Replies to post #120719 on Intel Corp (INTC)

[This Causes an Error]

DavidA2,

Okay, so a couple of things...

1. The Digitimes article is likely not accurate. Why? Well, take a look at this: http://vr-zone.com/articles/no-evidence-that-intel-has-pushed-oem-chips-into-retail/44743.html

While desktop Haswell isn't exactly a huge leap, I would be very careful in believing ANYTHING from Digitimes.

2.

Notebooks are not really better and the few Convertible Haswell Ultrabooks are at $1500.

The majority of Haswell designs haven't even rolled out yet...only a few have trickled out. How can you reach this conclusion just yet?

3.

The state of mobile benchmarking sucks, but what sucks more is that they, Intel potentially cheated on the benchmark and got caught. And Clover Trail+ seems no better than Medfield last year compared to competition.

What do you expect...it's a dual core based on an old core design. Bay Trail will go toe-to-toe with Snapdragon 800/Tegra 4, although I will say that the AnTuTu fiasco was...well, that wasn't right. The rest of the benchmarks showed CT+ solidly trailing the latest designs, so I'm not sure what Intel was trying to achieve here? OEMs are going to run a wide variety of benchmarks that reflect the intended use cases, and winning just one benchmark while losing all of the others isn't going to do much.

The real bummer is that on the corrected edition, CT+ still scores very well compared to quad A9's/dual A15/dual Krait, so I'm not sure why Intel even felt the need to cheat. If a dual Saltwell is faster in the non-BS version than a dual Krait or dual A15, then that's a pretty decent achievement and bodes well for Silvermont (2x the cores + significantly faster per core + higher clock speeds).

[Saturn V]

I am really puzzled by the Antutu hullabaloo, and I may be ignorant on some issues.

The hullabaloo is that the Icc ( Intel compiler) was used on the Intel benchmark, while the more common Gcc ( a generic compiler) was used for the ARM tests. The claim is that using the Intel compiler gave an artificial boost to the Intel hardware,since the Intel compiler generates more efficient code than the generic compiler. About 10 years ago Intel was accused of using its compiler in benchmark comparison, and the accusation was made that the Intel compiler was rigged to give better results for Intel products versus AMD products. Now the shoe is on the other foot. Intel is being accused of not using its compiler for competitors products. Did Intel run the comparison ? I thought that a third party ran the comparison. Obviously that third party must be an Intel lackey !

That is normal competitive mud slinging !

However I am puzzled by another fact. Why is the code being compiled on any compiler for an Android device ? A compiler generates assembly machine code from C or C++ source code.I thought that Android was a Java machine, which interprets Java code and thus no compilation is needed. Where does code compilation enter the picture at all? Does Android allow native machine language applications at all, and is AnTuTu one of those applications ? In that case it would be AnTuTu's responsibility to pick the best compiler for the different hardware architectures, and distribute the binary code to the user base. Would Intel be responsible for that decision ? I doubt that either Intel or any third party would have access to the AnTuTu source code

[wbmw]

Unannounced Asus Laptop featuring the U series Haswell, but Anand user review is saying 4.5-6 hours battery life

That's extremely good for a system with a 15.6" 1080p + touchscreen display, with a 4-cell battery. Ivy Bridge would likely have gotten less than that with a 6-cell.

If you want to see >10 hour battery life on a Haswell system, it's probably going to be on a 13" system with 50 Whr battery. More space constrained 11" systems with 40 Whr batteries will probably get to 9 hour. Obviously, battery life scales down with capacity, but 11" displays are also lower power, so it ends up balancing out.

With >15" displays, you'd expect OEMs to take advantage of the extra system space to put in 60-70 Whr batteries, which would help it to reach 10 hours. But 4-cells are 35-40 Whr, which is why you're at <6 hours with this machine.

[VeeCee]

Does Intel control this organization? No. Then how Intel cheated these results. Tests were run using the standard software.
If there is something wrong with the software then organization needs to take blame for that-which it has.
It has released latest version which proves exactly what I am trying to say.