ASUS Vivobook U38N

I wonder what may have caused the problems when testing the games. We already saw the A10-4655M having great results in gaming on the HP sleekbook. :/

 

Sry for the delay guys, here is my review: Asus VivoBook U38N review - the multimedia ultra-portable .

I didn't have time to check for undervolting or any overcloking possibilities, like I said, I only had this one for a short time and well.. I did the best i could in the little time I had. If I'll get my hands again in the near future, I'll check for those and any other questions you might have.

As for the gaming results... I blame it on Windows 8 drivers (have you seen other results on this platform on Windows 8? I couldn't find any... maybe someone who got Win 8 on the HP Sleekbook could share some details...).

Also, do not forget that the unit I've been playing is a press sample, so there's a slight it was still unbacked, maybe with some issues. I cannot tell. Besides those, I can't find any explanation.. the computer was always running on High Performance Mode, connected to the wall socket.

If I could get Windows 7 on it and try the games, that would have probably offered an answered, but once again... no time for that.

Anyway, Hope this helped at least some of you and if you have any questions i might be able to answer, shoot

 

the fans were very loud on the UX51 I tested a while ago... but I do hope this can be addressed with an update. I remember I used to play with the Toshiba Z830 many months ago and it was horrible, but shortly after that Toshiba released a BIOS update that mostly fixed the problem. O I hope Asus can do that as well.

 

Nice review. There are a .. couple of things, though. The entire "not as efficient as the intel platform" thing. The lowest ulv processors have an "average" tdp on the processor at 17w. They easily exceed that outside what's considered average use. So to actually get that power-use, it will actually have to run at lower than full burn.

(This comes from intel's way of defining power-draw - the philosophy being to complete tasks as quickly as possible and then idle the processor. So on "normal use", and what intel defines as normal use, the design will need to dissipate that amount of power. But under load - such as when playing games or film with high compression rates, and so on, then you're going to exceed that. Intel still gets the high benchmark score, of course - and it should, of course. But it won't run constantly at that performance while going inside the tdp budget).

The amd apus on the other hand are listed with the maximum tdw. And the way it's set up by default has that p-state manager enforcing this specific tdp. If the processor/gpu exceeds it, it will lower the p-states. And it will balance the two against each other as well. So under load, you only ever get that amount of power-draw.

Saw that on the Llano as well -- the backside being that you need to force p-states to get the peak performance you are actually able to get out of it, that... in some respects might be more comparable to what you can compare to on the intel variants.. for the purpose of benchmarks. Since they don't enforce a maximum tdp.

There's also the entire thing that you can underclock these processors ridiculously far to save battery life when just typing documents or watching film from the drive, etc.

So.. what is there is 17tdp "average", just for the cpu on the intel platform. Against 25tdp max for the gpu and cpu package together on the amd.

If you add an intel hd4000 to this while running something light, like hw-accelrated video, you can add 25w only to that. An hd4000 easily wants to draw 45w while playing a game on "battery saving". Or, the lowest frequency the kit allows. It also uses some power while just idling.

So you might want to, I don't know, word it differently. Instead of taking the tdw number to translate directly into lower power-draw. Because that's not how it works.

Best example of how ridiculous the numbers on this can be. The u32u with a Llano apu setup, and an u32 with an intel/635m combo were both listed to have 7h of battery... even at Asus' site. In practice, the amd variant lasted twice as long. Whether you did an idle, low loads or full burn, the results were consistently twice the battery life.

And the Llano was rated at 35w tdp. The intel at 25w tdp.

Which is funny, of course. When Intel caveats their TDP measurement with this phrase on their specification sheets:
"System and Maximum TDP is based on worst case scenarios. Actual TDP may be lower if not all I/Os for chipsets are used."

So everyone on the same page now? It's "based on" worst case scenarios. And "may be lower". But it's not actually listing the max draw the processor demonstrably allows. So if they wanted to be accurate, they would add that it "may be higher" as well. But they don't do that, of course.

 

Big block of text!

I agree with it though. Once you start tweaking the p-states with these processors, it puts these APUs in a whole other league. It's a shame most people don't recognize this.

 

Hey nipsen, thanks for all the details.

I wasn't actually aware of the difference in rating the TDPs by Intel and AMD. I'll look more into that. To be frank, I haven't had much contact with AMD platforms lately, besides their low-power ones (Zacate).

Anyway, I've been playing with a lot of Intel ULVs though and I haven't seen the kind of power drainage you're talking about. I've edited videos on them, I've even played games, I don't remember even getting past 35 mWh (average, not peak), not even on the Core i7s.

I'm not contesting that the AMD platform can get very versatile because of those tweaks you can make (underclock, overclock), but I doubt the average buyer will know much about it or will be able to get into such stuff. Yes, enthusiasts will, but based on the demographics on my site, those are only 10-15% od the visitors. So I have to write the review in a ay that everyone would understand...

Still, I should have done a better job in the review on this section, I should have played more with the APU and see what I could have squeezed out of it, both in terms of performances and efficiency. I'll be more careful next time, I'm learning as I go from the feedback I get

 

Mm. If you use hwinfo64 (or something similar) and look at the drain on the battery, the 35w tdp rated i7s can be forced to go fairly high But it works out because it will quickly go to the highest speeds, and then idle again normally. Even if a 35w rated processor plus a Hm76 chipset, for example, will be something around 20w at the absolutely lowest on "normal clocks" whenever something happens. And then hover around 45w at peak -- it won't look extremely bad.

Because what you see as a user is just that the intel processor uses less power when just sitting there, and that the average draw of the a10 to be higher. If you're editing a video, most of the time is not spent compiling the video, after all.

It's just that if you idle an a10 quad-core and get the same draw as an i5 on idle, it depends on what you're doing with the processor very quickly, even on standard clocks. (Even if the total package - mainboard, screen, apu on my n53 when underclocked ended up at 9w - lower than my EeePC). If the processor does idle around there, then that's low, right? Same when you can get around 1000 3dmarks in 3dmark11 out of a total 35w envelope on peak drain. That's very low, specially when you consider a 17w tdp dual core processor at 1.4Ghz with an intel hd card goes above that on "average" load with half the performance.

It's something to take a look at - because even if an intel processor -- specially the ulv ones - will idle low and idle low fast (which will give the generous battery estimates, right..), the drain when you actually run the battery dry is not as impressive. On my n53tk, I could beat 4h while running a constant 20% load on the processor. Was basically having mail-check, a typewriter, browser opening news, fb, and spotify playing in the background (streaming on wifi).

I have no way to beat that on my n56 (which has the same whr rated battery), even when forcing 800Mhz and actively demote the cores. At best I end up with squeezing out 3h of it when running the components, even with low light and dropping the speakers (which draw a lot of power). Could do it completely consistently as well. Set the i7 to 1.2Ghz max, components at low light, etc., and I would get a completely stable draw. That never would let the laptop come near the "up to 5h" asus (and intel) advertise with. But on the n53 I knew I would always have 4h on "low loads" while typing.

So as a "user", I expected to be able to squeeze out more battery of the intel chipset than I could. And I'm just saying you shouldn't say you expect better battery life from the tdp and stats in the whitepapers - because you can't really expect that when you sit down and use it in practice..

So I'm not suggesting you should start to include long treatises on what tdp means.

(Btw, seems I was wrong about the u32u - that it only ever came with a Brazos chipset. And the intel version was a variant with i3 and i5/630m. And the e-450 chipset of course would draw much less power than the other combo. Not on paper, though -- that was the point. If you looked at the tdp rating, and the asus estimate, the battery life was supposed to be the same. That's why it stood out.. i3+ integrated controller -> 17w/25w depending on model, e-450 -> 18w. But on actual work it's different. As in the brazos actually got you 7h, while the intel/nvidia combo ended up at max 5.. And you wouldn't expect that from just looking at the numbers.)

 

Great job with that review! Really balanced and thorough without all that subjective bullsh*t of this is a good notebook but "AMD chips are worse than Intel" without numbers to back that up.
And the comparison of the benchmark numbers of this unit with the Intel ultrabooks shows that this is actually a very competitive device and with a much better price point.
I'll be reading your next reviews from other notebooks.
And I hope you can review more AMD ultrathins ofcourse. Can I suggest the 14'' Samsung Series 5 wih the A10-4655M APU. It's been in the market for a while now but no site reviewed that notebook, only the weaker A6-4455M version.

 

Oh, and were you even using dual channel RAM? I'm not even sure if 2GB + 4GB would work... if you aren't running dual channel RAM, that A10 will be starved for memory bandwidth. Crippled, essentially, which would explain the bad game performance.

You should've used CPU-Z to check

How to Find My Memory Bandwidth | eHow.com

 

Thanks. Very informative.

One question, did you connect two external monitors to the laptop and see if the system allows you to show 3 desktops (1 laptop and 2 external)?

 

I think he said Asus didn't include the mini-VGA adapter in this review unit so therefore hdmi+vga+internal displays weren't tested.. It should be there in the retail version id imagine.

 

Well, I think you are wrong. Look at this test: it measure Task Energy and Total Energy over period which is the best way to asses efficiency. It's a clear measure for a specific task (x264 encoding).

 

And a desktop test where the intel system will run software compression only (and you can remove any draw from a gpu), will have "speedstep" and boost states running towards an unknown limit the test doesn't list.. with tasks running where they don't identify whether or not the apu graphics part is powered or not, or if it's actually used..

...this of course tells us /everything/ we want to know about the mobile processors, right?

I mean.. it would if we were talking about intel processors, since it's the same chip just clocked at a lower speeds, and with different methods to slow it down. But it's not the same between amd mobile and desktop chips. Specially because of the way the power-balancing is set up.

Have been different tests pointing out the very big discrepancy between the apu desktop and apu mobile processors as well. For example:
http://semiaccurate.com/2012/11/12/laptop-versus-desktop-amds-a10-apus/

Again, the thing is that if you expect an intel i5 mobile/igp to consistently draw less power than an a10 quad-core - you're going to be disappointed.

edit: i.e., the only reason a mobile intel processor draws less power is the overall less frequent use of boost-states and lower frequencies. The effect is not as flattering if you unlock it and force it to run constantly at the highest frequency states. But since most tasks don't truly demand that, on average this means good performance on reasonably good power-consumption on average.

Just worth pointing out.

It does actually work on controllers nowadays to have mismatched ram and still get dual-channel. You get dual channel on up to 2Gb, the rest in single-channel. Not entirely sure if you can consistently expect that performance for active memory, though, so this is kind of iffy.

And cpu-z is kind of useless when looking at a system that changes bus-speeds and ram-speed in each p-state as it clocks up and down. It might not report what you actually get.

 

Hmm, Semiaccurate tests are jokes, I check that site for forum and some good insight info, not for tests.
Well, if you compare gaming on notebooks of course AMD Trinity is better (power consumption, performance, price) than Intel combinations. But if you compare some business applications (Office, RDBMS, VM, ERP, GIS) which are well threadead AND doesn't know about assisted GPU, then Intel is far better in every aspect. I don't know about CAD, from my point opening a 90MB DWG in Draftsight uses 3GB RAM and some big CPU utilisation and no GPU (on my N56VZ).
So my initial link is reference: choose your application, compare similar specced notebooks (external display, same battery/adapter/RAM/SSD), choose your tasks and measure both power over task completion AND over idle for CPU that finishes first. But these are too many variables for all reviewers...

 

They at least explain their methodology, and specify what they actually find and test. In this case: the difference between Trinity at 3.8Ghz, and Trinity at 1.2-2.8Ghz.

..It's not really that complicated, though. As long as something multimedia runs with hardware acceleration, you can't beat it at the moment in that power-category, either on power use or computation power.

On the other hand, if you don't have programs that do run on hardware-acceleration, etc. Or run in software mode, depend a lot on single-threaded tasks completing before other routines can start, etc. Such as CAD. Then you will get a consistently better result off the intel platforms.

I don't know.. I don't think it takes prescience to know what will favour usage patterns for the normal facebook surfer and youtube-video watcher. If you wanted to run linux with compiz as well. Or run CS a lot - a laptop apu will stomp on intel whether you measure average power-draw, standby time, or peak power use on load.

Even if Intel still has the synthetic benchmarks, of course. I'm just saying it's not as impressive if you use hw-accelerated software (that does run smoother and more accurately with good hw-acceleration). Then the benchmarks won't reflect what you're actually getting.

 

Well, a little off-topic: please try Google Maps/Street View in OpenGL mode. A classic application for many users. On my desktop (Q6600/5870) or my notebook no matter what I do (all GPU acceleration flags on) it's a VERY CPU intensive and user very low GPU. So even a application that in thery can use at best GPU support it doesn't.
Moving to a very simple Youtube 1080p movies I never have low CPU usage that I have in MPC-Home Cinema. Flash support for hardware acceleration... what a joke!

 

..try adding the web-browser as an "nvidia high performance" application first. If you use Opera or Chrome, you can enable webgl as well then. Tends to reduce the power-draw a bit, at least with kepler cards.

This kind of thing is pretty annoying.. I mean, it's 2012, isn't it? No flying nanomachine robots yet... And you have to manually set that graphics has to be drawn with the dedicated graphics processor.

On the apu this sort of thing is handled transparently towards the OS, by the way. Acceleration of instruction sets on the low-level is a dead end, imo (over explicitly parallel assembly languages). But for now.. while waiting for flying nanomachine robots.. it's probably the most convenient solution.

 

Would've been nice to see "the Netflix test" too!

Sent from my SPH-M580 using Tapatalk

 

Yeah, it was already done, do you tried yourself? So for 1080p youtube playback there is a much better solution here: HTML5, no flash, real (somehow) GPU acceleration. So GPUZ: 1080p youtube play Nvidia/Intel 37/15 GPU usage. ? I already see a fokin' flying nanomachine saucer, Intel Graphics uses lower GPU than Nvidia. CPU usage was about 10%.
On the desktop I have 50% CPU (q6600) and 40%GPU (5870). Big flying saucer, I need a big CPU and GPU for a lousy online movie.
No wonder dedicated boxes are better for specific task.

 

The former.

But yes, this is a pretty good unit.

 

*nods* But yeah, tried a few different things. One thing about %busy first.. ideally you would have 100% activity on the lowest frequency, right? So high % use on the kepler card when running at 135Mhz, that's the way it should be. 10% on the intel graphics might mean few stream-processors are engaged, and that the frequency is high (which it is - still can't find a way to underclock the hd4000 somehow)..

So enabling hardware acceleration in flash, then launching the browser with the adaptive profile and the nvidia card, basically let the processor stay at the lowest frequency as well, and have the gpu running at 10-30% or so. The same worked for me with silverlight (or netflix) - but that may have been luck. Was pretty surprised to find out that I have less power-draw on battery all the way from idle and up for the web-browser when setting it to the nvidia-card, though.

(Just to point that out, again, an apu does similar things as what the kepler cards does transparently towards the OS. Just normal gpu-acceleration is great, of course. But when we do get nano-bots, OpenCL and more explicit parallelism to conserve power/lower clock-speed is where things will happen. Set-top boxes based on ARM, handhelds with things like tegra - it's all based on that. Specialized instruction sets to allow more complex instructions per cycle.

And you get this effect to some extent on the apus now - reducing bus and clock-speed on the core, reduce power-draw, drop volt, etc. And still have enough time to run the instruction sets you need on either gpu or cpu logic).

 

Hoping to see benchmarks of this using an A8-4555m soon.

I'm pretty curious how it would stack up against an A10-4655m considering the A8 only consumes 19w.

 

Wow. I was not expecting to ever see Trinity in this form factor, especially being nearly fully powered. I feel like I am dreaming, because this is what I have been looking for. A ultrabook like device with the ability to do light gaming, while having a touch screen with Windows 8. I'm more inclined towards GPU if possible on these as the most intense area I would be doing is gaming, I doubt I will notice the difference otherwise with a Intel ULV. I think this could even potentially replace my TF300, I rarely even take it out of the keyboard dock. Only part holding me back to the 3.5 hour battery life and the potential price tag. Is there both a SSD and HDD, also any idea on a definite release date?

 

The ASUS U38N with the A10-4655M is a great ultraportable laptop, probably the only better one is The 15'' Zenbooks that cost $2k.

 

Since I have a job now, I'm saving up for the Asus U38 series. I want the ips display but I wouldn't mind having the 8550M either. Unless AMD and the oems unleash something even better then I'm going to be deciding between this and what's to come. I hope other OEMs are noticing the balls Asus grew with this model.

I might also buy both the u38n and u38dt and swap the ips display over to the u38dt that way ill get both the ips display plus the Radeon 8550M, then finally sell the u38n as is with 1366x768 display for cheap. Still not a bad package if sold for an attractive price. What do you guys think? Am I crazy or what?

Sent from my SPH-M580 using Tapatalk

 

That is absolutely good idea, if you feel this is the way to having the perfect computer for yourself. Although I wouldn't go for the 8550M, Dual-graphics just plus noise and heat, when A10 probably stronger itself.

Reading the MikeTLB's review of U38N;

AMD... " a bit more energy hungry (25 Wh TDP), but packs significantly faster graphics, runs cooler"

Isn't strange more energy hungry but runs cooler? I'd like to measure with power meter if 25W TDP AMD is really more energy hunger than 17W TDP Intel...

 

Excellent point Atom Ant! Makes you wonder how a "more energy hungry device" performs better and runs cooler than the competition. Just like how on Anandtech the iPad Mini got a 22+ page review while AMD Trinity Buyers Guide was 4 pages max!!!! LOL seriously? Meh, whatever.

 

WOW! 22 pages for that garbage?

 

Garbage? Seriously? It's a 7" tablet. There's never been a 7" tablet before! Ever! That's innovation!

Oh...wait...

To make this on-topic...I would buy this if the touchscreen flipped like the XPS Duo 12 or something similar. I don't see the point of making a touchscreen but not adding a $5 hinge to make it function as an over-sized tablet while you're at it.

 

And I do not see the sense of touch-screen laptops... a little if has the function to use as an over-sized tablet, but mouse and keyboard always preferred. Will the televisions also coming with touch-screen? That will be cool to wake up from bed to change channel instead using the good old remote controller .

 

I would love it TBH. I hate using the damn track pad is why.

My desktop is my main system, but the laptop/tablet is for being on the couch or my bed. I just find the touch screen to be better to use vs the track pad. Especially when laying down.

 

Is there any news about a US release date?

 

I don't disagree. I'm just saying, if you're putting in a $100 touchscreen, please put in a $20 hinge so I can close it over the keyboard and just use it as a straight tablet.

Yes, any news on US/Canada.

 

More good news. Another Russian website has posted and article about ASUS' upcoming products, but this time the presentation was in Moscow.
http://blogs.pcmag.ru/node/1905 (Russian Edition of PC Mag)
It says that the U38 should be available in early December with a price tag of 28000 RUB which is roughly $900/700euro, and with the inflated prices of the Russian/European market it may end up costing even less.

Now lets hope that the US/NA version comes out as soon as the Russian version.

The Taichi was also priced 33% (52000RUB about $1685, way overpriced) higher in Russia than in US, so lets ope the U38 ends up cheaper as well. Also there is the ASUS X202 priced at 18000RUB which is at about $583 on par with the US pricing of $599 and cheaper with discounts. From what I can tell only the prices of the more expensive items were inflated.

 

Yes I agree, rather touchscreen than a terrible track pad, but I was talking about mouse (bluetooth). Easier than lift all the way up the hands and holding there, while with mouse you can keep your hands in comfortable position, also working faster...

 

In terms of price, it sits right in-between those models, so I'm guessing it'll be a little bit cheaper here in the US.


Even with it's minor flaws (the hinge, soldered RAM), this is the first piece of technology I've found myself wanting ever since I got an iPhone 4 and built a desktop awhile back... Come on Asus, my wallet is open!



EDIT: According to that article, the S400 was launched there and here in november.

 

How do you guys think it will do in programs like Illustrator and Photoshop compared to an i5 with a hd 4000? Considering all the specs are the same except cpu and gpu.

 

I just think that by reading the review MikeTLB should've given an 8/10 to the notebook instead of a 7/10. But excluding that, the review was excellent.

 

Well, you see.. ..here's that problem with the short sarcastic answer and the long complex one again.

But we're talking about the difference between 1. being able to do real-time editing, and between 2. not being able to do real-time editing.

 

I already posted this on the Trinity subforum, but I think it's more relevant here:

Première prise en main du Asus Vivobook U38N

First WEI for the U38N. Minimum index is 5.7 and comes from the graphics card. The model tested there had a 128 Gb SSD + 500 Gb HDD, hence the 7.6 at primary HDD. Cool, no? No gaming benchmarks though, so let's keep our fingers crossed the drivers update solves the issues seen before.

 

When will be released? Any news for Spain?

 

I have heard that they are already taking orders in Romania and should be available in Russia and Ukraine soon (withing next ten days or so). As for anywhere else, we have no idea.

 

In the some french forums the date of 27th December :-( was circulating as a release date (for France at least and probably for the rest of Europe).
I will try to get in touch with Asus here to see if they have any up to date information.
I was really excited to see the 128Gb SSD+ 500 Gb HDD sollution and if the gaming performance (after driver update) is as good as in Starcraft II, then we'll have one HELL of a machine...
If I would make a wild guess about the delay in the release, it would be due to the demand of the UX32VD which might have slown down the inflow of 13 inch IPS screens... So far only ASUS uses them (sleep well competition, we WON'T buy anymore TN crap) so their production might not be large... We'll see...

 

What I'm excited to see is the performance of the actual production units using the A8-4555m APU.
The units that were reviewed all uses A10-4655m. That's faster than the production units for sure.

 

Lets hope that AMD and Asus don't screw up this launch with immature drivers.
This notebook has everything to be the turn-around device for AMD.

 

Well, I wish I had a better way to test how "hungry" this laptop was, instead of relying on software like battery Bar Pro, HWMonitor and HWInfo.

But those, plus the run time I got with various scenarios (described in that article), point towards a less efficient platform. Those are just the numbers, I might have a faulty unit (that was a test sample, things like these happen), there might be drivers issues as well (see the numbers i got with various games), I might have interpreted some things wrong... But please, do not imply my conclusions were biased in any way, I don't have an affinity for either Intel or AMD or any brand whatsoever...

As for why this laptop runs cooler... do not forget that it uses two fans, while the Zenbooks only use one. PLus, the U38N shares the frame with the UX32A, which is slightly thicker than the UX31A, thus there's more room inside...

 

No, no, don't understand it wrong, obviously 25W is consumes more power than 17W and you made your right conclusions. However made me wonder how is possible the 25W APU has cooler operation than 17W Intel. Of course dual fan vs single fan is explain everything, just I didn't took enough attention for the difference between the two laptops .

 

You ever get in touch with Asus?

 

They claim 2nd half of December with highest probability having a first batch of laptops before the end of the year and a new (larger) batch after the first week of January. In the second batch we should expect probably A10 versions as well. Let's see...

 

I asked Asus Iberica on Facebook about the launch date. They told me that no release date yet for Spain.