By using this site, you agree to our Privacy Policy and our Terms of Use. Close

Forums - Gaming Discussion - xCloud reportedly bests Stadia in early latency tests at E3, 4 ms vs 21ms of added latency

kirby007 on 13 June 2019
SpokenTruth said:
kirby007 said:

It was a xb1x vs xb1s

Edit : i like how you dont want to see past your own viewpoint

So let's narrow down the specifics.  It's not that I don't want to see past my own viewpoint, it's that they are presenting data that can't possibly happen and I'm trying to address where the discrepancy can be understood.

Xbox One S local - 60 ms

Xbox One X remote - 64 ms

Are this the specifics?  If so, do we have any tests demonstrating the input lag between the two consoles so that we can deduce the transmission latency?  By the way, I also just read they did this test over Wi-Fi at the Microsoft Theater.  That's going to add at least 1-2 ms with a lot of fluctuations.

Now thats is a good question, now my next part is of memory from the stadia digital foundry vid so i could be wrong but i think the xb1x was at around 40-45ms locally



 "I think people should define the word crap" - Kirby007

Join the Prediction League http://www.vgchartz.com/predictions

Instead of seeking to convince others, we can be open to changing our own minds, and seek out information that contradicts our own steadfast point of view. Maybe it’ll turn out that those who disagree with you actually have a solid grasp of the facts. There’s a slight possibility that, after all, you’re the one who’s wrong.

Around the Network
haxxiy on 13 June 2019

SvennoJ has actually a pretty good point. A mere 4 ms of added latency would beat the speed of light for anything further away than some 600 km. And optic fibers are slower than the speed of light, plus all the hardware and streaming involved on both ends.

Unless servers are really, really decentralized, which seems... expensive for an affordable streaming service, there's no way even the optimal, best case added latency is just 4 ms. The xCloud being demonstrated was probably hosted from a neighboring borough, or something.



 

 

 

 

 

kirby007 on 13 June 2019
haxxiy said:

SvennoJ has actually a pretty good point. A mere 4 ms of added latency would beat the speed of light for anything further away than some 600 km. And optic fibers are slower than the speed of light, plus all the hardware and streaming involved on both ends.

Unless servers are really, really decentralized, which seems... expensive for an affordable streaming service, there's no way even the optimal, best case added latency is just 4 ms. The xCloud being demonstrated was probably hosted from a neighboring borough, or something.

He has if it would have been so simple 

Edit: there are many variables which make taking that 4ms at facevalue hilariously wrong

Last edited by kirby007 - on 13 June 2019

 "I think people should define the word crap" - Kirby007

Join the Prediction League http://www.vgchartz.com/predictions

Instead of seeking to convince others, we can be open to changing our own minds, and seek out information that contradicts our own steadfast point of view. Maybe it’ll turn out that those who disagree with you actually have a solid grasp of the facts. There’s a slight possibility that, after all, you’re the one who’s wrong.

psychicscubadiver on 13 June 2019

Begun, the Cloud Wars have,



SvennoJ on 13 June 2019
kirby007 said:
SpokenTruth said:

I thought about that too but didn't they say the remote rendering was basically just a server form of the Xbox One hardware?  So whether you render it locally or remotely, if it's largely the same hardware, you should largely get the same render time.  Only now you have all that extra stuff I just said above.  So again, how?

It was a xb1x vs xb1s

Edit : i like how you dont want to see past your own viewpoint

Microsoft has built custom hardware for xCloud in its data centers, combining four Xbox One S consoles into single server blades.

No XBox One X hardware. Perhaps overclocked server blades to bleed off a few milli seconds but it's still not realistic given the actual transmission time is much higher combined with video compression and decoding. It's not just ping time, it actually has to send data which takes time as well. Ping just sends 32 bytes back and forth. If you need 30 mbps and you have 60 mbps connection, transmission per frame already takes 8.3 ms under perfect conditions, on top of the travel time. That's just from the start of transmission until all data for 1 frame has been received on a connection that's twice as fast as the requirement. It's BS.



Around the Network
kirby007 on 13 June 2019
SvennoJ said:
kirby007 said:

It was a xb1x vs xb1s

Edit : i like how you dont want to see past your own viewpoint

Microsoft has built custom hardware for xCloud in its data centers, combining four Xbox One S consoles into single server blades.

No XBox One X hardware. Perhaps overclocked server blades to bleed off a few milli seconds but it's still not realistic given the actual transmission time is much higher combined with video compression and decoding. It's not just ping time, it actually has to send data which takes time as well. Ping just sends 32 bytes back and forth. If you need 30 mbps and you have 60 mbps connection, transmission per frame already takes 8.3 ms under perfect conditions, on top of the travel time. That's just from the start of transmission until all data for 1 frame has been received on a connection that's twice as fast as the requirement. It's BS.

4x xb1s=xb1x  so even if your technically right, i still count that as one for me... but listen if this all is such BS why are they working on it? And pouring more and more money in it? Right because it works. But you know what im going to agree with you and assume its BS, can only be pleasantly suprised that way



 "I think people should define the word crap" - Kirby007

Join the Prediction League http://www.vgchartz.com/predictions

Instead of seeking to convince others, we can be open to changing our own minds, and seek out information that contradicts our own steadfast point of view. Maybe it’ll turn out that those who disagree with you actually have a solid grasp of the facts. There’s a slight possibility that, after all, you’re the one who’s wrong.

SvennoJ on 13 June 2019
kirby007 said:
SvennoJ said:

Microsoft has built custom hardware for xCloud in its data centers, combining four Xbox One S consoles into single server blades.

No XBox One X hardware. Perhaps overclocked server blades to bleed off a few milli seconds but it's still not realistic given the actual transmission time is much higher combined with video compression and decoding. It's not just ping time, it actually has to send data which takes time as well. Ping just sends 32 bytes back and forth. If you need 30 mbps and you have 60 mbps connection, transmission per frame already takes 8.3 ms under perfect conditions, on top of the travel time. That's just from the start of transmission until all data for 1 frame has been received on a connection that's twice as fast as the requirement. It's BS.

4x xb1s=xb1x  so even if your technically right, i still count that as one for me... but listen if this all is such BS why are they working on it? And pouring more and more money in it? Right because it works. But you know what im going to agree with you and assume its BS, can only be pleasantly suprised that way

The 4ms over XBox One S is BS, 66ms latency for streaming is BS. Doesn't mean it still works at 120ms, some games even have that locally. Yes it works, but MS is always quick with fudging numbers. I even managed to play Trackmania turbo via remote play (in home) with 120ms lag, not very well, but it was playable.

Btw using all 4 xbox one S for the same player doesn't make it any faster. It's still receive input -> update game state -> render frame -> compress video -> transmit -> decompress and display. Using more consoles to render the frames doesn't make the frame appear faster, unless you rewrite the rendering software to render in tiles. Which is possible, just more costly to do.

That does mean that games specifically written with xCloud or Stadia in mind have the potential to run better.

Last edited by SvennoJ - on 13 June 2019

kirby007 on 13 June 2019
SpokenTruth said:
kirby007 said:

4x xb1s=xb1x  so even if your technically right, i still count that as one for me... but listen if this all is such BS why are they working on it? And pouring more and more money in it? Right because it works. But you know what im going to agree with you and assume its BS, can only be pleasantly suprised that way

He didn't say the concept of xCloud was bullshit, he said the latency comparison numbers are bullshit.  And he's right.  Both of us have given you real world information from the hardware to the process of remote delivery and it doesn't add up.  It's like adding 10 + 10 and claiming its 15.  

Let me put it this way if the server was in a case near the techdemo instead of 300miles away, thats still amazing the only step skipped is the 300 miles vs 20 meters



 "I think people should define the word crap" - Kirby007

Join the Prediction League http://www.vgchartz.com/predictions

Instead of seeking to convince others, we can be open to changing our own minds, and seek out information that contradicts our own steadfast point of view. Maybe it’ll turn out that those who disagree with you actually have a solid grasp of the facts. There’s a slight possibility that, after all, you’re the one who’s wrong.

kirby007 on 13 June 2019

I live 124miles from the nearest datapoint which means 720p will have a 1ms traveltime based on this test going with 69% of SoL
Seeing as it took about 7-8 seconds hmmm that 400m could be true



 "I think people should define the word crap" - Kirby007

Join the Prediction League http://www.vgchartz.com/predictions

Instead of seeking to convince others, we can be open to changing our own minds, and seek out information that contradicts our own steadfast point of view. Maybe it’ll turn out that those who disagree with you actually have a solid grasp of the facts. There’s a slight possibility that, after all, you’re the one who’s wrong.

SvennoJ on 13 June 2019
kirby007 said:
I live 124miles from the nearest datapoint which means 720p will have a 1ms traveltime based on this test going with 69% of SoL
Seeing as it took about 7-8 seconds hmmm that 400m could be true

You have a single fibre optic cable running from your tv to the data center?

Want to see how 'stable' and fast it actually is?
http://www.azurespeed.com/
That's only pinging 32 bytes of data back and forth, not actually transmitting full frames of video data.

That's my nearest data center, less than 60 miles from where I live.