The AMD Kabini Review: A4-5000 APU Tested

[Abwx]

No they won't cost that much. Do you know what actual processor costs are for the various OEMs? Or are you just assuming costs based on Intel's recommended customer price? Because, ya know, the i7 3770k's recommended customer price is $342 and yet Microcenter sells them for $229.

We re talking of mobile , where intel does most of its milk
in the consumer market , besides there will be only 4C
haswell for many months , i dont expect them to be
below 200$ , that s 60% of the laptops/netbooks retail
price targeted by Kabini.

 

[Sweepr]

Just add that $100 difference for an SSD and the Kabini laptop will be a better exipirence than the core i3. Slimmer design, longer battery life, lighter and with an SSD it would be the better choice.

Yeah right, why bigger laptops exist? Just add an SSD and the ultrabook will be a much better choice than a 4/8T Ivy Bridge... not really (at least not for everyone). An ULV i3 IB has a large performance advantage in MT tasks and a huge advantage in ST. Slimmer and lighter? Maybe, but at ~1,1-1,4kg the cheaper ULV notebooks are not what you'd call heavy/bulky. Price should favour Kabini a lot.. but that also depends on OEMs. $100 less than IB/Trinity/Richland would be worth it, <$100 cheaper doesnt justify the performance deficit.

 

[wlee15]

No they won't cost that much. Do you know what actual processor costs are for the various OEMs? Or are you just assuming costs based on Intel's recommended customer price? Because, ya know, the i7 3770k's recommended customer price is $342 and yet Microcenter sells them for $229.

There's an reason that half of the 17W Ivy Bridge cpu have a tray price of $225. Intel doesn't want an OEM to make a laptop with a 13" IPS touchscreen and a SSD with an i3 processor. Intel did this to Samsung and Leonovo when they built a 12.1" notebook with an Atom processor. So now we an opportunity to get high quality components down into lower price points and change up the standard features in lower end of notebooks.

 

[Khato]

There's an reason that half of the 17W Ivy Bridge cpu have a tray price of $225. Intel doesn't want an OEM to make a laptop with a 13" IPS touchscreen and a SSD with an i3 processor. Intel did this to Samsung and Leonovo when they built a 12.1" notebook with an Atom processor. So now we an opportunity to get high quality components down into lower price points and change up the standard features in lower end of notebooks.

What reason is that exactly? And why wouldn't Intel want an OEM to make such a laptop?

Intel's pricing has never been a barrier to making high quality 'inexpensive' laptops. You get to blame the OEMs for that. The only thing stopping them from making such a laptop based on a celeron is demand... well, that and the fact that OEMs can use the supposed 'cost' of the higher end Intel processors to increase their margins.

 

[Abwx]

TR review , curiously there s no power comsumption number.

http://techreport.com/review/24856/amd-a4-5000-kabini-apu-reviewed

 

[wlee15]

What reason is that exactly? And why wouldn't Intel want an OEM to make such a laptop?

Intel's pricing has never been a barrier to making high quality 'inexpensive' laptops. You get to blame the OEMs for that. The only thing stopping them from making such a laptop based on a celeron is demand... well, that and the fact that OEMs can use the supposed 'cost' of the higher end Intel processors to increase their margins.

Because every 1080p laptop out there has an i5 or i7 processor in it (excluding the occasional AMD one). If OEM can start reducing Intel's share of an laptop's component cost in one segment it'll start expecting it in another.

 

[monstercameron]

I wonder, currently intels i3s beat the a4-500 by ~40-150% but on can use up to 15W more...what if they were all operating in the same power envelope? what wuld the differential be then?

 

[Accord99]

I refuse to believe that the ULV Pentium is consuming more power under load than the 35W Celeron. That's ridiculous.

Those numbers are practically impossible for a laptop without a video card, in fact they're higher than a desktop system I have with a Asrock Z77 Extreme 3 and Pentium 2020 @ 2.9 GHz running Prime 95.

They're also double the maximum power consumption of the Intel NUC. The only explanation that I can think is that they accidentally multiplied the correct values by 3.

 

[Phynaz]

The shrinking market is the over 1000$ ultrabooks segment
as i pointed it a few posts earlier while thoses CPUs will end
in systems that cost 400-600$ , Kabini being devoted to the
250-400$ segment.

You are saying AMD is limiting themselves to a max price target of $600, what those of us in the IT business call craptops.

Do you have proof of any of this?

 

[frozentundra123456]

Price is key. The only currently available model is well over 400.00. There is no way I would pay that for that kind of performance. Perhaps after more models come out, the price will come down. It needs to get into the 350 to 400 dollar range to be an attractive buy.

 

[Abwx]

You are saying AMD is limiting themselves to a max price target of $600, what those of us in the IT business call craptops.

Do you have proof of any of this?

So expect thoses "craptops" to be 90% of the laptop/netbook
market in the the next times.

 

[SPBHM]

I wonder, currently intels i3s beat the a4-500 by ~40-150% but on can use up to 15W more...what if they were all operating in the same power envelope? what wuld the differential be then?

Kabini have a 25w model, considering it integrates more, it should be comparable to the 17w i3s/i5s in terms of power, BUT, performance is not that easy, it's a 500MHz increase, so it's going to look good in many MT tests, but ST still is a big factor for basic usage (web browsing..) I think... so focusing to much on MT benchmarks can be a little misleading, like Anand said: "In my own use, I can feel a performance difference between the 2020M and the A4-5000 in tasks like installing/launching applications, as well as bigger CPU bound activities." and it's easy to understand why looking at the ST benchmarks.


even so, Kabini clearly have the potential to be a problem for the lower cost Celerons, Pentiums and ULV i3s

to be honest I would be really curious to see how high AMD could go with Kabini and 35w... at the same clock perhaps ST performance would be lower than (2 module) Richland, but MT higher (for the CPU)!?

Anyway, I'm happy to see the pretty good power usage numbers and decent performance improvements...

as I said, I think this CPU will become popular to all sorts of low cost machines, desktops, 15" laptops (like this awful bobcat based thing), 11" laptops, tablets...
this is certainly a huge improvement over the current Atom tablets in terms of performance and a good improvement over the ULV i3/i5 tablets in terms of power usage and cost I would think...

 

[Phynaz]

So expect thoses "craptops" to be 90% of the laptop/netbook
market in the the next times.

When is "next times"? Pick a date and we will look at market share.

And also post proof of your other "facts" as requested.

 

[mavere]

I've seen plenty of places stating that Kabini will do great in the sub-$400 bottom-of-the-barrel notebook market, which strikes me as one hell of a backhanded compliment.

The last couple of years has been guided by the narrative that tablets and hybrids (mostly the former as far as sales go) are eating up netbooks' place in the mainstream. Now this new chip comes along and people go "yea good for you AMD, go take over netbooks!"

I think everybody want to say optimistic things because the performance numbers on the spreadsheets are optimistic. However, I don't think those numbers in the context of the bigger picture and, well, reality are as optimistic.

 

[fusion238]

TR review , curiously there s no power comsumption number.

http://techreport.com/review/24856/amd-a4-5000-kabini-apu-reviewed

That is very astonishing. The AMD A4-5000 even manages to outperform Ivy Bridge i3 and i5 in a number of app accelerations!

 

[strata8]

That is very astonishing. The AMD A4-5000 even manages to outperform Ivy Bridge i3 and i5 in a number of app accelerations!

I thought the gaming results were interesting:

 

[SomeGuyAtHome]

There is some TDP overlap between Kabini and Richland ULV. It'll be interesting to see a performance comparison between the 15W Kabini and 17/19W Richland parts.

Has pricing been released for these parts? I wonder how Kabini, Richland ULV and Ivy Bridge ULV compare.

 

[strata8]

There is some TDP overlap between Kabini and Richland ULV. It'll be interesting to see a performance comparison between the 15W Kabini and 17/19W Richland parts.

The 17W Trinity beats the 15W Kabini by 45% in single threaded, but Kabini has a 60% advantage in multi threaded.

Has pricing been released for these parts? I wonder how Kabini, Richland ULV and Ivy Bridge ULV compare.

$49-$72 for the embedded Kabini chips, but I'd expect the pricing of regular parts to be lower than that.

 

[SPBHM]

I thought the gaming results were interesting:

look at the i5 vs i3... this is ridiculous, because it's not a CPU test, but a notebook test (and they only use the CPU model to identify, totally misleading), the i5 is using dual channel 1600 the i3 single 1333 (both support dual 1600)... apart from that both should perform really close (oh well, maybe the i3 notebook is having some thermal throttling issues lol), considering it's the same IGP with the same clocks, and when all threads are loaded the clock should be the same...

if the 128GCN sps GPU from Kabini supported dual channel it would probably be very competitive against ulv HD 4000, but the lower CPU performance would always limit a lot anyway.

 

[386DX]

look at the i5 vs i3... this is ridiculous, because it's not a CPU test, but a notebook test (and they only use the CPU model to identify, totally misleading), the i5 is using dual channel 1600 the i3 single 1333 (both support dual 1600)... apart from that both should perform really close (oh well, maybe the i3 notebook is having some thermal throttling issues lol), considering it's the same IGP with the same clocks, and when all threads are loaded the clock should be the same...

It's also misleading because TR uses FRAPS to record there framerate and frame latency. Most games are written for 2 or more cores... you add FRAPS ontop of it, the CPU's that don't have enough threads (ie. the i3 and E-350) is going to take a bigger performance hit as FRAPS will be bouncing between the core that's being least utilize while on the i5 and Kabini with 4 threads this is less likely to happen, that's probably why you see the huge latency difference between the 4 threads and 2 thread CPU's which isn't going to show up if you are playing the game without FRAPS running.

 

[wlee15]

It's also misleading because TR uses FRAPS to record there framerate and frame latency. Most games are written for 2 or more cores... you add FRAPS ontop of it, the CPU's that don't have enough threads (ie. the i3 and E-350) is going to take a bigger performance hit as FRAPS will be bouncing between the core that's being least utilize while on the i5 and Kabini with 4 threads this is less likely to happen, that's probably why you see the huge latency difference between the 4 threads and 2 thread CPU's which isn't going to show up if you are playing the game without FRAPS running.

Core i3 have hyper-threading.

 

[AtenRa]

The moral of the story(TR review), stay away from single channel Core i3 Ultrabooks.

 

[ShintaiDK]

The moral of the story(TR review), stay away from single channel Core i3 Ultrabooks.

I think we can extend that to all dualchannel solutions, since its done on both CPU vendors.

 

[Abwx]

And also post proof of your other "facts" as requested.

It can be found in a PDF that was released at time
of bobcat launch with estimations of segments future evolution.

The paper clearly show the 200-500$ segment as being
the faster growing and targeted one.

http://www.amd.com/us/press-releases/Pages/amd-fin-analyst-day-2010nov09.aspx

 

[ShintaiDK]

That chart is only 2½ year old