InvalidError :
somebodyspecial :
They have more wins than 805 and 805 loses to K1 a LOT, so just stating the facts on this gen of socs.
I would wait to see real-world power consumption numbers before declaring that much of a win for Nvidia... people who have seen Nvidia's demos are quite worried about that - the demo prototypes had some pretty large heatsinks for "mobile" devices.
How many actual design wins did the K1 actually score so far? The Shield Tab is practically the only K1-based product with a fair amount of buzz about it.
On the 805 side, I can bet my shorts that just about every manufacturer with 800/801-based designs will refresh them with the 805 without much fanfare.
As far as 20nm goes, I would not count on that being some sort of magical equalizer since everyone except Intel and Samsung relies on GF/UMC/TSMC/etc. for chip fabbing and it looks like they are all going to get there at about the same time. Everyone who has been waiting for 20nm to become available will be doing their best to have their 20nm designs ready for when that happens. Whoever gets their first batch of 20nm wafers baked is not going to have that advantage for long.
http://www.anandtech.com/show/8329/revisiting-shield-tablet-gaming-ux-and-battery-life
Waited. Battery life worries over
Today's games won't push the K1 to max which caused the shorter life in recent tests as anandtech shows here. If you lock it at 30fps which as they note still beats everyone else, it's pretty darn good battery life. Temps ok too, throttling averted. GFXbench battery life lost to shield handheld a little, but remember it has a 45% battery capacity advantage over the tablet (28.8whr for shield handheld vs. 19.75whr for tablet), AND most importantly even winning with that battery by ~18%, K1 beat it in fps by OVER DOUBLE (same for lg g pro 2 with S800). So massive battery advantage to T4, but K1 smoked it with over 2x perf. If it wasn't impressive before the anandtech article over worries you mention, it's truly awesome now as proved. Doubters just need to give this one up. It's over. Don't forget the tablet is pushing a massive screen compared to the huawei phone or the shield screen (both far smaller). So again, even more slanted to K1 when judging what they are pushing in screens here. Note the huawei perf was quadrupled and then some. Considering huawei perf, it's battery life is easily discounted. If they even tried to cut the NV perf in half (still being double perf) that battery wouldn't be winning for it right? Never mind the fact that they'd need to quadruple perf to catch NV here.
Uh, NV will have an advantage for quite a while. This race is over at least for tablets with K1 vs. 805. Next race will be over also, as Denver is IN HOUSE vs. off the shelf IP from ARM for Qcom, a true reversal of roles and won't be fixed for 6-9 months AFTER S810 hits and loses. Off the shelf will mean lesser perf and worse battery vs their old Krait in house tech. So it's NV's game for at least a year or more in at least tablets which should mean it wins the next two nexus tablets. They already got Nov's HTC nexus in the bag from google and they'll do the same next year as both hit 20nm for the next rev. Qcom will be too late to the IN HOUSE 20nm fix to follow S810 so NV will win that Next year Nov war also. Who knows what samsung etc will do, but if nexus starts stealing huge numbers of sales due to gaming samsung will be forced to use K1. Since NV will license the gpu IP (and the whole point of this was to win apple and hurt imagination), apple may see themselves going with at least NV's gpu also. You can't let others steal your gaming sales especially with that being the focus now for everyone.
http://www.develop-online.net/news/report-majority-of-mobile-app-downloads-are-games/0191695
90% of sales at googleplay are games. 76% at apple, 61% at amazon.
"All of which adds fuel to the claim that games continue to be the strongest moneymakers on mobile."
This is why apple just got a patent for gamepad controls on the back of a device. Even they get it now. It's about gaming now, not so much the modem with 2GB caps (less than a single 720p movie) in most parts and only 2 places with 10GB caps (HK/Singapore). Faster than 150mbps is pointless with caps like this. My cable modem is only hitting 50mbps now, so I'm sure modems won't rule again unless the caps are removed and there is ZERO chance of that for a long time as the phone co's just want to milk mobile users to death just like cable caps.
One more thing, NV has popped out silicon at Samsung ages ago in a test run, and as apple kicks up some stuff at TSMC, what do you think will be filling samsung fabs?
http://www.brightsideofnews.com/2012/03/14/exclusive-nvidia-tapes-out-tegra-at-samsung-fab2c-ibm2c-globalfoundries-next/
March 2012. Tapeout. Nothing stopping Nvidia from pumping out some socs over there. I'm guessing the whole point of this collaboration was in preparation for apple taking some stuff to TSMC or even to their own fab they now own or the one they're building. If apple doesn't go TSMC, they will at least use their own fabs at some point. That is after all the point of buying or building them right?
Samsung knows the apple socs are on a limited time-frame here and it's running out fast. The only thing that would cover the amount of silicon apple is using would be NV GPU's (5x+ the size of socs) and NV socs themselves. I'd guess SOCS first to test the waters then the huge GPUS next. Also important is the fact that Samsung and GF can produce the same chips easily at each others fabs after taking IBM's tech and running with it. So sourcing from samsung means you can easily ramp at GF if needed in a pinch with pretty much zero design time. They are interchangeable now.
http://electronics360.globalspec.com/article/4171/samsung-globalfoundries-form-14nm-alliance-to-fight-tsmc
""All the fabs are 100 percent synched. Customers can take a design to one foundry or another or both," SW Jeong, executive vice president for foundry at Samsung, told Electronics 360.
The advantage for a customer is that a common manufacturing process makes it easier to transfer a design to a second foundry if a first foundry should have supply constraints or if a natural disaster should strike a fab."
http://blogs.barrons.com/techtraderdaily/2014/08/04/taiwan-semi-to-lose-apple-qualcomm-biz-to-samsung-global-say-bernstein-bluefin/
TSMC losing favor anyway it seems in 2015. At the very least there shouldn't be as many shortage problems running around next year for gpus or socs I'd guess.
Jetson TK1 development board, Lenovo ThinkVision 28, XiaoMi MiPad, HTC Nexus 9, Nvidia Shield Tablet, project Tango already selling from Google also (likely come with Denver later this year as an upgrade to 64bit also). Denver has Acer Chromebook CB5 on tap supposedly this year too. Since it's been benchmarked already, I'm sure it will be on time. My guess is a shield handheld rev2 is coming either shortly as they ramp up K1 or in Nov/Dec for xmas refresh at worst with Denver inside. If Google can get a Nov nexus out the door, shield R2 will be there too. It takes far less design to plop a K1 into your already done Shield platform than to make a new tablet switching socs for google. Clearly power is ok so shouldn't be much work changing socs for NV in shield handheld. The 1440x810 specs and 4GB of ram rumor shows it's not the tablet we see today and must be a slight upgrade to the specs of the shield handheld which makes sense. I'm sure the K1 version would far outsell the T4 version right? It's the top soc for quite a while now. Priced at $250 vita/3ds sales would stop in their tracks and even $300 is doable based on the power vs these two. You won't be running unreal 4 engine games on a vita/3ds at any point in time right?
Should I vote you down like you did my post for being an NV pessimist repeatedly on toms?...LOL. More NV fud from you
As usual. It would be different if my enthusiasm was unfounded, or their quarterly reports sucked like Amd etc, but that isn't the case. Did you catch their quarterly report? Very nice, tegra sales up 200% no doubt due to K1 designs coming and it hasn't even ramped, also took over google I/O (auto, tv, tablet, everything they had was K1 there). Margins 56%, profits fine etc. Guiding higher, thus stock price higher too now (up 9% from the quarter is pretty great correct?). Google showing they want AEP and AndroidL to be gaming centric and to close the gap on PC's shows they'll be leaning on NV as their IO show demonstrated. You'll go with the graphic leader from here on out for gaming, not Qcom who has ZERO game experience with devs or drivers (you can google the driver hate for them). NV/AMD have been working with devs for 20yrs and optimizing their drivers for games for the same length of time. Good luck to qcom now.
Facebook
Twitter
Instagram