Intel Broadwell Thread

[Sweepr]

If they don't it increases the chance you will not see the higher end benchmark scores due to thermal limitations,and you also need to consider most tablets have 10.1" screens and under in size not nearly 13" screens like the pre-production Core M tablet.

Their Llama Mountain reference tablet comes in two sizes, 12.5'' and 10''. Not only Core M fits a smaller tablet but it's also thinner and lighter.

The 12.5 inch tablet measures 0.28 inches(7.2mm) thick and weighs 1.48 pounds (670 grams) while the 10 inch model is even smaller, at 0.27 inches (6.8mm) thick and 1.2 pounds (550 grams).

There are rumours that Apple will introduce a 12'' iPad, and other Android/Windows vendors are interested in larger tablets as well. Those should more expensive their 10'' counterparts, so Core M could be a viable option. 10'' and smaller tablets are usually cheaper (up to $500), that's Bay Trail/Cherry Trail territory, I'm not sure lots of customers would be willing to pay $600-800 for a faster iPad sized tablet, targeting larger tablets/convertibles makes sense.

 

[Phynaz]

Dont know but i would have thought that tablets are builts with relatively light but rigid parts, of course a lot of aluminium, casted preferably for the cover, is a good thing quality wise but it s also aknowledgment that the devices are quite large thermal source, if there s no heating issues lighter materials will be prefered as lower weight is also a desirable feature.

Seriously, you didn't know an iPad is made of aluminum?

So then you must of missed the part where Intel said these chips are designed for use in aluminum tablets - premium products. Intel will leave the crumbs of the craplet devices to AMD.

BTW, where are the AMD based tablets?

 

[Homeles]

Apple and anodized aluminum are practically synonymous...

 

[Abwx]

Look at the iPad MK5:

http://cdn.macrumors.com/article-new/2013/08/fanaticfone_ipad_5_shell_inside.jpg

That is pressed aluminium plates, it looks cool but it s not professional, pressed aluminium is fragilised definitly and will often crack at the borders with time, Apple are very good at creating an appearance of good quality using cheap manufacturing processes and materials.

 

[ams23]

Okay, though now we also have 3DMark Cloud Gate results for the Core M tablet - surely that doesn't also fall under the category of not being adequately stressful?

3DMark Cloud Gate (unlike Ice Storm Unlimited) is obviously not a cross-platform benchmark for Windows, Android, and iOS, nor does it render at a fixed [720p] resolution (unlike Ice Storm Unlimited). Anyway, the point I was trying to make is that if Core M has similar GPU performance in a tablet compared to the Core i5 variant in Surface Pro 3 (which seems plausible given the data released so far), and since the Core i5 variant in Surface Pro 3 performs similarly to Tegra K1 in a GPU-limited benchmark such as GFXBench 3.0, then logically speaking, Core M should have similar GPU performance compared to Tegra K1 in GFXBench 3.0 too.

 

[Homeles]

That is pressed aluminium plates, it looks cool but it s not professional, pressed aluminium is fragilised definitly and will often crack at the borders with time, Apple are very good at creating an appearance of good quality using cheap manufacturing processes and materials.

I wasn't aware materials had anything to do with being "professional."

 

[Abwx]

I wasn't aware materials had anything to do with being "professional."

Try to find an old car cell phone like a Motorola or a Simonsen, dismount it and check the case manufacturing quality...

 

[Khato]

3DMark Cloud Gate (unlike Ice Storm Unlimited) is obviously not a cross-platform benchmark for Windows, Android, and iOS, nor does it render at a fixed [720p] resolution (unlike Ice Storm Unlimited).

Actually, all of the 3DMark benchmarks render at a fixed resolution - http://s3.amazonaws.com/download-aws.futuremark.com/3DMark_Technical_Guide.pdf page 3. Hence what does it matter that it's not a direct comparison when it's the best that we have and it clearly indicates that the 3DMark Ice Storm Unlimited results are similar to what we should expect in more graphically intense workloads as well?

Anyway, the point I was trying to make is that if Core M has similar GPU performance in a tablet compared to the Core i5 variant in Surface Pro 3 (which seems plausible given the data released so far), and since the Core i5 variant in Surface Pro 3 performs similarly to Tegra K1 in a GPU-limited benchmark such as GFXBench 3.0, then logically speaking, Core M should have similar GPU performance compared to Tegra K1 in GFXBench 3.0 too.

Surface Pro 3 results in GFXBench have been something that I've failed to find reported anywhere, mind sharing? All that they have reported in the results section on gfxbench are Iris Pro 5200, HD Graphics 5000, 'HD Graphics Family', 'HD Graphics', and HD Graphics 4600. I guess you could say that since Core M shows about 90% of HD Graphics 5000 performance in 3DMark then it should do similarly in GFXBench. Doesn't matter too much since on the end result we're saying the same thing - Core M should be similar to K1 in graphics performance. Even better, it'll actually have games available that can use that performance.

 

[Abwx]

It is not clear to me why this is something to complain about.

Don't we want smart engineers working smarter at giving us better performing products? Isn't the above just one example of that happening?

If I already need a sturdy rigid case to frame my tablet/phone/laptop/etc then why would I care if that sturdy rigid frame is optimized to be multi-purpose and helps keep the internal stuff cooler or running faster?

This convo is sliding into the "complaining for the sake of complaining" category IMO.

I m complaining that not a single commercial product will use such a massive heatspreeader with big thermal inertia whose evident pupose is to help get the device running at full speed for a longer time, very good for benches indeed, (although on sustained usage once it has heated the inertia will keep it warm for a longer time as well , that s double hedged sword) but nowhere representative of what the buyers will have on their hands.

Anyway look like all this was rushed to materialize the CEO promise of a 2014 launch but seems like nor the process itself neither the chip are completely finalized :

Par deux fois, on nous a confirmé que ce PCN annonce en réalité un nouveau stepping de Broadwell-Y, le stepping F, lié à des changements sur le process 14nm

"We were confirmed twice that this PCN is actualy announcing a new stepping of Broadwell Y, the F stepping, wich is linked to modifications on the 14nm process."

http://www.hardware.fr/news/13891/idf-stepping-f-core-m.html

 

[mikk]

In Cinebench 11.5 OpenGL the Surface Pro 3 with an Core i5-4300U scores 24% higher than Core M 5Y70. I would expect that it differs from workload to workload.

 

[Sweepr]

Broadwell IDF News Summary

Intel's Broadwell Core M-5Y70: The First Benchmarks

Speaking of heat generation, Intel mentioned that OEMs have the option to configure all Core M processors to one of three TDP targets: 3 Watts, 4.5 Watts, or 6 Watts. The purpose of this is to allow manufacturers to cater products for specific use-case scenarios. For instance, the top of the line Core M 5Y70 could be limited to a 3 Watt TDP in a product designed for maximum battery life.

[...] Intel strongly suggested that OEMs would probably apply TDPs to processors in a manner that makes sense from a marketing point of view.

The 3DMark scores demonstrate about a 3x performance increase for the Core M-5Y70. The interesting part is that this bump isn't restricted to graphics performance, but shows up in CPU physics computations, too.

www.tomshardware.com/news/intel-core-m-broadwell-benchmarks,27656.html#topcomments_27656

IDF 2014: Intel Shows Core M 5Y70 Performance Numbers

www.pcper.com/reviews/Processors/IDF-2014-Intel-Shows-Core-M-5Y70-Performance-Numbers

Intel Core M 5Y70 Broadwell-Y Benchmark Numbers Run At IDF 2014

http://www.legitreviews.com/intel-c...rs-run-at-idf-2014_150217#RHpM2YWDhVLSQEom.99

First Actual Intel Broadwell Core M Benchmarks From IDF 2014

Finally, we have some numbers from Futuremark’s 3DMark Ice Storm Unlimited benchmark. The Broadwell tablet’s score of 50,985 is, again, much higher than we’ve seen from other ultra-mobile platforms, whether they’re based on ARM, AMD, or Intel tech. In fact, Intel was quick to point out that the score is over 2X that of a Qualcomm Snapdragon 800 series high end chip. For an additional reference point, it also appears, at least this version of the new Core M processor, is faster than NVIDIA's Tegra K1 processor in the new SHIELD Tablet we recently tested here. A total IceStorm in excess of 50K is significantly faster in fact, where Tegra K1 current drops in at a little over 31K.

http://hothardware.com/News/First-A...ablet-Benchmarks-From-IDF-2014/#ixzz3D3Ii7IzM

A peek at Intel Core M tablet performance (14nm Broadwell)

http://hexus.net/tech/news/cpu/74557-a-peek-intel-core-m-tablet-performance-14nm-broadwell

Quick Look at Core M 5Y70 and Llama Mountain

Needless to say, the Core M 5Y70 is the fastest tablet that we have test results for. The Surface Pro line has 15W TDP parts, while Core M operates in a 4.5W TDP. However, for this reference design Intel is leveraging the large amount of surface area to drive a 6W TDP. Despite this, it seems that it manages to equal or better the Surface Pro line in performance. Intel also emphasized that only an aluminum back cover would be needed to dissipate the heat to keep costs down. In fact, there were versions of Llama Mountain with copper and gold-plated back covers, but had no noticeable effect in performance. I'm definitely looking forward to future tablets and 2-in-1 devices launching with this chip, as this could enable laptop levels of performance in a tablet formfactor without the compromise we see now.

www.anandtech.com/show/8515/quick-look-at-core-m-5y70-and-llama-mountain

 

[Eug]

So, basically, that implies these results could be about a third faster than what we may expect in smaller devices. These results aren't necessarily representative of real world performance for actual mainstream shipping devices. More benchmarketing.

That said, even with a setup with more limited heat dissipation, the result are very promising. Fanless OS X laptop or Windows 2in1, here we come.

However, I'm not really interested in this right now as a pure tablet solution. Performance on pure tablets is NOT a serious issue in 2014 IMO. ARM has that covered quite well, despite lower performance, with the appropriate OSes (ie. Android and iOS, the latter being my preference), at much, much lower cost.

I don't see the point of spending $300 more on a tablet to gain performance I don't need in a tablet. What am I going to do with that extra performance, run Cinebench all day? We need Atom type price points or lower for the true mainstream tablet market.

To put it another way, right now I am actually running the absolute latest version of iOS, iOS 8, on a dual-core 1 GHz ARM A9 - iPad 2 with Apple A5. It's not exactly fast, but it's adequate.

OTOH, I also have that running on a dual-core 1.3 GHz Apple A7. It absolutely screams with that CPU, and I have no pressing desire for more performance at this time. Core M brings absolutely no practical benefit in that segment.

 

[Enigmoid]

So, basically, that implies these results could be about a third faster than what we may expect in smaller devices. These results aren't necessarily representative of real world performance for actual mainstream shipping devices. More benchmarketing.

That said, even with a setup with more limited heat dissipation, the result are very promising. Fanless OS X laptop or Windows 2in1, here we come.

However, I'm not really interested in this right now as a pure tablet solution. Performance on pure tablets is NOT a serious issue in 2014 IMO. ARM has that covered quite well, despite lower performance, with the appropriate OSes (ie. Android and iOS, the latter being my preference), at much, much lower cost.

I don't see the point of spending $300 more on a tablet to gain performance I don't need in a tablet. What am I going to do with that extra performance, run Cinebench all day? We need Atom type price points or lower for the true mainstream tablet market.

To put it another way, right now I am actually running the absolute latest version of iOS, iOS 8, on a dual-core 1 GHz ARM A9 - iPad 2 with Apple A5. It's not exactly fast, but it's adequate.

OTOH, I also have that running on a dual-core 1.3 GHz Apple A7. It absolutely screams with that CPU, and I have no pressing desire for more performance at this time. Core M brings absolutely no practical benefit in that segment.

Absolutely. More caveats. However, intel is showing the type of performance that is possible. There really is nothing to prevent OEMs from choosing to build similar devices.

I agree that pricing is important but cherry trail should cover that.

I see in the next year or so.

Bay trail continues to be sold for the low end.

Cherry trail is introduced for the mid end.

Core M takes the premium space.

 

[Khato]

So, basically, that implies these results could be about a third faster than what we may expect in smaller devices. These results aren't necessarily representative of real world performance for actual mainstream shipping devices. More benchmarketing.

While this is true here's an equally appropriate query - are benchmarks on ARM devices more representative of real world performance? They've been in the state of inflating benchmark results by usage of unsustainable power states for years now. And some of them are worse in that they specifically identify benchmarks in order to initiate that behavior - at least with Intel's superior hardware power and temperature management scheme consumers will benefit from those potentially unsustainable turbo boosts in real world scenarios.

 

[Qwertilot]

Mostly in said laptops and convertibles where there should be room for this to stretch its legs properly.

Does anyone know if we're getting 35/45W desktop versions of this? When you see the percentage of the performance they can cram down this low it does make you wonder how much extra the 85/90W versions will have over those.

 

[Eug]

While this is true here's an equally appropriate query - are benchmarks on ARM devices more representative of real world performance? They've been in the state of inflating benchmark results by usage of unsustainable power states for years now. And some of them are worse in that they specifically identify benchmarks in order to initiate that behavior - at least with Intel's superior hardware power and temperature management scheme consumers will benefit from those potentially unsustainable turbo boosts in real world scenarios.

Depends on the platform. In general the A-series of chips from Apple don't throttle much over time. What you see is what you get.

The benchmarketing shenanigans came more from the Android side, but only specific vendors, esp. Samsung.

http://arstechnica.com/gadgets/2013...rking-adjustments-inflate-scores-by-up-to-20/

 

[witeken]

I see in the next year or so.

Bay trail continues to be sold for the low end.

Cherry trail is introduced for the mid end.

Core M takes the premium space.

SoFIA will be sold for the low-end.
Cherry Trail takes the mid-end.
Broxton takes the high-end.
Core M the premium space.

 

[liahos1]

SoFIA will be sold for the low-end.
Cherry Trail takes the mid-end.
Broxton takes the high-end.
Core M the premium space.

is that right? I thought sofia was purely a integrated 3g low end part for phones and that we'd get an integrated lte part also for phones next year?

 

[mikk]

SoFIA will be sold for the low-end.
Cherry Trail takes the mid-end.
Broxton takes the high-end.
Core M the premium space.

Broxton takes the highend? Why do you think so?

 

[mikk]

is that right? I thought sofia was purely a integrated 3g low end part for phones and that we'd get an integrated lte part also for phones next year?

Sofia is for lowend Smartphones afaik yes. Not for tablet. I think he is wrong.

 

[bullzz]

@mikk - do u know that there will only 1 sofia die with integrated modem. how difficult is it to cut out the modem. intel would want to sell into $50 tablets too

 

[liahos1]

they said sofia was a pragmatic solution for getting an integrated baseband apps process out asap for low end phones. seems like it would be counter intuitive to pay tsm a margin to build a discrete aps processor version for tablets when they could keep the foundry margin for themselves

 

[bullzz]

@liahos1 - paying tsmc a margin is what all ARM competitors are doing while making a profit. considering the time and money spent on designing the chip and porting silvermont to an outside foundry, it only makes the sense to sell it wherever possible

anyway this is OT

 

[witeken]

is that right? I thought sofia was purely a integrated 3g low end part for phones and that we'd get an integrated lte part also for phones next year?

Broxton takes the highend? Why do you think so?

Sofia is for lowend Smartphones afaik yes. Not for tablet. I think he is wrong.

Have you seen my avatar ?

 

[mikk]

Bay Trail is high-end according to this lol. This slide is nonsense.