Brazos Testing Requests

[JarredWalton]

Hi guys,

So we're finally starting to get some AMD Brazos laptops in house for testing. I've got one, Dustin has another (with one more on the way), and even Vivek is getting in on the action. We're probably going to focus on the normal stuff for most of the reviews, but I want to do a larger comparison between ultra-mobile platforms where we do some extended testing. To that end, I'm taking requests.

The plan right now is that I'm going to put a 60GB budget SSD into each platform so that everything is equalized. Then we will test things like boot speed, battery life, internet performance, application performance, and gaming/graphics performance. Most of those areas will be about the same as with an HDD, but anything that really hits the drive during a test will benefit from the SSD.

So beyond the above, what specific tests would you like me to run? I'm taking game requests and any other benchmarks that you feel would add value. And since this thread will probably drop off the main page if there aren't a bunch of responses, email me as well if you can: jarred.walton@anandtech.com

Thanks!
Jarred Walton
Mobile Senior Editor
AnandTech.com

 

[Hacp]

Gaming while on battery is a test I think some people would like to run.

 

[Emulex]

runtime for casual (couch browsing). compare to macbook 11/13" current gen please. they have the same ssd size too. charging time. heat when in use and charging. heat when in use and not charging. extended run benchmarks in lap where you are blocking heat vents (if applicable) to see if the built in throttling kicks in and how much. Block the vents and make it run hot and see how it goes the distance or if it melts down.

many systems have extremely aggresive throttling (macbook especially) when things are toasty. If its 120F in balmy beirut and you want to sit by the pool you shouldn't have an ipad that goes into thermal shutdown or de-clocks itself too quickly now should it?

 

[RyanGreener]

Would love to see more in depth benchmarks of SC2....haha. AT has one, and so do many other sites, but they don't really list the specifics of how they did the test, and they didn't really do many configurations with the game settings.

 

[cbn]

I'm not sure if you have ASUS 1015B in yet, but it would be great to see that one tested. (representing the classic 10.1" Windows 7 Starter Netbook with all the platform optimizations ASUS is known for.)

 

[IntelUser2000]

Hacp made a good point. I don't know why more reviewers don't do it. For certain systems, the battery life when gaming is slightly better than all out CPU. That I'd like to see, if you guys can, make it a more common practice among laptop reviews.

A more thorough architectural review would be nice. Like run some graphics specific tests like fillrate, and system memory bandwidth.

More general tests would be nice. Like do a SSD vs HDD battery life and performance comparison. I know it was done already, but having much slower CPUs and on a far more power efficient platform changes the game.

 

[JarredWalton]

Just a quick reply on a few of these items.

Charging time is heavily dependent on the size of the power adapter. In this case, I've got an MSI X370 and it has a 40W adapter. The result is that it charges quite slow, especially if it's in use.

I'll run some gaming (3DMark06 loop) while on battery to see how that goes, though truthfully the gaming potential for E-350 is quite limited. Batman does "okay" (30FPS average at low detail, but minimum frame rates drop into single digits). Half-Life 2 was much better; nearly all of the regular suite of modern games that we benchmark fall short of 30FPS. I don't think it's the GPU so much as the Bobcat CPU holding things back.

Regarding thermal shutdowns... well, maybe I'll give that a shot when I'm finished with the main review. I'd hate to kill the laptop before I'm done testing!

SSD testing will come in part two, and again it can be heavily influenced by the SSD you choose -- not to mention the tests you run. In many battery life tests, the HDD gets to go to sleep a lot of the time (or at least it should), so there the difference would be negligible. Still, I think most 2.5" laptop drives are pretty power thrifty; max power draw might spike up as high as a couple watts, but in general they're probably in the milliwatt range (judging by battery life tests I've run in the past).

StarCraft II testing for laptops is a 1 vs. 1 battle between me and a computer. I replay a sequence where the Zerg attack my Terran base, and it's representative of most battle scenes. Anand has another test where it's a 3 vs. 3 free-for-all in the center of the map. There are more than 100 units at the start, and the test is hugely bottlenecked by the CPU. Really, Brazos isn't going to be very good for SC2 unless you're only playing casual single player. (The last Protoss mission where you're defending a never-ending Zerg rush is another example of a map similar to Anand's 3v3 battle royale; I remember getting low 20s in frame rates on a beefy desktop when I played that mission!) Anyway, if someone wants to email me a specific SC2 replay they'd like me to check out, I'll be happy to do so.

Finally, the comparison netbooks/laptops are... well, lacking. I've got Atom N550 but without ION. No one wants to send a current ION netbook right now (probably with good reason). I do have an old CULV setup that I'll look at (with an SSD as well), and an AMD Nile unit as well. We might have to see about buying an ION system for comparison if we want to get a real head-to-head.

 

[themillak]

how about some not so demanding games or games that are otherwise several years old? lotro doesn't seem to have high system requirements and is free, wow shouldn't be too bad on low except in big cities.

maybe something like torchlight or oblivion, or whatever games were used like 3 or 4 years ago in the games testing suite might not be a bad choice.

 

[Emulex]

ssd drive shutdown time 1sec versus hard drive - power versus annoying spin up time factor

apple uses aggressive power down to avoid the cost of ssd (it uses MORE power than a hard drive).

 

[LiuKangBakinPie]

Well your messing your battery up while leaving it in if you run of the AC. Dont expect the battery to last long

 

[flensr]

if it has a webcam make sure you test webcam quality impact on cpu. built in mic test would also be good. lots of people use skype or other video chat with small computers so it's nice to know.

 

[ElFenix]

Well your messing your battery up while leaving it in if you run of the AC. Dont expect the battery to last long

that may have been true 15 years ago but i doubt it's true now.


i know heavy gaming isn't going to be a strong suit of any brazos platform, but checking your mailbox, auction house, etc., of an MMO and running a couple solo quests is something i could see doing. i'm not aware of one that has a built in benchmark or allows a trace to be recorded, though. and running around in some vacant area just isn't the same as running around in a town. :\

 

[Zap]

Thanks for polling us for suggestions, Jarred. Here are mine:

Battery tests while completely idle, while "normal" web browsing, while watching HD videos, while gaming. Battery tests at full LCD brightness as well as the dimmest usable in a dark room. Tests with everything factory included enabled (WiFi/BT/3G) as well as disabled.

Gaming performance, especially in stuff like RTS games (Starcraft II and more?), WoW, less demanding FPS games. Also, compare gaming performance with similar thin/light notebooks such as ones using Intel Core 2 CULV chips (GS45 chipset IGP?) and Intel Clarkdale IGP. Heck, toss in Sandy Bridge IGP numbers for grins. I'd prefer ones that can actually be found in thin/light notebooks, though Sandy Bridge might be rare at this moment.

Does the manufacturer give software controllable optimizations such as overclocking/underclocking? For instance my MSI netbook offered overclocking when on AC power, plus allowed locked down underclocking (would lock CPU at 800MHz) for extended battery life. Asus did that too with netbooks and even CULVs.

Comparison with older AMD Nile platform (I see you're on that already) to see any improvements in speed and battery life.

Maybe list available options. Seems like most are one-size-fits-all, but would be nice if some manufacturers allowed LCD upgrades, or even stuff like internal BlueTooth or extended batteries. I know some manufacturers offer custom built while others just have model variations. If the latter, maybe list a matrix of offerings.

Weight. I'm not talking manufacturer specs, but actually use a postage scale to weigh them. I did that with some of my own devices and found a few variations from specs.

Any notebook bags or especially sleeves (for frequent fliers) that fit properly? Sometimes tough to find a properly fitting sleeve.

Wow, I ended up with a big wish list!

 

[Bateluer]

Gaming performance, especially in stuff like RTS games (Starcraft II and more?), WoW, less demanding FPS games. Also, compare gaming performance with similar thin/light notebooks such as ones using Intel Core 2 CULV chips (GS45 chipset IGP?) and Intel Clarkdale IGP. Heck, toss in Sandy Bridge IGP numbers for grins. I'd prefer ones that can actually be found in thin/light notebooks, though Sandy Bridge might be rare at this moment.

Piggy back on this, older games should be tested. Its a given that the latest FPS is going to tank on any thin/light notebook. I'd like to see how it performs in various RTSs.

Also, how do you test video playback reliably? I did a quick test last night and was able to play back 720p True Blood episodes, but the dmz1 coughed up a lung trying to play back a 1080p BD Rip of Golden Compass, x.264. Flash played smoothing at 720p, but I didn't check any 1080p Flash. The thing only has a 720 screen regardless, so I'm not overly concerned at this point. But, I am thinking of building an HTPC around an E350.

 

[Eyeless Blond]

Also, how do you test video playback reliably? I did a quick test last night and was able to play back 720p True Blood episodes, but the dmz1 coughed up a lung trying to play back a 1080p BD Rip of Golden Compass, x.264. Flash played smoothing at 720p, but I didn't check any 1080p Flash. The thing only has a 720 screen regardless, so I'm not overly concerned at this point. But, I am thinking of building an HTPC around an E350.

Are you running the newest graphics drivers/codecs? Zacate is so new that a lot of the video drivers are getting leaps and bounds better with every release; I've heard others say that the 11.1 Catalyst drivers that came with the dm1z laptop would sometimes choke on 720P Flash, and others saying that 11.2 Catalyst can play 1080P Flash with no issues.

Anyway, second (or third, or fourth, whatever) on the battery life / longevity / total power draw tests, under a variety of conditions (HD video playback, web browsing, gaming, idle, sleep mode). That's one of the main reasons to buy something like a Brazos laptop, yet most reviews only devote a page, at most, to power consumption while spending ten or more doing gaming/video encoding benchmarks. Why would I do video encoding on a netbook? That's what my real computer is for.

 

[jaydee]

Would it be possible to run Anand's storage bench (light and heavy) as a standard benchmark for battery life of laptops/netbooks?

 

[flensr]

power consumption while running skype

 

[ArchAngel777]

Gaming while on battery is a test I think some people would like to run.

100% Agreed. I'd love to see you loop 3DMark06 or whatever until the battery dies. Then, provide the benchmarks for 3DMark06 while running on battery.

My UL30VT with dedicated graphics 210m would last about 2.5 hours under load while gaming.

 

[gsaldivar]

With data security becoming a increasing concern for a mobile workforce, and with the introduction of (and software support for) hardware-accelerated encryption on many newer processors (such as Intel's AES-NI), it would be nice to see a Truecrypt benchmark included as a standard part of any future platform benchmarks.

 

[Bateluer]

With data security becoming a increasing concern for a mobile workforce, and with the introduction of (and software support for) hardware-accelerated encryption on many newer processors (such as Intel's AES-NI), it would be nice to see a Truecrypt benchmark included as a standard part of any future platform benchmarks.

Pisses me off when my main desktop no longer makes it onto benchmark charts.

 

[DaveSimmons]

Video playback of rental or purchased content from iTunes and Amazon. My guess is 480p will be OK but their HD might not be.

 

[chewietobbacca]

More general CPU things, since that was where Atom was decidedly weak, and where Brazos promises a big step up

 

[gsaldivar]

Pisses me off when my main desktop no longer makes it onto benchmark charts.

Don't feel bad I'm in the same boat, and looking to purchase a new laptop soon

 

[RyanGreener]

More general CPU things, since that was where Atom was decidedly weak, and where Brazos promises a big step up

Well, Brazos made a step up by having slightly better CPU performance, but I think the main idea was to have the GPU help the CPU out. The atom was stuck with the crappy intel graphics for the most part.

 

[JarredWalton]

Note: the HP dm1z review just went up from Dustin. That is not taking into account any of the requests from this thread. I'm still working on the MSI X370 review, and I'll do my best to include any reasonable requests. Also, for the interested:

MSI recharges really slow (2.25 hours for the 32Wh battery; about twice that for the 64Wh). It's slow whether the system is off or powered on, so it seems like the charging circuit is just limited to doing around 15-20W.

Gaming battery life is pretty low, particularly with the 32Wh battery. I clocked 106 minutes looping 3DMark06 and 112 minutes with 3DMark03. Bump up to the 64Wh battery, though, and I got 270 minutes in 3DMark03. That's with WiFi enabled, but at 100nits.

As for LCD impacting battery life, at maximum brightness I got 164 minutes in the Internet test compared to 201 minutes at ~100 nits (actually 70 nits on the initial test; I need to retest at 112 nits; the 40 nits step between 70% and 60% brightness is really huge). So that works out to 11.7W power draw calculated for maximum brightness and 9.55W at 70 nits. That puts the LED backlight power draw at around 2W for double the brightness.

Gaming, as I mention in the HP dm1z review from Dustin, is going to be better if you stick to older titles. HL2 was fine (32FPS at low, and 27FPS even with most of the quality settings up -- just stay away from enabling shadows!). HL2EP2 drops to 27FPS min quality, or 23 FPS with most of the settings at high; there are definitely parts in the game where frame rates will drop into the low teens. Quake 4 ran well at low (57FPS) and still managed 33FPS with most quality settings enabled -- remember Q4 was one of the first titles to support SMP, and that definitely helps here! Unreal Tournament 3 managed 30FPS at minimum detail, and quickly drops to ~21FPS at even moderate settings. Most of these games are going to be bottlenecked by the CPU first, memory bandwidth second, and GPU last.

Still running more tests....