You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
Tegra K1-64 continues to hold the top spot in mobile device (smartphones and tablets) PCMark for Android; but close behind are two Atom Z3580 devices (Zenphone 2 outperforms Nokia N1; perhaps due to having 4GB of DRAM). Given the high expectations the Samsung Galaxy S6 performs surprisingly poorly in the real world and comes in at #12. Perhaps most interesting of all is the first Snapdragon 810 device ranking #48 and underperforming Snapdragon 800 (ranked #28)
In real-world devices, Finfets are near the top of the list, with the exception of Nvidia non-Finfet devices. Nvidia Denver+Finfet should be unbeatable (with Maxwell graphics, OMG!!). ARM A57 is looking shaky as a basis for leading devices. Intel Atom Finfet (22nm) has proven it's worth with the Atom Z3580 and X7 devices should perform similarly, but with better battery life. Qualcomm's position has slipped badly. Where is the successor to Krait?
The problem is that your are comparing tablet scores with phone scores.
Samsung Galaxy S6 is actually pretty good for a phone: 5180 according to Anandtech. The Nexus 9 is a tablet and has a bigger power envelope, that's why it's faster.
You should not limit yourself to the results one single benchmark if you want to see fair results. PCMark is weighed heavily towards multimedia, while geekbench is pure CPU, for instance.
The benchmarks surprised me also, that's why I put it up. But it gibes with reality. Qualcomm Kryo is at least a year away, Krait is 3.5 years old, and while some benchmarks can hide it, Qualcomm is hurting in the real world.
@astilo
The second best score on that list is an Intel phone as is number 15. From the review I have read, the real world performance is very good. I also have a cheap Android tablet with an Intel Atom, which is surprisingly good both in performance and battery life, especially considering how little I paid for it. I would descibe it as 90% of the performance of an iPad (for my needs) for 25% of the price.
Have justed looked at the website of one of the largest electroniucs retailers in the UK, almost every tablet they had for sale apart from Apple and Samsung now has Intel inside.
I also have a cheap Android tablet with an Intel Atom, which is surprisingly good both in performance and battery life, especially considering how little I paid for it. I would descibe it as 90% of the performance of an iPad (for my needs) for 25% of the price.
Have justed looked at the website of one of the largest electroniucs retailers in the UK, almost every tablet they had for sale apart from Apple and Samsung now has Intel inside.
For that we can just say thank you to the Intel contra revenue policy (as a customer). Intel silicon is not cheaper than that of the foundries absolutely.
Let's be fair, with these scores it is not all contra revenue. Also consider, the original Intel Inside campaign involved a fair bit of contra revenue and I think we can all agree this was a successful program.
Let's be fair, with these scores it is not all contra revenue. Also consider, the original Intel Inside campaign involved a fair bit of contra revenue and I think we can all agree this was a successful program.
Hi Fred, I got one of those windows 8.1 2to1 convertible because it was cheap and even if it is not very fast, I still find it quite decent for what I have to do. So, as a customer, I'm satisfied.
But I wouldn't call that a bit of contra revenue and definitely not a successful program: they lost 4B$!!!
No question that on snap shot basis Intel is not succeeding in mobile but I have been around long enough to see how they adapted from DRAM to processors. Its not showing now but the mobile market plays to Intel's strengths. I live in QCOM country and can tell you they are having going concern worries over there.
2015 seems to be shaping up to be the year it was "safe" to buy an Android smartphone that didn't have Qualcomm inside. The 14nm Exynos benchmarks better than the Snapdragon 810 and in reality it only needed to match or closely approximate it.
2015 seems to be shaping up to be the year it was "safe" to buy an Android smartphone that didn't have Qualcomm inside. The 14nm Exynos benchmarks better than the Snapdragon 810 and in reality it only needed to match or closely approximate it.
I think we have to give Samsung credit. I remember when they said that they would start 14nm early 2015, most people were sceptical. But the fact that they have an actual finished product in the shops already is quite impressive and has proved the doubters wrong.
Anandtech has had some quality articles (with no Anand) on Samsung's Exynos progress. The Anandtech thesis is interesting: Samsung has become a process optimizer that gets the most out of "stock" ARM and Mali cores (eventually, not always at first). TSMC and Intel, in contrast, are better bringing innovative, differentiated designs to market.
This is understandable as Samsung engages in many commodity chip markets (NAND, DRAM) and they are applying the same model to logic. The implications are huge though. As the ARM market becomes more commoditized, more NAND and DRAM-like, there may be a strong competitive devaluation of the Intel/TSMC model (if they don't adapt). Samsung has the dominant position in NAND and DRAM, proving their model is quite successful.
