The Intel compiler broke the AnTuTu benchmark and AnTuTu "fixed" it so the code could not be auto-vectorized. End users will benefit from the ICC vector optimizations even if the benchmarks do not show it. When a compiler breaks a benchmark, that is "good".
No, the issue was that ICC was removing loops from the benchmark. This "optimization" was specific to Antutu.
That is quite different than geekbench 3.0 which intentionally chose benchmarks (SHA1/SHA2) that explicitly give an edge to native instructions.
Yes, although only one of those SHA1 algorithms use instructions that speed up the calculation. The AES benchmark also has the same problem (but both ISA's have those when running 64bit code). However, it's easy enough to strip these scores from the numbers.
I've seen people say it won't happen because MS said they wouldn't offer full Windows on devices with small screens. Nonsense. MS is offering Windows on 7" screens already and they will be offering it on cheap Intel HDMI-sticks with no screen at all. How much smaller a screen can you get?
I'm not arguing against your broader point. The FACT, however, is that any Windows 10 device under a certain screensize doesn't have access to the desktop, regardless of what chip it has or if it's docked or not. This is MS policy for Windows 10.
Perhaps MS will change this overtime?
angs: You have asked a direct question, but don't expect a straight answer from "theblueredmonk". He has always made vague and ambiguous comments giving the benefit of doubt to ARM, never for Intel. I suspect he works for or with ARM in some capacity. Sorta like what the Nenni guy does for TSMC.
:) I've never worked for ARM or anyone in the ARMy.
You are obviously biased against Intel, making it your business to try and refute any even mildly positive comment. May I ask why?
This comment was for will_amd_yu and is more about our history of talking about benchmarks. Basically will_amd_yu doesn't believe in benchmarks and thinks Samsung is always cheating while GB3 is somehow rigged (Intel was found cheating with the Antutu).
legacy apps that do not depend on screen size such as security apps, will still work on all screen sizes.
Nope. I don't think you understand what MS is doing.
These devices (limited by screen size) are locked down and you can only run software from the MS store.
As a developer you can only put software on the store if your App meets MS requirements. ANY legacy app does not meet these requirements.
I trust Antutu more than geekbench which is rigged toward ARM chips.
Antutu, the benchmark that Intel cheated in? Once a cheater...:)
Sorry buddy, but you are wrong. Why use ARM when you can get a Windows phone running all legacy apps..
Full featured Windows 10 phones will be x86. While the Windows ARM phones will be castrated versions of windows 10 which no one will buy.
No, I am not wrong.
It's really quite simple - and this comes from MS - google it yourself - no device under a certain screen size will have access to the legacy API (the desktop). Any smartphone, any tablet, regardless of ISA (x86 or ARM) under that screen size running windows 10 can't run a traditional windows application.
Check out the Seeking Alpha article by Alcatraz Research. Says it's an INTC processor
Then the article is wrong:)
I'm aware the current version of the Mi 4 is ARM based, but the comment suggests Xiaomi will be introducing a new version of the phone. If it's not Intel based it begs the question what the comment was supposed to mean. Irrespective of the article, Microsoft needs to more clearly label the different versions of Windows 10
The next version of the Mi4 is also ARM based. You can look on MS's website, under "Windows 10 Technical Preview" for supported phones, all ARM based.
MS is using "Universal Apps" for touch, so it doesn't matter if your device has an ARM core or an x86 one.
Note the comment about "Windows 10 Technical Preview for phones". The Technical Preview is referring to full Windows 10. It's logical to conclude it's an Intel phone. What a game changer this will be.
FYI, this is an ARM ISA device (QCOM 801).
Also Microsoft isn't allowing access to the desktop for any device with a screen less than a certain size.
Also, the Note 4 shipped for the U.S uses a Qualcomm chip. The Exynos chip is for Korea. I'll leave you to Weaver to take your argument apart from now. LOL.
It's true that the Note 4 shipped for the U.S uses a Qualcomm chip, funnily enough the Qualcomm chip is also used in some Note 4 variants in Korea...
The Exynos is used in other parts of the world. I think you'll find that it's your argument that is full of holes.
Note 4 is not a smartphone. It has a wide latidude for TDP. Like I said once a cheater on benchmarks, always a cheater.
What would you call it? A big screen smartphone? A Phablet?
As for cheating, just like Intel then eh?
How do you know the 20nm Exynos doesn't overheat and throttle down in the real world? They are sold only in Korea and China.
There have been a couple of 20nm Exynos's SoC's which have shipped in various devices all over the world. The key example is the Note 4 with the 20nm Exynos and the A57 CPU. Sweepr over on anandtech has done lots of testing (no overheating or throttling like the 810) and you can ask him to run some 'real world' tests if you aren't happy with his analysis. Let me know if you want links to his testing.
Oh yeah, Samsung does well in benchmarks especially geekbench.
Well, it would do, given that it's using a high clocked A57.
especially geekbench for which it is rigged. But in real life apps, it will overheat and throttle down same as Qualcom's 810. Just watch.
Given that the 20nm Exynos doesn't overheat, why would the Samsung 7? No, GB isn't rigged.
Yeah, I know about the purity law, mine wouldn't pass. Have an Australian stout in secondary.
Given that another Liverpool ship, the Titanic, went down rather easily...The Titanic was also newer...
I know quite a bit about the Lusitania, which is why I said," you can never be sure what it's carrying". I didn't want to go down the road of arguing about whether the shells contained explosives or not. Put yourself in the U boat's captains shoes. You *know* for certain that you'll be taking thousands of lives if you target an unarmed liner. Why did he treat cargo vessels differently?
Do you also believe the UK knew about Pearl Harbor in advance, but didn't say, just to suck the US into WW2?
Have you been drinking your home brew again?
Are you really trying to justify the sinking of an unarmed passenger liner? Unless you stop the ship and search it, you can never be sure what it's carrying. But as a liner, you know for sure it's got thousands of people on board. Days earlier that same U boat stopped and sank a merchantman - after allowing the crew to get into the lifeboats...
While I agree with you on Versaille (as most historians would), I struggle to see any major similarity between Russia today and Germany of 1936? Care to elaborate?
Do you know what chip is in the iWatch? I tried googling to find out, but found no definitive info. I didn't think Apple had designed a chip that small.
Almost certainly ARM ISA based, due to the watch running a version of iOS.
In terms of CPU, it could be a low clocked version of Apples own Swift, or a bog standard ARM core. If it is running a 64bit version of iOS, then it's probably ARM's A53.
Do you foresee any 'game changers'?
Depends on what you define as a game changer. While this is good news for Intel, it's had this socket win before. Apple would be paying, what $10 for this chip?
A real game changer would be Apple dropping it's own App processor for one designed by Intel (ISA switch), but that's not going to happen. What about Intel fabbing for Apple? Possible.
You can love all you want today. All i'm saying is IF the rumor is true (which seems to be since no one is denying it) then it will be a game changer.
It's the modem...Not a game changer at all.
I think arm said this would never happen because Intel phone chips use too much power are too big are too slow bla blah blah. Imagine Brits being so full oh it. That is less than polite blokes.
You seem to think this win is the app processor, it's not it's the baseband/modem. I
Intel (infineon) has had this socket win before in the iphone. While good news for Intel, this win has no impact on ARM (the company) as Intel's modem contains an ARM CPU. It's bad news for QCOM.
You should also keep an eye out for the S6 launch, as Intel might be inside them as well (Intel won a few Samsung baseband sockets for 20nm SoC phones).