I'm out of ideas for news-titles

[Swordfish II]

It's not the device that's wrong; in a way it's your point of view. The modern mobile device isn't built to run under full load for a long time. That's what servers are built for. The device is built to deliver a smooth handling experience at moderate power consumption. To achieve that, there's the possibility of firing short spikes of performance way above the usual load. Like nitro boost in your car. It gives you more power than it could normally, not less than it should. But it can do that only for a short time.

Then it should be marketed that way.

Saying "dual core 1.5ghz" is not true when the OMAP can only run 1.5ghz for a few seconds before throttling.

An accurate statement would be "dual core 1ghz, with boost to 1.5ghz"

This isn't anyone fault except TI and the rest of the mobile chip makers, since they are the only ones that count seconds as "mission accomplished." Imagine if computers couldnt sustain their max clock speed?

 

[Linux-SWAT]

The OMAP5 _runs_ even at 1.7GHz when properly cooled.
So "dual core 1.5ghz" is a valid statement.

 

[Swordfish II]

The OMAP5 _runs_ even at 1.7GHz when properly cooled.
So "dual core 1.5ghz" is a valid statement.

Then why isn't it marketed as "dual core 1.7ghz?"

 

[Linux-SWAT]

It was in the very first days. At 1.7GHz.
They probably switched back to 1.5GHz because it was engineered in the pre-active cooling era, the CPU was hotter than expected, and the customers designs were on the way, using only passive.

This is where I'm also a bit upset for the Pyra because I was one of the very few who warned ED about the temperature problem. It was years ago.
I'm not an electronics expert, but I have enough knowledge to anticipate that.
Not to mention the modular design of the Pyra. It's blatant that if the CPU can be upgraded for a better one, the better will produce more heat...

That said, don't underestimate the OMAP5, even not at fullspeed. I have two devboards and this CPU really rocks.
Adding to this, look at all the miracles done by notaz, ptitSeb and many others on the Pandora's OMAP3. There's nothing to worry about.

 

[Swordfish II]

Tbh I don't worry, and I don't underestimate.

I still use my Pandora daily, and I don't care about clock speed, I care about abilities and usage. Which the Pyra will rock at.

I do think mobile chip manufacturers are misleading. No phone will run at its touted clock speed longer than seconds, so that isn't it's realistic sustainable speed, now is it?

Saying "proper cooling" is also misleading. What is proper? Fans? Water? CO2? The OMAP can probably be tons quicker if you add in extreme cooling.

It would be like saying this "cars speed is 80" then you find out 2 minutes after hitting 80 it knocks things back to 50. So really 50 is the speed with burst to 80.

 

[fusion_power]

It's not the device that's wrong; in a way it's your point of view. The modern mobile device isn't built to run under full load for a long time. That's what servers are built for. The device is built to deliver a smooth handling experience at moderate power consumption. To achieve that, there's the possibility of firing short spikes of performance way above the usual load. Like nitro boost in your car. It gives you more power than it could normally, not less than it should. But it can do that only for a short time.

Yes, this seems the overall philosophy with todays mobile chipsets. It may make sense for these kind of devices and running a CPU low when not under load is just logical. But in case of the advertised top-speed parameters it still leads easily to some confusion. And it does not help to keep small devices cool of course. I don't have a smartphone and I am actualy a PC only user (except for the Pandora of course). So I come from a platform that can run at the target CPU speed for a long time, which makes it a little bit weird for me when I see the mobile chip speed-philosophy. I hope the Pyra get's propper power-management to keep the SoC cool and efficient. This took a while in case of the Pandora I remember.
So or so, I prefer cool'n quiet devices so I can live with an chipset that may not be the most powerful. But it should have enough capacity to do every task it should do, including longer gaming sessions without melting the device.

 

[vcoleiro1]

I do think mobile chip manufacturers are misleading. No phone will run at its touted clock speed longer than seconds, so that isn't it's realistic sustainable speed, now is it?

Well IMO, SOC manufacturers are not being missleading at all. SOCs are made for a wide range of devices and not just phones. They are found in all sorts of applications other than phones such as Auto to handheld gaming devices like the Shield portable. By the way, the Shield Portable has a Tegra 4 in it which is a quad core ARM Cortex A15, it uses both heatsink and fan to prevent it throttling. Anyway, it's fair for them to market the top speed of the SOC, because it can be achieved in some applications where better cooling can be applied

I thnk if you were going to point a finger at something, then it would be fairer to point it at phone makers and resellers who tout missleading CPU speed specs in their marketing .for smartphones

 

[Linux-SWAT]

I totally agree.

 

[Swordfish II]

Well IMO, SOC manufacturers are not being missleading at all. SOCs are made for a wide range of devices and not just phones. They are found in all sorts of applications other than phones such as Auto to handheld gaming devices like the Shield portable. By the way, the Shield Portable has a Tegra 4 in it which is a quad core ARM Cortex A15, it uses both heatsink and fan to prevent it throttling. Anyway, it's fair for them to market the top speed of the SOC, because it can be achieved in some applications where better cooling can be applied

I thnk if you were going to point a finger at something, then it would be fairer to point it at phone makers and resellers who tout missleading CPU speed specs in their marketing .for smartphones

So then CPU makers would be OK in your eyes saying "7ghz clock speed." Because that is possible with liquid nitrogen. Guess it's not their fault you can't afford it and it is unrealistic.

No that is not achievable in normal use, not sustainable and a load of crap. Luckily CPU mfg don't pull that shady crap.

Strap a rocket engine on a Yugo and you don't get to suddenly advertise a normal Yugo's top speed as 400mph (because in some cases it can be).

 

[ClockworkCoder]

I thnk if you were going to point a finger at something, then it would be fairer to point it at phone makers and resellers who tout missleading CPU speed specs in their marketing .for smartphones

Absolutely. If phones did run full tilt all the time, the average battery life would be in minutes, not hours.

Trouble is that people are brainwashed into thinking that it's reasonable to expect ultra powerful phones that only need charged once a day.

Sometimes I still miss my 3310 that could go weeks without a charge...

 

[vcoleiro1]

So then CPU makers would be OK in your eyes saying "7ghz clock speed." Because that is possible with liquid nitrogen. Guess it's not their fault you can't afford it and it is unrealistic.

No that is not achievable in normal use, not sustainable and a load of crap. Luckily CPU mfg don't pull that shady crap.

Strap a rocket engine on a Yugo and you don't get to suddenly advertise a normal Yugo's top speed as 400mph (because in some cases it can be).

Do you see SOC makers totuing speeds that are only accessible via liquid nitrogen cooling - no, you don't.

They are only touting speeds which are accesible with cooling that can be achived in the realistic gamut of the SOCs applications.

To get back on point, you made the comment that TI were being untruthfull in saying the OMAP5 can hit 1.5ghz. IMO, they were being truthfull. As I said, in the range of applications SOCs can be used these days, a number of them allow for heatsinks and fans (like for example Auto and handheld gaming devices - shield portable)

As I said before , if you wont to blame someone , blame the missleading marketing of some phone makers and resllers.

 

[Swordfish II]

Absolutely. If phones did run full tilt all the time, the average battery life would be in minutes, not hours.

That's fine... But that isn't even possible
[doublepost=1504704770,1504704629][/doublepost]

Do you see SOC makers totuing speeds that are only accessible via liquid nitrogen cooling - no, you don't.

They are only touting speeds which are accesible with cooling that can be achived in the gamut of the SOCs applications.

Keep giving them a cop out which is not feasible for the use case of their chips and they will continue to jerk you around.

When was the last (or first) time a phone had active cooling?

So active cooling might as will be liquid cooling or liquid nitrogen.

 

[vcoleiro1]

That's fine... But that isn't even possible
[doublepost=1504704770,1504704629][/doublepost]

Keep giving them a cop out which is not feasible for the use case of their chips and they will continue to jerk you around.

When was the last (or first) time a phone had active cooling?

So active cooling might as will be liquid cooling or liquid nitrogen.

You seem stuck on this idea that SOCs are just for phones - that's what I tried to explain to you isn't the case. When you look at the range of things SOCs are used for beyond phones, then yes it is perfectly fine for TI to say the top speed is 1.5ghz

 

[Swordfish II]

You seem stuck on this idea that SOCs are just for phones - that's what I tried to explain to you isn't the case. When you look at the range of things SOCs are used for beyond phones, then yes it is perfectly fine for TI to say the top speed is 1.5ghz

... That is 90+% of the use case

 

[vcoleiro1]

... That is 90+% of the use case

Name a phone that uses the Omap5?

For that matter, name a phone that uses the Nvidia Tegra X1? Yet Nvidia are raking it in on sales almost solely to Nintendo and to Auto

By the way, do you remember when ED mentioned early on that he chose the Omapa5 (despite TI more or less getting out the SOC business), becuase they told him since they supplied the SOC to Auto makers, they guaranteed 7 years of supply. In the end , that was TI's focus for their SOCs -> Auto. Who knows if it actually eventuated and if TI actually did supply Omap5s to Autos

 

[Swordfish II]

Name a phone that uses the Omap5?

For that matter, name a phone that uses the Nvidia Tegra X1? Yet Nvidia are raking it in on sales almost solely to Nintendo and to Auto

By the way, do you remember when ED mentioned early on that he chose the Omapa5 (despite TI more or less getting out the SOC business), becuase they told him since they supplied the SOC to Auto makers, they guaranteed 7 years of supply. In the end , that was TI's focus for their SOCs -> Auto. Who knows if it actually eventuated and if TI actually did supply Omap5s to Autos

OMAP 5 intentions:
https://www.google.it/amp/gizmodo.c...-core-omap5-chips-are-gonna-do-crazy-shit/amp

http://www.forbes.com/sites/elizabe...i-says-first-omap-5-device-could-debut-in-q4/

Know why they didn't? Power consumption. Again, marketed for mobile, you don't get to claim the world when fans are not a reality for a phone, the pyra, or other ACTUAL mobile use cases.

The Tegra was a victim of many factors.

You are actively trying to shift the goal posts from the intended use of mobile chips to the fringe cases.

Mobile chips should be marketed with sustainable speeds for confined, non actively cooled uses. Anything else is false advertising at best and fraud at worst.

 

[Wrath Of Khan]

Yeah, would take a 2 gigabyte unit in place of a 4 gigabyte one. 4 gig later could be an option if it worked, via an upgrade or something. Or just 2 gigabyte and I will put up and shut up

 

[Letalis Sonus]

Again, marketed for mobile, you don't get to claim the world when fans are not a reality for a phone, the pyra, or other ACTUAL mobile use cases.

The mobile variant has never hit the market, we're using an automotive grade OMAP5. Guess what, the automotive industry has totally different requirements.

 

[Grench]

SizeWeightVolume Vs MassiveComputingPower

The Pyra is a mobile computing device. What it will be able to do by itself is pretty amazing. However, there will always be applications that are better performed using the capabilities of desktops, servers and supercomputers.

So many of the, 'but I need huge RAM to do X application' calls tend to be for applications that aren't overly suited to stand alone mobile. Many of them could/should/will work - sort of, but the OMAP 5 CPU simply isn't going to keep up very well with application loads that genuinely require 4GB of active RAM. That will result in application 'lag' and thermal throttling. It isn't a bug, it isn't a deficit, it is simply inherent.

That is where the mobile connectivity of the Pyra kicks in. At some point it will be faster and more efficient to remotely take over a desktop or server or other computing environment and transport the commands and screen data.

It isn't that difficult to set up a home 'server' with a dyndns (or similar) and SSH -X into it and launch graphical applications. Alternatively these can be done through hosting providers (AWS, Rackspace, etc). Not cheap. Yes, you could count on some interface lag, but you would have processing capabilities far beyond what could be expected in a handheld device.

So, when dealing with big computing applications, there is a trade off between mobile interface lag (remote computing) with it's inherent setup costs and processing lag (overloading the local CPU).

I don't see the Pyra as a 'do everything device'. I see it as a 'do 90+% device'. Having more RAM than the SoC can process against efficiently does not make it a 'do 100% device'.

 

[Swordfish II]

The mobile variant has never hit the market, we're using an automotive grade OMAP5. Guess what, the automotive industry has totally different requirements.

Please provide a source to back that
[doublepost=1504719379,1504718804][/doublepost]

SizeWeightVolume Vs MassiveComputingPower

The Pyra is a mobile computing device. What it will be able to do by itself is pretty amazing. However, there will always be applications that are better performed using the capabilities of desktops, servers and supercomputers.

So many of the, 'but I need huge RAM to do X application' calls tend to be for applications that aren't overly suited to stand alone mobile. Many of them could/should/will work - sort of, but the OMAP 5 CPU simply isn't going to keep up very well with application loads that genuinely require 4GB of active RAM. That will result in application 'lag' and thermal throttling. It isn't a bug, it isn't a deficit, it is simply inherent.

That is where the mobile connectivity of the Pyra kicks in. At some point it will be faster and more efficient to remotely take over a desktop or server or other computing environment and transport the commands and screen data.

It isn't that difficult to set up a home 'server' with a dyndns (or similar) and SSH -X into it and launch graphical applications. Alternatively these can be done through hosting providers (AWS, Rackspace, etc). Not cheap. Yes, you could count on some interface lag, but you would have processing capabilities far beyond what could be expected in a handheld device.

So, when dealing with big computing applications, there is a trade off between mobile interface lag (remote computing) with it's inherent setup costs and processing lag (overloading the local CPU).

I don't see the Pyra as a 'do everything device'. I see it as a 'do 90+% device'. Having more RAM than the SoC can process against efficiently does not make it a 'do 100% device'.

It seems like at least part of this is directed at me.

Let me be very clear, even though I thought I was multiple times. The Pandora is fantastic, the Pyra will be even better. The Pyra WILL meet all my use cases. Seriously, I use a Pi 3, with a custom raspbian as my normal desktop (only video editing or AutoCAD fire up the real one).

I don't care about specs. I care that it can actually do the things I want/need, and the Pandora is already 98% there.

I dislike predatory advertising. Many people these days are spec junkies. Just look at the hype surrounding the newest phones. Small companies like ED's develop innovative products and are hamstrung by chip developers. Either they follow the company line or they sink.

How do you think the market would respond, as a whole, if a dual core 1.5ghz chip can't stay at 1.5ghz for more than seconds?

They would feel cheated, especially since they buy a dual core 1.5ghz chip for their desktop that can achieve max speed for hours.

This is not on ED. This is on TI, Qualcomm, Samsung, etc.

Phones are regularly marketed and sold, even from Samsung itself with capabilities they cannot hold for more than seconds.

If their marketing is true, and OK, than maybe I should start calling myself an Olympic Gold medalist. I can hit the speed required, for 3 steps, I just can't maintain it. But according to some here, that is all that matters "with proper things you could"