Jump to content

New display server configuration


Ioannis Glantzis

Recommended Posts

Hello all!

I would like to build a new display server with 6 x 4K/60 outs.

I haven't previous experience so I need your help to choose a robust configuration.

This is my first attempt from what I have read from the forum.

Motherboard ASUS ROG STRIX Z390-F
CPU Intel Core i9 9900K
RAM Crucial Ballistix Sport DDR4 2x16Gb 3200MHz
OS SSD Samsung 860 Pro 512GB
MEDIA NAND M2 Samsung 970 Pro 1Tb
GPU RADEON PRO WX9100
GENLOCK AMD ATI FirePro S400
OS Win LTSC 2019

I really appreciate for your opinions and comments.

Link to comment
Share on other sites

if i were you i prefer x299 platform because of max pci lanes . 9900k has only 16 pci lanes(wx9100 use 16 pci lanes) for example i7 9800x has 44 pci lanes it means in the future if you want to add uour system capture  card your gpu works 8x speed . normally it works 16x . and other reason is ram  z390 has dual channel  x299  platform has quad channel ram  .as a result it depends on what you need  if you dont need use capture card  you can use this system. finally samsung m2 ssd  works hot .you should buy m2 ssd cooler .

 

Link to comment
Share on other sites

  • 3 weeks later...
  • 1 month later...
On 9/6/2019 at 12:05 AM, bambam said:

if i were you i prefer x299 platform because of max pci lanes . 9900k has only 16 pci lanes(wx9100 use 16 pci lanes) for example i7 9800x has 44 pci lanes it means in the future if you want to add uour system capture  card your gpu works 8x speed . normally it works 16x . and other reason is ram  z390 has dual channel  x299  platform has quad channel ram  .as a result it depends on what you need  if you dont need use capture card  you can use this system. finally samsung m2 ssd  works hot .you should buy m2 ssd cooler .

 

Currently any chipset available for Intel platform supports a maximum of 24 PCI-e lanes. Therefore, it doesn't matter how many lanes the CPU supports. Any reasonably priced Intel setup be it consumer or server grade, is limited to 24 lanes.

For instance, the display PC in piccinaezio's example has 24 on board. AMD Pro WX9100 uses 16, Datapath VisionRGB-E-S2 uses 4, BM Declink uses 8, each Samsung NVMe SSDs use 4 lanes of PCI-e lanes, in total one would need 36 lanes supported by the chipset in order to realize the full potential of the hardware listed.

So, there always be compromises in setup and performance of any WO server, including Dataton's own.

This is why it is crucial to get the OS and BIOS perfectly tuned for achieving least amount of bottleneck, especially for complex setups with demanding media files.

Link to comment
Share on other sites

On 10/30/2019 at 12:26 PM, HouseofGusto said:

Currently any chipset available for Intel platform supports a maximum of 24 PCI-e lanes. Therefore, it doesn't matter how many lanes the CPU supports. Any reasonably priced Intel setup be it consumer or server grade, is limited to 24 lanes.

For instance, the display PC in piccinaezio's example has 24 on board. AMD Pro WX9100 uses 16, Datapath VisionRGB-E-S2 uses 4, BM Declink uses 8, each Samsung NVMe SSDs use 4 lanes of PCI-e lanes, in total one would need 36 lanes supported by the chipset in order to realize the full potential of the hardware listed.

So, there always be compromises in setup and performance of any WO server, including Dataton's own.

This is why it is crucial to get the OS and BIOS perfectly tuned for achieving least amount of bottleneck, especially for complex setups with demanding media files.

Thats not correct  check Asus ws sage x299 İts support 4 ways slı and crossfire full performance with correct cpu 

you can check it from https://www.asus.com/us/Motherboards/WS-X299-SAGE/

Link to comment
Share on other sites

Hi Bambam.

Let me make things clearer then;

1. PCI-e lanes provided by CPU provides communication only for GPUs at the moment.

2. Therefore, total count of lanes and resulting data bandwidth is only relevant to tasks between the CPU and GPU/S such as, gameplay, rendering, CAD, etc. where a very limited dataset is given to the GPU (usually and algorithm and accompanying parameters) and it is expected to spit out huge data(video if you will) output as the result. Not a streaming output fed by storage.

3. As for the chipset provided PCI-e lanes, they are used for data transmission between CPU, peripherals, storage, network I/O, in short; everything except the GPU/S.

3. In WO's case it is more of a matter of how fast and steady the storage can provide content for the GPU rather than how fast GPU can render the content. This is true especially for challenging content with huge dimensions, odd frame rates, transparency data and high color depth. This makes chipset provided/managed bandwidth critical and limiting factor with WO.

4. This is why WO currently does dot support multiple GPUs running simultaneously. You can easily do the math online for yourself. There are plenty of technical resources on many websites for this. This is also why people go to multiple server configurations for very large resolution or multiple screen applications instead of sticking their servers with four WX9100 cards to get 24 4K outputs.

5. Mind you, I did not take capture and sync cards into account above which are sharing bandwidth, too.

Of course, I am referring to hardware usage in reference to WO. If we talk about playing CS on 160 Hz monitor with HDR support then that's another story 🙂

Hope this helps.

Link to comment
Share on other sites

  • 2 weeks later...
  • Moderator
On 11/5/2019 at 1:24 AM, HouseofGusto said:

...

4. This is why WO currently does dot support multiple GPUs running simultaneously....

While those are very valid arguments, the main reason is the DirectX 3D multiple output native Windows function that provides multiple outputs on all gpu types is restricted to a single graphics card / six outputs by Windows.To get multiple output cards / more than six outputs under DirectX 3D, a different multiple output model that is not native to widows is required (eyefinity, mosaic, etc.).

Link to comment
Share on other sites

  • 3 weeks later...
On 11/19/2019 at 9:01 PM, jfk said:

While those are very valid arguments, the main reason is the DirectX 3D multiple output native Windows function that provides multiple outputs on all gpu types is restricted to a single graphics card / six outputs by Windows.To get multiple output cards / more than six outputs under DirectX 3D, a different multiple output model that is not native to widows is required (eyefinity, mosaic, etc.).

In which case users of Eyefinity and Mosaic primarily get a tiled large image of non-standard resolution and/or shape not increased numbers of full resolution and independent output streams. Those technologies both designed to utilize software mapping of a single video stream output up to six screens/projectors, while allowing a total resolution limited to when using a single display. They provide simple edge blending, display tiling for basic use and distributing Windows desktop into multiple screens.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...