The NVIDIA DGX Spark and the NVIDIA RTX 4090 are two distinct systems with different focuses and architectures, which results in different memory bandwidths.
The NVIDIA DGX Spark is a desktop AI supercomputer powered by the NVIDIA GB10 SoC, featuring a 20-core Armv9 processor and a Blackwell GPU. It includes 128 GB of 256-bit LPDDR5x memory, which provides a memory bandwidth of 273 GB/s[1][4]. This system is designed for AI workloads, offering up to 1,000 TOPS of AI performance and is optimized for tasks like training and inference of large AI models[1][4].
In contrast, the NVIDIA RTX 4090 is a high-end graphics card primarily aimed at gaming and graphics rendering, though it also supports AI tasks. It uses the Ada Lovelace architecture and includes 24 GB of GDDR6X memory with a 384-bit memory bus, achieving a memory bandwidth of 1,008 GB/s[2][5]. This makes the RTX 4090 significantly faster in terms of memory bandwidth compared to the DGX Spark.
Overall, while both systems are powerful in their respective domains, the RTX 4090 offers nearly four times the memory bandwidth of the DGX Spark, making it more suitable for applications requiring high-speed data transfer, such as high-resolution gaming and complex graphics rendering. However, the DGX Spark is specifically optimized for AI tasks and provides a more integrated solution for AI development and deployment.
Citations:
[1] https://www.cnx-software.com/2025/03/19/nvidia-dgx-spark-a-desktop-ai-supercomputer-powered-by-nvidia-gb10-20-core-armv9-soc-with-1000-tops-of-ai-performance/
[2] https://vast.ai/article/nvidia-rtx-4090-vs-5090
[3] https://www.fibermall.com/blog/nvidia-a100-vs-4090.htm
[4] https://www.pcmag.com/news/what-is-nvidias-dgx-station-a-new-specialized-desktop-line-for-ai-work
[5] https://www.itcreations.com/nvidia-gpu/nvidia-geforce-rtx-4090-gpu
[6] https://beebom.com/nvidia-rtx-5070-vs-rtx-4090-comparison/
[7] https://www.youtube.com/watch?v=krBh0Von-2A
[8] https://coinpoet.com/ml/learn/gpu/nvidia-geforce-rtx-4090
[9] https://www.reddit.com/r/LocalLLaMA/comments/1jef1dd/dgx_spark_previously_digits_has_273gbs_memory/
[10] https://forums.developer.nvidia.com/t/i-need-to-know-about-how-long-it-take-to-fill-the-geforce-rtx-4090s-24gb-memory-from-computers-ram-os-memory/251738
[11] https://chipsandcheese.com/p/microbenchmarking-nvidias-rtx-4090