Would love to hear from others using the spark for model training and development.

I'm utterly shocked at the article saying GPU inference (PyTorch/Transformers)isn't working. Numerical instability produces bad outputs, Not viable for real-time serving, Wait for driver/CUDA updates!

My job just got me and our entire team a DGX spark. I'm impressed at the ease of use for ollama models I couldn't run on my laptop. gpt-oss:120b is shockingly better than what I thought it would be from running the 20b model on my laptop.

The DGX has changed my mind about the future being small specialized models.

No mention of the monstrous 200GbE NIC, seems like a waste if people aren't finding a use for it.

I absolutely love it. I’ve been up for days playing with it. But there are some bleeding edge issues. I tried to write a balanced article. I would highly recommend for people that love to get their hands dirty. Blows away any consumer GPU.

The llama.cpp issues are strange.

There are official benchmarks of the Spark running multiple models just fine on llama.cpp

https://github.com/ggml-org/llama.cpp/discussions/16578

There are bleeding edge issues, everyone dials into transformers so that's generally pain proof.

I haven't exactly bisected the issue but I'm pretty sure convolutions are broken on sm_121 after a certain size, getting 20x memory blowup from a convolution from a 2x batch size increase _only_ on the DGX Spark.

I haven't had any problems with inference, but I also don't use the transformers library that much.

llama.cpp was working for openai-oss last time I checked and on release, not sure if something broke along the way.

I don't exactly know if memory fragmentation is something fixable on the driver side - this might just be the problem with kernel's policy and GPL, it prevents them from automatically interfering with the memory subsystem to the granularity they'd like - see zfs and their page table antics - or so my thoughts on it is.

If you've done stuff on WSL, you have similar issues and you can fix it by running a service that normally compacts and clean memory, I have it run every hour. Note that this does impact at the very least CPU performance and memory allocation speeds, but I have not have any issue with long training runs with it (24hr+, assuming that is the issue, I have never tried without it and put that service in place since getting it due to my experience on WSL).