AMD's Kabini: Jaguar And GCN Come Together In A 15 W APU

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.

SteelCity1981

Distinguished
Sep 16, 2010
1,129
0
19,310
don't look into this being the same as what's in the ps4 and xbox one because it isn't. it may share the same jag cores as these but everything else has been greatly modified.
 

GeoN

Honorable
May 24, 2013
1
0
10,510
Whatever AMD does is bad here on TH. Why such a tough word like mediocre? I think performance is great for its category. Please compare apples to apples not some processor which consumes twice as much power and costs twice as much.
Please consider that for desktop the power is not so relevant while for a laptop this is huge.
Bottom line the AMD will be at least 100$ cheaper than any i3 and battery life will be at least 50% better than any i3. Even if they will be in the same price bracket for sure AMD will offer better equipment like (15" screen vs 11" and/or 1TB hdd vs 500 and so on)

BTW a more fair comparison can be found here: http://www.computerbase.de/artikel/prozessoren/2013/amd-temash-und-kabini-im-test/8/
 

tomfreak

Distinguished
May 18, 2011
1,334
0
19,280
the biggest problem bobcat have is the poor single threaded performance. adding 22% isnt really enough. Kabini gone quad core is a poor choice. I would much prefer a more powerful dual core under the same TDP.
 

martel80

Distinguished
Dec 8, 2006
368
0
18,780
It makes little sense to compare the whole laptops at a specific price point. There's so much in a laptop where money can be saved or burned. This can heavily sway the comparison result in favor of one CPU vendor or another.

You could have picked some imbalanced laptop with a powerful Intel CPU, having everything else absolute crap (cheapest chassis, display and storage possible, small battery etc.) and it would totally annihilate the AMD laptop in every single of your benchmarks, except for the power consumption.

When you publish mostly performance benchmarks, you should compare at the price level of components responsible for the benchmark's results ONLY. No matter what AMD tells you, there's always common sense.
 

americanbrian

Distinguished
I for one agree with the author.

AMD sent a prototype LAPTOP. not a netbook. You say compare apples with apples. It would not be fair to compare a 14" Laptop to a atom netbook.

The people who made a mistake is AMD in choosing a laptop form factor for this chip. If they provide a 14" laptop then the author is obligated to look for a similar form factor.

Intel don't put atoms in laptops for a reason. IT IS STUPID. AMD should think hard about putting a stupid cheap APU alongside expensive FULL HD screens and what have you. Dumb move.

 

emad_ramlawi

Distinguished
Oct 13, 2011
242
0
18,760
****AMD*** TOLD US TO COMPARE IT TO $300 - $500 LAPTOPS.

Sorry but that wont fly right here, its your job as reviewer to make the right call, and produce a balanced review regardless of what the manufacturer says.

Cause who knows what AMD said and what AMD didnt say, the truth is in the numbers, and in this review there are numbers missing (ATOM)
 

mvoinea

Honorable
May 24, 2013
6
0
10,510
@americanbrian
No, you are wrong. A 14" good 1080p screen is mandatory for office, browsing and light programming+portability, if the CPU is good enough for that tasks. And Kabini is good for that and IF the price of 500USD for this prototype is right is much better for THESE tasks than a crap lowres HP i3. Of course, for VM, WEB, RDBMS and GIS (like I do), more power (i7) is mandatory.
 
I remember AMD newspage stating that Tamesh and Kabini targeted ULV markets in notebook/ultrabooks and Tablets giving the buyer copious amounts of features and performance for an affordable price tag. This review shows just that albeit again the bar for measuring AMD is excessively high.

 

pangolin_user

Honorable
Oct 23, 2012
24
0
10,510
actualy i am little bit disapointed with this review. i think it would be better if toms could update this review with intel pentium/celeron ulv in the same segment (price and tdp/basically pentium/celeron ultrabook) as :

Intel® Celeron® Processor 1037U (2M Cache, 1.80 GHz) $ 86
Intel® Celeron® Processor 1007U (2M Cache, 1.50 GHz) $ 86
Intel® Pentium® Processor 987 (2M Cache, 1.50 GHz) $ 134
Intel® Pentium® Processor 2117U (2M Cache, 1.80 GHz)

i think the celeron/pentium ulv based ultrabook was what amd kabini aimed for (low cost ultrathin), not full fledget 35 watt pentium or celeron, theres kabini 25watt n richland 35watt which is fair to compare.

sure there's $ 350 pentium n $ 500 i3, but in what form factor, brand, and build quality? HP Pavilion 11 TouchSmart hass $ 399 price point, with 10 point capacitive touch screen. good luck to get similiar spec with intel i3.
 

killabanks

Distinguished
Oct 27, 2011
104
0
18,680
[citation][nom]cleeve[/nom]Once again, please read carefully: ****AMD*** TOLD US TO COMPARE IT TO $300 - $500 LAPTOPS.[/citation]
should've thrown an atom in there for good measure i see them in laptops as high as $500
 

pangolin_user

Honorable
Oct 23, 2012
24
0
10,510
some thing like this would be more interisting :
http://www.computerbase.de/artikel/prozessoren/2013/amd-temash-und-kabini-im-test/6/
 

Steveymoo

Distinguished
Jan 17, 2011
227
0
18,680
I'm starting to appreciate AMD's direction atm. Great value, decent efficiency, and most importantly, well balanced all-round chips. There's not much point making one side of an APU way faster than the other, really.
 
http://www.techspot.com/review/671-amd-a4-5000-kabini/

There is an official AMD slide which shows the Kabini A4's comparitive competitor is the Pentium not i3. The 20-30% faster clocked A6 is for that segment.
 

whyso

Distinguished
Jan 15, 2012
689
0
19,060
Fact of the matter like it or not kabini will have to compete with intel. In the budget market the most important metric is price and perf/$. Kabini doesn't really compete well in those metrics.
From newegg i can get a 11.6 inch i3 ULV notebook with a touchscreen for $400.
http://www.newegg.com/Product/Product.aspx?Item=N82E16834230874
Ivy bridge is going EOL and prices are dropping (there are SV i3 parts for even cheaper). That is the market that kabini is going to have to compete in.

Official competitor or not this is a free market and the consumer and price determines competition.

Its absolutely bananas to think the manufacturers are paying anything near $200 for an i3 ULV chip. They are getting a massive discount (because how then could they sell a $200 chip + windows license + rest of notebook for $400 and make a profit--and remember its newegg selling the notebook for $400, the manufacturer is selling it for less than that; newegg is making a profit). Furthermore laptops are sold as systems; you cannot compare the individual price of the cpu, you must compare the price of the system as a whole.

As for power. In a thermally limited device tdp comparisons make perfect sense but if the notebook can handle the heat then power is not a very important metric; battery life is. In these types of notebooks (14 inches) it is perfectly fine to put a 17 or 35 watt tdp cpu because it can be cooled properly. Battery life must be measured and compared though. Kabini makes perfect sense for smaller thermally limited devices but as you move toward devices which can handle the heat it makes less (like in a desktop comparison between a E-450 and a 8350). If I have a 15.6 inch notebook in two versions; assuming the notebook has no thermal constraints I am going to take the notebook with a 40 watt cpu vs a 20 watt cpu if they have the same battery life but the 40 watt cpu is twice as powerful.

This review measured power consumption poorly (weighing it much heavier than battery life). The notebooks should have been unplugged (because obviously battery life makes sense when on the battery) and the time to drain the battery measured and normalized to battery capacity. These reviews should also state what power state the notebook was in during testing and for battery life tests put it on the lowest state.
 

RupertJr

Honorable
Mar 20, 2013
27
0
10,530
¿Why are you displaying and A6 logo picture on an A4 chip review? That is at least misleading. Can you put the proper one?
 

ojas

Distinguished
Feb 25, 2011
2,924
0
20,810
Don, excellent review imo. Anand wasted too much time comparing this with ARM stuff and Atom. He did include an i5 ULV, was flamed for it like you're being for comparing it to a Sandy Bridge Pentium.

People, let's face facts.

There are two ways you can measure these chips: their target market based on price, or their target market based on TDP. You don't say that "OH BUT LOOK THE CORE i3 ULV COSTS THIS MUCH", because hell, a Core i5 3570K costs the same if you go by Intel's *recommended* price, which they they've clearly written may vary.

So let's round it up:

1. The Kabini SoC tested here is a quad-core, 15w part with a decent IGP. Now it'll go into devices with either low power requirements or low cost requirements.

2. Tablets require low TDP. 15w isn't tablet-class, it's Surface Pro aka hybrid/tablet PC class. Here it'll be competing with Core i3 ULVs for price, or Core i5s for performance. It trails in performance for both, but consumes less power absolutely. However, since the other 2 finish workloads faster, efficiency and thus most likely battery life is almost going to be the same.

3. A 15W PART CANNOT BE COMPARED TO A 2W ATOM SOC. Period. You compare the 4w, 1GHz, dual core Tamesh SoC to a Clover Trail or CT+ Atom. If you read AnandTech's review, he's made a comparison with Atom, and if you scale down for clocks and cores, you'll see better single threaded but equal multi-threaded performance as a CT Atom from the lowest-end Tamesh that i'm talking about.

4. The Pentium here was a Sandy Bridge part. Ivy Bridge performance would have been a good 10% more, even more on the GPU, and the product that it would be in would be cheaper than the i3 ULV.

5. AMD will sell these in products competing with Pentium/IVB ULV chips, so the comparison is valid.

In fact, seeing that Trinity/Richland post much better performance than Kabini/Tamesh in similar thermal limits, i'm really not sure what the point of Jaguar is above 8w.

Heck AnandTech reported seeing better performance from a i5 ULV clocked at 800MHz, and you'll realize that part could fit within a 12w TDP.

Haswell has long been rumored to go down to as low as 8w, with Silvermont likely to match a quad A15 within a 2-4w TDP.

I'm really underwhelmed with Jaguar, to be honest. Netbooks/hybrids might be the only place where Jaguar succeeds, and that too if build quality is superior to an i3 model.

That's my take. Bracing for down-votes.
 

bustapr

Distinguished
Jan 23, 2009
1,613
0
19,780
looks like this new apu has an edge on battery life, which is very important in a laptop. but the performance is underwhelming. I dont really care about all this TDP crap thats being argued in the comments as if it were the most important thing in the world. this looks like itd be perfect for a college student who uses it mainly for web surfing and word apps, but it doesnt look too good for someone who uses the laptop for everything.
 

cleeve

Illustrious
[citation][nom]kyuuketsuki[/nom]That's fine, but did you guys seriously expect a low-power architecture like Jaguar to outperform Intel's full-power architecture? [/citation]

All we expect is for a product to compare to similarly priced products. :)
 

cleeve

Illustrious
[citation][nom]mvoinea[/nom]@cleeve, please provide a link for HP Sleekbook with i3-3217U at ,,starting around $400". [/citation]

As we stated, we used the same hard disk and memory across all platforms, and eliminated the display from our testing by using external. We tested the i3-3217U, not the HP sleekbook.

The i3-3217U performs in these tests just as it would in a $360 Dell:

http://www.newegg.com/Product/Product.aspx?Item=N82E16834200702
 

yhikum

Honorable
Apr 1, 2013
96
0
10,660
[citation][nom]RupertJr[/nom]Official competitor or not this is a free market and the consumer and price determines competition. [/citation]

Can you define social norms and practices which govern free market? And while you are puzzled by this question let me give you some chunk of reality. Free market is governed by laws. Laws are created by government bodies in countries. And did you check patent laws? Intel has patent on x86 instruction set, which AMD has to license. Licensing is regulation of market, be it free or not.

As for this new APU, I would love to have one in HTPC. It is quite good deal compared to what even Intel has due to factor of SoC and graphics versus power draw.

How many Atom chips go out in devices? And how many AMD APU units are going as well?

This chip represents scaled down APU for x86 instruction set. Arm is not x86, nor it plans to be. Reason is simple, Arm is much more than desktop user can deduct from limited usage of desktop CPUs, so Arm modules turning on/off will not be easily matched to desktop CPU technology. Now ARM vs ATMEL vs PIC would be a good topic to discuss.
 

cleeve

Illustrious
[citation][nom]emad_ramlawi[/nom]****AMD*** TOLD US TO COMPARE IT TO $300 - $500 LAPTOPS.Sorry but that wont fly right here, its your job as reviewer to make the right call, and produce a balanced review regardless of what the manufacturer says.[/citation]

Wow.

So, if Ford made a car that cost as much as a Ferrari, but told automotive reviewers to compare it to a hyundai accent because the engine is the same displacement, you'd be good with that?

Yeah, we're going to have to agree to disagree on that. Products compete with what's similarly priced, not based on an arbitrary specification.
 
All in all, The efficiency of cpus and gpus is just getting so damn good. I mean not at the TOP end, but the middle ground is very good.

Think of how ALL of these notebooks beat out any of the older single core models from years past.
 
Status
Not open for further replies.