Although, as Asker has mentioned in the discussion box, translations in Microsoft Learn are unreliable (blame MT for that!), it is nonetheless clear from the following extracts which explain the purpose of of DDA in both English and Spanish, that 'DDA' refers to the 'discrete assignment' / asignación discreta of 'devices' / dispositivos. IWO, 'discrete' qualifies 'deployment', not 'devices'.

EN
https://learn.microsoft.com/en-us/windows-server/virtualizat...
Starting with Windows Server 2016, you can use Discrete Device Assignment, or DDA, to pass an entire PCIe Device into a VM
...
There are three steps to using a device with Discrete Device Assignment:
Configure the VM for Discrete Device Assignment
Dismount the Device from the Host Partition
Assigning the Device to the Guest VM

ES
https://learn.microsoft.com/es-es/windows-server/virtualizat...
A partir de Windows Server 2016, puede usar la asignación discreta de dispositivos o DDA para pasar un dispositivo PCIe completo a una máquina virtual.
...
Hay tres pasos para usar un dispositivo con asignación discreta de dispositivos:
Configuración de la máquina virtual para la asignación discreta de dispositivos
Desmontar el dispositivo de la partición de host
Asignación del dispositivo a la máquina virtual invitada

--------------------------------------------------
Note added at 3 hrs (2023-06-17 14:10:23 GMT)
--------------------------------------------------

@Asker's comment below:
La que más me convence es intel
(https://www.intel.es/content/www/es/es/support/articles/0000...
en el sentido de que los gráficos (tarjetas) son periféricos. Aquí se ve
discreto/periférico vs integrado.

In the context of Windows Server, and DDA in particular which is concerned with optimising the use of resources in a virtual machine, it is irrelevant whether the device is physically integrado or a periférico.


--------------------------------------------------
Note added at 7 hrs (2023-06-17 17:45:52 GMT)
--------------------------------------------------

Further clarification (although this really isn't the right place for explanations of how this IT stuff works...)

Referring to the extracts from Microsoft Learn posted above:
'discrete' (discreta refers to the fact that when a VM requires the resources of a PCIe device (GPU, sound processor, or whatever), it will trigger an assignment protocol which disconnects (dismount/desmontar) that PCIe device from its current host (user) and transfers access to the VM. This process necessarily transfers the entire resource to the VM, and the original host loses all access; hence the term 'discrete' to refer to the assignment process - it assigns on an 'all or nothing' basis - i.e. it is not possible to assign only a part of the resource (and less still is it possible to share the resource between the old and new hosts).

https://learn.microsoft.com/en-us/windows-server/virtualizat...
acceleration-in-windows-server#discrete-device-assignment-dda
Discrete Device Assignment (DDA), also known as GPU pass-through, allows you to dedicate one or more physical GPUs to a virtual machine. In a DDA deployment, virtualized workloads run on the native driver and typically have full access to the GPU's functionality.
A DDA deployment can accelerate only a limited number of virtual machines, since each physical GPU can provide acceleration to at most one VM.

ES
La asignación de dispositivos discreta (DDA), también conocida como paso a través de GPU, permite dedicar una o varias GPU físicas a una máquina virtual. En las implementaciones de DDA, las cargas de trabajo virtualizadas se ejecutan en el controlador nativo y normalmente tienen acceso total a la funcionalidad de la GPU.
Una implementación de DDA solo puede acelerar un número limitado de máquinas virtuales, ya que cada GPU física no puede proporcionar aceleración a más de una máquina virtual.

We can note in passing the following, on the same page under the heading 'RemoteFX vs vGPU':
RemoteFX vGPU is a graphics virtualization technology that allows a single physical GPU to be shared among multiple virtual machines.
IOW, PCIe resource sharing is possible with Remote FX (deprecated - see the purple warning note), but not with DDA.