NVIDIA Tegra Architecture

Again, you have not yet provided any evidence that this is the case. Furmark power consumption is not evidence. If you don't understand why it is not evidence, I am not sure myself or anyone else here can help you. Indeed, the only actual compute benchmarks in this thread, that I am aware of, show Maxwell to be considerably more efficient than Kepler in compute workloads, sometimes over 100%+.

The only person here I see ignoring evidence is you.

The Tegra K1 in the Nexus 9 is based on Kepler, not Maxwell.
Here is the article I was talking about, where they talk about how maxwell's power consumption is due to extremely good voltage control and power gating.



http://www.tomshardware.com/reviews/nvidia-geforce-gtx-980-970-maxwell,3941-12.html


I am not debating that it's efficient IN PRACTICE, but there is nothing that is fundamentally more efficient than kepler in the architecture. It doesn't actually compute using less energy.


I am disappointed but not surprised that Tegra is all over the place performance wise. As I stated in an earlier post, if you can depend on one thing it's nvidia to screw up their mobile offerings. Their desktop GPUs are top notch and usually higher performance (in single gpu) than anybody else, but I think nvidia is finally realizing that qualcomm and samsung and Apple are not just stooges riding on the backs of ARM. It takes a lot more than a good GPU architecture to make a good mobile chip.
 
Jesus...





This is actually not really nvidia's fault. Android does this with all devices, and the faster and more power hungry the processor the worse it gets. Its a problem with the sleep states of programs, and android constantly turning wifi on and off while on standby.



I would have no problem criticizing nvidia if standby time was bad because of their processor, but it really has nothing to do with them.
 
I am not debating that it's efficient IN PRACTICE, but there is nothing that is fundamentally more efficient than kepler in the architecture. It doesn't actually compute using less energy..
Not only are you not debating, you are completely ignoring. Do us a favor, don't come back with your theories until you go through the effort of trying to understand what people have been telling you.
 
This is what Ars Technica said about Nexus 9 web browser performance:

"The Nexus 9 performed well when focusing on a single task, not a surprise given its excellent single-core performance (see our browser scores for evidence). Our iPad Air 2 review included an informal test of Web browser performance by loading and perusing an endless Facebook page, scrolling for days' worth of feed content, and we did the same thing again with HTC's tablet. The Nexus 9 handily beat the iPad Air 2 in that regard; it simply never crashed, nor did it lose track of any content it loaded during scrolling. (This stands in stark contrast to its surprisingly lousy Sunspider test score, by the way.) "
 
This is actually not really nvidia's fault. Android does this with all devices, and the faster and more power hungry the processor the worse it gets. Its a problem with the sleep states of programs, and android constantly turning wifi on and off while on standby.

I would have no problem criticizing nvidia if standby time was bad because of their processor, but it really has nothing to do with them.

You would be right if other devices behaved the same way. My Nexus 5 uses about 7% of it's battery capacity during 10 hours of standby.

I'm a fairly light user of my phone though, I only use my Nexus 5 for checking email, WhatsApp, Calendar, some Google Now and a some light browsing. By the end of the day yesterday, my battery was at 91% (I had to turn my phone off in the morning for an exam) and it was at 84% capacity when I checked this morning at 10 AM. That includes getting updates from the Play Store. Also, since yesterday I've been constantly checking for system updates. Fingers crossed for a quick update to Lollipop here.

From my usage pattern, getting 2 days of use, even 3, isn't that unusual. Yes, this is only anecdotal, but my own experience already proves you wrong. The question is, will proving you wrong actually change your opinion?
 
All of my android devices, provided that they only have the default apps or no misbehaved apps, running only on WiFi (no cell connection), it could last a long time on standby. Especially true for tablets because they have bigger battery. I haven't yet experienced any of my devices drain 30% of its power on standby with WiFi on. Heck, my phone can do better on standby using cell connection (3G/HSPA) with poor signal! And I have various apps that needs Internet connection in the background (multiple mail apps, multiple messenger apps, steam, etc). So telling me that leaving a device overnight will drain 30 percent of its battery as somewhat normal is misleading.
It is something that they have to fix because it's not normal. I don't think it's a hardware problem. Probably they forgot to tone down WiFi power on standby or maybe something wrong with throttling.
I'm aware that there use to be a bug where an android device can't enter deep sleep. If it's a bug on the base android package, then this behavior (using relative lots of power on standby) should also happen on other devices using the same android version (should be worse on N6 or the older Nexus because of their smaller battery), but haven't heard about this behavior on those other devices.
 
All of my android devices, provided that they only have the default apps or no misbehaved apps,
I think its most likely a misbehaving app, Ive seen sone wake up 24hrs a day check the wifi for some info, then go back to sleep. So I'ld check if theres not a rogue app first
 
Overall that Nexus 9 was quite a step forward for all the actors involved:
HTC has not built a tablet in a while
Google is pushing and new Android version as well as a new runtime
Nvidia is pushing a new CPU architecture.

It will take some time for things to get ironed out and we can have a more proper look at the product as a whole, being a NExus means long term support.

For Denver and Nvidia, it is going to take time too, now it is a bit all over the place but I would agree with Ams, I think perceptions are a bit too negative, Honestly I don't think the Nexus 9 is the most attractive nexus product ever, or most attractive Android tablet at the time, I see many reasons and quite a lot have nothing to do with Tegra K1 prowess (or lack of it). I would say the same for the Nexus 6. Now I think that when it is all said and done it is pretty strong showing for Nvidia, if in depth reviews and a couple months of maturing confirm the promise of their CPU.
Denver is their first try at a CPU and it seems it is far from a bad one.

I hope Nvidia continue to produce its own line of devices and that those tegra k1-64 are going to make it to some of their in-house product, Nvidia tv would be nice for example.
 
If perceptions on Denver are a bit too negative, it's mostly because nVidia set expectations high. If I remember correctly nVidia promised that the K1 Denver would completely outpace other ARM mobile processors or something to that effect. And now by the time it's available, the iPad Air 2 has been shipping in volume for a few weeks already and the A8X looks like the overall (slightly) faster chip. Yes the A8X uses a more advanced 20 nm market, but sticking with a mature 28 nm process could have allowed nVidia with a time-to-market advantage, instead Denver K1 availability ended up being after A8X availability. Denver K1 is an impressive chip, but nVidia set the perception that it'd be a home run, which it doesn't seem to achieve.
 
Out of curiosity, anyone have any ballpark figure on the price of a Denver K1, in says 100K quantity ? are we talking $10 or $30 or what ?
 
If perceptions on Denver are a bit too negative, it's mostly because nVidia set expectations high. If I remember correctly nVidia promised that the K1 Denver would completely outpace other ARM mobile processors or something to that effect. And now by the time it's available, the iPad Air 2 has been shipping in volume for a few weeks already and the A8X looks like the overall (slightly) faster chip. Yes the A8X uses a more advanced 20 nm market, but sticking with a mature 28 nm process could have allowed nVidia with a time-to-market advantage, instead Denver K1 availability ended up being after A8X availability. Denver K1 is an impressive chip, but nVidia set the perception that it'd be a home run, which it doesn't seem to achieve.

True..Denver really should have come out sooner. Heck even Samsung beat them to it with their 20nm A57 based SoC.

Out of curiosity, anyone have any ballpark figure on the price of a Denver K1, in says 100K quantity ? are we talking $10 or $30 or what ?

I dont have concrete information but based on some reports I've seen I would say its in the $20-30 range.
 
Having 30% drain overnight or any unusually high drain during light use is clearly a software issue rather than a hardware issue, and may be due to an issue with the software drivers or with a misbehaving app.

Here is what Computer World said about Nexus 9 in their review:

"The [latest OTA] update went a long way in improving the Nexus 9's stamina: While I had poor results during my first days with the device, I'm now consistently able to get a solid eight to nine hours of active use per charge. That's a bit short of Google's 9.5-hour estimate but still quite respectable and within the realm of reason."

"The tablet's standby power usage, meanwhile, is top-notch: Leaving it on all night, with background data running and notifications trickling in, the device typically dropped a mere three percentage points over as much as 11 hours. Google says the tablet can last up to 30 days on standby, and with that kind of low idle power consumption, I wouldn't hesitate to believe it."
 
Back
Top