AMD Q4/2013 Desktop Roadmap

[AtenRa]

The Kabini is a SOC, the Haswell Celerons/Pentiums will also have a chipset in the motherboard that will add to the overall power consumption of the system.

 

[Sweepr]

Actual power comsumption difference will be minor, totally worth the extra performance for an entry-level desktop user. Not that you care about power comsumption, you already made that clear in 220W Vishera vs 84W Haswell comparisons.

Oh and let me use one of your own quotes here:

You know most of the time the system is Idling and/or working at lower power states. At those workloads the power consumption difference between AMD and Intel is a few Watts(depending on the system and workload) .

 

[ShintaiDK]

LOL!

 

[AtenRa]

Actual power comsumption difference will be minor, totally worth the extra performance for an entry-level desktop user. Not that you care about power comsumption, you already made that clear in 220W Vishera vs 84W Haswell comparisons.

Oh and let me use your own quote here:

Even in Idle the chipset will use power adding to the overall system. Why are you trying to be a smartass troll in every AMD related topic ?? This is a Low Power product, it is only natural to care about power consumption more than High-End Gaming desktops.

 

[Arachnotronic]

Even in Idle the chipset will use power adding to the overall system. Why are you trying to be a smartass troll in every AMD related topic ?? This is a Low Power product, it is only natural to care about power consumption more than High-End Gaming desktops.

Bwahahaha, oh boy...if power consumption matters then Kabini/Temash are DOA. Oh, wait...they're so DOA that AMD is now pumping Beema and Mullins ;-)

 

[NTMBK]

Bwahahaha, oh boy...if power consumption matters then Kabini/Temash are DOA. Oh, wait...they're so DOA that AMD is now pumping Beema and Mullins ;-)

They're bringing out a refresh a year later... if I remember correctly, quite a lot of companies do this. Including one starting with "I"... can't remember quite which one

 

[mrmt]

Bwahahaha, oh boy...if power consumption matters then Kabini/Temash are DOA. Oh, wait...they're so DOA that AMD is now pumping Beema and Mullins ;-)

While Kabini/Temash actually failed to open new markets for AMD, there is nothing wrong with the yearly release cadence. Brazos was a heavily synthesized design and one of the benefits here is the smaller TTM.

Since its conception it was supposed to have a yearly refresh cadence, and AMD would be able to sustain it if the subpar foundry of them had a 28nm process ready on time.

 

[AtenRa]

Bwahahaha, oh boy...if power consumption matters then Kabini/Temash are DOA. Oh, wait...they're so DOA that AMD is now pumping Beema and Mullins ;-)

Are you trying to be a smartass too ?? So BayTrail is DOA because Intel will bring a new ATOM earlier than 12 months of BT release ?? IS THAT WHAT YOU ARE SAYING ???

 

[Shivansps]

come on stop it already, i welcome to have cheap options on desktop, i have at least 3 old pc that are on dire need of a cheap reemplacement, and if AMD has something better than the today Sempron 140+780G AM3+ MB its welcome, right now the other options im considering where the G1610+H61 and a BT based mb...

So lets wait and see the final price of the entire platform...

 

[Homeles]

Are you trying to be a smartass too ?? So BayTrail is DOA because Intel will bring a new ATOM earlier than 12 months of BT release ?? IS THAT WHAT YOU ARE SAYING ???

Well that's a false analogy if I ever saw one. How many marketing slides have you seen about Airmont? Compare that to the number of slides on Beema/Mullins that have been circulated.

I rest my case.

 

[mrmt]

Well that's a false analogy if I ever saw one. How many marketing slides have you seen about Airmont? Compare that to the number of slides on Beema/Mullins that have been circulated.

I rest my case.

This is because both Silvermont and Airmont are aimed to be strong contenders, products that Intel wants to spearhead in new markets and they are far better than anything Intel had before, both in terms of raw performance and comparatively with competitor's products. Intel has every incentive to tell world + dog how good Silvermont and Airmont are.

Beema and Mulins are just architectural refreshes on a node that will be fairly old stuff when they arrive. It's AMD's move to offer competition in very bad terms instead of uterly disastrous as if they didn't do the architectural refresh. If these chips won't open new markets and won't put AMD in a better competitive position, why bother advertising it?

 

[Homeles]

This is because both Silvermont and Airmont are aimed to be strong contenders, products that Intel wants to spearhead in new markets and they are far better than anything Intel had before, both in terms of raw performance and comparatively with competitor's products. Intel has every incentive to tell world + dog how good Silvermont and Airmont are.

Beema and Mulins are just architectural refreshes on a node that will be fairly old stuff when they arrive. It's AMD's move to offer competition in very bad terms instead of uterly disastrous as if they didn't do the architectural refresh. If these chips won't open new markets and won't put AMD in a better competitive position, why bother advertising it?

You missed my point. To my knowledge, there haven't been any publicly circulated marketing materials on Airmont. There has been, however, publicly circulated marketing material on Beema/Mullins. I.e., AMD is "pumping" Beema. Intel is not "pumping" Airmont, therefore AtenRa's analogy is completely false.

 

[mrmt]

You missed my point. To my knowledge, there haven't been any publicly circulated marketing materials on Airmont. There has been, however, publicly circulated marketing material on Beema/Mullins.

There was. Intel has been talking about 14nm Atom circa 2011, that on IDF, on EC Q&A, IDF and other events. In fact, Intel is taking a very heavy-handed approach to market Atom chips. What Intel didn't disclose is Airmont performance figures, something AMD already did. But for what reasons?

In a marketing presentation, why would you write a slide with the title "Why partner with AMD?"? You wouldn't, unless the target audience was "partners who don't want or don't seem to want to work with AMD anymore", so AMD has to cajole them back. But AMD's word shouldn't be enough anymore. Llano, Krishna/Wichita, Bulldozer, Steamroller, all this should have burnt their reputation to ashes in the channel, so they must back their claims with something more concrete, so we have two presentations. One public, showing that Beema and Mullins are descent contenders, and another one making the business case for AMD line up.

AMD is basically trying to tell partners that Beema and Mullins will be something they can sell and not choke on inventory like in Llano and Bulldozer's case. As the chip won't be the next big thing AMD discloses the chip and stays mum afterwards.

 

[AtenRa]

Well that's a false analogy if I ever saw one. How many marketing slides have you seen about Airmont? Compare that to the number of slides on Beema/Mullins that have been circulated.

I rest my case.

You shouldn't even open your case,

IDF 2013 September 2013,
Intel Officially talks about 14nm Broadwell only 3 months after Haswell released in June 2013. I guess Haswell is DOA as well then ???

Also,

Intel officially released a 14nm Airmont slide before even Baytrail was Officially released. Clearly by Intel17's and your logic BayTrail was DOA, actually BT was dead before even got released

 

[Homeles]

Not seeing performance numbers there, sorry. Your argument is invalid.

Come back when you understand what "pumping" a product means.

 

[AtenRa]

So by what you saying, when Intel let AT do a preview of SandyBridge half a year before the official release, that meant that Westmere was DOA ???

http://www.anandtech.com/show/3871/the-sandy-bridge-preview-three-wins-in-a-row

To get on topic again, like it or not, Kabini is a Low Power Low Cost APU. We care about power consumption in that segment. Yes Intel Haswell Pentium has more CPU Performance but it also has higher power consumption. Simple as that.

 

[ShintaiDK]

So by what you saying, when Intel let AT do a preview of SandyBridge half a year before the official release, that meant that Westmere was DOA ???

http://www.anandtech.com/show/3871/the-sandy-bridge-preview-three-wins-in-a-row

To get on topic again, like it or not, Kabini is a Low Power Low Cost APU. We care about power consumption in that segment. Yes Intel Haswell Pentium has more CPU Performance but it also has higher power consumption. Simple as that.

BT also got more CPU performance, at 2/5ths the TDP.

And Kabini is not cheap, the quadcores start at 70$.

 

[Homeles]

So by what you saying, when Intel let AT do a preview of SandyBridge half a year before the official release, that meant that Westmere was DOA ???

http://www.anandtech.com/show/3871/the-sandy-bridge-preview-three-wins-in-a-row

To get on topic again, like it or not, Kabini is a Low Power Low Cost APU. We care about power consumption in that segment. Yes Intel Haswell Pentium has more CPU Performance but it also has higher power consumption. Simple as that.

Okay, way to dodge the fact that you have absolutely no clue what "pumping" means.

 

[AtenRa]

Okay, way to dodge the fact that you have absolutely no clue what "pumping" means.

If you want to educate us, open a new topic about what "pumping" means. On this thread we are talking about AMD Desktop roadmaps.

 

[Homeles]

If you want to educate us, open a new topic about what "pumping" means. On this thread we are talking about AMD Desktop roadmaps.

If you had any confidence in your argument, you'd stick up for it, rather than running from it, no?

So by what you saying, when Intel let AT do a preview of SandyBridge half a year before the official release, that meant that Westmere was DOA ???

Sandy Bridge was a huge improvement over Westmere and Nehalem. So yeah, relative to Sandy Bridge, Westmere and Nehalem weren't worth purchasing. I do believe Intel's marketing paid off -- look how popular Sandy Bridge was. If I recall correctly, Intel had record profits after the Sandy Bridge launch, even with the chipset recall.

Compare this situation to Kabini. Kabini is yet another lackluster successor in a lackluster line of offerings from AMD. The only reason why it had any value was because of AMD's opportunism. This also applies to their Bobcat-based designs. AMD exploited a weakness in Intel's lineup -- a weakness that has been since addressed, leaving AMD in the dark. Now the tables have turned, except there's no chance for AMD to steal the spotlight again, with Intel now paying attention to the market segment. AMD won't even be able to hold onto their graphics lead for long, with the overhaul that Gen8 will bring, in addition to Intel's 14nm process.

Knowing how bad off they are, AMD has started "pumping" Kabini's successor up. It's damage control -- plain and simple. They've even had many of their engineers working on the project(s) picked off by Samsung. The Jaguar core will be known for little more than the CPU architecture powering the XB1 and PS4, which were products that most would argue that Intel didn't even have interest in. AMD will be completely shut out in markets where it actually has competition to face.

Trumpeting an upcoming product isn't an inherently bad thing, but if you're unable or unwilling to assess the history and market conditions surrounding a company's decision to cannibalize sales of current products in favor of upcoming ones, then you have no business participating in such a discussion. It's great an all that AMD was able to spot a weakness in Intel's offerings, but it's a shame that they weren't able to exploit it for very long.

 

[AtenRa]

Actual power comsumption difference will be minor,

A6-5000 15W TDP vs Pentium 2020M 35W TDP. Kabini has HALF the idle power consumption of 22nm Pentium 2020m. The power consumption difference in the rest of the applications is HALF as well.
To conclude, the Pentium is faster but it also uses 2x the power.

http://www.anandtech.com/show/6981/...ality-of-mainstream-pcs-with-its-latest-apu/2

 

[USER8000]

So,when AMD talks about future CPU models,it is all doom and gloom,but when Intel does the same it isn't. Wow,just wow.

 

[Homeles]

So,when AMD talks about future CPU models,it is all doom and gloom,but when Intel does the same it isn't. Wow,just wow.

If that's what you think I said, you should feel awfully embarrassed for yourself.

It's pretty sad that there can't be any intelligent discussion on this forum board. Every time someone writes something well thought out, inane garbage such as this gets posted.

 

[Sweepr]

BT also got more CPU performance, at 2/5ths the TDP.

And Kabini is not cheap, the quadcores start at 70$.

Precisely. Even if both (Haswell Celeron & Sempron/Athlon Kabini) ran @ full load 24/7 (most people won't) we're talking about minor additional cost per month. Certainly much lower difference than 220W Vishera vs 84W Haswell that double standard fanboys try to downplay.

Bottom line: if you want Kabini-like CPU performance and good enough GPU performance for basic office use & other light tasks Bay Trail-D gives you that at 2/5 the competition's TDP.
If you need to step up do real work on a budget system Haswell Celeron gives you more than twice Kabini's single-thread performance and easily beats it even in MT tasks (Haswell GT1 GPU should have no problem here too).

 

[Sweepr]

A6-5000 15W TDP vs Pentium 2020M 35W TDP. Kabini has HALF the idle power consumption of 22nm Pentium 2020m. The power consumption difference in the rest of the applications is HALF as well.
To conclude, the Pentium is faster but it also uses 2x the power.

http://www.anandtech.com/show/6981/...ality-of-mainstream-pcs-with-its-latest-apu/2

Unsurprisingly, an apples and oranges comparison - Mobile Ivy Bridge vs Mobile Kabini.
In the desktop many (probably most) people would give up a minor ''few watts'' difference for a noticeable performance jump. Also, theres a 30-35W gap between idle and load with Core i3 (HT enabled) Haswell's. Haswell Celerons running @ 20-30% lower clocks without HT (and GT1 instead of GT2 IGPs) should draw even less power, so dont try to make them look like power hogs, they are not.