Early Nvidia earnings show growth, beat forecasts

[Marine-RX179]

Nvidia making a greater profit? Good for them...but I don't see how it is relevant to the graphic forum if they are just gonna pour most of that money back into the development for mobile. This is graphic card forum not general hardware or general discussion or Speaker's Corner forum is it not? I think this topic is in the wrong part of the forums

 

[Stanners]

Nvidia making a greater profit? Good for them...but I don't see how it is relevant to the graphic forum if they are just gonna pour most of that money back into the development for mobile. This is graphic card forum not general hardware or general discussion or Speaker's Corner forum is it not? I think this topic is in the wrong part of the forums

And I think you are wrong, Of course it effects the graphics card division of the company and it is good to know they are making profits to keep developments up. Same goes for AMD.

 

[andybird123]

And I think you are wrong, Of course it effects the graphics card division of the company and it is good to know they are making profits to keep developments up. Same goes for AMD.

exactly, AMD making losses as they were is no good for anyone, competition breeds advancement, nice to know both sides are still in the fight

 

[layte]

I guess this is as good as anywhere to post this information seeing as Tegra is taking a kicking in this thread. The first K1 based tablet has been announced, the 192 CUDA core variant seems to have double the GPU performance of Apple's A7 with comparable battery life.

http://www.engadget.com/gallery/xiaomi-tablet-launch-event/2618907/#!slide=2618921

http://www.engadget.com/gallery/xiaomi-tablet-launch-event/2618907/#!slide=2618928

battery life information of MiPad (from Xiaomi.com)
6700mAh battery,
11 hours online-video,
17 hours reading ebooks,
86 hours music,
1300 hours idle.

Looks pretty good to me.

Xiaomi is a good catch for NV. They are huge in China.

 

[Orangey]

Saw it on Charlie's site, they actually sounded optimistic. Charlie wasn't the author obviously.

Also this is unrelated but Xiaomi just announced a $650 4K panel as well.

 

[drunkenmaster]

Will be interesting to see K1 performance vs others over time. We've already seen Nvidia try to link 5W with 326gflops by... putting them close together in sentences but never actually linking them. We've seen since Nvidia slides with 290gflops sustained at 10-11W.

11W on a tablet would eventually get too hot to actually hold. Apple/sensible companies just clock these chips appropriately, others can bump clocks up and run a benchmark which has nothing to do with sustained usage.

maybe those GFX3.0 benchies are at already reduced and sustainable clocks, but I suspect not comparing specs/power usage and what we know about the chip.

http://www.anandtech.com/show/7519/apple-ipad-mini-with-retina-display-reviewed/2

The ipad mini retina drops what 30-40% performance within a few tests due to the cooling limits of a smaller tablet. Noticeably more performance drop off than a bigger tablet with basically the same chip. These are effectively 5W chips while max performance for K1 is around 11W. Performance in the first minute isn't nearly as important as performance after 5-10 mins, particularly for a chip targeted at improving gaming on tablets.

The real issue is that rather like the Nexus 7(i think the 7) with Tegra 3 it could easily be obsolete within 3 months, there will be a new ipad with the A8 in it before the end of the year on 20nm. Rumours it could appear as early as August really don't bode great for it in general(as a chip). A $240 decent tablet however is good, in that sense everything generally beats Apple and their prices though. Within 6-12 months there will likely be a dozen + 20nm chips with better cores(a57/A8) better performance and battery life(20nm) and reduced cost for performance due again to process.

I still can't believe after seeing Tegra 3 get one significant win then being obsolete then they are going with a literally last second 28nm new chip so so close to 20nm, causing them to be very late to 20nm.

 

[jh30uk]

@ layte , Good news but have you made a thread for it so peeps will see topic and enter to read it?

 

[_Alatar_]

Will be interesting to see K1 performance vs others over time. We've already seen Nvidia try to link 5W with 326gflops by... putting them close together in sentences but never actually linking them. We've seen since Nvidia slides with 290gflops sustained at 10-11W.

11W on a tablet would eventually get too hot to actually hold. Apple/sensible companies just clock these chips appropriately, others can bump clocks up and run a benchmark which has nothing to do with sustained usage.

Those 11W numbers were for the entire jetson tk1 board, sometimes with PSU inefficiencies in there as well.

SOC power (what other companies are using and what's used for TDP numbers) is actually much lower.

See this whitepaper for example:

http://developer.download.nvidia.com/embedded/jetson/TK1/docs/Jetson_platform_brief_May2014.pdf

Some quotes from there:

6
Apple A7 consumed 2560 mW to deliver 13 fps on iPhone 5S. An equivalent Tegra K1 mobile platform consumes
2605 mW to deliver 1.45x higher performance.

Since Jetson uses DDR3L and OS is not tuned for mobile, it
consumes 3658 mW to deliver an equivalent 1.45x performance advantage over Apple A7.

 

[drunkenmaster]

Odd choice for the cut off in that quote.

Apple A7 consumed 2560 mW to deliver 13 fps on iPhone 5S. An equivalent Tegra K1 mobile platform consumes
2605 mW to deliver 1.45x higher perform
ance. Since Jetson uses DDR3L and OS is not tuned for mobile, it
consumes 3658 mW to deliver an equivalent 1.45x performance advantage over Apple A7

So the jetson platform uses a full.... 1W extra, which is not massive at all and still puts it as a 9.8W platform vs other guys running much lower power.

It also shows in the paper for instance that it's delivering 2.5x the fps performance, but it's only 1.45x better in performance/watt. Assuming 13fps at(according to Nvidia 5.08fps per watt) then the A7 is using 2.55W, while the Nvidia chip is using 4.45W.

Either way this is Nvidia and they are rarely particularly honest about these things, losing most of the OEM industry who dumped design wins because Nvidia couldn't produce what they claimed. We'll see(hopefully) from some good reviews showing performance after a warm up period(Anandtech weren't particularly clear if they benchmark only after a warm up, they had a performance during warm up test, but I don't know if they then did each test at peak or throttled down performance, then the question is who will get a chinese tablet and benchmark it). Anandtech refused to do a K1 review when Nvidia refused to tell them the power usage, this is after Anandtech compared T3 or T4(forget which) to Apple's chip at the time and posted bogus power numbers Nvidia gave them. When they found out the numbers later on, their preview would have read completely differently.

I wouldn't actually be surprised if K1 was pretty efficient, they are using pretty much the latest process(further tweaked) and the latest and most improved version of A15. It's effectively using 18 months of improvements TSMC has made and that Arm has made on the process and core. Most companies could produce an improved 28nm chip, but everyone with half an ounce of sense is working to make much better, faster, cheaper A57 based 20nm chips.

 

[_Alatar_]

You have the exact same quote I have... It says that since the jetson TK1 isn't a mobile platform and uses different type of memory it consumes more than a mobile K1 platform would.

Why you would want to compare the jetson numbers to the ipad instead of the mobile numbers I have no idea. Jetson is a board for developers, not a mobile device.

The mobile K1 platform is giving 1.45x the performance of the A7 for 55mW less power.

The mipad that just announced by Xiaomi also has similar battery life as the ipad mini as well as performance that's up to 2x as good as the ipad in 3d applications.