updating image data through bokeh server is slow

Hi,

I would like to setup a continuously updating camera view. It does not have to have the actualy device frame rate (60fps) but right now I am getting only 1fps.

Specifically, the image data is 256x320px with one byte per pixel. The JSON serialization seems to be quite fast. The bottleneck appears to be communicating with the bokeh-server, i.e.,

Using this as an example http://bokeh.pydata.org/en/latest/docs/user_guide/server.html#reacting-to-user-interactions ,

each invocation of

cursession().store_objects(ds)

takes up to a second to return when transferring images.

Is there some setting to change how fast the server responds?

I am not using Blaze. Would that help?

Best,

Nikolas

Nikolas,

Yes, there are a few recent messages regarding ongoing work on the bokeh server you can look up on the mailing list. TLDR: There will be a new server that uses a web socket wire protocol exclusively and supports binary transmission of numpy, pandas, etc. directly into JS typed arrays, as well as streaming/partial updates as first class messages in the protocol. But, this is a couple of months away (targeting 0.10 release end of September, but this might slip)

Bryan

···

On Aug 18, 2015, at 4:16 PM, [email protected] wrote:

Hi,

I would like to setup a continuously updating camera view. It does not have to have the actualy device frame rate (60fps) but right now I am getting only 1fps.
Specifically, the image data is 256x320px with one byte per pixel. The JSON serialization seems to be quite fast. The bottleneck appears to be communicating with the bokeh-server, i.e.,
Using this as an example http://bokeh.pydata.org/en/latest/docs/user_guide/server.html#reacting-to-user-interactions ,
each invocation of
    
    cursession().store_objects(ds)

takes up to a second to return when transferring images.
Is there some setting to change how fast the server responds?

I am not using Blaze. Would that help?

Best,

Nikolas

--
You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/a98bb88a-6e14-4b0d-a003-2d77e706d4d1%40continuum.io.
For more options, visit https://groups.google.com/a/continuum.io/d/optout.

Hey Bryan,

thanks for the heads up! That sounds good!

Nikolas

···

On Tuesday, August 18, 2015 at 2:22:39 PM UTC-7, Bryan Van de ven wrote:

Nikolas,

Yes, there are a few recent messages regarding ongoing work on the bokeh server you can look up on the mailing list. TLDR: There will be a new server that uses a web socket wire protocol exclusively and supports binary transmission of numpy, pandas, etc. directly into JS typed arrays, as well as streaming/partial updates as first class messages in the protocol. But, this is a couple of months away (targeting 0.10 release end of September, but this might slip)

Bryan

On Aug 18, 2015, at 4:16 PM, [email protected] wrote:

Hi,

I would like to setup a continuously updating camera view. It does not have to have the actualy device frame rate (60fps) but right now I am getting only 1fps.

Specifically, the image data is 256x320px with one byte per pixel. The JSON serialization seems to be quite fast. The bottleneck appears to be communicating with the bokeh-server, i.e.,

Using this as an example http://bokeh.pydata.org/en/latest/docs/user_guide/server.html#reacting-to-user-interactions ,

each invocation of

cursession().store_objects(ds)

takes up to a second to return when transferring images.

Is there some setting to change how fast the server responds?

I am not using Blaze. Would that help?

Best,

Nikolas


You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/a98bb88a-6e14-4b0d-a003-2d77e706d4d1%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.