Post reply

Name
Email
Subject
Message icon

Other options

Return to this topic
Don't use smileys

Verification:

Please leave this box empty:

Shortcuts: ALT+S post or ALT+P preview

Topic summary

Posted by NXG

- Today at 18:29:02

It's super in the fact it does 1000 TOPs compared to all the other platforms that come with 50-100 TOP NPUs. Or I guess that's the point they'd likely make.

A bit pointless to compare to M Ultra which isn't even a laptop chip.

Nvidia and AMD are both data center companies. They no longer really care what happens in consumer market.

Apple is different in that they're a consumer facing company. It's also a bit funny in that they were amongst the first companies to go AI and introduce npu hardware in products, yet even they no longer care about it besides like, integrating Siri functionality that they've been doing for over a decade, not much else has been done.

When one of the richest consumer facing companies in the world that was a pioneer in the tech isn't going all in heavy in AI anymore, you know it has no real use case for the masses and the average person, as they probably realized it too. ;)

P.S. u need to update ur LLM cuz ur still getting it wrong, RTX spark is 300 GB/s. (..for the 25th time already)

Posted by not a superchip -- and

- Today at 10:51:52

"Superchip"
The NV Spark is a 256-bit APU, running at 273 GB/s. For comparison:

AMD Strix Halo 256-bit APU, 256 GB/s (came out first, so NV is copying it).
AMD Radeon 660M-890M 128-bit APU, at up to ~100 GB/s.
Apple M1-M3 Ultra 1024-bit APU, 819 GB/s. (en.wikipedia.org/wiki/Apple_M3).
RTX 5070 Ti desktop GPU, 896 GB/s.

Let's continue, why it's not a superchip: For the same file-size and hence memory requirement, a dense AI LLM model outperforms a MoE AI LLM model (see 122B MoE vs 27B dense: huggingface.co/Qwen/Qwen3.5-122B-A10B#benchmark-results), while being 4-5 times smaller.

There needs to be more ~120B MoE models for these up to 128 GB RAM APUs to be usable, as Qwen3.6-27B dense model outperforms them. Basically, only more than 120B MoE models outperform dense Qwen3.6-27B and they require more than 128 GB RAM: Look where 27B dense model sits vs the larger MoE models:
artificialanalysis.ai/?models=qwen3-5-122b-a10b-non-reasoning%2Cqwen3-5-122b-a10b%2Cmimo-v2-5-0424%2Cqwen3-6-27b-non-reasoning%2Cminimax-m2-7%2Cnvidia-nemotron-3-super-120b-a12b%2Cqwen3-6-35b-a3b-non-reasoning%2Cstep-3-7-flash%2Cqwen3-6-35b-a3b%2Cdeepseek-v4-flash-high%2Cqwen3-6-27b%2Cdeepseek-v4-flash-non-reasoning%2Chy3%2Cgemma-4-31b%2Cgemma-4-31b-non-reasoning%2Cdeepseek-v4-flash%2Cmistral-medium-3-5%2Cgpt-oss-120b%2Cmotif-0714&intelligence=artificial-analysis-intelligence-index
(Even a 120B model, at a 8-bit, requires 120 GB, so to fit these up to ~300B models, a 4- to 5-bit quant is used.)

What is really needed is a 192 or 256 GB RAM APU. AMD knows this and this is why they announced Gorgon Halo (a Strix Halo refresh, but with up to 192 GB RAM (160 GB allocatable on Windows, and maybe more on Linux, just like with Strix Halo)):
notebookcheck.net/AMD-Ryzen-AI-Max-400-lineup-now-official-with-up-to-192-GB-RAM.1301777.0.html

Posted by Redaktion

- Yesterday at 21:32:34

We already know that there will be multiple configurations of the RTX Spark "Superchip," but Nvidia has not shared those details. However, a new driver version has shed some light on a second configuration.

https://www.notebookcheck.net/RTX-Spark-Second-lower-end-config-confirmed-via-latest-driver.1348349.0.html