Was brazos a bad product?

[mrmt]

It was a bad product in the sense that it found its way into 15" class notebooks that sat on the store shelf next to notebooks with Intel Core chips, and the discount was not enough to justify the atom-class performance. People who bought these things expecting reasonable performance were basically screwed, and they were sure to avoid AMD products in the future. Brazos was a bridge burner.

If Brazos was an Intel or Nvidia product, meaning that it would have company's support with market, usage guidelines, the whole smash, then it would have been a much more successful product.

But it was an AMD product, and that means an screwed relationship with OEMs, and Brazos spent most of its useful life in the beginning of the Bulldozer crash, when AMD sales were skidding and the company was in the red. With this, AMD choice was really simple: Selling an inadequate processor for 15" notebooks or send the OEMs to Intel's arms. Since the big core line was FUBAR'ed anyway, they burned Brazos too.

 

[Shivansps]

It was a bad product in the sense that it found its way into 15" class notebooks that sat on the store shelf next to notebooks with Intel Core chips, and the discount was not enough to justify the atom-class performance. People who bought these things expecting reasonable performance were basically screwed, and they were sure to avoid AMD products in the future. Brazos was a bridge burner. So no, there is no way it was a good product. If they kept it strictly restricted to netbooks then it wouldnt have been so bad. But they had no way of doing that.

That's the problem through, im not sure if Brazos was ever intended to be on netbooks or ultraportables, im not really sure for what was intended, it looks like a proof of concept of Fusion that they decide to produce and sell.

If you look at the APU itselft, it has a x4 PCI-E link intended for a dgpu, another x4 pci-e for chipset link, the chipset itselft has 4 sata, 14 USB and 4 x1 links, way too much for netbooks and ultraportables.
And yet it lacked a critical thing like DC, and they still lack.

Its also hard to explain why it performed less than Athlon Neos considering both are K8.

As for OEM, again they where given no choice from AMD by not making a new chipset for the Neo 2, Athlon 2 and Phenom 2 for their mobile lineup, Brazos had more features and it was cheaper, its no wonder why OEM decided to go Brazos way, it was better for them to use Brazos on AMD models and just forget about everything else AMD had, that also allowed Intel to gain too much terrain by the time Llano launched.

So before we blame a OEM here think about what other option from AMD it had.

 

[NTMBK]

Its also hard to explain why it performed less than Athlon Neos considering both are K8.

Bobcat is not K8. It's a totally separate design.

 

[Dribble]

There was a market for netbooks. They were selling well with Intel atom's and whatever that Nvidia gpu chipset was. Both companies milked it until the market rapidly died with the advent of tablets, both companies rapidly moved onto those new markets.

AMD manages to miss the whole netbook boom, and finally releases something just as it was dying out. No Brazos had no competition in netbooks, but that's because Intel was focusing on tablets - ms surface was their aim because that's where the future was. AMD meanwhile had a great netbook chip only no one was buying netbooks.

Brazos was not suitable for tablets. Hence in an attempt to find something to do with Brazos it started ending up in full size laptops, where it was rubbish. This made AMD look really bad and I am sure put plenty of people off using AMD again. This is not OEM's fault, this is AMD's - they were trying to sell a chip that didn't fit into any of the markets at the time. It's classic AMD, they don't see future markets. Once a market's established only then do they re-actively do something but by then it's too late and everyone has moved on, and AMD is left out in the cold again.

 

[Shivansps]

Bobcat is not K8. It's a totally separate design.

They designed something new that performed less than K8 to launch in 2011? how was that possible? i always belived it was K8 based.

 

[Shivansps]

There was a market for netbooks. They were selling well with Intel atom's and whatever that Nvidia gpu chipset was. Both companies milked it until the market rapidly died with the advent of tablets, both companies rapidly moved onto those new markets.

AMD manages to miss the whole netbook boom, and finally releases something just as it was dying out. No Brazos had no competition in netbooks, but that's because Intel was focusing on tablets - ms surface was their aim because that's where the future was. AMD meanwhile had a great netbook chip only no one was buying netbooks.

Brazos was not suitable for tablets. Hence in an attempt to find something to do with Brazos it started ending up in full size laptops, where it was rubbish. This made AMD look really bad and I am sure put plenty of people off using AMD again. This is not OEM's fault, this is AMD's - they were trying to sell a chip that didn't fit into any of the markets at the time. It's classic AMD, they don't see future markets. Once a market's established only then do they re-actively do something but by then it's too late and everyone has moved on, and AMD is left out in the cold again.

AMD did not missed the netbook boom, they had the Neos, but platform TDP was way too high for anything below 11.6"
Brazos also had huge tdp as well, thats why they created the C series, but C series where just marginally better than Atoms, and some cases like the C-30 where just awfull.

They just did a bad decision when they choose not to make it SoC and not to make a new chipset for their other mobile cpus that ended killing the entire AMD mobile lineup except for brazos.

 

[NTMBK]

They designed something new that performed less than K8 to launch in 2011? how was that possible? i always belived it was K8 based.

It was a much smaller, low-cost design, built from the ground up to be low-power and cheap. Same as how Jaguar is lower performance than K10.

 

[Insert_Nickname]

also not forget the ram
some cheap solutions were very limited in ram that also hurt performance

very bad choices from the oems really damage a product
too low ram
bad hd

i bought a laptop with an i3 which come with 1 ram module installed bought a kit and went from single channel to double channel. and the performance of the igp doubled at 3d
oems are stupid they cut 10$ and that ruins the product

OEMs really like their margins. If they can save $10, they'll do it rather then pass it on to customers. Who (mostly, except for the few in the know) in turn are also cheap. If they can save $10 they'll also do it.

That is, sad to say, the state of the mass-market.

Edit; autocorrect was apparently a little too quick yesterday...

 

[ALIVE]

OEMs really like their margins. If they can save $10, they'll do it rather then pass it on to customers. Who (mostly, except for the few in the know) intern are also cheap. If they can save $10 they'll also do it.

That is, sad to say, is the state of the mass-market.

the fun is the consumer pay the cheapest they can
and then the curse the company

both oems and consumers go cheap
the problem is that the consumers are pissed and change boat
so consumers can be cheap oems dig their grave

 

[poofyhairguy]

Brazos is the last AMD CPU I have owned, so no they weren't bad products.

 

[coffeemonster]

It was a much smaller, low-cost design, built from the ground up to be low-power and cheap. Same as how Jaguar is lower performance than K10.

even if both were clocked ~2Ghz? I thought Jaguar was pretty strong in that range compared with other AMD cpus

 

[ALIVE]

even if both were clocked ~2Ghz? I thought Jaguar was pretty strong in that range compared with other AMD cpus

i think the real problem was not the first bravos 350
the problem was what followed after that

amp tried to lower tdp to less and less but the cost was that the performance of the new chips was tiny better. great engineering feet but the problem was the world was moving the software was getting heavier the videos also the web pages with the flash.

if they tried to keep the tdp the same and focus raising the performance primary and secondary the power consumption then now we would have a different talk. so when e450 was out i was disappointed because it offered nothing, then the c series which was how to destroy a company name too weak to be used for windows but yet they found their way there

next generation e1000 e2000 stayed only in mobile search for desktop but could not find a product

and why not they did not make for desktop a cpu with 4 cores and the same igp with highter tdp but still much much lower than the athlonII phenomII that were sold that era or even make a 4 core with no igp.

then intel regrouped and made some fantastic chips in the e3 xeon series the L had chips 25 watt that creamed any small core architecture, also the i3 or pentium in light scenarios were consuming comparable power to a brazos but have the possibility to go higher if needed.

amp focused on tdp and lost
low tdp is a nice feature but in the end of the day its all about perfomance

 

[mindbomb]

I have a brazos. The chip is alright. The big issues were the accompanying components that came with it in my HP netbook. The wireless card was slow, even for it's time (1x1 wireless n 2.4ghz). SSD would have been nice, but to be fair, it wasn't mainstream at the time. Most egregious is that the cooler somehow struggles with the 20w tdp. Also, 1080p 60 fps video has since become popular format on youtube, and it can't play it.

 

[ALIVE]

also the portion of the igp on the chip was too unbalanced focusing merely in the igp rather than the cpu. in a product that was sold in single channel memory configuration and thus limiting the igp only from that

amp simple did not manage to get the balance cpu gpu power in their chips

i know have an hp laptop with an i3 that was sold with a single channel memory configuration thus halving the igp in 3d. i upgrade it to double channel and the test i run before and after show almost double performance

so yes put a monster igp on the chip and then starve it in memory bandwidth

 

[Shivansps]

also the portion of the igp on the chip was too unbalanced focusing merely in the igp rather than the cpu. in a product that was sold in single channel memory configuration and thus limiting the igp only from that

amp simple did not manage to get the balance cpu gpu power in their chips

i know have an hp laptop with an i3 that was sold with a single channel memory configuration thus halving the igp in 3d. i upgrade it to double channel and the test i run before and after show almost double performance

so yes put a monster igp on the chip and then starve it in memory bandwidth

And dont forget to add a x4 PCI-E for dgpus when the cpu perf is awfull low.

Personally ill never undestand what AMD was trying to do with Brazos, i do belive Neo/Neo 2 was a better product, and today they hold their own better than brazos, no idea of why AMD never trought about making Brazos a SoC, and develop a new chipset for Neos and his other mobile lineup, that whould have make a lot more sence, hell AMD released a refresh to Neo 2 at the same time they launched E-350, what to me makes no sence.

 

[mrmt]

And dont forget to add a x4 PCI-E for dgpus when the cpu perf is awfull low.

Personally ill never undestand what AMD was trying to do with Brazos, i do belive Neo/Neo 2 was a better product, and today they hold their own better than brazos, no idea of why AMD never trought about making Brazos a SoC, and develop a new chipset for Neos and his other mobile lineup, that whould have make a lot more sence, hell AMD released a refresh to Neo 2 at the same time they launched E-350, what to me makes no sence.

AMD was pursuing lower power consumption, lower platform costs and lower design and manufacturing costs. Brazos was better business for AMD and partners than Neo.

 

[Shivansps]

AMD was pursuing lower power consumption, lower platform costs and lower design and manufacturing costs. Brazos was better business for AMD and partners than Neo.

But a the same time they decided things like putting in a big IGP and them starve it(and they still do!), and the dedicated pci-e lanes intended for dgpus, the big chipset that supported 14 USB, 6 SATA and another 4 PCI-E

Not sure if OEM ever figure it out that was better to just use the pci-e of the APU for devices instead of the chipset ones.
Anyway, its like they where not sure of what the intended market was(Desktop, AIO, Netbook, Notebook? lets do it all), thats why it looks more like a concept prototype that they decided to sell.

And personally, i think AMD let all his other mobiles platforms die, in what i suspect it was an attempt to sell this, i cant blame OEMs for placing Brazos where i did not fit, there just was no other option. Its a shame the nvidia decided to burn its chipset division, maybe they could have come out with something viable to use along other AMD Mobile cpus.

 

[mmntech]

My HTPC has the E-350 in it. Plays everything I thow at it flawlessly. Like all low power CPUs of the era, it wasn't exactly a champ at desktop performance. That GPU made a big difference though for media playback. Only issue I have with them is they do seem to run hot.

 

[ALIVE]

But a the same time they decided things like putting in a big IGP and them starve it(and they still do!), and the dedicated pci-e lanes intended for dgpus, the big chipset that supported 14 USB, 6 SATA and another 4 PCI-E

Not sure if OEM ever figure it out that was better to just use the pci-e of the APU for devices instead of the chipset ones.
Anyway, its like they where not sure of what the intended market was(Desktop, AIO, Netbook, Notebook? lets do it all), thats why it looks more like a concept prototype that they decided to sell.

And personally, i think AMD let all his other mobiles platforms die, in what i suspect it was an attempt to sell this, i cant blame OEMs for placing Brazos where i did not fit, there just was no other option. Its a shame the nvidia decided to burn its chipset division, maybe they could have come out with something viable to use along other AMD Mobile cpus.

thats the problem when you have managers to make the decision
they have no idea of the product they are selling the look in numbers features and numbers that really does not make sense.

but brazos could fit in a market the original e350 to make a computer as media center. where atom failed because it could not play the videos. so the product as it was delivered it had a market and was okey.

the problem is what followed of the e350
e450 was a good improvement but the gains in cpu power was so little that did not make the need to upgrade from the e350 and many after that

for desktop the cooling solution was not good enough to provide a cool experience thank you oems for cheaping out there.

more or less amp copy intel atom in features but was able to put a better igp in the chip.

they had no vision of the future we made a good product we sold like crazy then what?????

igp must be good enough to do some basic stuff play the media files be able to surf the net.
for the rest the people will go to a dgpu

insted of all that usb if the put more sata
thay could sell it as nas solution for home and small business
maybe even have a more expensive model with ecc

what amp did not done is intel doing now with i3 supporting ecc from sandy bridge pair an i3 with a professional chip-set and you have a low cost ecc platform

the 4 pci lanes could help to put a raid card for that direction
also if they focus later to improve the cpu then it would even make sense to use a dgpu to get a bit more graphic power

definitely they should have a double channel memory even as a more
option

but amp focus in more gpu
as it was focusing in more cores
they focus in the complete wrong aspect

 

[NTMBK]

the problem is what followed of the e350
e450 was a good improvement but the gains in cpu power was so little that did not make the need to upgrade from the e350 and many after that

The problem is that AMD were crippled by their foundry partner, GlobalFoundries. AMD had plans to make a four-core Bobcat shrink on 28nm:

http://www.techpowerup.com/150691/amds-next-generation-wichita-and-krishna-apus-detailed.html

This was going to be a single-chip SoC, with the southbridge integrated into the die- like Kabini, but with Bobcat cores and coming out in early 2012. But their foundry partner had problems with their 28nm manufacturing process, and could not deliver in time to bring their chips to the market: https://semiaccurate.com/2011/11/15/exclusive-amd-kills-wichita-and-krishna/

This left AMD only able to sell their old Brazos chips, while they got Kabini ready to ship on TSMC's 28nm process.

 

[ALIVE]

The problem is that AMD were crippled by their foundry partner, GlobalFoundries. AMD had plans to make a four-core Bobcat shrink on 28nm:

http://www.techpowerup.com/150691/amds-next-generation-wichita-and-krishna-apus-detailed.html

This was going to be a single-chip SoC, with the southbridge integrated into the die- like Kabini, but with Bobcat cores and coming out in early 2012. But their foundry partner had problems with their 28nm manufacturing process, and could not deliver in time to bring their chips to the market: https://semiaccurate.com/2011/11/15/exclusive-amd-kills-wichita-and-krishna/

This left AMD only able to sell their old Brazos chips, while they got Kabini ready to ship on TSMC's 28nm process.

well glofo what problems all the time when was the last time that glofo kept a schedule? or nto have problems in production?

amp tried to play aggressive where they could not
they should design their cpus with the process glofo could offer them and when glofo got a new one make a die shrink

instead they design something that they could not build
a cpu is good if you launch it now but bad if you launch it 6 months later.

you need to refresh your line as fast as the competition
especially if you are behind

 

[NTMBK]

well glofo what problems all the time when was the last time that glofo kept a schedule? or nto have problems in production?

amp tried to play aggressive where they could not
they should design their cpus with the process glofo could offer them and when glofo got a new one make a die shrink

instead they design something that they could not build
a cpu is good if you launch it now but bad if you launch it 6 months later.

you need to refresh your line as fast as the competition
especially if you are behind

Thing is, GloFo didn't have anything competitive to offer. 32nm was a high cost, high performance process (SOI) designed for high end CPUs. It wasn't suitable for cheap chips like Brazos.

AMD would have been better off trying to ship Wichita on TSMC 28nm from day one, instead of messing around with GloFo, but they were contractually obliged to try by the WSA.

 

[Insert_Nickname]

also the portion of the igp on the chip was too unbalanced focusing merely in the igp rather than the cpu. in a product that was sold in single channel memory configuration and thus limiting the igp only from that

amp simple did not manage to get the balance cpu gpu power in their chips

i know have an hp laptop with an i3 that was sold with a single channel memory configuration thus halving the igp in 3d. i upgrade it to double channel and the test i run before and after show almost double performance

so yes put a monster igp on the chip and then starve it in memory bandwidth

I remember someone doing a test comparing performance between the Brazos (E350) IGP and a comparable discrete GPU (HD5450). Which meant the GPU had its own dedicated memory bus. The GPU performance was almost identical. So I don't think the single channel bandwidth limitation of Brazos to be that bad in practice.

 

[ALIVE]

I remember someone doing a test comparing performance between the Brazos (E350) IGP and a comparable discrete GPU (HD5450). Which meant the GPU had its own dedicated memory bus. The GPU performance was almost identical. So I don't think the single channel bandwidth limitation of Brazos to be that bad in practice.

yes but which 5450 they used
5450 made a range of cards
form ddr3 64bit cards with 256 mb
to gddr5 128bit 1 gbyte ram

so which one was use in the test??

while single channel is okey for 2d for 3d you will see a difference going double channel also cpu will get a small boost going double channel.
which will help to get better frame rates in the end

the problem is that gpu of today are designed to use a high bandwidth ram and thus they need the fast ram. if they could have design it different well thas another topic but as for now we need the speed

so going dd3 single channel will have an impact on the performance

 

[ALIVE]

Thing is, GloFo didn't have anything competitive to offer. 32nm was a high cost, high performance process (SOI) designed for high end CPUs. It wasn't suitable for cheap chips like Brazos.

AMD would have been better off trying to ship Wichita on TSMC 28nm from day one, instead of messing around with GloFo, but they were contractually obliged to try by the WSA.

well so options
1. build something a bit more expensive but have a product in the market
2. wait and in the end cancel chips because of the delay and if launched they are too late

we are not talking about an ideal world but about the reality
you know you have to do the best in a situation which all your options can be consider bad

also remember that brazos was very tiny so the more expensive process would not translate that much per chip

in the end you make money by selling chips not waiting to build them

making even half the profit in 32nm would be better than waiting

company value is increased as the consumer base is increased and a person that bought amd and was happy has more chances to buy amp also in the future so that means having a product and selling it not waiting

and as you said because of glofo they canceled many products
if that products made the market then today situation for amp could have been different

he could compete intel in an area they are loosing billions with atom
so how many years do you thing intel can keep up?? loosing? and destroying the name intel and atom at the same time?

remember the first introduction of the celeron how bad they performed? in the end you were hearing celeron and you were running away