NodeValid

Published video-model weights, licences and hardware notes

Model sharding, and what it does not divide

Sharding splits one model across several cards. It lowers what each card has to hold and does not divide the requirement by the number of cards, because activations and buffers are duplicated while only part of the load is split. As of 2026-09-22.

What four cards actually buyOne release here states a single-card peak and a four-card peak for the same output, and the second is about a quarter lower rather than a quarter of the first.What sharding lowersWhat it does notPer-card memorySubstantiallyTotal memory across cardsIt risesInterconnect requirementIt appears, and mattersReproducibilityNeeds the same link between cardsNot a division by the card count
Fig. 1 The measured saving, against the saving a reader assuming linear scaling would expect.
What a multi-card figure has to name to be reproducible. Recorded 2026-09-22.
Part of the statementWhy it matters
The card countThe split is over that many devices
The card modelIt fixes the interconnect as well as the memory
The outputThe frame and the frames are still the workload
The per-card figureA total tells a reader nothing actionable

Inclusion rule. Components of a reproducible multi-card statement. A component whose absence makes the figure unreproducible is listed as required. Order. From the count outward to the figure itself.

1The naive arithmetic fails badly

One release in this register states 60.3GB on one card and 44.3GB across four or more. A reader dividing by four would budget about fifteen gigabytes per card and be short by nearly thirty.

That is the single most useful thing a vendor can publish about sharding, and almost nobody does. The register keeps both figures in the cell rather than choosing.

2A card count is itself a hardware statement

Another release states eight named data-centre cards for its large model and one consumer card for its small one. Read as gigabytes that is a large number and a useless one; read as a node it specifies the interconnect too.

Sharding across cards is bandwidth-bound, so the same figures on four cards without a fast link between them would not reproduce.

3The one measured answer in the register

One family publishes a single-card figure and a multi-card figure for the same output, and the second is about a quarter lower rather than a quarter of the first. Another names a count of eight data-centre cards without a per-card figure.

Between them those are the only multi-card statements here. Four families say explicitly that their figure is for one GPU, which is cheap precision and worth having.

A reading note, not an entry: no vendor value appears on this page. Where the register records this term for a particular release, it is on the hardware column and on that family's own page. Nearby terms: GPU architecture, Batch size.