Trinity Provides Up To 29% Faster Productivity, 56% Faster Visuals Than Llano: AMD

[btarunr]

A marketing slide by AMD for industry partners, which sums up what the company's 2012 Mainstream Platform led by "Trinity" APUs will offer, got leaked to the web. In it, AMD claims its next-generation APUs to offer up to 29 percent higher productivity performance (read: CPU performance), and up to 56 percent higher visual performance, compared to current-generation (Llano). At least the graphics performance figures seem to be consistent with early test results.

Apart from these, the slide claims Trinity to be optimized for Windows 8 (with AVX, AES-NI, SSE4.2, and DirectX 11.1 graphics, it could very well be). The processor is said to feature third-generation auto-overclocking technology, TurboCore 3.0. The mobile version of the chip will be designed to offer over 12 hours of resting battery-life. Lastly, there's mention of new media-acceleration features. AMD is expected to launch its new line of APUs in this quarter (before July).



View at TechPowerUp Main Site

 

[reverze]

no need to consider anything else anymore for office PCs

 

[robal]

Thanks for heads up !

Well done AMD. Now give me my Vishera

 

[NC37]

12hr battery life. Yikes! Even if it is under controlled environment, I'd like to get me some of that. Theres the battery life I want + the graphics performance I need. Now give me it in a Mac and I'll return to the platform . If not, i7 Qosmio ftw!

 

[phanbuey]

12hr battery life. Yikes! Even if it is under controlled environment, I'd like to get me some of that. Theres the battery life I want + the graphics performance I need. Now give me it in a Mac and I'll return to the platform . If not, i7 Qosmio ftw!

this

 

[HillBeast]

I'm not actually that impressed by these figures. The chip will be 28/22nm, so of course it's going to be faster. Where is the impressive part? In fact I'm actually ashamed of these numbers. With 28nm, these numbers should be more like '100% faster'. Fail.

 

[kasp1js]

I'm not actually that impressed by these figures. The chip will be 28/22nm, so of course it's going to be faster. Where is the impressive part? In fact I'm actually ashamed of these numbers. With 28nm, these numbers should be more like '100% faster'. Fail.

Trinity is still 32nm. And if the performance increase is this big in general usage not just in some specific tasks, then this will be the biggest jump in performance for the desktop CPU industry in quite a while.

 

[xenocide]

I'm not actually that impressed by these figures. The chip will be 28/22nm, so of course it's going to be faster. Where is the impressive part? In fact I'm actually ashamed of these numbers. With 28nm, these numbers should be more like '100% faster'. Fail.

It's 32nm still. They just took Llano and incrementally improved it. Swapped Stars (I believe) for Piledriver-based cores, and upgraded the integrated GPU. I expected more, but it is a decent upgrade.

 

[THE_EGG]

I'm not actually that impressed by these figures. The chip will be 28/22nm, so of course it's going to be faster. Where is the impressive part? In fact I'm actually ashamed of these numbers. With 28nm, these numbers should be more like '100% faster'. Fail.

just look at the gtx 580 transition to the 680. It certainly wasn't 100%, not even 50% faster.

 

[DarkOCean]

Being made on the same node 32nm as llano any kind of performance gain is good.

 

[TheoneandonlyMrK]

I'm not actually that impressed by these figures. The chip will be 28/22nm, so of course it's going to be faster. Where is the impressive part? In fact I'm actually ashamed of these numbers. With 28nm, these numbers should be more like '100% faster'. Fail.

even if it had a die shrink ,100% is in retard land of expectations, from any company, intel inclusive

 

[brandonwh64]

If it would perform well on BF3 at 720P then I would be all over this for a new laptop.

 

[reverze]

hoping to see these in ultrathins asap, since ultrabooks are way too expensive

 

[alienstorexxx]

just look at the gtx 580 transition to the 680. It certainly wasn't 100%, not even 50% faster.

and that is really a lot considering past generation steps.. the problem is that amd is really behind intel on cpu's so, that ~25% maybe isn't enough.

on igp's side, it pretty good, 50% more than something that was already fast competing with intel, is good news.

 

[Steevo]

No mention of clock speed to achieve the increase?

 

[Mussels]

my llano laptop already kicks ass, improvements to performance AND battery life at the same time? how the hell can anyone not love that?

 

[THE_EGG]

and that is really a lot considering past generation steps.. the problem is that amd is really behind intel on cpu's so, that ~25% maybe isn't enough.

on igp's side, it pretty good, 50% more than something that was already fast competing with intel, is good news.

agreed, an OCed 480 could almost compete with a 580 and so on. I usually wait for new architecture before upgrading even though it is on the risky side (achem, fermi). But yes i want more competition because usually that translates into lower prices for us

 

[Kärlekstrollet]

An overall improvement by ~42% with the same 32nm process is a pretty damn good job.

 

[sergionography]

just look at the gtx 580 transition to the 680. It certainly wasn't 100%, not even 50% faster.

and gtx680 was a new architecture + new process node
trinity uses the same 32nm process as llano so getting these numbers in the same power envelope is pretty impressive.
however we do have to mention that llano was a step down in cpu performance than previous laptop cpus due to the lower clock speeds to meet the tdp when integrated with graphics
so overall amd will need this much increase per generation if they plan to catch up with intel or atleast close the gap a bit, so if amd can do 30% now and 30% next gen with the 28nm node they will pretty much become on par, tho that is very unlikely

 

[ensabrenoir]

Was hoping that someone got a hold of one and benched it.....however after reading the words marketing slide ....well u know...here come the mini(plausible...maybe) salt trucks .Marketing tends to add that one in a million best case scenario magic to situation.

 

[Mussels]

first time around i missed the "all data based on projections" part, so this could be all crap.

 

[Vulpesveritas]

No mention of clock speed to achieve the increase?

Quite honestly, clock speeds don't matter, whether they're up or down. What matters is performance / price and performance / watt. If it is more power efficent than Llano, and is the same price, and performs faster, then I believe that is a win no matter what.
Unless you're hoping for the IPC gains we're all hoping for. Well, everyone who wants AMD to survive for the long haul hopes for.

 

[happita]

Give me a reason enough to buy you AMD!! I wanted to get a laptop in recent years, but want a semi-powerful system that plays all the latest games with decent battery life. This thing gives both it seems. However, I'm still looking to upgrade my sandy to ivy first anyway....so, 1 thing at a time

 

[TheoneandonlyMrK]

No mention of clock speed to achieve the increase?

upto and beyond 4Ghz according to the rumour mill, and this is helped due to the resonant clock mesh tech they have bought in to spruce their chip up

eg of rumour mill
http://www.engadget.com/2012/02/22/amd-piledriver-cores-will-employ-resonant-clock-mesh/

looking good in a realistic kind of way too

being realistic also means that in swapping from stars arch to piledriver arch they might have run into trouble since BD was a bit frequency happy, i am expecting some extreme Ocin results with these ,no comment on pciex allocation ever though, that to me could be the deal breaker , 1 or 2 pciex3 slots would be v nice

 

[Benetanegia]

just look at the gtx 580 transition to the 680. It certainly wasn't 100%, not even 50% faster.

Yeah except the GTX680 uses a chip almost half the size. Compare it to the GTX560 Ti and in 2560x1600 is in fact twice as fast. And with a significantly smaller die.

A CPU is not the same, it does not scale like that, but it still holds true to the iGPU. 56% over Llano is far from impressive IMO. Especially when we are talking about a marketing slide. Real difference is not going to be more than 20% on the GPU and 10% on the CPU most probably atributable to higher clocks.