Friday, January 9th 2009

Samsung Introduces World’s First PC-over-IP LCD Desktop Display

Samsung Electronics America, a subsidiary of Samsung Electronics Corporation, today introduced an innovative PC-over-IP (PCoIP) display solution in conjunction with Teradici Corporation. Teradici’s PCoIP technology is incorporated into the Samsung SyncMaster 930ND 19-inch display, which delivers a true integrated desktop computing experience over standard IP networks. Such a solution allows users to operate a remote host PC and use the 930ND as a local desktop monitor. The integrated PCoIP display features a 1000:1 contrast ratio, 5ms response time and a resolution of 1280 by 1024 pixels. The monitor also has four USB connectors, 2-way HD audio and DVI-out for a second display, and supports both PCoIP and RDP modes. The 930ND is currently compatible with Microsoft Windows Vista and Windows XP operating systems.

Source: Samsung
Add your own comment

27 Comments on Samsung Introduces World’s First PC-over-IP LCD Desktop Display

#1
Roph
I'm guessing it uses a remote desktop client then? I'm sure it would sell better (and cost them nothing extra to implement) if they included a VNC client too.
Posted on Reply
#2
lemonadesoda
I dont think that is what they mean. I think they essentially have a docking station over IP. So you would have USB, Keyboard, Mouse, Keystick, and VGA out, at the desk... but the PC would be some long way away. Rather than proprietary wiring... it just used the TCP/IP (ethernet) cabling already in place to communicate. It will mean that USB data speeds are constrained by available network bandwidth. Dont expect to play games or push/pull large data over USB devices. But perfect for simple terminals, esp. server control. No need to be "in the server room", and no need for separate computer over Remote Access.
Posted on Reply
#3
srobinson
3D and HD Video Performance of Samsung 930ND

Here is a short video demo of the 3D graphics and HD video performance of the Samsung 930ND PCoIP display portal - http://www.youtube.com/watch?v=HzGmwNIpFG4

Stuart Robinson, Director of Business Development, Teradici
Posted on Reply
#4
srobinson
lemonadesoda said:
I dont think that is what they mean. I think they essentially have a docking station over IP. So you would have USB, Keyboard, Mouse, Keystick, and VGA out, at the desk... but the PC would be some long way away. Rather than proprietary wiring... it just used the TCP/IP (ethernet) cabling already in place to communicate. It will mean that USB data speeds are constrained by available network bandwidth. Dont expect to play games or push/pull large data over USB devices. But perfect for simple terminals, esp. server control. No need to be "in the server room", and no need for separate computer over Remote Access.
Thanks for the explanation! One point to add is that this is for a full user desktop including full frame rate 3D graphics - design engineering, medical, and video games. Here is a link to a video demo of Crysis on an HDTV running over a wireless network (802.11n) - http://www.youtube.com/watch?v=7FAoJfU4Iwo. One of our OEM partners has been testing PCoIP on Crysis/Call of Duty4 with dual NVIDIA GPU's that are SLI connected to increase the frame rate.

To your point, the USB may be limited by the available network bandwidth. But if there is bandwidth available PCoIP supports up to USB 1.1 rates. A next generation device will support USB 2.0 going forward (most IT prefer USB1.1, but some need USB2.0). We have not found a USB peripheral that does not work (assuming it can work with 11Mbps throughput) and we support isochronous USB data transfers - webcams, USB speakers/mic etc.

The longest test an OEM partner has done to date is 8,000Km (Houston, USA to London, UK or 125ms round trip network latency). The latency at these distances would be an issue for most gaming apps.

Cheers, Stu
Posted on Reply
#5
Darkrealms
srobinson said:
Here is a short video demo of the 3D graphics and HD video performance of the Samsung 930ND PCoIP display portal - http://www.youtube.com/watch?v=HzGmwNIpFG4

Stuart Robinson, Director of Business Development, Teradici
Nice to see vendors showing up to represent their products!

Interesting concept. If it supports RDP then it would be great for an exame room concept that RDPs into a WTS anyway.
. . . But support for XP and Vista only. Needs to have at least Server 2003 capability as well.

This could become cumbersum if it requires PCs stored in a closet somewhere.
Posted on Reply
#6
srobinson
Darkrealms said:
Nice to see vendors showing up to represent their products!

Interesting concept. If it supports RDP then it would be great for an exame room concept that RDPs into a WTS anyway.
. . . But support for XP and Vista only. Needs to have at least Server 2003 capability as well.

This could become cumbersum if it requires PCs stored in a closet somewhere.
Thanks for your comments. I agree, Server2003/Terminal Services support is a must. The Samsung 930ND does support Terminal Services (Server 2003/2008) via RDP connection.

The RDP mode is also in the process of being certified for VMware View (aka VDI).

The benefit for Vmware View is that companies can deploy a highly integrated and stateless client that connects to Vmware View via RDP today, but then a seamless switch to PCoIP when available in a subsequent release of VMware View (in 2009 - but you would need to talk to Vmware for a more specific date). PCoIP supports all applications, all USB and All graphics (including Vista Aero, DirectX, OpenGL etc).
Posted on Reply
#7
Darkrealms
srobinson said:
Thanks for your comments. I agree, Server2003/Terminal Services support is a must. The Samsung 930ND does support Terminal Services (Server 2003/2008) via RDP connection.

The RDP mode is also in the process of being certified for VMware View (aka VDI).

The benefit for Vmware View is that companies can deploy a highly integrated and stateless client that connects to Vmware View via RDP today, but then a seamless switch to PCoIP when available in a subsequent release of VMware View (in 2009 - but you would need to talk to Vmware for a more specific date). PCoIP supports all applications, all USB and All graphics (including Vista Aero, DirectX, OpenGL etc).
Thats good to hear about the Server 2003/2008 support.
Pitty this wasn't already out. I just spent 700/800 per exam room to use them as RDP clients for one of our setups. Granted these are somewhat smaller monitors but I think it would have been worth it. How well would this support High Res/High Quality digital Xray/CT images?
Posted on Reply
#8
srobinson
Darkrealms said:
Thats good to hear about the Server 2003/2008 support.
Pitty this wasn't already out. I just spent 700/800 per exam room to use them as RDP clients for one of our setups. Granted these are somewhat smaller monitors but I think it would have been worth it. How well would this support High Res/High Quality digital Xray/CT images?
Extremely well when in PCoIP mode (the Samsung 930ND supports RDP and PCoIP). PC-over-IP supports all graphics and full display frame rates (DVI supports up to 60frames per second). Briefly, PCoIP bridges the user desktop over standard IP networks. There are no special drivers in the remoted PC and there are no drivers in the desktop appliance. Essentially, PCoIP compresses the user display and sends over IP to a dumb and stateless appliance on the desk (ie the Samsung display). USB and HD audio are transparently bridged over the network as well for a full desktop experience for the user. In most cases, the user cannot tell that their PC/workstation is not at their desk anymore, but is somewhere back in a datacenter. So Xray/CT image data never leaves the datacenter - only a compressed and encrypted image of the XRay/CT image. This saves a lot of bandwidth by not sending the large image files themselves and is very secure.

A key feature for medical applications is the ability for PCoIP to support progressive image refinement that builds to a fully lossless image. In a constrained network environment (ie WAN) PCoIP will get a highly compressed image to the user quickly then builds the image to an exact replica of the source.

The benefit is that a radiologist could flip through CT scan slices very quickly and then know that as soon as the image stops moving it is an exact replica of the source. There are ways to configure a "direct-to-loss-less" mode that ensures the image is always lossless even when the images are in motion - this mode takes more network bandwidth.

Teradici has worked with a Heathcare provider to test PCoIP in a radiologist workstation for MRI/CT scans - these tests were very successful. Here is a link to EVGA's site for the PCoIP kit that we used in the workstation: http://www.evga.com/articles/00448/default.asp

You may be interested in the unique USB security - a key reason why PCoIP was deployed at various government security agencies. Here is a video demo on Youtube: http://www.youtube.com/watch?v=amv_6OemfD4

To your earlier point, there are a variety of datacenter PC hardware, some more dense than others:
Rack workstations - extreme compute performance, any graphics cards etc (Dell R5400)
Blade workstations, blade PC's - up to extreme PC performance, but typically mobile graphics (IBM HC10, Verari Connexxus and Clearcube I9400)
Virtualized Servers - highest density with scalable user performance. Teradici licensed PCoIP to VMware (http://www.vmware.com/company/news/releases/teradici_vmworld08.html) for use in their upcoming VMware View products (VMware's new name for VDI).

A range of equipment in the datacenter would enable highly dense servers for nursing stations (delivered via RDP or in future VMware View with Software PCoIP) and extreme workstations for radiologist machines (delivered via hardware-accelerated PCoIP). Any PCoIP enabled desktop appliance could be used to dynamically access a range of compute resources in the datacenter via free-seating. So a radiologist could take over the station briefly to view a scan and when the nurse logs back in their session would come back instantly.

If you would like to contact me directly my email is srobinson@teradici.com.
Posted on Reply
#9
srobinson
XRay/CT images are handled extremely well when in PCoIP mode (the Samsung 930ND supports RDP and PCoIP). PC-over-IP supports all graphics and full display frame rates (DVI supports up to 60frames per second). Briefly, PCoIP bridges the user desktop over standard IP networks. There are no special drivers in the remoted PC and there are no drivers in the desktop appliance. Essentially, PCoIP compresses the user display and sends over IP to a dumb and stateless appliance on the desk (ie the Samsung display). USB and HD audio are transparently bridged over the network as well for a full desktop experience for the user. In most cases, the user cannot tell that their PC/workstation is not at their desk anymore, but is somewhere back in a datacenter. So Xray/CT image data never leaves the datacenter - only a compressed and encrypted image of the XRay/CT image. This saves a lot of bandwidth by not sending the large image files themselves and is very secure.

A key feature for medical applications is the ability for PCoIP to support progressive image refinement that builds to a fully lossless image. In a constrained network environment (ie WAN) PCoIP will get a highly compressed image to the user quickly then builds the image to an exact replica of the source.

The benefit is that a radiologist could flip through CT scan slices very quickly and then know that as soon as the image stops moving it is an exact replica of the source. There are ways to configure a "direct-to-loss-less" mode that ensures the image is always lossless even when the images are in motion - this mode takes more network bandwidth.

Teradici has worked with a Heathcare provider to test PCoIP in a radiologist workstation for MRI/CT scans - these tests were very successful. Here is a link to EVGA's site for the PCoIP kit that we used in the workstation: http://www.evga.com/articles/00448/default.asp

You may be interested in the unique USB security - a key reason why PCoIP was deployed at various government security agencies. Here is a video demo on Youtube: http://www.youtube.com/watch?v=amv_6OemfD4

To your earlier point, there are a variety of datacenter PC hardware, some more dense than others:
Rack workstations - extreme compute performance, any graphics cards etc (Dell R5400)
Blade workstations, blade PC's - up to extreme PC performance, but typically mobile graphics (IBM HC10, Verari Connexxus and Clearcube I9400)
Virtualized Servers - highest density with scalable user performance. Teradici licensed PCoIP to VMware (http://www.vmware.com/company/news/releases/teradici_vmworld08.html) for use in their upcoming VMware View products (VMware's new name for VDI).

A range of equipment in the datacenter would enable highly dense servers for nursing stations (delivered via RDP or in future VMware View with Software PCoIP) and extreme workstations for radiologist machines (delivered via hardware-accelerated PCoIP). Any PCoIP enabled desktop appliance could be used to dynamically access a range of compute resources in the datacenter via free-seating. So a radiologist could take over the station briefly to view a scan and when the nurse logs back in their session would come back instantly.

If you would like to contact me directly my email is srobinson@teradici.com.
Posted on Reply
#10
srobinson
The short answer: Xray/CT images are handled extremely well when in PCoIP mode (the Samsung 930ND supports RDP and PCoIP).

The long answer:
PC-over-IP supports all graphics and full display frame rates (DVI supports up to 60frames per second). Briefly, PCoIP bridges the user desktop over standard IP networks. There are no special drivers in the remoted PC and there are no drivers in the desktop appliance. Essentially, PCoIP compresses the user display and sends over IP to a dumb and stateless appliance on the desk (ie the Samsung display). USB and HD audio are transparently bridged over the network as well for a full desktop experience for the user. In most cases, the user cannot tell that their PC/workstation is not at their desk anymore, but is somewhere back in a datacenter. So Xray/CT image data never leaves the datacenter - only a compressed and encrypted image of the XRay/CT image. This saves a lot of bandwidth by not sending the large image files themselves and is very secure.

A key feature for medical applications is the ability for PCoIP to support progressive image refinement that builds to a fully lossless image. In a constrained network environment (ie WAN) PCoIP will get a highly compressed image to the user quickly then builds the image to an exact replica of the source.

The benefit is that a radiologist could flip through CT scan slices very quickly and then know that as soon as the image stops moving it is an exact replica of the source. There are ways to configure a "direct-to-loss-less" mode that ensures the image is always lossless even when the images are in motion - this mode takes more network bandwidth.

Teradici has worked with a Heathcare provider to test PCoIP in a radiologist workstation for MRI/CT scans - these tests were very successful. Here is a link to EVGA's site for the PCoIP kit that we used in the workstation: http://www.evga.com/articles/00448/default.asp

You may be interested in the unique USB security - a key reason why PCoIP was deployed at various government security agencies. Here is a video demo on Youtube: http://www.youtube.com/watch?v=amv_6OemfD4

To your earlier point, there are a variety of datacenter PC hardware, some more dense than others:
Rack workstations - extreme compute performance, any graphics cards etc (Dell R5400)
Blade workstations, blade PC's - up to extreme PC performance, but typically mobile graphics (IBM HC10, Verari Connexxus and Clearcube I9400)
Virtualized Servers - highest density with scalable user performance. Teradici licensed PCoIP to VMware (http://www.vmware.com/company/news/releases/teradici_vmworld08.html) for use in their upcoming VMware View products (VMware's new name for VDI).

A range of equipment in the datacenter would enable highly dense servers for nursing stations (delivered via RDP or in future VMware View with Software PCoIP) and extreme workstations for radiologist machines (delivered via hardware-accelerated PCoIP). Any PCoIP enabled desktop appliance could be used to dynamically access a range of compute resources in the datacenter via free-seating. So a radiologist could take over the station briefly to view a scan and when the nurse logs back in their session would come back instantly.

If you would like to contact me directly my email is srobinson@teradici.com.
Posted on Reply
#11
srobinson
The short answer: Xray/CT images are handled extremely well when in PCoIP mode (the Samsung 930ND supports RDP and PCoIP).

The long answer:
PC-over-IP supports all graphics and full display frame rates (DVI supports up to 60frames per second). Briefly, PCoIP bridges the user desktop over standard IP networks. There are no special drivers in the remoted PC and there are no drivers in the desktop appliance. Essentially, PCoIP compresses the user display and sends over IP to a dumb and stateless appliance on the desk (ie the Samsung display). USB and HD audio are transparently bridged over the network as well for a full desktop experience for the user. In most cases, the user cannot tell that their PC/workstation is not at their desk anymore, but is somewhere back in a datacenter. So Xray/CT image data never leaves the datacenter - only a compressed and encrypted image of the XRay/CT image. This saves a lot of bandwidth by not sending the large image files themselves and is very secure.

A key feature for medical applications is the ability for PCoIP to support progressive image refinement that builds to a fully lossless image. In a constrained network environment (ie WAN) PCoIP will get a highly compressed image to the user quickly then builds the image to an exact replica of the source.

The benefit is that a radiologist could flip through CT scan slices very quickly and then know that as soon as the image stops moving it is an exact replica of the source. There are ways to configure a "direct-to-loss-less" mode that ensures the image is always lossless even when the images are in motion - this mode takes more network bandwidth.

Teradici has worked with a Heathcare provider to test PCoIP in a radiologist workstation for MRI/CT scans - these tests were very successful. Here is a link to EVGA's site for the PCoIP kit that we used in the workstation: http://www.evga.com/articles/00448/default.asp
Posted on Reply
#12
srobinson
and a bit more...

You may be interested in the unique USB security - a key reason why PCoIP was deployed at various government security agencies. Here is a video demo on Youtube: http://www.youtube.com/watch?v=amv_6OemfD4

To your earlier point, there are a variety of datacenter PC hardware, some more dense than others:
Rack workstations - extreme compute performance, any graphics cards etc (Dell R5400)
Blade workstations, blade PC's - up to extreme PC performance, but typically mobile graphics (IBM HC10, Verari Connexxus and Clearcube I9400)
Virtualized Servers - highest density with scalable user performance. Teradici licensed PCoIP to VMware (http://www.vmware.com/company/news/releases/teradici_vmworld08.html) for use in their upcoming VMware View products (VMware's new name for VDI).

A range of equipment in the datacenter would enable highly dense servers for nursing stations (delivered via RDP or in future VMware View with Software PCoIP) and extreme workstations for radiologist machines (delivered via hardware-accelerated PCoIP). Any PCoIP enabled desktop appliance could be used to dynamically access a range of compute resources in the datacenter via free-seating. So a radiologist could take over the station briefly to view a scan and when the nurse logs back in their session would come back instantly.

If you would like to contact me directly my email is srobinson@teradici.com.
Posted on Reply
#13
Darkrealms
srobinson said:
...
If you would like to contact me directly my email is srobinson@teradici.com.
Thank you, I think I will do that today or tomorrow.

I'd be very interested in more details about the lossless video transfer. I'm glad to see Teradici has taken the steps in testing with CT/Xray imaging. This is becoming very important in the medical industry as PACS systems are popping up everywhere (one of my locations has the Fuji PACS system).

Another thought . . .
Perhaps a USB based Cisco VPN client that allows the system to connect to RDP or VMware remotely (yes I know now I'm getting greedy ; )
Posted on Reply
#14
srobinson
Darkrealms said:
Thank you, I think I will do that today or tomorrow.

I'd be very interested in more details about the lossless video transfer. I'm glad to see Teradici has taken the steps in testing with CT/Xray imaging. This is becoming very important in the medical industry as PACS systems are popping up everywhere (one of my locations has the Fuji PACS system).

Another thought . . .
Perhaps a USB based Cisco VPN client that allows the system to connect to RDP or VMware remotely (yes I know now I'm getting greedy ; )
VPN - we do not have a VPN client integrated at this point but it is under consideration. PCoIP is already IPSec traffic. But for RDP or PCoIP we work with external VPN routers - like the Cisco 871 etc. I expect there are lower cost routers that would work too, but we have only tested with small enterprise VPN routers at this point.

Some of the connection broker partners we work with have intelligence to dynamically select the connection type. If you connect via a PCoIP-enabled portal then it configures for PCoIP, if you connect from a TC or notebook it selects RDP. This is good for DR scenario's too.
Posted on Reply
#15
Jizzler
Nice. Can't wait to see the VMware implementation.

And once I see the 930RD and the EVGA kit for sale, new toys to play with :D
Posted on Reply
#16
lemonadesoda
Stu,

That's a great product you have there! Very nice. I can think of many applications for this... good stuff.

However, can you offer some further stats. I'm a bit confused.

1./ You say it can deliver HD video. However, what is the PCoIP rendered framerate for this? 1920x1080x3x8x24(fps)= 1000mpbs uncompressed. You could *just* deliver this on gigabit network assuming ZERO contention and 100% efficiency. You couldnt do that over wireless. However, perhaps you have some stats for average compression rates for video data oIP?

2./ "A key feature for medical applications is the ability for PCoIP to support progressive image refinement that builds to a fully lossless image" OK, that's a good example. What is the data in a typical radiology chart (24 bit greyscale on 3840x2400?). How long does it take to build up the full lossless image?

3./ For Crysis over .11n, what is the PCoIP rendered framerate. Obviously the framerate of the server is full speed. But what is being delivered at the user screen? (Assuming 1GB network and you can QoS 10% of that bandwidth i.e. data stream at 100mbps)

Questions are not criticisms. Not at all. I think the tech is fabulous. Just want to bring some real world data to the case studies so I can use objective data to understand where and how this tech can be used.
Posted on Reply
#17
srobinson
lemonadesoda said:
Stu,

That's a great product you have there! Very nice. I can think of many applications for this... good stuff.

However, can you offer some further stats. I'm a bit confused.

1./ You say it can deliver HD video. However, what is the PCoIP rendered framerate for this? 1920x1080x3x8x24(fps)= 1000mpbs uncompressed. You could *just* deliver this on gigabit network assuming ZERO contention and 100% efficiency. You couldnt do that over wireless. However, perhaps you have some stats for average compression rates for video data oIP?

2./ "A key feature for medical applications is the ability for PCoIP to support progressive image refinement that builds to a fully lossless image" OK, that's a good example. What is the data in a typical radiology chart (24 bit greyscale on 3840x2400?). How long does it take to build up the full lossless image?

3./ For Crysis over .11n, what is the PCoIP rendered framerate. Obviously the framerate of the server is full speed. But what is being delivered at the user screen? (Assuming 1GB network and you can QoS 10% of that bandwidth i.e. data stream at 100mbps)

Questions are not criticisms. Not at all. I think the tech is fabulous. Just want to bring some real world data to the case studies so I can use objective data to understand where and how this tech can be used.
Thanks for your detailed questions. I'll dig up some info and respond shortly.

A key note for the build to lossless is that the overall data sent is the maximum resolution*color depth of the original loss-less image (assuming a static image for all or part of the screen). Other technologies send a compressed JPEG of the display and the send subsequent JPEGs of the whole screen that are progressively less compression. That results in a significant increase in network bandwidth.

I'll get back to you shortly with more info.

Stu
Posted on Reply
#18
lemonadesoda
Sorry, this comment "(Assuming 1GB network and you can QoS 10% of that bandwidth i.e. data stream at 100mbps)" should also apply to the medical example #2.
Posted on Reply
#19
srobinson
lemonadesoda said:
Stu,

That's a great product you have there! Very nice. I can think of many applications for this... good stuff.

However, can you offer some further stats. I'm a bit confused.

1./ You say it can deliver HD video. However, what is the PCoIP rendered framerate for this? 1920x1080x3x8x24(fps)= 1000mpbs uncompressed. You could *just* deliver this on gigabit network assuming ZERO contention and 100% efficiency. You couldnt do that over wireless. However, perhaps you have some stats for average compression rates for video data oIP?

2./ "A key feature for medical applications is the ability for PCoIP to support progressive image refinement that builds to a fully lossless image" OK, that's a good example. What is the data in a typical radiology chart (24 bit greyscale on 3840x2400?). How long does it take to build up the full lossless image?

3./ For Crysis over .11n, what is the PCoIP rendered framerate. Obviously the framerate of the server is full speed. But what is being delivered at the user screen? (Assuming 1GB network and you can QoS 10% of that bandwidth i.e. data stream at 100mbps)

Questions are not criticisms. Not at all. I think the tech is fabulous. Just want to bring some real world data to the case studies so I can use objective data to understand where and how this tech can be used.
1./ PCoIP is a real-time protocol that ensures a responsive and interactive desktop delivery. Compression rates vary widely based on the characteristics of the images on the screen. However, what we typically see for HD video's is a little better than 12:1 compression ratio. Text and office graphics exhibit significantly higher compression ratio's than this (also see note in 3./ that talks about additional compression during periods of network constraints which increases the compression ratio but may result in a temporal image quality reduction - which most users cannot see.)

Keep in mind, that there is a difference comparing compression ratio's of real-time compression and non-real time/off-line compression schemes. Non-real time and Off-line compression can compress data across multiple frames which increases the compression ratio - however, you need to hold on to the frames for a little while in order to compress across the multiple frames. This added latency significantly impacts the user experience and system responsiveness. So real time compression ratio's are a bit lower, but highly responsive compared to non-real time compression schemes.

2./ The very high-end color displays are 3840x2400 at 24 bit pixels. The current PCoIP hardware (called 'TERA1') does not support displays of this resolution. The maximum supported today is 2560x2048 at 30Hz which is sufficient for most of the medical display market. Grayscale for typical medical displays are 10bit pixels. The amount of time it takes for an image to build to loss-less depends on the network bandwidth available. With a 100Mbps link the image will build to loss-less in a fraction of a second. It would take a couple of seconds on a 10Mbps link. Note that PCoIP builds to loss-less with the minimum bandwidth possible (unlike ICA speedscreen which sends progressively higher quality images of the whole screen and multiplies the bandwidth required.).

3./ Computer generated images like CAD wireframes and high-end computer games are more difficult to compress than HD video, so the compression ratio's tends to drop to between 6:1 to 5:1 to be perceptually lossless. With this compression ratio a 100Mbps data stream would deliver between 20-25Million pixels per second. However, when the PCoIP system detects network constraint it begins to increase the compression which will still allow support for around 24fps but with some loss of visual quality. This drop in visual quality would be almost unnoticeable for a 1280x720 display. To support 24fps at the perceptually lossless quality level would require more than 200Mbps for a HD display (1920x1080). Higher frame rates are possible depending on the bandwidth available and the quality settings. The TERA1 PCoIP hardware accelerator can generate more than 200Mbps if needed. When I have played action games on an HD TV I set the quality below perceptually loss-less and could see some compression (ie fuzzy-ness) when I specifically looked for it. However, when actually playing the game I did not notice it at all.

While PCoIP can take advantage of network QoS, we do not require it since our compression adapts to the available network bandwidth.

I hope this information helps. Let me know if you have more questions, I'd be happy to respond.

Cheers,
Stu
Posted on Reply
#20
Jizzler
EVGA lists CDW as a seller of their Host/Portal setup, though the sales rep I talked to yesterday had no idea of when/if it should arrive.

Stu, you wouldn't happen to know more, would ya? :)
Posted on Reply
#21
lemonadesoda
Thanks for the info, Stu!

Sounds like you have a serious video compressor/decompressor HARDWARE there? Is that correct? Hardware driven video data compression? "Live 12:1 video compression at 24-30fps" on the server, stream oIP, and decompression at the terminal is pretty serious and needs some (relatively) serious hardware.

Any data on what technology/chips are used for that?
Posted on Reply
#22
srobinson
Jizzler said:
EVGA lists CDW as a seller of their Host/Portal setup, though the sales rep I talked to yesterday had no idea of when/if it should arrive.

Stu, you wouldn't happen to know more, would ya? :)
My understanding is that inventory is enroute and should be available in the next couple of weeks.
Posted on Reply
#23
srobinson
lemonadesoda said:
Thanks for the info, Stu!

Sounds like you have a serious video compressor/decompressor HARDWARE there? Is that correct? Hardware driven video data compression? "Live 12:1 video compression at 24-30fps" on the server, stream oIP, and decompression at the terminal is pretty serious and needs some (relatively) serious hardware.

Any data on what technology/chips are used for that?
Yes, serious hardware and a custom compression that is optimized to handle any PC display content (video, text, still images, computer generated graphics) - but in a tiny, low power chip package of 31mm x 31mm and 5Watts typical. The TERA1 chipset is in 130nm process and supports dual DVI display inputs with up to 1900x1200 (or single display to 25x20) resolution. Our chip team comes from Intel, Broadcom, PMC-Sierra and nVidia.

The TERA1 also supports line-rate encryption and network offload (similar to a TOE) with a bandwidth range of 1Mbps peak and can go over 200Mbps peak. The average bandwidth used is far less than the peak - in-house we use an average of ~200Kbps per desktop (we only count hours of office operation for the average calcualtion). We have about 1/3 of our company with their design workstations back racked (a mix of workstation blades "HC10s" from IBM and R5400 Precision rack workstations from Dell) and the average bandwidth during office hours is about 200Kbps per user since we only send changed pixels and then apply compression.
Posted on Reply
#24
Jizzler
Cool. I'll keep an eye out for it.

Also, with all these video card manufacturers as partners, suggest to them that they sell Hosts with a GPU onboard. With five of these twin 1U setups I could replace the 10 workstations in IT :)

Everyone else can share a server with a VMware View setup :D
Posted on Reply
#25
srobinson
Jizzler said:
Cool. I'll keep an eye out for it.

Also, with all these video card manufacturers as partners, suggest to them that they sell Hosts with a GPU onboard. With five of these twin 1U setups I could replace the 10 workstations in IT :)

Everyone else can share a server with a VMware View setup :D
:D Great idea - actually there is a program for this but it is not announced yet. If you want, I may be able to introduce you to the graphics company product manager responsible. srobinson@teradici.com

Would you be interested in a host card that has a Media Expansion Card (PCIe+sDVO) interface that allows the use of Intel integrated graphics? (http://www.intel.com/Assets/PDF/whitepaper/307620.pdf). I don't think it fits the Twin motherboard you linked, but there are 1U PC's that have the integrated graphics that it could fit into. Would that be dense enough for you?
Posted on Reply
Add your own comment