NVIDIA Tegra Architecture

[A1xLLcqAgt0qc2RyMz0y]

Because attaining max performance would require unacceptable compromises on other very important things like power consumption, temperature and battery life. And even then you still may not get to max given inherent limitations of the form factor.

You can argue for giving the user software knobs to crank performance to the limits of the cooling system but that's a very bad idea for a mass market device like a tablet.

Ailuros wanted to scale back the performance of both the MiPad and Shield and the Shield can actually handle that performance by design so that throws out your argument of unacceptable compromises.

Originally Posted by Ailuros
IMHO both the Shield tablet and the MiPad are a bit on the too aggressive power settings side

 

[A1xLLcqAgt0qc2RyMz0y]

Since you obviously aren't a spokesperson neither for NVIDIA nor Google I'd suggest you wait for the relevant Google device to ship and then we'll see again. As for my own preferences they aren't negotiable. I know what I want and for which reason thank you.

You seem to be advocating that Nvidia neuter the performance on the Shield Tablet to less than it can handle to please your belief system.

Originally Posted by Ailuros
IMHO both the Shield tablet and the MiPad are a bit on the too aggressive power settings side;

Actual Shield Tablet Review Data:

When taken at face value, NVIDIA seems to have created a perfect storm for themselves: the Tegra K1 is one of today’s fastest mobile processors and it will spend the vast majority of its life running demanding applications. Meanwhile, the SHIELD Tablet’s external skin is coated in a soft-touch finish rather than aluminum which can act as a quasi external heatsink. By all sane thinking, this SHIELD should run extremely hot….but it doesn’t. Nor does the K1’s performance throttle, even when it’s being used in Max Performance mode.

After two hours of continual gaming in Trine 2 (which is arguably one of the most demanding games available right now) the SHIELD Tablet remains relatively cool to the touch. Granted, there is a hot spot over the K1 itself but we can see that NVIDIA’s internal heatsink design is able to disperse the heat across several square inches. The hottest section never went above 56°C while the area directly behind the battery was only slightly warm.

http://www.hardwarecanucks.com/foru...iews/67076-nvidia-shield-tablet-review-7.html

Battery life was very good with four straight hours of gaming in Trine 2 or nearly a day of mixed browsing, gaming and basic app use. Since all but a small token amount of processing is done on the host PC during GameStream, we were easily able to achieve 5 hours playing PC games. In addition, there wasn’t any noticeable battery consumption increase when using the SHIELD Controller. Anyone who thinks a purpose-built tablet like this one will sacrifice battery life will have their beliefs changed dramatically after a few hours with the SHIELD.

http://www.hardwarecanucks.com/foru...iews/67076-nvidia-shield-tablet-review-8.html

 

[Silent_Buddha]

Why would anyone want a device neutered where the max performance of the SOC could never be obtained?

That capability may be needed on a future game and leaving that performance less that it could be is downright dumb on a Gaming Tablet.

If you want less aggressive power then enable the Nvidia software that caps performance.

So what you are saying is that the Shield device is unacceptable because it is neutered. You do realize that the Shield device can never attain maximum performance because the software stack and drivers limit how much power it can consume?

So basically your own words are saying that noone should want to buy the Shield tablet because it has been neutered (cannot attain max performance of the SOC). As well implying that Nvidia are "dumb" for not allowing the SOC to run at max performance.

Regards,
SB

 

[Nebuchadnezzar]

Nebu, those CPU voltages don't use the DFLL voltage regulator. The real values on a production system will be significantly lower (e.g. the table I'm looking at on the Jetson K1 is 1190mV with DFLL instead of 1360mV without DFLL for 2320MHz).

There's not a lot of specifics on DFLL but e.g.: http://comments.gmane.org/gmane.linux.ports.tegra/15273

DFLL is a frequency generator, not a voltage regulator... I said that those were the nominal voltages without binning and the real ones are lower. Not sure where you get those 1190mV from because the runtime voltages depends on actual parameters read out from the SoC itself; as they're calculated with the coefficients in the first triplet of columns:

/* cvb_mv = ((c2 * speedo / s_scale + c1) * speedo / s_scale + c0) / v_scale */
...
	mv = DIV_ROUND_CLOSEST(cvb->c2 * speedo, s_scale);
	mv = DIV_ROUND_CLOSEST((mv + cvb->c1) * speedo, s_scale) + cvb->c0;

And they also use temperature voltage scaling as it seems;

/* cvb_t_mv =
   ((c3 * speedo / s_scale + c4 + c5 * T / t_scale) * T / t_scale) / v_scale */

...	mv = DIV_ROUND_CLOSEST(cvb->c3 * speedo, s_scale) + cvb->c4 +
		DIV_ROUND_CLOSEST(cvb->c5 * t, t_scale);
	mv = DIV_ROUND_CLOSEST(mv * t, t_scale);

So as long as somebody doesn't go in to create some kind of interface to read out the generated voltage tables from the device itself, you can't really retrieve the values for discussion. The DFLL nominal voltages are the next best thing.

 

[Ailuros]

Ailuros wanted to scale back the performance of both the MiPad and Shield and the Shield can actually handle that performance by design so that throws out your argument of unacceptable compromises.

I suggested its power settings are too aggressive for my taste and I am entitled to my opinion. Stop putting words in my mouth I never meant or implied or else read the sentence you just quoted.

You seem to be advocating that Nvidia neuter the performance on the Shield Tablet to less than it can handle to please your belief system.

I made a suggestion and that's all; other than that I am entitled to my opinion which you don't have to like but still will have to tolerate.

..........

I don't care what each pissat marketing bullshit has to say wherever it comes from. If you by yourself have something substantial to add to any current conversation (which I haven't seen yet outside of a ton of copied/pasted marketing material) then so be it. Not everyone has the same preferences and you will have to live with that.

As I said even with more conservative power settings the K1 is still screamingly fast and way out of reach of any Android competition this year and in that regard a combination of performance per Watt in combination with as long as possible battery life is obviously what I'm looking for. Too bad you can't force my money out of my pocket and stuff a MiPad or Shield tablet down my throat.

 

[Plano]

A15 variant of K1 seems to be widely tested. Talking about Volantis and Denver, is there any new info around the latter?

 

[Ailuros]

A15 variant of K1 seems to be widely tested. Talking about Volantis and Denver, is there any new info around the latter?

Not yet, but I am very much looking forward for it. I expect to hear more about it closer to this year's fall?

 

[ams]

A 64-bit Denver TK1 variant should have pretty impressive performance in the rumored 8.9" aluminum clad chassis. At the very least, the CPU perf. per watt should be even better than the 32-bit R3 A15 TK1 variant. Apparently NVIDIA will be speaking more about Denver at Hot Chips later this month.

 

[Ailuros]

I wonder if they just switched CPU Blocks or if they had time for a few further refinements.

 

[ams]

IMHO both the Shield tablet and the MiPad are a bit on the too aggressive power settings side; a wee bit less wouldn't had hurt and the device would still had been screamingly fast and damn hard for any Android competition to catch up.

Shield tablet has a power control center where one can adjust frame rate limit, max CPU frequency, and max CPU cores. It also has a feature where one can choose to optimize on an app by app basis. So that is quite a lot of flexibility.

A game like Trine 2 needs every bit of Shield tablet's sustained GPU performance to run smoothly @ 720p30. The render resolution for this game on the MiPad is actually a bit lower at 1024x768 (probably motivated in part by the fact that this tablet has no "console" mode and has a native 2048x1536 resolution).

Shield tablet should have ~ 10 hours battery life (ie. all day) with web browsing and video playback, ~ 5 hours battery life with casual Android gaming and game streaming, and ~ 3 hours battery life with the most graphically intensive games such as Trine 2 that are not even playable on any other thin fanless tablet. So I am pretty happy with that. That said, it would be nice to see more battery life testing done in the power saver modes.

 

[Ailuros]

If it lasts somewhere over 2 hours under stressful gaming the road for less than an hour to reach 3 hours battery life shouldn't be that long. I' d truly wonder how much less performance it would reach under such a power scenario.

 

[ams]

It appears that most people are getting close to ~ 3 hours battery life with Trine 2 running in NVIDIA optimized mode (and it would last even longer in "console" mode with the tablet screen powered down). I see what you are saying, but if the performance was lower by even 25%, then they may not be able to achieve a smooth 720p30 in a game like Trine 2 (this particular game uses real-time physics and advanced lighting and particle effects). Note that the battery size capacity is not particularly large on Shield tablet (the MiPad has close to 30% more battery capacity in comparison), so that may be one area that needs some improvement next time around.

 

[A1xLLcqAgt0qc2RyMz0y]

So what you are saying is that the Shield device is unacceptable because it is neutered.

No I am not saying that at all.

 

[trinibwoy]

Ailuros wanted to scale back the performance of both the MiPad and Shield and the Shield can actually handle that performance by design so that throws out your argument of unacceptable compromises.

Then I must have misunderstood your point about max performance cause neither of those products allow the SoC to run full tilt.

 

[Ailuros]

It appears that most people are getting close to ~ 3 hours battery life with Trine 2 running in NVIDIA optimized mode (and it would last even longer in "console" mode with the tablet screen powered down). I see what you are saying, but if the performance was lower by even 25%, then they may not be able to achieve a smooth 720p30 in a game like Trine 2 (this particular game uses real-time physics and advanced lighting and particle effects). Note that the battery size capacity is not particularly large on Shield tablet (the MiPad has close to 30% more battery capacity in comparison), so that may be one area that needs some improvement next time around.

Wouldn't that depend on a number of factors? Is performance truly lower by as much as 25% for less than an hour of battery life? Does N demanding game feel smooth only at 30fps average or is an average of say 22-23 fps also tolerable (highly subjective of course)?

On a sidenote I'd be personally also interested how K1 performs in Real Racing 3 with extreme settings; it's nice that different websites investigating ULP SoC performance give us quick and lean synthetic benchmark results but there must be a way to include a couple of demanding games into the mix, irrelevant if those would be inevitably vsynced. In such a case what I personally care most about is if the game experience is smooth enough and less about the last possible inch of performance.

OT but I was close to ordering a MiPad from an online retailer since at 220+ Euros it's a complete steal; I was turned off because I thought it was some sort of standard Android 4.4 with a different GUI (MIUI); from what I've been told it's a custom Android version where you'd have to install things like the Playstore manually amongst others.

 

[ams]

Anandtech has done a followup on GFXBench T-Rex Onscreen battery life and temps using a 30fps framerate cap: http://www.anandtech.com/show/8329/revisiting-shield-tablet-gaming-ux-and-battery-life

 

[silent_guy]

Nebu,

In your interesting update at Anandtech (http://www.anandtech.com/show/8329/revisiting-shield-tablet-gaming-ux-and-battery-life), you write the following:

We see max temperatures of around 85C, which is edging quite close to the maximum safe temperature for most CMOS logic.

85C may be pushing it for system issues (user experience, certain limits imposed by product manufacturers, etc.) but it's not in anyway considered close to the maximum safe temperature of most CMOS logic: that temperature is universally considered to be 125C.

 

[Alexko]

As technically interesting as chip temperatures are, they're not terribly relevant to end-users. What about the temperature of the outer skin of the tablet itself?

 

[silent_guy]

As technically interesting as chip temperatures are, they're not terribly relevant to end-users. What about the temperature of the outer skin of the tablet itself?

I don't disagree. But it's incorrect to point out that it's uncomfortably close to CMOS limits when it's 40C away from that limit.

 

[A1xLLcqAgt0qc2RyMz0y]

As technically interesting as chip temperatures are, they're not terribly relevant to end-users. What about the temperature of the outer skin of the tablet itself?

Review with Thermal Imaging:

http://www.hardwarecanucks.com/foru...iews/67076-nvidia-shield-tablet-review-7.html

After two hours of continual gaming in Trine 2 (which is arguably one of the most demanding games available right now) the SHIELD Tablet remains relatively cool to the touch. Granted, there is a hot spot over the K1 itself but we can see that NVIDIA’s internal heatsink design is able to disperse the heat across several square inches. The hottest section never went above 56°C while the area directly behind the battery was only slightly warm.