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

Page 5 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
This is a stupid review. This is an Intel bought one considering they choose the weakest processor out of the batch. AMD has a few 18 watt processors whats do 1.7 ghz / 2.3 ghz turbo that are quad core, they literally destroy an i5 in benchmark performance. Their GPU plays all games at medium settings at relatively high resolutions. They are capable of being put into a sleek portable clamshell/sleek laptop design with a massive battery. Last years model a8-4555m, it has 384 pixel shader processors. Capable of running Bioshock Infinite at 50-60 fps at medium settings at 1366 x 768. That's just the iGPU. I learned not to trust this website as I nearly had to allow 100 scripts through just to fucking posting. Tom's Hardware is collecting dirty money from Intel to diffuse the slowly growing AMD laptop market. Intel wants to close out them out of laptops because they already did it with Desktop. However AMD is getting even that back, the newer IPC performance of the recent chips had nearly a 1:1 performance ratio per mhz at lower frequencies. That's scary for Intel because their current laptop i3/i5/i7's are only doing about 0.8:1 ratio. I want to see some GDDR5 laptops.
 
[citation][nom]whyso[/nom]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/Prod [...] 6834230874Ivy 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.[/citation]
its has 2 cell battery cough..cough..
and what prices do you expect from amd kabini a4 5000 powered laptop with similiar specs from newegg? i think it would cheaper.

2. you forgot the fact that amd kabini is soc, thats why 2 watt TDP difference between kabini and i3 translate into 14 watt difference in real life. i3 with southbridge will go up 25 watt or higher in the range of amd a6 segment.

3. this is not budget market but budget ultrathin/ultrabook market. i have seen people prefer less wight than performance, like college and school girl.

4. I think toms is international site, who knows their reader come from across the globe, there's market outside USA which cannot access NewEgg or any other deals, like my country, theres only atom celeron and pentium in $300-400 with cheap plastic and bad screen. an asus x202e with intel i3 will cost $ 699 ($300 difference with newegg) and the same laptop with pentium 847 (1.1Ghz) will cost $ 479. whooping $ 120 difference. and i would expect amd kabini equipped laptop at $ 449 (like liano a6 here)

[citation][nom]whyso[/nom] 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).[/citation]

the same situation goes to manufacturer buying amd chips. so whats your point?

[citation][nom]whyso[/nom]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.[/citation]

then why when we change the hdd with sdd? why not benchmarked with what it got in the first place.
This is processor benchmark, not laptop benchmark.

OTOH amd didn't sale laptop, its up to manufacturer and retailer to price its product, give its discount and sale. so this is outside off amd hand. what do you think about comparison between amd kabini equiped with ssd and intel i5 with hdd (similiar price) from subjective snapinnes/respons? its fair?

[citation][nom]ojas[/nom]There are two ways you can measure these chips: their target market based on price, or their target market based on TDP. [/citation]

third Both of them. I want ultrathin low power notebook which also lowcost. or best performance ultrabook and i dont gave a damn about the price.

[citation][nom]ojas[/nom] AMD will sell these in products competing with Pentium/IVB ULV chips, so the comparison is valid.[/citation]

but the review is pentium 35watt cpu not pentium/celeron ivb ulv chip. the the comparison is not valid.

[citation][nom]ojas[/nom]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.[/citation]

based on your logic, then there's no point on intel pentium ulv, celeron ulv, and atom 10w as well, because it will cost as much as amd kabini, and will be trashed by intel core i3/i5 ulv. so in your logic all we need is atom
 
[citation][nom]SiliconWars[/nom]You've served up the most negative review on the web by far Cleeve. You might want to have a look at your methods and wonder why that is.[/citation]
AnandTech followed up with a comparison to an IVB Pentium

Don (aka Cleeve) posted perhaps the most realistic review on the web.

[citation][nom]Blandge[/nom]Atom chips range from 1W-10W, so comparing Atom to 15W Kabini would be just as wrong as comparing Kabini to a 35W Core product. The fairest comparison is to an similarly priced product in the same category.[/citation]
Exactly. Atom is sub 1-3w BTW (that too Clover Trail+).
 
[citation][nom]pangolin_user[/nom]but the review is pentium 35watt cpu not pentium/celeron ivb ulv chip. the the comparison is not valid.based on your logic, then there's no point on intel pentium ulv, celeron ulv, and atom 10w as well, because it will cost as much as amd kabini, and will be trashed by intel core i3/i5 ulv. so in your logic all we need is atom[/citation]
The Pentium is there for the price range. There isn't any 17w pentium to compare with. I repeat, the pentium is there because it's competing at the same price, the i3 is there because it has similar TDP.

The only celeron/pentium ulv parts are y-series parts, which are in the 10-13w range, but i'm not aware of any product that has them. Though i think i do remember a win 8 tablet with a celereon.

If you feel the comparison is invalid, you're free to test and post benchmarks, promote AMD, buy 10 of them yourself, but if you put your fanboism behind you (or stop trolling, whatever you're doing), you would see that this isn't probably what they were looking to show.

Intel Pentium/Celeron Y-series parts are rated at a 10-13w TDP with a 8w SDP, and they're hardly in any tablet models so there isn't much uptake, because you need forced air cooling at this TDP. So yeah, there isn't a point.

Atom, on the other hand, hasn't been 10w for quite a few years, and current single core models are sub 1w and are in a number of phones (Medfield/Lexington) while dual core parts (Clover Trail) are sub 3w (typically sub 2w). So again, you either have incomplete data, don't want to accept it, or are just trolling.

Reduce this APUs performance by 33%, and divide that by half, that's the performance you'll see in a 4w Tamesh (roughly).

FFS fanboism has its limits. So does trolling.

I generally like AnandTech but i found their review overtly optimistic. Even he later states that the only point of Kabini is to give OEMs more budget headroom to add stuff like a better screen within a $400 device price tag.

And another thing. Haswell is 8 days out.
 
Tech Report:
We asked AMD whether this whitebook was representative of a typical Kabini configuration. We were told that it's "in the middle of what you might see." The company expects the most inexpensive Kabini notebooks to be priced at just $399. Thanks to the processor's tight power envelope, PC makers will have a wide range of display sizes to choose from—and there will no doubt be some touch screens in the mix, as well.

Our testing methods
We compared the performance of the A4-5000 whitebook to that of four systems:

A premium ultrabook, the Asus Zenbook Prime UX31A, which has a 17W Core i5 processor and is priced at $1,100 right now. Retail notebooks based on the A4-5000 shouldn't cost anywhere near that much, but the Zenbook Prime gives us a high-water mark for performance in the ultrathin category.

A low-end ultrathin laptop, the Asus VivoBook X202E. This system has a 17W Core i3 CPU backed by single-channel memory, and it costs $399. In terms of both pricing and performance, this should be one of the most direct competitors to upcoming laptops based on the A4-5000.

An Atom-powered Windows 8 tablet, the Asus VivoTab Smart ME400C, which is priced just south of $430. This is one of the lowest-power Windows 8 systems on the market today. Its Atom Z2760 processor manages to squeeze dual 1.8GHz, Hyper-Threaded cores into a Lilliputian 1.7W power envelope. The ME400C is obviously not in the same league as the A4 whitebook, but it provides us with a performance baseline for an ultra-low-power x86 config.

A Mini-ITX desktop build based on AMD's E-350 mobile APU. The E-350 is the A4-5000's predecessor. It has two Bobcat cores, integrated Radeon HD 6000-series graphics, and an 18W thermal envelope. We were hoping to procure a notebook based on the E-350 (or the slightly quicker E2-1800) to run battery life comparisons, but we weren't able to get one in time for our review. This desktop build is the next best thing; it will let us see how much Kabini has moved the ball forward.
Click to expand...
http://techreport.com/review/24856/amd-a4-5000-kabini-apu-reviewed/3
 


kabini is a great product but its in the ultraportable categories that its shines in. By the time we are talking 14 inch notebooks its lustre is fading.

2. Very true, but this really only matters when thermally limited. kabini has an edge in power consumption which really will mostly be seen in internet and video playback (and its less as a % there because those values do not include the screen--about 10-15% more battery life it looks like).

3. Its both. There were E2-1800/ E-450 desktops and 15.6 inch E-350 laptops as well as ultraportables. Kabini has a major edge in ultraportables but in a 14 inch laptop that edge is lessened (thermals aren't a concern, performance is lasklustre--pretty much no gaming, but better battery life). Comparison in a 15.6 inch notebook where thermals aren't a concern comes down the performance, price and battery life. Kabini loses performance, possibly does better in price and does a little better in battery life.

4. If you think you are going to get a $300-400 notebook without cheap plastic and shoddy quality you are mistaken.

5. My point is cutting the price of a $200 chip in half is a $100 reduction (which intel can afford given their margins). AMD first can't afford to cut their prices like that and second the differential shrinks because their chip is cheaper (a $60 amd chip at half price is $30 [which amd couldn't do and make money] a $200 intel chip at half price is $100. Price difference is $70 vs $140 before). The intel system looks slightly more appealing now that chip prices have been cut in half.

6. Like was mentioned in another toms article you cannot buy the cpu itself. You must buy the cpu as part of a system. Furthermore this means that perf/$ must be compared as a system cost.

Say CPU A costs $50 and CPU B costs $100. CPU B is 40% faster than CPU A. From a strict perf/$ comparison CPU B loses. However in a laptop there is a large essentially fixed cost that has nothing to do with the cpu (chassis, battery, win license, screen, etc). If for instance we assume that the other BOM + other costs (marketing,shipping, profit, etc) is $300, then the same notebook with CPU A costs $350 and with CPU B costs $400. Now all of a sudden that 40% increase in performance only costs 50/350 =15% more.
 
Wait, so 31 fps is "quite smooth" and 26 fps is "not really playable"?

Ignoring the fact that your "acceptability threshold" always seems to be exactly one frame below whatever the Intel result is, are you guys aware that 99% of movies use 24 fps?

Tom's Hardware, praising whoever buys the most ad space since 2001...
 
frankly FPS means little... i've seen smooth playable experiences at 20FPS, and choppy ugly experiences at 40FPS.

it comes down to how the frames are paced.

so it is plausible the 26fps was choppy while the 31fps was smooth...
 


Rule of thumb for desktop gpus

60 fps - very smooth (assuming no bad lag/MS)
>=40 fps - decent, not ideal
~30 fps - borderline of playability (playable with many compromises)

for mobile igp standards are lower
(igps can also have stuttering problems so its much looser as well)

60 fps - very good
30 fps - quite good/ okay
25 fps - borderline (playable with compromises)

But generally you need 30 fps to really be playable.

Movies are 24 fps and you can see it. They also use motion blur the 'smooth out' the transitions between frames. Also movies are simply being watched, you are not playing (no player input) and so you can't 'feel' that it is not smooth.

 
[citation][nom]whyso[/nom]Rule of thumb for desktop gpus60 fps - very smooth (assuming no bad lag/MS)>=40 fps - decent, not ideal~30 fps - borderline of playability (playable with many compromises)for mobile igp standards are lower(igps can also have stuttering problems so its much looser as well)60 fps - very good30 fps - quite good/ okay25 fps - borderline (playable with compromises)But generally you need 30 fps to really be playable.Movies are 24 fps and you can see it. They also use motion blur the 'smooth out' the transitions between frames. Also movies are simply being watched, you are not playing (no player input) and so you can't 'feel' that it is not smooth.[/citation]

I think this article is still pulling magically non-existant numbers out of it's hat. I'm currently using an E-350 for a netbook. I can play Dead Island at 20-30 fps on low settings, I could play Bioshock Infinite on low settings. I even left some of my games on native resolution. I've got Skyrim running at 20-35 fps. Seeing as I was getting that, I think this article writer is fibbing his results, I bet he doesn't even have an A4-5000. Also note peak wattage of the processor remains near the TDP where Intels nearly does double.
 

Measured powers are not CPU wattage; they are system wattages and in Ivy/Pentium's case, the CPU TDP does not include the IO bridge and other component TDPs included in Kabini SoC. That alone can account for a fair chunk of the increased power under load.

Broadwell will have integrated IO hub. Eliminating the DMI bus by bringing all IO on-die should drop platform power by a few more watts.
 
I don't want information removed, Cleeve. I want more information added. More chips tested, and tested with optimal memory setups. Combining the data from TH and Anand gave me a bigger picture, but their articles are just better in my opinion.
[citation][nom]ojas[/nom]AnandTech followed up with a comparison to an IVB Pentium Don (aka Cleeve) posted perhaps the most realistic review on the web.
Exactly. Atom is sub 1-3w BTW (that too Clover Trail+).[/citation]I read both Anandtech articles. They both paint an interesting picture, that puts Kabini at the power consumption level of laptop/netbook Atoms, but performance is considerably better - in some cases about as good as hotter, more power hungry, more expensive chips. Look at the power consumption in the Anand follow up article. In CPU benches, total platform power (not just the SoC) on the Kabini unit didn't even hit 12W. In GPU benches total power barely broke 15W (again not just the SoC). It's really quite efficient, and will compete nicely with Atom (at least for now).

I have no idea why AMD wants to pit it against higher-thermal and higher-cost chips, I would say the upciming A6 Kabini would be better suited for that task, or better yet Richland. But the thing is... it's up to the OEMs. The AMD prototype uses a rather decent display. I think in the end we'll see Kabini in cheaper setups, and Richland will end up in the pricier models. That changes Kabini's targets around to make more sense.

Also, not all Atoms are super low 1-3 watt models. They span a wide range of power and performance targets. That's like saying IVB chips are 17W - some are, some aren't. So I'm not sure what your point is there... Temash has a 3.9W model - and that's a SoC so that's really pretty good, especially with the 20-22% IPC increase over Bobcat.
 
[citation][nom]tourist[/nom]Based on AMD expected retail pricing, A kabini laptop/netbook is over priced. At least don took the time to show readers why it is so. A quck check on newegg shows Atom netbooks are 250 us, pent notebooks 300-350, i3 notebook 350. AMD said between 350 500 dollars common sense dictates that is where you would look for competition.[/citation]The prototype tested has a good 1080p display and is 14". The ones you're talking about are generally smaller, like 13" i3s. Anyway, just like Brazos, Kabini will end up in very affordable budget models and will likely do well at lower prices. OEMs dictate components and prices.
 


Nah, all review sites show barely playable to non playable performance. (anandtech, techreport, etc).

From anandtech

Oblivion - 1366 x 768 Medium Diablo III - 1366 x 768 Low
AMD E-350 (Radeon HD 6310) 20.1 fps 21.9 fps
AMD A4-5000 (Radeon HD 8330) 26.1 fps 25.8 fps

Not really a big jump there. (generally seems around 20-40% jump)

Skyrim

skyrim-fps.png


(note the i3 used in this test had throttling issues which can be seen in reviews of that laptop--expect a normal i3 to be around 10% less than the i5). Other reviews mirror the trend (absolute fps differ but the % differences remain similar)
 
[citation][nom]alextheblue[/nom]I don't want information removed, Cleeve. I want more information added. More chips tested, and tested with optimal memory setups. [/citation]

We can agree on that.

What we can't agree on is testing a 15W Kabini against Atom. That doesn't make sense.

When we have a Temash on hand, that will make sense. No sooner.


 
[citation][nom]alextheblue[/nom]Also, not all Atoms are super low 1-3 watt models.[/citation]
I rechecked, you're right, however:

The ones intended for smartphones and tablets are 1-3w (Clover Trail costs $41, Medfield would be less), and both anand tech and tech report compared against a 2w part.

The ones intended for *desktop* usage are 4-13w (priced at $29-$63), with the Saltwell based ones being 10w.

The ones intended for NAS and storage server applications are 11-15w (priced at $80-$120).

The ones intended for microservers are 6-8.5w (priced at $54-$64).

What you'll ultimately end up choosing between is either a 5 year old Bonnell part, or Pentium, Core i3, Llano, Trinity Richland or Kabini. I'm not sure those Bonnell based netbooks are sold anymore.

So what do you think should be the valid atom comparison?

[citation][nom]cleeve[/nom]We can agree on that.What we can't agree on is testing a 15W Kabini against Atom. That doesn't make sense.When we have a Temash on hand, that will make sense. No sooner.[/citation]
+1. Though i'm sure now people will wanted it tested against the desktop/NAS chips XD.

Which might be fine on its own for a pure architecture comparison, ignoring stuff like what you can actually buy.
 
[citation][nom]cleeve[/nom]We can agree on that.What we can't agree on is testing a 15W Kabini against Atom. That doesn't make sense.When we have a Temash on hand, that will make sense. No sooner.[/citation]
This ...
I am agree with you like i stated on previous post, i just hope theres boarder processor to be tested, so reader could have boarder picture of what they could expect and what compromise of processor they selected.
but after reading review on anandtech, I realized that reviewer didn't have all the processor they or their reader want to be compared, there are limited by time and by laptop at hand at the moment. and its up to the reader to broaden their knowledge by combining other review by the other site.
so my conclusion :
amd kabini a4/a5 you got slightly worse threaded application, better multi threaded application and battery life, similiar price
intel pentium/celeron ulv you got better ipc, better single threaded application, slightly worse multi threaded application (2core no ht) and battery life, similiar price with kabini.
intel pentium/celeron non ulv you got better ipc, better single threaded application, similiar or better multi threaded application (higher frequency) and battery life, similiar price with kabini. worst TDP so you got bulky laptop
intel i3/i5 ulv best performance (ipc,single threaded app, and multi threaded app) but higher price
 
[citation][nom]ojas[/nom]The Pentium is there for the price range. There isn't any 17w pentium to compare with. I repeat, the pentium is there because it's competing at the same price, the i3 is there because it has similar TDP.The only celeron/pentium ulv parts are y-series parts, which are in the 10-13w range [/citation]

as i stated in previous post :
Pentium ulv 17 watt
intel® Pentium® Processor 2117U (2M Cache, 1.80 GHz) 17 watt IVB
http://ark.intel.com/products/71469/Intel-Pentium-Processor-2117U-2M-Cache-1_80-GHz
Intel® Pentium® Processor 987 (2M Cache, 1.50 GHz) 17 watt SB
http://ark.intel.com/products/67194/Intel-Pentium-Processor-987-2M-Cache-1_50-GHz
Celeron 17 watt
intel® Celeron® Processor 1037U (2M Cache, 1.80 GHz) 17 watt IVB
http://ark.intel.com/products/71995/Intel-Celeron-Processor-1037U-2M-Cache-1_80-GHz
Intel® Celeron® Processor 1007U (2M Cache, 1.50 GHz) IVB 17 watt
http://ark.intel.com/products/72061/


[citation][nom]ojas[/nom]If you feel the comparison is invalid, you're free to test and post benchmarks, promote AMD, buy 10 of them yourself, but if you put your fanboism behind you (or stop trolling, whatever you're doing), you would see that this isn't probably what they were looking to show.[/citation]

my first laptop intel pentium m then intel core 2 duo and i sold my brazos laptop cause its to slow for my need. I sell a lot of intel powered notebook, because my customer need best suited with it (IT college with casual gaming). so wheres is my fanboism and trolling? i never stated amd kabini is better than i3 or pentium. although i probably buy kabini 2.0 if equiped with turbocore (like trinity to Richland) 11,6 screen with well build quality (thinkpad x201e successor maybe). what i stated is i little disappointed with the review.
 
Cleeve: TechReport OpenCl Benchmarks

Regarding OpenCl acceleration and Kabini vs Ivy Bridge:

TechReport has the best coverage on the capabilities of the Kabini and you should revise your review to show the ALL the results. Such as in LuxMark App ICD + IGP. The Atom-priced AMD Kabini outperforms the $200+ Ivy Bridge I3 and in the MuseMage outperforms the i5.

Luxmark 2.0 - OpenCL




MuseMage - OpenCL



 
[citation][nom]ojas[/nom]I rechecked, you're right, however:The ones intended for smartphones and tablets are 1-3w (Clover Trail costs $41, Medfield would be less), and both anand tech and tech report compared against a 2w part.The ones intended for *desktop* usage are 4-13w (priced at $29-$63), with the Saltwell based ones being 10w.[/citation]You're right. I ran across 3W when looking at that model, but you're right, the higher-power Atoms are just for desktop/microservers now. So I withdraw any comparison to Atom. But comparing it to a 35W chip is no better, especially if the justification is pricing alone. There are lower-power Pentiums and Celerons to compare it to, that will fall more in line with both the price AND power consumption of this chip. Also Brazos, which this chip replaces.
 


Depends really on the form factor in which the chip is used. In a 15.6 inch laptop 35 or 45 watt competition is fair game (though will lose horribly in power usage). In a netbook/11.6 inch subnotebook then atom/17 watt cpu comparisons is a must. This reference design came in a 14 inch form factor and so it makes sense to compare to a similar form factor and price.
 
I once had a Brazos based laptop (e-350) and found it to be intolerable compared to even a Pentium M from 2005. I have yet to look at other reviews but I am wondering if Kabini has a single or dual channel memory controller. Bobcat (Brazos) had only a single and if Kabini is the same then I won't bother as it won't cope with day to day tasks and very light gaming. I expected better results in gaming so there is some disappointment.
 


We can't test every laptop out there. The Pentium B960 and Core i3-3217U were an ideal best- to worst- case scenario, the whole spectrum of what the A4-5000 is going up against in the $350-$500 laptop market.

The A4-5000 does amazingly well against the Pentium, no doubts there. Our tests show that, given a choice, the highly-clocked Pentium is hardly a viable option. Regardless of power draw, people with $350 are going to have to make this choice. The Pentium is an obvious loser here... yes, there are lower powered pentiums, but performance goes down, not up. The A4-5000 is the obvious choice.

The Core i3-3217U, though, that might be the high end of the $400-$500 spectrum, but it shows the worst case scenario. It's a good indicator why, if you're considering an A4-5000 in a laptop, you would want to stay in the low end of the price range. Because if you can afford the Core i3, it's probably the better bet, a combination of low power use but still some more performance if needed, particularly when it comes to gaming.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom