The Brazos Performance Preview

[Hornet331]

yeah i stopped reading the anand review, it was bs. none of the cpus or tests made any sense. all the other reviews were great though! i thought the cpu would be a bit more powerful, but given its power draw...well you just cant beat it.

lol wut.. so anad reviews didn't make since for the last 3-4 years.
Hes using this benchmark setup for pretty much 3-4 years now, i like how he gets bashed more and more for doing the same thing bractically changing nothing with his benchmark setup. Plus most of the other sides pretty much use the same benchmarks and they show nearly the same results...

If you want to critizise a review then that of legit reviews, it has no consistency, they add and remove systems or do no compression at all.

Unfortunately, as much as people would like, you can't really compare scores across different sites (different methodologies for example) and assume that they'd get the same score for the same platform

You can if you concentrate on certain sub systems like the CPU, most scores of the CinebenchR10 benchmark are very consisten accros the review sites, same goes for the x264 benchmark. But if you start to compare the whole system I agree with you.

@ reviews
Amd really did some magic with the idle power consumptionn, its really really good. Compared to the Atom platform its needs ~70% less power.
Idel runtime swill be crazy for this platform, I hope the OEM don't use this fact and save on the capacity of the batteries.

The cpu part on the other hand is not that impressive, it needs 2-2,5x (~10W compared to ~4W on atom) more power then atom and delivering more performance in the range from 10-80%, avarage performance advantage is hard to tell but i guess its somewhere ~40%. But at least this really doen't matter that much because of the awesome idel power consumption the barazos platform has and overall load power consumption is in the same range as atom.

The IGP is very good for its intended market, thought the limited memory bandwidth eats more performance then I thought.

Overall the plattform is a real winner for the ultra mobile market.
Low cost entrie desktop and lowcost main stream books (15") i am not so sure if the performance is enough, since in this segement power consumption is still not the main selling point and there are already dirty cheap Pentiums and Celerons batteling (and even some i3s) the sub 500€ market.

Now what i really want to see is how the C-50 performs, cause that is the CPU that will go against the N550.

[duploxxx]

http://www.netbooknews.com/6742/dual...n-benchmarked/

In techreports review most of the scores between bobcat & N550 were pretty close. But keep in mind that bobcat was tested with SSD, so I would want to see how it performs with 5400 RPM HD and 1GB of memory (like in 1015PN)

N550 is lower speed then D510 right? 1,5 vs 1,66ghz at least according to official intel specs:
http://ark.intel.com/Compare.aspx?id...54,49669,49491,

but yet if you look at PCperspective review the D510 is kicked very badly on any part against brazos, with same amount of ram and same ssd disk. So how can a N550 then be even close to the one reviewed????

http://www.pcper.com/article.php?aid=1039&type=expert

so your source has 0 added value to this compare.

[hurleybird]

Seems a bit slower than expected, but also way less power usage than expected. Even the "high end" zacate seems more of a netbook than notebook part. Seems to blow away everything in perf/watt.

Do want.

[-Boris-]

http://www.netbooknews.com/6742/dual...n-benchmarked/

In techreports review most of the scores between bobcat & N550 were pretty close. But keep in mind that bobcat was tested with SSD, so I would want to see how it performs with 5400 RPM HD and 1GB of memory (like in 1015PN)

Can you please stick to the reviews that actually compare power consumption and performance between the two?

In techreports review the difference is very big between the two.

Zacates performance compared to Atom 550 according to the Techreport:

Javascript: 175%
7Zip: 111/108%
Tryecrypt: 132%
x264: 133/118%
Call of Duty: 293/235%
Far Cry 2: 144%/230%


Never slower, and up to 3 times at fast at graphics intensive tasks, and that is when Atom has Ion. In CPU intensive tasks Zacate is 8-75% faster. In singlethreaded benches we see bigger differences than these multithreaded shown above.

When it comes to power consumption Zacate is overall lower than dual core Atoms.

And yes, the comparision isn't entirely fair, but it's safe to say that the performance difference is huge, and general usage with singlethreaded apps probably would increase the lead. Lower power consumption, especially in idle is also a big advantage.

You simply can't say that the performance is similar, or hint that Atom would have much lower power consumption. You don't have any solid base for these claims.

[generics_user]

looks like a real winner, and much better than i expected in power consumption (idle power draw is impressive)
it completely destroys ATOM in performance but i didn't expect it to do the same in power consumption!

PS: close up of that 10+hours MSI netbook:
MSI Netbook in Techreport review


i'm 100% sure that i'm going to get a netbook based on this platform next year as i don't need the power and screen of my E6410 all the time, maybe dell is going to release a latitude or vostro system based on this

[Lightman]

This will be great upgrade for my wife's Aspire One with Atom N270 and G965.
Really impressive perf/power/die area considering usual AMD offering in these markets.

Looking forward to proper nettops and laptops test soon!

[informal]

Was about time . Performance was expected but power draw numbers are amazing! With the L2 working at the half the CPU clock,having "lowly" 2 issue decoder/front end ,sharing the memory controller and many other power optimizing features,it's actually amazing to see this level of performance. I hope AMD manages their goal of close to 10 hours of resting battery life,there is still time to tweak the platform a bit.

[Dimitriman]

These will get SERIOUSLY interesting at 32nm.
Im waiting for an Alienware M11 based on Llano, 400+ shaders should wipe the floor with that wimpy nvidia 335M. Will do lots of shopping next year =D.

[eric66]

sometime i think that anandtech reviews are written by complete idiots who don't have slightest idea what they are talking about lol

[informal]

sometime i think that anandtech reviews are written by complete idiots who don't have slightest idea what they are talking about lol

Like testing netbook chip in heavily multithreaded applications?The ones that won't be used on this platform in real usage scenario?
In any case,Bobcat runs circles around Atom with lower power draw.He can use 3DS Max and Cinebench all they if we likes,it won't change this fact,even in those cases .

These will get SERIOUSLY interesting at 32nm.

AMD will use 28nm node for "Brazos II" ,codenamed Wichita and Krishna .These will use the "Enhanced Bobcat cores" and up the core count to 4. I suppose the GPU will be beefed up,to say 160SPs ? So,all in all,more cores,faster cores(both x86/GPU),less power.

[Lokinhow]

http://www.tomshardware.com/reviews/...ance,2790.html

[Dimitriman]

Like testing netbook chip in heavily multithreaded applications?The ones that won't be used on this platform in real usage scenario?
In any case,Bobcat runs circles around Atom with lower power draw.He can use 3DS Max and Cinebench all they if we likes,it won't change this fact,even in those cases .

AMD will use 28nm node for "Brazos II" ,codenamed Wichita and Krishna .These will use the "Enhanced Bobcat cores",and up the core count to 4.I suppose the GPU will be beefed up,to say 160SPs ? So,all in all,more cores,faster cores(both x86/GPU),less power.

by then it will be time to ask: can it play crysis?

[eric66]

by then it will be time to ask: can it play crysis?

seriously i am suprised that they didn't test crysis and call brazos dissapointment

[informal]

seriously i am suprised that they didn't test crysis and call brazos dissapointment

Ohh no man,don't give them ideas

[generics_user]

These will get SERIOUSLY interesting at 32nm.
Im waiting for an Alienware M11 based on Llano, 400+ shaders should wipe the floor with that wimpy nvidia 335M. Will do lots of shopping next year =D.

Bobcat is going to switch to 28nm Globalfoundries or TSMC Bulk process next year with 4 cores, ipc improvements and a more powerful igp + higher DDR3 speeds, with a refresh of the APU platforms every year they should be able to put some real preassure on intel in this market

[Kai Robinson]

From the tests, it looks like the nano dual core beats it in every way, except graphics...

[Dimitriman]

Bobcat is going to switch to 28nm Globalfoundries or TSMC Bulk process next year with 4 cores, ipc improvements and a more powerful igp + higher DDR3 speeds, with a refresh of the APU platforms every year they should be able to put some real preassure on intel in this market

ye so I just learned I thought they would use the same process as bulldozer but 28nm should alow for some serious improvement indeed. Damn 2011 will bring somewhat of a revolution on netbooks, soon well be talking about gaming performance on these platforms and thats awesome.

[flyck]

From the tests, it looks like the nano dual core beats it in every way, except graphics...

and power/die size... which is much lower for Zacate.

[generics_user]

From the tests, it looks like the nano dual core beats it in every way, except graphics...

nano is a power hog, idle power usage of nano is very high, nano is good for nettops but i just don't see this thing in a a netbook because it draws way too much power in its current state, additionally it is a 3 chip design with a huge formfactor on both the NB and SB (real estate of the VN10000 northbridge alone is bigger than Zacate + hudson M1)

[Kai Robinson]

The current Nano *is* a 65nm design - you can't blame them for process size. TSMC/GloFo have their hands full as it is without VIA trying to get on the 45/40nm boat...

[Boissez]

Was about time . Performance was expected but power draw numbers are amazing! With the L2 working at the half the CPU clock,having "lowly" 2 issue decoder/front end ,sharing the memory controller and many other power optimizing features,it's actually amazing to see this level of performance. I hope AMD manages their goal of close to 10 hours of resting battery life,there is still time to tweak the platform a bit.

Indeed - I initially thought performance woud be higher but considering the level of neutering that has been applied to keep power low it's still amazing performance.

Now I'm just waiting for Sony to put one inside one of these:

[LightSpeed]

Brilliant, seriously. Also, the huge single threaded advantage over Atom counts a lot. When a program is multithreaded, it doesnt feel that sluggish on an Atom, but single threaded programs feel SLOW. Thats where the performance difference will be most felt.

Great job AMD

[[XC] Synthetickiller]

So why do all the reviews, besides hothardware, compare this to the atom D510? I was hoping to see a D525 w/o ion (at least hothardware compared the ion2 D525 against this) comparison.

I only skimmed the articles, but I didn't see if AMD will release mini-itx type baords to compete w/ ATOM/NANO. I guess that'll happen later?

[marten_larsson]

The current Nano *is* a 65nm design - you can't blame them for process size. TSMC/GloFo have their hands full as it is without VIA trying to get on the 45/40nm boat...

The Nano isn't that amazing when you look at the total system power or silicon to get there. The APU by AMD is 74mm² in 40nm which include NB and GPU. The VIA platform has 136mm² CPU and a 100mm² GPU/NB. If it were on 40nm process with perfect scaling these combined would be 89mm² (236 x (40/65)² or 51 + 38mm²). And that is assuming both the CPU and GPU/NB gets to shrink. It is however somewhat competitive and will offer an alternative which is good IMO.

[kl0012]

N550 is lower speed then D510 right? 1,5 vs 1,66ghz at least according to official intel specs:
http://ark.intel.com/Compare.aspx?id...54,49669,49491,

but yet if you look at PCperspective review the D510 is kicked very badly on any part against brazos, with same amount of ram and same ssd disk. So how can a N550 then be even close to the one reviewed????

http://www.pcper.com/article.php?aid=1039&type=expert

so your source has 0 added value to this compare.

When I said close to N550, I meant that the perf of bobcat still in the Atom category. It is barely faster then lowest end C2D 1.2GHz celeron, so I don't really understand what AMD meant when said "90% of today's mainstream performance" 1.4GHz DC celeron or 1.6 GHz Athlon is not a "mainstream performance" at least for 4 last years already.
Any way my point was that in all reviews pwr consumption was compared to desktop systems (Atom D-series doesn't support SpeedStep for example). So it worth wait and see an actual netbooks on bobcat before draw a conclusion. 9W-11W at idle seems to me a bit too high for tablets. My Asus T101 (N450) consumes ~4.5W from the wall socket at idle.