NVIDIA Developer Forums, 'The Tensor core units in the GA102 chip architecture do not contain the FP64 path that is present in e.g. the Tensor core units in the A100 GPU' (https://forums.developer.nvidia.com/t/what-is-the-fp64-rpeak-value-for-a5000/196056). Canonical Ampere baseline (TF32, BF16, FP16, INT8, INT4) applied per gpu-researcher.md; FP64 Tensor Core support is reserved for GA100-based SKUs (A100, A30) only.
18. fp8TFLOPS and fp4TFLOPS are left null (field omitted) and not added to unverifiedSpecs — this is a confirmed hardware negative, not a research gap. The A10's Ampere 3rd-generation Tensor Cores predate both FP8 (introduced in Ada Lovelace/Hopper) and FP4 (introduced in Blackwell) tensor formats, per the canonical architecture table in .claude/agents/gpu-researcher.md.
19. fp32TFLOPS of 62.5 is TF32 Tensor Core throughput without sparsity, per NVIDIA's A10 page (nvidia.com/en-us/data-center/products/a10-gpu/). FP32 vector (CUDA core) throughput is 31.2 TFLOPS.