AMD Steps Up Investment in Virtualized Graphics, Launches FirePro R5000
AMD is launching its new FirePro R5000 today in a bid to gain traction in the remote graphics market. The company has partnered with the workstation graphics virtualization company Teradici to create a GPU that's specifically designed for this type of offloading in hardware, as well as software. To make that happen, AMD added Teradici's hardware compression chip to a workstation graphics card meant for server deployment. The idea behind Teradici's PCoIP technology is to integrate all of a businesses' workstation hardware into a single datacenter.
The PCoIP compression chip makes that possible by transmitting pixel information rather than raw data streams, substantially saving on bandwidth. All of the transmissions are encrypted (which can theoretically improve security), and the R5000 can drive up to four 1920x12090 monitors or two 2560x1600 displays. According to AMD/Teradici, this type of solution lets companies save space in small environments where packing dozens of desktops or laptops is impractical, but thin clients (with modest power requirements) can fit more easily.
AMD noted that it's only selling the R5000 (not the thin-clients that go along with it), but it's also certified two older cards based on Radeon HD 4000 technology to work with Teradici as well. It's not clear if these cards contain the TERA host processors as well, since the application can apparently also be emulated in software (though this might obviate the use of a thin client).
Virtualized Workstations and Remote Graphics: The Next Big Thing?
Both AMD and Nvidia have made major pushes into virtualized workstation graphics in the past year. AMD has mostly repurposed workstation hardware, though its S10000 is the most powerful GPU the company currently ships, while Nvidia has VDI and a specialized board capable of driving four separate workstations from a single PCB. Each time one of the two companies launches this type of initiative, we're left wondering just how much demand exists for these solutions.
Thin clients, after all, aren't new. They've existed since the early days of computing, typically as terminals that connected to a mainframe. The term itself dates back to 1993 and the launch of Oracle 7 -- but historically, fat clients have continued to dominate. This is partly due to the fact that so-called fat clients have steadily lost weight and dropped in price. Thin client deployments emphasize network bandwidth and infrastructure more than fat client distributions, which means moving from one to the other can require significant shifts in compute power allocation.
Finally, there's the fact that the R5000 is still a one-to-one solution. If you need to equip 50 thin clients with virtualized PCoIP graphics, that means you need 50 R5000's to do so. This bites rather deeply into claims of cost savings -- hopefully AMD will address this need in future iterations of its virtualized graphics solutions.
The PCoIP compression chip makes that possible by transmitting pixel information rather than raw data streams, substantially saving on bandwidth. All of the transmissions are encrypted (which can theoretically improve security), and the R5000 can drive up to four 1920x12090 monitors or two 2560x1600 displays. According to AMD/Teradici, this type of solution lets companies save space in small environments where packing dozens of desktops or laptops is impractical, but thin clients (with modest power requirements) can fit more easily.
AMD noted that it's only selling the R5000 (not the thin-clients that go along with it), but it's also certified two older cards based on Radeon HD 4000 technology to work with Teradici as well. It's not clear if these cards contain the TERA host processors as well, since the application can apparently also be emulated in software (though this might obviate the use of a thin client).
Virtualized Workstations and Remote Graphics: The Next Big Thing?
Both AMD and Nvidia have made major pushes into virtualized workstation graphics in the past year. AMD has mostly repurposed workstation hardware, though its S10000 is the most powerful GPU the company currently ships, while Nvidia has VDI and a specialized board capable of driving four separate workstations from a single PCB. Each time one of the two companies launches this type of initiative, we're left wondering just how much demand exists for these solutions.
Thin clients, after all, aren't new. They've existed since the early days of computing, typically as terminals that connected to a mainframe. The term itself dates back to 1993 and the launch of Oracle 7 -- but historically, fat clients have continued to dominate. This is partly due to the fact that so-called fat clients have steadily lost weight and dropped in price. Thin client deployments emphasize network bandwidth and infrastructure more than fat client distributions, which means moving from one to the other can require significant shifts in compute power allocation.
Finally, there's the fact that the R5000 is still a one-to-one solution. If you need to equip 50 thin clients with virtualized PCoIP graphics, that means you need 50 R5000's to do so. This bites rather deeply into claims of cost savings -- hopefully AMD will address this need in future iterations of its virtualized graphics solutions.
