Qualcomm Cloud AI 100 Ultra – 150W AI Accelerator Card with 128GB at 548GB/s

https://www.qualcomm.com/data-center/products/cloud-ai-100-ultra

37•peter_d_sherman•1h ago

Comments

pearl_panel5•47m ago

Curious how the memory bandwidth actually holds up under sustained inference loads compared to the spec sheet numbers.

varispeed•44m ago

Aren't these specs already outdated? Seems meh. I would think players would like to bring something new to the table instead of bringing what was okay two years ago or so.

Give me 1TB RAM and 5TB/s or something that exciting.

tempest_•33m ago

5 year old GPUs are still at crazy prices.

Plenty of businesses want in local inference and are priced out.

entrope•28m ago

For most purposes, this seems reasonable? This looks like a single-height card, so fitting four of these in a workstation or server would give 512 GB RAM and 2 TB/s total bandwidth. That's pretty attractive for a lot of current LLMs, even if one wishes for 1 TB capacity at 5 TB/s bandwidth.

mciancia•17m ago

8 of those cards give you 1TB of ram and 4.4TB/s

PhunkyPhil•11m ago

It may be a semantic thing but I don't think bandwidth pooling works like that? Unless your blocks are distributed perfectly between cards there's going to be data passed between each card and from VRAM to compute. Correct me if I'm wrong though

aetherspawn•35m ago

Ok so it’s significantly worse than Apple M6 and we don’t know if it’s cheaper or not

Edit: Ah sorry you’re right, the bandwidth is lower but the TOPS is a lot higher. I’m not sure how much effect you get from higher TOPS with such low bandwidth though.

dist-epoch•25m ago

Apple M6 has 80-100 TOPS.

Qualcomm Cloud AI 100 Ultra has 800 TOPS, almost 10 times more.

zitterbewegung•29m ago

Sort of interesting design It has half a terabyte of SRAM on the accelerator with a 128 GB of RAM replacing HBM with SRAM but keeping regular RAM.

2001zhaozhao•29m ago

half a gigabyte not terabyte lol

zrm•9m ago

Yeah, this thing is unusual. It resembles a consumer GPU but using normal DRAM with more channels, so then you get more RAM but less memory bandwidth than GDDR/HBM, which is the same thing Apple does. And the TDP is pretty attractive for what it is.

But what is that amount of SRAM for? It simultaneously seems like a lot, because SRAM is so expensive, but still not very much, because what can you fit in 0.5GB? Is there a specific workload that needs something that looks like this?

tokepi•27m ago

If the specs seem outdated, it’s because this chip launched in 2024. It’s available at cloud providers including Cirrascale

thway15269037•16m ago

What's the price?

simlevesque•5m ago

According to their own website, they cost 1 cent.

Residential Proxies Detected on White House Network

Ripple-65B Preview: U.S. Open-Weight SOTA on Agentic Coding

Insects Don't Swarm at LED Lights

AI is forcing us to pay off 40 years of technical debt

A new Go API for JSON

Progress on provable inference for large language models

Qualcomm Cloud AI 100 Ultra – 150W AI Accelerator Card with 128GB at 548GB/s

Nvidia RTX Spark encrypted firmware slows inference speed to 35% on Linux

Launch HN: Kino AI (YC S23) – AI media manager and video editor

Have I Befriended a Murderer?

Show HN: After a decade of reversing, I got co-op working in DOS Wing Commander

OpenAI to acquire Pierre Computer Company

Tencent Hy4

AMD expects to outsell Intel in Q4 for the first time in history

Everything is an AI scam now

Conductor Is Joining Meta

AR guitar with just a laptop and scissors

North Korean fake job interview led to $6.3B Binance heist

Verification Is Not Knowledge: Science needs more than answers

Guerrilla bike lanes: San Francisco makes illicit infrastructure permanent (2016)

Show HN: I built a tool that prices a prebuilt PC part by part

Incident with Pull Requests, Git Operations and Issues

Kimi K3.1 Pro model weights are out

Tell HN: Gemini trains on chat history, even for subscribers (personal accounts)

Jelly Sheet

From wild to mild: Friendly foxes show changes in brain region

Benchfmt.rs: Bringing Go's benchmarking and benchstat workflow to Rust

Evergreen: A Provably Safe Systems Language

Letting users trust their own MCP clients, without opening a phishing hole

Kitty 0.50

Residential Proxies Detected on White House Network

Ripple-65B Preview: U.S. Open-Weight SOTA on Agentic Coding

Insects Don't Swarm at LED Lights

AI is forcing us to pay off 40 years of technical debt

A new Go API for JSON

Progress on provable inference for large language models

Qualcomm Cloud AI 100 Ultra – 150W AI Accelerator Card with 128GB at 548GB/s

Nvidia RTX Spark encrypted firmware slows inference speed to 35% on Linux

Launch HN: Kino AI (YC S23) – AI media manager and video editor

Have I Befriended a Murderer?

Show HN: After a decade of reversing, I got co-op working in DOS Wing Commander

OpenAI to acquire Pierre Computer Company

Tencent Hy4

AMD expects to outsell Intel in Q4 for the first time in history

Everything is an AI scam now

Conductor Is Joining Meta

AR guitar with just a laptop and scissors

North Korean fake job interview led to $6.3B Binance heist

Verification Is Not Knowledge: Science needs more than answers

Guerrilla bike lanes: San Francisco makes illicit infrastructure permanent (2016)

Show HN: I built a tool that prices a prebuilt PC part by part

Incident with Pull Requests, Git Operations and Issues

Kimi K3.1 Pro model weights are out

Tell HN: Gemini trains on chat history, even for subscribers (personal accounts)

Jelly Sheet

From wild to mild: Friendly foxes show changes in brain region

Benchfmt.rs: Bringing Go's benchmarking and benchstat workflow to Rust

Evergreen: A Provably Safe Systems Language

Letting users trust their own MCP clients, without opening a phishing hole

Kitty 0.50

Qualcomm Cloud AI 100 Ultra – 150W AI Accelerator Card with 128GB at 548GB/s

Comments