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

[xenocide]

All else being equal if I were to build a basic box with light weight gaming needs the APU would win. An extra $90 to spend on a GPU would get you more bang for the buck than the CPU's 10-15% increased efficiency.

If you think the difference between a 2500 and even the best APU is only 10-15%, I hate to say it, but you're misinformed. The CPU side of Quad-Core SB's is upwards of 50% faster than Llano;

http://www.hardwarecanucks.com/foru...2-amd-a8-3870k-unlocked-llano-apu-review.html

Even with a 20% overclock, Llano wasn't really even close to the i5-2400. In games where the CPU plays any role, the i5's crush the Llano offerings, with the same discrete GPU. That's also comparing the $140 A8-3870K, the i5-2400 is currently $190, and you could even get an i5-2300 for $180, both of which are better than that CPU in every way (sans iGPU). If you look at the charts and see the ~$100 APU's, they are far worse.

Saying a Llano-based solution is as good as a Quad-Core SB solution, for anything short of day to day use, is dishonest at best.

 

[sergionography]

If you think the difference between a 2500 and even the best APU is only 10-15%, I hate to say it, but you're misinformed. The CPU side of Quad-Core SB's is upwards of 50% faster than Llano;

http://www.hardwarecanucks.com/foru...2-amd-a8-3870k-unlocked-llano-apu-review.html

Even with a 20% overclock, Llano wasn't really even close to the i5-2400. In games where the CPU plays any role, the i5's crush the Llano offerings, with the same discrete GPU. That's also comparing the $140 A8-3870K, the i5-2400 is currently $190, and you could even get an i5-2300 for $180, both of which are better than that CPU in every way (sans iGPU). If you look at the charts and see the ~$100 APU's, they are far worse.

Saying a Llano-based solution is as good as a Quad-Core SB solution, for anything short of day to day use, is dishonest at best.

yea that kinda worries me, i just looked up llano benchmarks and they are a good deal slower than a phenom II 980
and even when looking at 3dmbark vantage results(i heard this is were amd based its 29%increase) adding 30% more to llano makes it match phenom II 980
which is a great deal slower than sb
but knowing that a10 5600k will have 3.8 ghz similar to phenom II 980 only strengthens my assumption that amd might have finaly matched phenom II ipc with piledriver XD lol

edit: the sad part is, this is what bulldozer shouldve been, piledriver is supposed 2 be 10% better than this

 

[xenocide]

yea that kinda worries me, i just looked up llano benchmarks and they are a good deal slower than a phenom II 980
and even when looking at 3dmbark vantage results(i heard this is were amd based its 29%increase) adding 30% more to llano makes it match phenom II 980
which is a great deal slower than sb
but knowing that a10 5600k will have 3.8 ghz similar to phenom II 980 only strengthens my assumption that amd might have finaly matched phenom II ipc with piledriver XD lol

edit: the sad part is, this is what bulldozer shouldve been, piledriver is supposed 2 be 10% better than this

I'm only really interested in Trinity for a decent "Light-Gaming" Laptop. I just don't see a place for it outside of a day-to-day use PC. I'd rather take a low-end SB offering and a good discrete GPU than an APU.

 

[HTC]

This is taken from Hardware Canucks review of 3850.

How about the CPU section:

WOW!!! Look @ that SuperPI: 117% better for the I3 2120.
WOW!!! Look @ that WinRAR: 84% better for the I3 2120.

Obviously, this makes this E2120 @ least 84% better then A-3850, right?


But wait: something's not right. What's this?

Interesting! That cinebench R10: they're almost dead even.
Interesting! That Cinebench R11.5: 7% better for the A-3850.
Interesting! That POV-Ray: 15% better for the A-3850.

Does this mean the CPU portion of A-3850 is better then I3 2120? See how a few hand picked results can skew performance? I have zero doubts them 29% faster productivity come from such "handpicked" applications.


How about the graphics section:

WOW!!! Look @ that L4D: 123% better for the A-3850.
WOW!!! Look @ that 3DMark Vantage: 109% better for the A-3850.
WOW!!! Look @ that Crisys: 109% better for the A-3850.

Obviously, this makes this A-3850 @ least 109% better then I3 E2120, right?


But wait: something's not right. What's this?

Interesting! That World in Confict: 56% better for the A-3850.
Interesting! That X3 Terran Conflict: 54% better for the A-3850.


Does this mean the GPU portion of A-3850 is only 54% better then I3 2120? See how a few hand picked results can skew performance? I have zero doubts them 56% faster visuals come from such "handpicked" applications.



Check this pic: notice anything? I'll give you a hint: "reversed".

With any of Intel's depicted (exception being I5 2400), the GPU portion is always smaller meaning it's the CPU that's the best. With A-3850, it's the other way around: it's the GPU that's the best of the 2.

It would have been interesting to see the associated CPU results from this test, IMO.

There's no reason to beat around the bush here, the "Stars" K10.5 cores simply cannot compete with the newer Sandy Bridge microarchitecture when it comes to lightly threaded applications. The Llano chip does occasionally score the odd victory thanks to its four native cores, but even those victories are often by a very slight amount. To make matters worse, in all of our gaming benchmarks it was a slaughter in favour of the Sandy Bridge processor, but only when using a discrete GPU.

When we benchmarked the integrated GPU's, Llano's raison d'être became crystal clear. The Radeon HD 6550D graphics solution that AMD have embedded in the A8-series leaves in the dust anything that Intel have produced so far in the IGP realm. The A8-3850 had on average 80% higher frame rates than the Core i3-2120. If we increased the Llano's memory frequency from DDR3-1333 to DDR3-1866, we could have easily realized a further 15% framerate improvement. This synergy between the GPU and CPU that has been personified in AMD's APUs is becomingly increasingly important as software manufacturers begin harnessing the parallel processing horsepower of modern graphics architectures. While there was once only a handful of GPU-accelerated programs available, now everyday programs like Microsoft Word, Internet Explorer and Flash can benefit from GPU acceleration.

Sources: pics 1, 2, text, pic 3.

 

[Vulpesveritas]

@ben, learn math. 35% faster is 1.35 times faster.

 

[sergionography]

This is taken from Hardware Canucks review of 3850.

http://images.hardwarecanucks.com/image/mac/reviews/amd/Llano/Llano_A83850_98.jpg


How about the CPU section:

WOW!!! Look @ that SuperPI: 117% better for the I3 2120.
WOW!!! Look @ that WinRAR: 84% better for the I3 2120.

Obviously, this makes this E2120 @ least 84% better then A-3850, right?


But wait: something's not right. What's this?

Interesting! That cinebench R10: they're almost dead even.
Interesting! That Cinebench R11.5: 7% better for the A-3850.
Interesting! That POV-Ray: 15% better for the A-3850.

Does this mean the CPU portion of A-3850 is better then I3 2120? See how a few hand picked results can skew performance? I have zero doubts them 29% faster productivity come from such "handpicked" applications.


How about the graphics section:

WOW!!! Look @ that L4D: 123% better for the A-3850.
WOW!!! Look @ that 3DMark Vantage: 109% better for the A-3850.
WOW!!! Look @ that Crisys: 109% better for the A-3850.

Obviously, this makes this A-3850 @ least 109% better then I3 E2120, right?


But wait: something's not right. What's this?

Interesting! That World in Confict: 56% better for the A-3850.
Interesting! That X3 Terran Conflict: 54% better for the A-3850.


Does this mean the GPU portion of A-3850 is only 54% better then I3 2120? See how a few hand picked results can skew performance? I have zero doubts them 56% faster visuals come from such "handpicked" applications.



Check this pic: notice anything? I'll give you a hint: "reversed".

http://images.hardwarecanucks.com/image/mac/reviews/amd/Llano/Llano_A83850_53.jpg

With any of Intel's depicted (exception being I5 2400), the GPU portion is always smaller meaning it's the CPU that's the best. With A-3850, it's the other way around: it's the GPU that's the best of the 2.

It would have been interesting to see the associated CPU results from this test, IMO.





Sources: pics 1, 2, text, pic 3.

nah it won't be to this extreme, overall a quad core apu has more processing power than a dual core sb, so in multithread its only normal the apu is 7% faster, and cinebench is multithread friendly, now if u look at cinebench single thread then u will see i3 way better
First things first is that AMD are comparing with their last Gen not against competition, and that 30% is most likely something like 3d mark or so that checks for general performance

 

[TheoneandonlyMrK]

i think HTC was just trying to point out that by picking and choosing the benches ,you want to show off, you can make something look better or worse then it is , i think there is little point compareing such benches as trinity will have a new arch ,and some new tech to wave about the place (resonant clock mesh ,PD etc) so its performance and ipc and oc ability arent known yet.

 

[HTC]

i think HTC was just trying to point out that by picking and choosing the benches ,you want to show off, you can make something look better or worse then it is , i think there is little point compareing such benches as trinity will have a new arch ,and some new tech to wave about the place (resonant clock mesh ,PD etc) so its performance and ipc and oc ability arent known yet.

That's exactly what i was trying to get across.

 

[Steevo]

If you think the difference between a 2500 and even the best APU is only 10-15%, I hate to say it, but you're misinformed. The CPU side of Quad-Core SB's is upwards of 50% faster than Llano;

http://www.hardwarecanucks.com/foru...2-amd-a8-3870k-unlocked-llano-apu-review.html

Even with a 20% overclock, Llano wasn't really even close to the i5-2400. In games where the CPU plays any role, the i5's crush the Llano offerings, with the same discrete GPU. That's also comparing the $140 A8-3870K, the i5-2400 is currently $190, and you could even get an i5-2300 for $180, both of which are better than that CPU in every way (sans iGPU). If you look at the charts and see the ~$100 APU's, they are far worse.

Saying a Llano-based solution is as good as a Quad-Core SB solution, for anything short of day to day use, is dishonest at best.

Thanks, but honestly users don't run Super PI all day. So back in the real world it will still be the better overall buy, as again.


Even another $40 makes the difference between a 6770 and a 6850, and that is a large enough increase in performance the CPU will never be able to make up the difference.


Users care about

Boot speed.
Can it play "X"
How many jiggawatts does it have.

All those factors users can't tell the difference between AMD and Intel if they don't know at the desktop, most actually choosing AMD, and once a game turns to a slideshow on a equally priced Intel rig they will go for the AMD.


I have both at work, and have been building both, but the AMD wins overall as the price/performance for running non-cpu dependent software means the GPU needs to be able to do more than basics. Horribly optimized web based training videos means a C2D at work needed an additional graphics card a few years ago. video conferencing with a clean install using large TV's in the conference rooms means we needed HDMI with audio. AMD won that when comparing, plus the ability to hardware upscale the video was a bonus.


There is the enthusiast segment where price/performance doesn't mean the same thing as it does to the rest of the world. We are it, we choose what is going to give us the best performance in the 2 sigma, the rest of the world if fine with 1.

 

[sergionography]

I'm only really interested in Trinity for a decent "Light-Gaming" Laptop. I just don't see a place for it outside of a day-to-day use PC. I'd rather take a low-end SB offering and a good discrete GPU than an APU.

that is very true, at 720p the apus do very well in medium high settings. Getting Intel with discrete costs 700+
With an apu for that much u get apu + discrete witch beats the Intel in fps and gaming in general, I got my Asus k53ta for 450$ and don't think I can be happier, and I doubt I can get the same fps in games elsewhere

 

[xenocide]

that is very true, at 720p the apus do very well in medium high settings. Getting Intel with discrete costs 700+
With an apu for that much u get apu + discrete witch beats the Intel in fps and gaming in general, I got my Asus k53ta for 450$ and don't think I can be happier, and I doubt I can get the same fps in games elsewhere

You can build an acceptable Intel system for $450/500, it won't be an i5 with a GTX680 by any means, but it will get the job done.

@Steveo

The i5 beats Llano in every single application by at least 10-20%, and when it comes to CPU demanding games the difference grows to an astounding 50%. How is that niche benchmarks and unrealistic applications? Who cares if it can beat it in SuperPi? I never said it was the only thing, I said that the i5 is greatly superior to the APU's when you're looking at the CPU side of the equation. The GPU is noticably more powerful, and nobody can deny that, but your statement was that the APU was just as good on a CPU level, which is simply false.

If Trinity can improve on Llano, then it's a huge advancement, but I still think the only place for APU's is in Laptops and barebones day to day systems (Office work and Grandmas).

 

[Vulpesveritas]

@xeno so if AMD came out with an APU this year with say a CPU on the level of a SB i5, and had an iGPU that was as fast as a radeon hd 6670 with 1866+ RAM, and could overclock reasonably well under air, had an unlocked multiplier, and cost less than $150, would it still be truly niche in your mind?

That is, of course a best case scenario for Trinity, and likely won't happen that way. But it would be interesting if it did.

 

[bencrutz]

that is very true, at 720p the apus do very well in medium high settings. Getting Intel with discrete costs 700+
With an apu for that much u get apu + discrete witch beats the Intel in fps and gaming in general, I got my Asus k53ta for 450$ and don't think I can be happier, and I doubt I can get the same fps in games elsewhere

You can build an acceptable Intel system for $450/500, it won't be an i5 with a GTX680 by any means, but it will get the job done.

am pretty sure sergio there meant laptop, unless youre talking bout building your own laptop, of course

 

[ensabrenoir]

@xeno so if AMD came out with an APU this year with say a CPU on the level of a SB i5, and had an iGPU that was as fast as a radeon hd 6670 with 1866+ RAM, and could overclock reasonably well under air, had an unlocked multiplier, and cost less than $150, would it still be truly niche in your mind?

That is, of course a best case scenario for Trinity, and likely won't happen that way. But it would be interesting if it did.

If amd could have they wouldve already....i like their apus but jeeez these super apu fantasies are just plain silly

 

[Super XP]

Can't wait to Trinity in real action. I am confident Piledriver will perform the way it is meant to.

 

[Vulpesveritas]

If amd could have they wouldve already....i like their apus but jeeez these super apu fantasies are just plain silly

Main point is I'm hoping for something as fast as an i3 on the CPU end, with an acceptable iGPU for entry-level gaming, and priced about the same as current generation APUs. Lol.
Not impossible... i hope.

 

[Mussels]

just one q: are the people bashing APU performance aware of how well and how easy they are to OC? my laptops APU (1.4GHz quad core) runs at 2.4GHz just fine, and on models with better cooling, 3GHz is possible.

 

[ensabrenoir]

Main point is I'm hoping for something as fast as an i3 on the CPU end, with an acceptable iGPU for entry-level gaming, and priced about the same as current generation APUs. Lol.
Not impossible... i hope.

Very possible...would be great to see tbat come to light. My pipe dream is an i5 with an amd igp mow thats a.more perfect world. but thats impossible.

 

[xenocide]

@xeno so if AMD came out with an APU this year with say a CPU on the level of a SB i5, and had an iGPU that was as fast as a radeon hd 6670 with 1866+ RAM, and could overclock reasonably well under air, had an unlocked multiplier, and cost less than $150, would it still be truly niche in your mind?

That is, of course a best case scenario for Trinity, and likely won't happen that way. But it would be interesting if it did.

I would say it was an incredible bargain and advise anyone who wants a CPU in that price range to get one, but the fact is, AMD doesn't--and can't--make such a product. I refuse to give praise where it isn't necessarily deserved, but if AMD did create an APU that were that good, they would deserve mountains of it.

I'm not saying the concept is incredibly niche, I'm saying the current offerings are. They only seem good for Laptops, because on desktops the CPU's are just too underwhelming. Maybe if GPU-acceleration where standard in all software it would be amazing, but we're still years from that.

just one q: are the people bashing APU performance aware of how well and how easy they are to OC? my laptops APU (1.4GHz quad core) runs at 2.4GHz just fine, and on models with better cooling, 3GHz is possible.

Which is impressive, but if the cores themselves are really weak having a really high clock rate means next to nothing, and that's the issue I see. The CPU's themselves (devoid of the iGPU) are just too weak. It's kind of like with current FX Quad-Cores, where you can overclock them very well, but at the end of the day they are still very lackluster.

 

[Vulpesveritas]

well, if AMD is telling the truth somehow with this, and the supposed leak from before is true, then piledriver will have about the same IPC as llano, and be able to clock higher, and be more power efficient than llano.

But for at least a couple months we wont know for certain and we wait for wiz's review.

everything else till then are speculation, dreams, hopes, and fanboyism.

 

[sergionography]

well, if AMD is telling the truth somehow with this, and the supposed leak from before is true, then piledriver will have about the same IPC as llano, and be able to clock higher, and be more power efficient than llano.

But for at least a couple months we wont know for certain and we wait for wiz's review.

everything else till then are speculation, dreams, hopes, and fanboyism.

well there is one thing thats really interesting if amds claims are true
and thats bulldozer is fixed and sharing resources no longer cripples performance(4 core bulldozers could barely catch up to 4core phenom II even with much higher frequency) so in other words smt scaling is much higher

however one thing is for sure, when amd claims 30% increase in performance, is that comparing the top llano with the top trinity? or in general ? because different tdps lead to different results, and if frequency is the factor here well lets just say that is more effective on lower tdps as the higher u move up, tdp and performance dont nessesarly increase/scale at a constant rate, that explains how amd is claiming 35watt llano performance at 17watts

and another thing to note is that amds 35watt apus have a mediocre 1.4ghz, and honostly its really annoying unless the turbo is kicking in, and overclocking the whole time isnt really the best thing for a laptop, i only oc if needed for gaming or so
but my point is here, 30% increase over 1.4 ghz performance is still very bad and way behind intel, and if that is what amd is basing its numbers on then i would be worried

 

[Steevo]

I would say it was an incredible bargain and advise anyone who wants a CPU in that price range to get one, but the fact is, AMD doesn't--and can't--make such a product. I refuse to give praise where it isn't necessarily deserved, but if AMD did create an APU that were that good, they would deserve mountains of it.

I'm not saying the concept is incredibly niche, I'm saying the current offerings are. They only seem good for Laptops, because on desktops the CPU's are just too underwhelming. Maybe if GPU-acceleration where standard in all software it would be amazing, but we're still years from that.



Which is impressive, but if the cores themselves are really weak having a really high clock rate means next to nothing, and that's the issue I see. The CPU's themselves (devoid of the iGPU) are just too weak. It's kind of like with current FX Quad-Cores, where you can overclock them very well, but at the end of the day they are still very lackluster.

I never said it was equal, but when all you do is run 10% of your CPU time there is no way to tell a difference from a user standpoint other than how much they paid.

Once you start demanding more of your system it shows.


And lastly, if it takes AMD 3Ghz to equal a 2.4 Ghz Intel, but my laptop overclocked with stock voltage to over 3Ghz and it cost $200 less than the closest Intel equivalent I think I won. And i can game on it.

 

[xenocide]

And lastly, if it takes AMD 3Ghz to equal a 2.4 Ghz Intel, but my laptop overclocked with stock voltage to over 3Ghz and it cost $200 less than the closest Intel equivalent I think I won. And i can game on it.

It takes a hell of a lot more than 600Mhz to be equivalent. It would take like a 200% overclock for it to be on par for Intel Quad-Cores...

 

[sergionography]

I never said it was equal, but when all you do is run 10% of your CPU time there is no way to tell a difference from a user standpoint other than how much they paid.

Once you start demanding more of your system it shows.


And lastly, if it takes AMD 3Ghz to equal a 2.4 Ghz Intel, but my laptop overclocked with stock voltage to over 3Ghz and it cost $200 less than the closest Intel equivalent I think I won. And i can game on it.

with phenom II it needs atleast a 40% higher clock to match intel, so around 3.4ghz matches 2.4 sb
Edit: that's taking into account typical cpu applications, k10 architecture lacks real bad in newer instruction sets like AES and others, bulldozer closed that gap tho and did well, and also fpu dependant applications shows a different story, so that being said its really hard to generalize how much frequency is needed and be right on the point