Can the AMD Radeon PRO W7800 run granite-4.2-3b?

granite-4.2-3b · AMD Radeon PRO W7800 · 8,192 ctx · whole model residentFP16 sizepublished dataSpeedestimate

IT FITS

Yes. 9 of 9 formats evaluated fit in 32 GB, at 8,192 tokens. Every row is sized from a published file. The best quality that fits is FP16, needing 8.79 GB and running at an estimated 60 tokens per second.

Fitting in memory is not the same as loading. Whether the runtime and version you have supports this architecture and format on this machine has not been tested here.

Find the AMD Radeon PRO W7800: Amazon ↗ · eBay (new and used) ↗Store links may pay us a commission. They never decide which card is suggested — the memory arithmetic does.

9/9formats that fit

32 GBdevice memory

8.79 GBneeded at best quality

~60tokens per second, estimatedFaster than you read

576GB/s bandwidth

How long a conversation. At FP16, this pair holds 131,072 tokens of context — roughly 98,304 words — needing 18.9 GB. That is IBM's own configured maximum of 131,072 tokens, not the card running out. A larger device does not extend it; a documented RoPE or YaRN extension might, and is a separate question from this one.

Every format evaluated

Every row is sized from a published file. The Basis column says which is which for each row. Decode and first-token figures are calibrated estimates, not runs on this card.

Format	Needs	Of 32 GB	Fits	Decode	First token	Basis
FP16	8.79 GB	27%	yes	~60 tok/s	no comparable peak	published data
Q8_0	5.36 GB	17%	yes	~105 tok/s	no comparable peak	published data
Q6_K	4.48 GB	14%	yes	~129 tok/s	no comparable peak	published data
Q5_K_M	4.08 GB	13%	yes	~144 tok/s	no comparable peak	published data
Q5_0	4.03 GB	13%	yes	~147 tok/s	no comparable peak	published data
Q4_K_M	3.72 GB	12%	yes	~162 tok/s	no comparable peak	published data
Q4_0	3.60 GB	11%	yes	~169 tok/s	no comparable peak	published data
Q3_K_M	3.31 GB	10%	yes	~186 tok/s	no comparable peak	published data
Q2_K	2.93 GB	9%	yes	~224 tok/s	no comparable peak	published data

Sized at 8,192 tokens of context with the whole model resident — weights, the KV cache and the runtime reserve, offload off. Speed is only quoted for a format that fits: a rate for a configuration that cannot load is not a fact about anything.

Why the first-token figure is the same on every row

First token is modelled from the arithmetic the prompt requires, and that count does not change with the weight format — which is why it reads the same on every row. Real prefill does vary by format, because a quantized matmul is a different kernel; this model does not capture that, and the figure should be read as an order of magnitude rather than a ranking between formats.

What the labels mean

published data: Read from a published source — a file's byte count, a model's configuration or a manufacturer's specification — or exact arithmetic on such values. Not a measurement on a machine.
reconstructed size: Weight size reconstructed from the pinned architecture, because no published file exists.
size range: Only a lower and an upper bound are claimed for this weight size.
estimate: Calculated from sourced inputs by a stated method; an estimate, not a measurement.

Change anything

This page fixes the context at 8,192 tokens and one device. Batch, concurrent users, KV-cache format, clusters and rental cost are all in the calculator, already set to this pairing.

Open in the calculator →Everything the AMD Radeon PRO W7800 runs Best models for 32 GB

Where these numbers come from

granite-4.2-3b’s architecture is read from its publisher’s own config.json at a pinned revision, and the AMD Radeon PRO W7800’s 32 GB and 576 GB/s come from the manufacturer’s specification. How far each figure can be trusted is published, per format and worst case included, on the accuracy scorecard.

Building this into your own product? A free API key returns exactly these figures, and the widget puts this answer on a product page with one script tag.

Use this answer in your own product

As JSON, from the API

curl -s https://llmbottleneck.com/v1/analyze \
  -H "Authorization: Bearer $LLMB_KEY" -H "content-type: application/json" \
  -d '{"model":"ibm-granite-granite-4-2-3b","quantization":"FP16","context":8192,"hardware":"amd-radeon-pro-w7800"}'

Same engine, same evidence, every field sourced. Free key in one step →

On your page, as a widget

<script src="https://llmbottleneck.com/widget.js"
  data-model="ibm-granite-granite-4-2-3b" data-quantization="FP16"
  data-hardware="amd-radeon-pro-w7800" data-context="8192"></script>

No key needed. Unbranded, with your own buy button, on Pro and Business →