The Intel Atom Thread

[VengenceIsMine]

what I dont understand is how are baytrail tablets suddenly coming down to 349 and 299 price points when clovertrail ones was $450. I expect the chip cost to be similar. windows cud have cut the price down by say $30. and the 8 inch displays may cut the cost down by $10 or 20. not sure how they achieved such a low price

Intel has done a lot of work with the PCB's for a lot of baytrail machines, greater degree of integration on board and a little bit more is on the SOC itself is a big reason. You will see it once someone like iFixit tears down a BT tablet. 2nd, cost of storage & RAM has come down significantly, most BT tablets are still 2gb ram and 32/64 storage, same as covertrail but cost of those items is less. & yes, MS license fee is smaller but it was never that high on clovertrail tablets, the $90 Win 8 license for a Lenovo Thinkpad tablet 2 has always been a myth.

 

[Sweepr]

SanDisk Announces Optimized iNAND Extreme eMMC for Bay Trail

We've started looking more closely at the embedded storage used in smartphones and tablets, and have mostly come away disappointed. Thankfully there appears to be some progress being made in the space. I remember being relatively impressed with the behavior of the eMMC in Intel's Bay Trail FFRDs at their benchmarking event a week and a half ago. The IO performance wasn't perfect, but it was definitely much better than I had been expecting.

Today SanDisk announced that it will be bringing an optimized version of its iNAND Extreme solution to Bay Trail tablets. Architecturally iNAND Extreme is a combination of NAND and eMMC controller in a single package. The device supports eMMC 4.51 (HS200) and uses SanDisk's own 19nm MLC (2bpc) NAND. Capacities go all the way up to 128GB for a single device, which SanDisk arrives at by stacking 16 x 64Gbit 19nm MLC NAND die.

Sustained performance isn't too shabby. SanDisk promises sequential reads/writes of up to 150/45MBps and 4KB random read/write speeds of up to 4K/800 IOPS. I suspect these numbers are based on the largest configurations, but I would expect similarly good performance even for the 64GB and maybe even 32GB versions. The 4KB random write results are sustained, not peak, and are measured by looking at performance across a 1GB LBA range. In the case of the 128GB drive that test leaves a ton of spare area, which helps explain the relatively good performance compared to what we're used to in the market. I'd love to see true worst case scenario performance for SanDisk's iNAND. The promised sequential read performance is outstanding. We've seen mobile devices break the 100MB/s sequential read barrier, but none have hit 150MB/s yet. Sequential write performance is also pretty good.

The Bay Trail optimizations come via some custom tuning on the firmware. I suspect Intel has its own performance targets and behaviors it wants to encourage on Bay Trail tablets and SanDisk is likely just responding to those requests. You can also find iNAND Extreme on other, non-Intel platforms as well.

http://www.anandtech.com/show/7347/sandisk-announces-optimized-inand-extreme-emmc-for-bay-trail

 

[bullzz]

now that apple has started this craze for 64 bit in phones/tablets, it might help baytrail look good for android
http://fudzilla.com/home/item/32570-intel-is-ready-for-a-kit-kat

 

[Exophase]

Can't wait for Android NDK fat binaries that have ARMv6, ARMv7, ARMv8/AArch64, x86, x86-64, and MIPS all in one. Maybe they'll have to change the term to "morbidly obese binaries."

 

[BallaTheFeared]

It's sad to see that what was once enjoyed on desktop, has now been reduced to smart phones and tablets.

That said I'll still enjoy watching the fight.

 

[bullzz]

its not making me sad. its good this is happening. this shud really push intel/amd to increase performance in desktop to keep them relevant

 

[Sweepr]

Edit: According to Hans: Haswell core measures 14.5mm², 2x Apple Cyclone cores is ~18mm². I wonder what is the size of a Silvermont core, here is Avoton's die shot (8 silvermont cores):

 

[Nothingness]

According to Hans: Haswell core measures 14.5mm², Apple's Cyclone core is larger (than Haswell) ~18mm². I wonder what is the size of a Silvermont core, here is Avoton's die shot (8 silvermont cores):

Discussed in another thread...
http://forums.anandtech.com/showthread.php?t=2341762&page=7

BTW your figure for Cyclone is for 2 cores and 1MB L2.

 

[Sweepr]

Discussed in another thread...

BTW your figure for Cyclone is for 2 cores and 1MB L2.

Thanks for the correction. Pretty large core, both Silvermont and Jaguar seem much smaller.

 

[liahos1]

Thanks for the correction. Pretty large core, both Silvermont and Jaguar seem much smaller.

this is very interesting indeed. the a7 cores are pretty big relative to silvermont. given merrifield is going to be a 2core + Rogue solution. and given single threaded performance seems to be slightly higher with silvermont i wonder what the performance is going to be like on smartphones. anyone want to hazard an educated best guess?

 

[Nothingness]

Hard to say how good Merrifield will be: will it have 64-bit? What frequency?

 

[Arachnotronic]

Hard to say how good Merrifield will be: will it have 64-bit? What frequency?

Yes, 2.4GHz. Whether the underlying OS will support it or not is an ENTIRELY different question 😉

 

[Nothingness]

Thanks for the info 🙂

I really hope Android will soon support 64-bit. For the kernel, it should be no issue, but Dalvik is a different story...

 

[Imouto]

Believe it or not but this is the future, cheap as it can get hardware right in the front of the eyes of the average customer.

Someone was laughing at Rockchip just a few pages ago.

 

[seitur]

There is no single future. During last 20 years I've seen "this is the future" too many times. This is simply product for certain sizeable market segment that is and will be important market segment.

Not sure why anyone would laught at Rockchip. They always targetted lower market segments and they do specialize in them.
Just a company making good business. Certainly not first and not last to make money out of budget products.

 

[Sweepr]

Dell Venue 11 Pro 5130 (based on Bay Trail-T - Atom Z3770) @ Geekbench: http://browser.primatelabs.com/geekbench3/53855

According to rumours this is a 10.8'' Full HD tablet (with keyboard dock) running Windows 8.1 and priced $399. Rumoured 9 hours battery life without dock and 19 hours with the keyboard dock (integrated battery).

I'd rather buy this Dell than a $449 Windows 8.1 RT Surface 2 (up to 10 hour battery life). 🙂

 

[sequoia464]

Dell Venue 11 Pro 5130 (based on Bay Trail-T - Atom Z3770) @ Geekbench: http://browser.primatelabs.com/geekbench3/53855

According to rumours this is a 10.8'' Full HD tablet (with keyboard dock) running Windows 8.1 and priced $399. Rumoured 9 hours battery life without dock and 19 hours with the keyboard dock (integrated battery).

I'd rather buy this Dell than a $449 Windows 8.1 RT Surface 2 (up to 10 hour battery life). 🙂

Any idea if the screen is 16x9 or 16x10?

 

[Sweepr]

Any idea if the screen is 16x9 or 16x10?

1920x1080 16:9

 

[bullzz]

nice. dell devices are really looking good so far

 

[Nothingness]

Dell Venue 11 Pro 5130 (based on Bay Trail-T - Atom Z3770) @ Geekbench: http://browser.primatelabs.com/geekbench3/53855

This gives an interesting comparison of how Android NDK gcc performs against VS: http://browser.primatelabs.com/geekbench3/compare/52725?baseline=53855

According to rumours this is a 10.8'' Full HD tablet (with keyboard dock) running Windows 8.1 and priced $399. Rumoured 9 hours battery life without dock and 19 hours with the keyboard dock (integrated battery).

Let's wait for independent reviews. I hope the BIOS is not locked...

 

[NTMBK]

Did anyone listen to the new podcast?

Apparently these preview benchmarks were taken in a room controlled by Intel which was incredibly cold- it's described as "literally it was a refrigerator in there". I wonder how much that helped Atom's turbo performance?

 

[Nothingness]

Did anyone listen to the new podcast?

Apparently these preview benchmarks were taken in a room controlled by Intel which was incredibly cold- it's described as "literally it was a refrigerator in there". I wonder how much that helped Atom's turbo performance?

Yep read that too. This enforces what I have been saying since the first "independent" results were given: let's wait for consumer devices reviews 😉

 

[liahos1]

putting your tinfoil hats aside. Why would the Dell's processor description be different than the FFRD?


DELL
Intel Atom Z3770 @ 1.46 GHz
1 processor, 4 cores

vs

FFRD
Intel Atom Z3770 @ 2.39 GHz
1 processor, 4 cores

???

 

[Shivansps]

Did anyone listen to the new podcast?

Apparently these preview benchmarks were taken in a room controlled by Intel which was incredibly cold- it's described as "literally it was a refrigerator in there". I wonder how much that helped Atom's turbo performance?

The temperature is the not primary limit on Turbo boost, its more like a secondary limit, the turbo boost works inside a calculated TDP window, while is true that the calculated TDP its affected by temperature by a bit, Turbo Boost does not care for temperature itselft unless its critical.

Anyone with a Intel notebook can test this using openhardware monitor, the turbo boost decreases only when the "CPU Package power" reachs a certain number, that could happen at 60°C or 85°C, turbo boost does not really care, but increasing temperature also increases the cpu pakage power slighly, especially when it goes over 75°C.

So yeah a cool room its gona help, but unless the temperatures are WAY too high its not gona be a huge difference.

 

[Nothingness]

putting your tinfoil hats aside. Why would the Dell's processor description be different than the FFRD?


DELL
Intel Atom Z3770 @ 1.46 GHz
1 processor, 4 cores

vs

FFRD
Intel Atom Z3770 @ 2.39 GHz
1 processor, 4 cores

???

Just different OSes reporting things differently.