Fabric switched graphics modules within storage enclosures

1. A data system, comprising:

a first assembly comprising a chassis having a plurality of modular bays populated with one or more graphics processing modules each including a graphics processing unit (GPU), wherein the plurality of modular bays each comprise a bay connector that includes a bay Peripheral Component Interconnect Express (PCIe) connection;

the first assembly further comprising PCIe switch circuitry configured to communicatively couple the bay PCIe connections to a PCIe fabric over one or more external PCIe links; and

in the PCIe switch circuitry, providing an isolation function to form a peer-to-peer arrangement between at least two GPUs populated in the chassis, wherein the isolation function isolates a first address domain associated with the GPUs from at least a second address domain associated with a host by at least establishing synthetic devices representing the GPUs in the second address domain and redirects traffic transferred by the host for the GPUs in the second address domain for delivery to corresponding ones of the GPUs in the first address domain.

2. The data system of claim 1 , comprising:

a second assembly configured to provide at least a portion of the PCIe fabric and comprising one or more compute modules coupled over the PCIe fabric and configured to communicate with one or more GPUs housed in the first assembly.

3. The data system of claim 1 , comprising:

in the PCIe switch circuitry, adding a selected GPU into the peer-to-peer arrangement by at least instantiating the selected GPU in the second address domain and initiating a synthetic device representing the selected GPU in the first address domain.

4. The data system of claim 1 , wherein the one or more graphics modules each comprise an associated PCIe root complex.

5. The data system of claim 1 , wherein the one or more graphics modules each comprise an associated PCIe endpoint.

6. The data system of claim 1 , wherein the one or more graphics processing modules each comprise a carrier insertable into a single one of the modular bays and couple to an associated bay connector.

7. The data system of claim 1 , wherein the one or more graphics processing modules each comprise a carrier that spans more than one of the modular bays and couple to more than one bay connector.

8. The data system of claim 7 , wherein the carrier that spans more than one of the modular bays provides combined power of the more than one bay connector to the carrier, and wherein the carrier that spans more than one of the modular bays provides more than one bay PCIe connection to the carrier to couple to a quantity of PCIe lanes to the carrier above that of a single bay PCIe connection.

9. The data system of claim 1 , comprising:

the first assembly further comprising a backplane assembly that includes the bay connectors of the plurality of modular bays to communicatively couple a back side of the one or more graphics processing modules to the PCIe fabric; and

peer connectors on a front side of the one or more graphics processing modules to couple at least two of the one or more graphics processing modules over a point-to-point communication link separate from the PCIe fabric.

10. A computing system, comprising:

a chassis comprising a plurality of bays populated with one or more graphics processing modules each including at least one graphics processing unit (GPU), wherein the plurality of bays each comprise at least one bay connector that includes a bay Peripheral Component Interconnect Express (PCIe) connection;

the chassis further comprising PCIe switch circuitry configured to communicatively couple the bay PCIe connections to a PCIe fabric over one or more external PCIe links; and

the PCIe switch circuitry configured to provide an isolation function to form a peer-to-peer arrangement between at least two GPUs populated in the chassis, wherein the isolation function isolates a first address domain associated with the GPUs from at least a second address domain associated with a host by at least establishing synthetic devices representing the GPUs in the second address domain and redirects traffic transferred by the host for the GPUs in the second address domain for delivery to corresponding ones of the GPUs in the first address domain.

11. The computing system of claim 10 , comprising:

a computing chassis comprising one or more compute modules coupled over the PCIe fabric and configured to communicate with one or more GPUs housed in the chassis over the one or more external PCIe links.

12. The computing system of claim 11 , comprising:

the PCIe switch circuitry configured to add a selected GPU into the peer-to-peer arrangement by at least instantiating the selected GPU in the second address domain and initiating a synthetic device representing the selected GPU in the first address domain.

13. The computing system of claim 10 , wherein the one or more graphics modules each comprise an associated PCIe root complex.

14. The computing system of claim 10 , wherein the one or more graphics modules each comprise an associated PCIe endpoint.

15. The computing system of claim 10 , wherein the one or more graphics processing modules each comprise a carrier insertable into a single one of the bays and couple to an associated bay connector.

16. The computing system of claim 10 , wherein the one or more graphics processing modules each comprise a carrier that spans more than one of the bays and couple to more than one bay connector.

17. The computing system of claim 16 , wherein the carrier that spans more than one of the bays provides combined power of the more than one bay connector to the carrier, and wherein the carrier that spans more than one of the bays provides more than one bay PCIe connection to the carrier to couple to a quantity of PCIe lanes to the carrier above that of an individual bay PCIe connection.

18. The computing system of claim 10 , comprising:

the chassis further comprising a backplane assembly that includes the connectors of the plurality of bays to communicatively couple a back side of the one or more graphics processing modules to the PCIe fabric; and

peer connectors on a front side of the one or more graphics processing modules to couple at least two of the one or more graphics processing modules over a point-to-point communication link separate from the PCIe fabric.

19. A data processing system, comprising:

a first enclosure comprising one or more compute modules coupled over a Peripheral Component Interconnect Express (PCIe) fabric; and

a second enclosure comprising a chassis having at least one PCIe switch circuit and a plurality of graphics processing modules each including a graphics processing unit (GPU) and coupled into associated bays via bay connectors;

the PCIe switch circuit configured to communicatively couple the plurality of graphics processing modules to the PCIe fabric of the first enclosure over at least one external PCIe link communicatively coupling the first enclosure to the second enclosure; and

the PCIe switch circuit configured to provide an isolation function to form a peer-to-peer arrangement between at least two GPUs populated in the second enclosure, wherein the isolation function isolates a first address domain associated with the GPUs from at least a second address domain associated with a host by at least establishing synthetic devices representing the GPUs in the second address domain and redirects traffic transferred by the host for the GPUs in the second address domain for delivery to corresponding ones of the GPUs in the first address domain.

20. The data processing system of claim 19 , comprising:

the PCIe switch circuit configured to add a selected GPU into the peer-to-peer arrangement by at least instantiating the selected GPU in the second address domain and initiating a synthetic device representing the selected GPU in the first address domain.