I bought a Thinkpad 8 last week. Why? This high-end Windows 8 tablet with LTE and 4GB got an update to the Z3795 CPU, has much improved performance over the original and was retailing in the UK for an absolute bargain price. Since I bought it I’ve done a lot of testing and a lot of travelling with it. I’ve also treated it to a hard case and a USB 3.0 OTG adaptor which says a lot about how I plan to use it.
The Lenovo Thinkpad 8 was the only early 2014 Atom-based Windows 8 tablet that I didn’t do a deep-dive review on. It was, as it turned out, one of the best and with its 1920×1200 display and USB 3.0 quite unique. A recent update to the range saw the original Z3770 processor being replaced with a Z3795 CPU (1.59 Ghz – 2.39 Ghz) CPU, 4GB of RAM, 128 GB SSD and a true 64-bit Windows Pro build. The improvements in performance are easy to measure.
I’ve read a number of first impressions posts today that show the Microsoft Surface 3 (Intel Atom X7, Windows 8.1) has gone out to reviewers in the USA. A few more pieces of the puzzle have been slotted into place and there’s now a huge race to get the first full review out. Don’t expect the first reviews to be too in-depth (battery life tests might have to be added later!) but do expect some performance results and thoughts on-screen and keyboard. We’re also looking out for the Surface 3 eMMC SSD speed test results but in the meantime, at least we have the first performance test results and the battery capacity.
The lab-rats at Notebookcheck have just published their full review of the Dell Latitude 7350. This 1.6KG 13-inch device isn’t ultra-mobile but we’ve put it our database as we’re implementing a strict 1.3KG minimum operating weight limit. This is a 2-in-1 detachable with a 13-inch fanless tablet that weighs just 860 grams. There’s also something important inside that we need to know about – the Core M 5Y10. It’s also a 2-in-1 which means you can use the tablet on its own. It weighs 860 grams (1.9 pounds) which is OK for a 13.3-inch Core-based fanless tablet.
Dell Latitude 13 7350. Core M 2-in-1
The Dell Latitude 7350 is business-focused and priced at well over $1000 with entry-level specifications. There’s a split battery (30Wh in tablet, optional 20Wh in dock) and a large number of features and options including ExpressCharge, LTE, VPro and the like. Dell calls it an Ultrabook but with Core M inside, we’re not. Here’s why.
Core M, at best, when the gods of heat and cooling are on your side, performs as well as a Haswell-generation (2014) Core i5 but there’s a huge range of throttling that can kick-in when things warm up. A theoretical maximum clockrate of 2.0Ghz is impressive but the base clock is just 800Mhz.
Notebookcheck, a site I do reviews for, has a strict process when it comes to reviews so when I see their performance test results I take note. They’ve just published the full review for the Dell Latitude 7350 (Core M 5y10, 4GB RAM) and the limits of Core M are clear to see.
The Cinebench Single-Thread tests show that the CPU can maintain a clock of 2 GHz while the Multi-Thread tests are executed with 1.3 up to 1.4 GHz. This behavior is identical for mains as well as battery power. According to the benchmark results, the single-thread performance is between the ULV Core i3 and ULV Core i5 processors of the Haswell generation, but the Core M is beaten by Core i3 processors in multi-thread applications because it cannot utilize its maximum performance.
Have a look at the Sunspider, Cinebench and Peacekeeper scores on the Notebookcheck review and you’ll see sub-Ultrabook performance, at least where 2014 Ultrabooks are concerned. What the Dell Latitude 7350 does bring is 2012-era Ultrabook performance in a fanless design and that’s worth thinking about when it comes to tablets.
In terms of battery life the Dell Latitude XPS 13 does quite well on the 50Wh battery configuration when compared to Haswell-based devices but again note that under load, the Core M CPU won’t be getting as much done as a Core i5 Haswell-generation. The web-browsing performance is a good comparison to use though and here we see the 50Wh battery giving 522 minutes of battery life. Again, web page loads may not be as fast as on a Core i5 but the small delay is probably not going to concern most users. The average power usage in the web browsing scenario is 5.7W which is slightly higher than I’ve seen on 10 and 11.6-inch Core-based laptops. Again the screen backlight takes more power for the same brightness on a larger screen compared to a smaller screen.
Core M 2-in-1 line up for Q4 215. (Click for more details.)
So what can we learn from this Core M product test? Firstly we have to bear in mind that this is a single reference point form a single device. Core M performance relies heavily on good thermal design and benchmarks will vary a lot across different testing scenarios. Even a few degrees increase in ambient temperature will affect results. Benchmarks themselves can heat up a device such that the following benchmark can be negatively affected and you can see that in the sequence of Cinebench tests done on the ASUS UX305 by Ultrabookreview. In that review the Cinebench result varies from 140 down to 107 points. In the Notebookcheck review of the Dell Latitude 7350 the CPU score is 139 points. In a full review of the HP Envy X2 15 c000ng, another fanless Core M 5Y10 device, the max Cinebench R15 score is 167 points. Incidentally the Core M 5Y70 (1.1Ghz base clock) as seen in the Lenovo Yoga 3 Pro (which has a fan) is reaching over 170 points on this CPU-only test.
Core M is difficult to test but I believe that the results you see here are representative of a fanless Core M device and although performance isn’t as good as an Ultrabook, Core M is allowing lightweight fanless PCs in the 1KG-class with performance that we’ve never seen before. This brings ‘enough for everything’ and with the Lenovo Yoga 11, Acer Switch 12 and ASUS Transformer Book T300 Chi coming with Core M at the $700 price point there’s value in the equation too.
I’ve been testing Windows tabletPCs with 1GB RAM recently. It’s a truly poor experience if you plan to use all the features on a Windows PC. Even with 2GB RAM you can hit limits and start seeing swap usage rise and performance hit the floor. The performance degradation isn’t ‘graceful’ at all. Chromebooks face similar issues of RAM availability but in my tests with 2GB RAM I see it matching Chromebook usage scenarios well and don’t see performance degrading as quickly as it does on a Windows PC. If you’re switching between 10 HTML5 pages then 2GB is enough and if you’re prepared to accept a sub 1-second delay in switching tabs, you can go to 20 tabs without a problem. 4GB could, however, be a requirement in the near future. Watch the video demo below…
The Ramos i10 Pro Windows 8 dual-operating system tablet will be given a press launch in China tomorrow which means we’ll have new information in under 24hrs. Why is this important? Ramos are launching a dual-OS 10-inch tablet on Baytrail and the product development has been supported by Intel. We could get the first glimpse of how Intel plans to architect its Windows / Android Dual OS solution. Have they found a way round the Google issue?
We had some hands-on with the Ramos i10 Pro when we met Ramos at CeBIT last week but more information is now leaking via the pphz.com website.
‘Spy photos’ show the first Quadrant scores: 32099, beating the Galaxy S4 and coming in behind the Galaxy Note 3. Not bad for a production sample. Clearly the Android 4.2 build is working well on Baytrail-T.
We’re in contact with Ramos and trying to get more formal information for you. In the meantime, view our Ramos i10 Pro hands-on below.
In 2008 we were measuring netbooks with a test called Crystalmark. It was a quick and simple test that allowed us to do compare across the devices we used it with. Two CPU tests, three GPU tests, a memory test and a disk test were all we needed to get a feel for the performance of a device. In late 2011 we got hands-on with a Core i5 Windows 7 tablet from Samsung and were impressed with the scores. The cost and weight were high and the battery life was low. Today we’re seeing Atom-based platforms beating that Core i5 from 2.5 years ago.
I’ve been using a Bitlocker encrypted drive for a month now and it’s been totally transparent in terms of speed. I’m surprised. I’m also surprised that it was available on my Windows 8.1 (not Pro) OS. Inspired to boost security on my Ultrabook I’ve also enabled secure boot, increased the security level, made sure Defender and Firewall are working and, this is contentious, made sure my login is only via Windows Live account so the password can be changed remotely. Given the reporting and password / device management in the Microsoft Live account though, it seems worth it. Here’s how you can do it too.
As the year draws to a close I’m frantically trying to finish the full review of the Toshiba WT8. It’s taking longer than expected for a number of reasons, not least is the fact that there’s just so much to this tablet. So many usage scenarios, so much dynamic range and even a few surprises! For example, I didn’t realize that Bitlocker encryption is available.
In order to demonstrate a few of the unique features I’ve put together a couple of videos for you that I hope will keep you interested until the full review is available in a few days. You’ll see Miracast, a USB display setup, the Bitlocker encryption and a few other things. Below the videos I’ve also published a paragraph from the full review; A teaser!
Extract from the full review of the Toshiba Encore WT8 Windows tablet:
The performance advantage over previous, Clovertrail-generation tablets really shows up in web browsing. It’s much more of a desktop experience in terms of speed and quality. Program startup times are good too and after measuring the SSD, an eMMC-based soldered module, I was pleased to see an improvement over many Clovertrail-generation test results.
Cinebench tests results for CPU and OpenGL are as good as an Ultrabook I tested recently and, just for fun, the CrystalMark04 scores are 10 times the first netbook I ever had and about 4 times that of a good quality netbook from 2010.
..and a demo image from the 8MP auto-focus rear camera which I find surprisingly good in daylight usage. One more demo pic here.
We like CloverTrail. It’s the only connected-standby capable X86 platform on the market and it blows every other Windows-capable platform out of the water when it comes to battery life but in terms of processing power it’s not quite enough to satisfy most people on a day-to-day basis. I’ve recently been testing AMD Temash and with about 2X the all-round processing performance it’s enough to satisfy most people. It’s efficient, but unfortunately you lose that CS-capability, the mark of an energy efficient Windows platform. At the top end of the efficiency-focused platforms are the Intel Core U processors used in Ultrabooks. They’re expensive and don’t offer CS capability but if its processing power you want, they are the king.
I took three devices based on each of these platforms and ran some tests to try and get some sort of guideline performance differential between the three and I’ve come up with 1:2:5. Where Clovertrial 1.8Ghz is the baseline, AMD Temash (high-end A6-1450) comes in at around 2X the processing power and a high-end Intel Core i7 at 2.0-3.2Ghz will hit 5X the performance of Clovertrail. Obviously these figures will vary across different tests and system builds but after running a number of tests, this is a ratio I’m happy to use as a guideline.
Peacekeeper, the browser performance test, is a good example as it addresses one of the most common usage scenarios for a PC. In the tests above (done with Chrome) you can see that 1:2:5 ratio showing up. (1:1.6:6 in this case) In a Floating point test I saw 1:2:5 and in Cinebench, 1:2:5. The Sunspider ratio was 1:1.4:6.3. The average across all the tests I performed was 1:1.9:5.1
In two days we’ll have another platform to consider. Intel’s 4th-Gen Core, aka Haswell, will eventually offer sub-10W parts (timescale unknown) with CS capability and a performance ratio that probably fits into the 4X range. In Q4 we’ll see BayTrail that will offer lower TDP and something like 2X the current Clovertrail performance seen here. Again it will be CS capable. From AMD we also need to consider Kabini, the higher-clocked, and higher TDP version of Temash although that’s really only for laptops and Ultrabook-alternatives.
Performance ratio of mobile Windows 8 capable platforms (guideline.)
Clearly for Ultra Mobile PC fans looking to 2014, the BayTrail, Temash and low-TDP Haswell parts will be the most important and with BayTrail offering connected-standby, that’s the one that could work best in your next Windows 8.1 handheld.
Back to today though, here’s a (long) video showing three devices on three platforms with commentary and some visual indicators of performance.
For some people it’s more important to be mobile with all-day, all-scenario capability at the expense of processing power or speed. For others, the most important thing is to be able to carry desktop power. Intel Atom and Intel Core separate these two areas of computing cleanly but how big is the difference in platform performance? I took the chance to test the Acer W510 alongside an Ultrabook convertible â€“ the Lenovo Thinkpad Twist. Both devices have strengths, and weaknesses.
A copy of this article also appears on our sister site, Ultrabooknews.
We focus on Ultrabooks here but also keep a close track on what’s happening in the ultra mobile PC space through our work with UMPCPortal. For some people it’s more important to be mobile with all-day, all-scenario capability at the expense of processing power or speed. For others, the most important thing is to be able to carry desktop power and that’s where a Core-based Utrabook comes in. But how big is the difference in platform performance? I’ve had a Atom-Clovertrail based tablet convertible for a few weeks now and so I took the chance to test it alongside an Ultrabook convertible – the Lenovo Thinkpad Twist. Both devices have strengths, and weaknesses.