Trinity review

[Abwx]

When a freaking IGP gets touted as the best feature in a CPU...you know the CPU has failed.

I ve got a few laptops in the last ten years...
The only ones still usable are the one with acceptable GFX.

All the others , even with way better CPUs are too slow
for current usages , particularly surfing the net..

GPU is what decide the lifecycle length , not the CPU...

 

[Kalessian]

Nobody games on a laptop... should be nobody SHOULD game on a laptop, but tons of my friends still use core2 based laptops to play mmos/rpgs/rts/mobas. It makes me cringe. I bet any of them would love a $600 trinity laptop upgrade.

 

[dguy6789]

A laptop is significantly more convenient to take to a lan party or a friend's house. Lugging a complete desktop setup around every weekend gets very old very fast.

 

[Nemesis 1]

Do I look like I care about Midget MMA?
I don't care about IGP's too...waste of diespace for anything else but office programs.
Give a GPU...and cut the IGP from the CPU die, I don't need my CPU held back by a stupid IGP.

And if I wanted to retort I would say that both Intel and AMD's IGP's are to slow for serious gaming...but Intel makes AMD encoding feature look like something that died a slow death in the corner...

You go all *beeep* about useless diespace...because it's ALMOST usable....kinda funny to observe.

Its funny U and I in agreement . For now I do believe come haswell we will be in disagreement once again.

 

[Olikan]

Nobody games on a laptop... should be nobody SHOULD game on a laptop, but tons of my friends still use core2 based laptops to play mmos/rpgs/rts/mobas. It makes me cringe. I bet any of them would love a $600 trinity laptop upgrade.

nobody should game on a laptop?

please explain yourself!

 

[Nemesis 1]

A laptop is significantly more convenient to take to a lan party or a friend's house. Lugging a complete desktop setup around every weekend gets very old very fast.

Are we talking about online gaming with 64 players . If so I would like to see that .

 

[LOL_Wut_Axel]

If an i3 + GT540m barely consumes any more power then why would manufacturers bundle it with the bigger and heavier 90W adapter? I'm not saying that a dead silent laptop is an achievement, I'm just saying that a laptop with a discrete GPU is just overall more noisy and heavier.

Hmm... let me guess... because both components [CPU+dGPU] under full load would consume more power than the CPU+IGP at full load? During normal use, again, the power consumption would be the same because the dGPU can be turned off. And weight? Have you seen how small a GT540M inside a chassis is? The GPU itself would perhaps raise weight by 0.1 pounds and the necessitated additional cooling for when it's running at full load by 0.1 pounds. That's nowhere near what you're making it out to be. As it stands, the Intel CPU + AMD/NVIDIA switchable dGPU combo is much better than an APU, even if it costs a bit more. Talking about a difference in 0.2 pounds is nitpicking at best.

 

[Kalessian]

nobody should game on a laptop?

please explain yourself!

Haha, just hyperbole, but really they get hot, loud, they have crummy screens, the performance is lacking... If you're mobile a lot I can see it having to be done but if your laptop sits on a desk all day like my friends who use them as desktop replacements, then why didn't you just buy a desktop?

 

[Olikan]

Haha, just hyperbole, but really they get hot, loud, they have crummy screens, the performance is lacking... If you're mobile a lot I can see it having to be done but if your laptop sits on a desk all day like my friends who use them as desktop replacements, then why didn't you just buy a desktop?

i agree about the bad screens... but the rest is just..... "first world problems"

 

[Global688]

I'd count this as a victory for AMD. AMD released a product that was faster than its last iteration, lower power, more features.

 

[Gikaseixas]

i agree about the bad screens... but the rest is just..... "first world problems"

the screen on my Samsung Series 7 is amazing, very bright and at 1600x900 is the best fit for 15.6 laptops

 

[-Slacker-]

Computerbase also have a review of this Asus wth an A8-4500M (And that's what I was reading since the Anandtech one still hasn't come through in RSS for me.):
http://www.computerbase.de/artikel/notebooks/2012/bericht-amd-a8-4500m-trinity/
http://translate.google.co.uk/trans...y/&hl=en&safe=off&biw=1024&bih=607&prmd=imvns

Still not finished reading either of them, but for instance Computerbase only benched a few games and the A8-4500M didn't do that well on them. Intel's iGPU may be poor but HD4000 has two advantages: the age old Intel process advantage and a better memory controller. AMD need to seriously improve the memory controller but I doubt they have the capital or manpower to do so at this stage.

As it happens, Computerbase also recently did an article on IB and RAM speeds:
http://www.computerbase.de/artikel/arbeitsspeicher/2012/test-welchen-ram-fuer-intel-ivy-bridge/
http://translate.google.co.uk/trans...e/&hl=en&safe=off&biw=1024&bih=607&prmd=imvns

Intel HD4000 (unlike Llano) seems to not be too bothered about high-speed RAM. Obviously we'll have to wait for desktop Trinity reviews (unless some reviewer can overclock laptop RAM), but Trinity's iGPU will most likely be bandwidth starved again. I think this graph tells a lot:

Hm, disappointing overall I think. Unless AMD find some other markets (emerging markets etc.) I cannot see Trinity selling well if even at games in often ties with HD4000. This does not look good for competition in the CPU market - nor even for the GPU market since GNC is gCompute heavy precisely because of APU / Fusion and Nvidia have a cheaper to make design because they're leaving gCompute to big Kepler.

Wonder how much more stuff Intel will lock down without any competition? Or Nvidia for that matter: GK104 already has limited overclocking and with less competition they might feel bolded to lock down more too: if you want 680 performance you might no longer be able to buy a 670 and overclock it.

Yeah, the hd 4000 was running along side a 45w i7 3720 qm, and it still managed to be almost 20% slower than the A10...

 

[LOL_Wut_Axel]

One thing to keep in mind is that the model Anandtech reviewed has the highest-end APU. This probably won't be what you'll find in $500 laptops months from now, but on $600-700 laptops like the previous A8-3510MX and 3530MX. I didn't see any mention in the review of IF there will be a lower-end A10 like there was a lower-end A8 (3500M), but if there's not, the APU you'll find in $500 laptops will be a higher-end A8.

The conclusion from this is that it changes nothing except making Intel more competitive, basically. There's still a huge gap between Intel and AMD when it comes to CPU performance (which is more important), and the fastest mobile CPU from AMD doesn't come close to the lowest-tier mainstream CPU from Intel. In other words, if the Core i3 is faster, the Core i5 will make a clean swoop.

The other part of the equation is that for the very few that do game, Intel made a bigger step going from Sandy to Ivy in IGP performance than AMD did going from Llano to Trinity: like I mentioned before, an HD 4000 is now enough to play something like DiRT 3 on Medium, and that will appease the few people that want to game on the go for cheap. What's AMD's argument gonna be now since obviously, the A8, which will be the APU in the $500 range laptops, will have lower IGP and CPU performance than the A10? What's their argument gonna be when what we'll find in $600-650 laptops from them will be either the A10 or A8+dGPU in CF? In either of those cases they lose because if manufacturers are having no problems now making laptops with Core i3s or i5s + GT540Ms or HD 7670Ms then what's gonna stop them from pricing at the same range when Ivy Bridge is priced the exact same way as Sandy? An Ivy Core i3 or i5 + HD 7670M will obliterate an A10 or A8+dGPU CF in every way because part of AMD's problems when it comes to getting higher IGP performance is that they're almost running into a CPU bottleneck. Certainly an A8+dGPU CF will mean a CPU bottleneck.

I know Anandtech is making sound Trinity like if it were great, but in comparison to where we were at with Llano vs. Sandy Bridge it's back to square zero except Intel now have an okay-ish IGP for, again, the very small part of the market that wants it.

 

[LOL_Wut_Axel]

I'd count this as a victory for AMD. AMD released a product that was faster than its last iteration, lower power, more features.

Isn't that what you're supposed to do with new products... improve upon your older ones?

 

[LOL_Wut_Axel]

Yeah, the hd 4000 was running along side a 45w i7 3720 qm, and it still managed to be almost 20% slower than the A10...

You do realize the i7 and i5 have very similar single-threaded performance and IGP frequencies, meaning they'll also have very similar IGP performance?

I like how you word everything, forgetting the fact that the HD 4000 is still enough for most games at Medium settings and laptop resolutions (1280x720, 1366x768). Even with that, there's still the small fact that many people here forget because of their personal wants: the vast majority of people don't care about gaming on their laptops, and there's still a huge gap between Ivy and Trinity when it comes to the more-relevant-to-more-people CPU.

 

[pelov]

I think the reason 3DMark gains are so much better than on actual games is due to the new Turbo Core on Trinity. The gain over Llano's 6620G is similar to going from 5870 to 6970. The iGPU Turbo is probably running at base 494MHz rather than top 685MHz lot of the time.

On 3DMark11, the game tests are running at single digit frames and CPU matters not a bit, so it has more headroom for GPU. But in games its running at playable frames and CPU needs more power, thus less power for iGPU.

That of course varies with how well multi-threaded the game too. It's same with HD Graphics 4000 gain over HD Graphics 3000. Less in games that HD 3000 is already decent at, more in the ones HD 3000 is doing bad at.

The stark difference between the synthetic benchmarks and the gaming performance FPS struck me as shocking at first but looking at it again it looks to be a case of shitty drivers. AMD's numbers should pick up in DX11 games due to the architecture and threaded nature of DX11 games. Seeing the HD4000 closer in DX11 than DX9 just doesn't make much sense to me unless something is screwie with the drivers. The integer performance isn't likely holding it back considering at 4 threads it seems to do quite well. I think we'll see Trinity pull ahead even further in DX11 games when AMD gets a proper driver out... which you'd think they'd have learned this by now but apparently not. nVidia seems to at least understand that first impressions mean a lot and make sure to get working drivers on launch day but AMD is content for us to do the legwork for them and post updated benchmarks and figures months after the initial release. Judging by the synthetic workloads there's still some room for improvement in the gaming figures.

This is what I mean about the DX11 gaming

The Metro and Batman results point to driver immaturity rather than Intel being amazing. I think we'll see those results look drastically different with some proper drivers...

I don't know what they're thinking with the price, though. They're apparently pricing these, the A10s, between the i5 and i7 IB's. Is this comparing ultrabook/ultrathin with equal specs? If the A10s are selling for $700 then I might as well pick up an A8 Llano or an i3 IB (whenever they come out) with discrete GPU for a cheaper price. The CPU performance is fine and the GPU performance seems up to snuff and the battery life is great but that price is too damn high.

BTW, the 3730QM costs nearly $400 by itself so comparing that chip to anything Trinity is rather dumb. For the price of a single 3720QM chip I can pick up a brand new A6 Llano laptop.

 

[MrTeal]

Pathetic is that you have read a review compering a 45W CPU to a 35W CPU and you actually believe that Intels HD4000 closed the gap. Even with 10W lower TDP Trinity is 20% on average faster than Intels High end 45W CPU.

Do the math again against a 35W Core i5

1) TDP != Power Consumption
2) Performance does not scale linearly with TDP
3) The 35W i5s and 45W i7s have the same GPU with the same base clock and same turbo boost
4) The system is almost certainly GPU limited while gaming as opposed to CPU limited; the extra two cores of the i7 vs the i5 will likely make very little difference in gaming performance.
5) TDP isn't power consumption
6) TDP is not the same as power consumption

So, If 35W Trinity is 20-30% faster on avg than 45W IB, then it will be 30-50% faster than 35W IB.

I know, disappointing

Care to walk us through the math on that one?

 

[AtenRa]

You do realize the i7 and i5 have very similar single-threaded performance and IGP frequencies, meaning they'll also have very similar IGP performance?

http://www.anandtech.com/bench/Product/348?vs=327

Actually, in case of Sandy Bridge, there's usually only 10-20% difference in games between the 45W quad core and 35W dual core.

http://ark.intel.com/products/52229/Intel-Core-i5-2520M-Processor-(3M-Cache-2_50-GHz)

http://ark.intel.com/products/52227/Intel-Core-i7-2820QM-Processor-(8M-Cache-up-to-3_40-GHz)

Lets see if IB core i5 will have the same performance as Core i7.

 

[LOL_Wut_Axel]

http://ark.intel.com/products/52229/Intel-Core-i5-2520M-Processor-(3M-Cache-2_50-GHz)

http://ark.intel.com/products/52227/Intel-Core-i7-2820QM-Processor-(8M-Cache-up-to-3_40-GHz)

Lets see if IB core i5 will have the same performance as Core i7.

It'll be within 5% of its gaming performance probably. But again, how is this relevant?

People care more about CPU than GPU performance in laptops, and the few that want to game on them still have the option for the HD 4000 which, again, can play most recent games decently. What exactly is AMD gonna do when Haswell comes out and AMD's IGP advantage is either completely gone or they're actually left behind? After all, Intel will still have their big CPU advantage, and now an IGP advantage as well. Ivy Bridge to Haswell will bring a higher IGP performance difference than SB to IB, and that was an improvement of 30-40%.

Intel has been perfecting X86 CPU architecture for some six years now since Conroe, and they've gotten to a point where they're years ahead from AMD in that discipline which more people care about. Since they've largely perfected given what's possible with current technology, they've focused on the IGP now because they can get much larger gains from it while still gaining CPU efficiency. Intel has essentially largely closed the gap to AMD in IGPs in a single year and will be either on par or overtaking them by next year with Haswell, while AMD has been hugely slower when it comes to laptop CPUs for, what is it now, six years? They're slowly making gains, but there's still a huge gap--and that huge gap is a gap where it matters most, the CPU.

But at least their power consumption is excellent like Intel's.

 

[AtenRa]

1) TDP != Power Consumption
2) Performance does not scale linearly with TDP
3) The 35W i5s and 45W i7s have the same GPU with the same base clock and same turbo boost
4) The system is almost certainly GPU limited while gaming as opposed to CPU limited; the extra two cores of the i7 vs the i5 will likely make very little difference in gaming performance.
5) TDP isn't power consumption
6) TDP is not the same as power consumption


Care to walk us through the math on that one?

When reading a review dont skip the first pages, they have useful technical information

http://www.anandtech.com/show/5831/amd-trinity-review-a10-4600m-a-new-hope/2

Improved Turbo

Trinity features a much improved version of AMD's Turbo Core technology compared to Llano. First and foremost, both CPU and GPU turbo are now supported. In Llano only the CPU cores could turbo up if there was additional TDP headroom available, while the GPU cores ran no higher than their max specified frequency. In Trinity, if the CPU cores aren't using all of their allocated TDP but the GPU is under heavy load, it can exceed its typical max frequency to capitalize on the available TDP. The same obviously works in reverse.

Under the hood, the microcontroller that monitors all power consumption within the APU is much more capable. In Llano, the Turbo Core microcontroller looked at activity on the CPU/GPU and performed a static allocation of power based on this data. In Trinity, AMD implemented a physics based thermal calculation model using fast transforms. The model takes power and translates it into a dynamic temperature calculation. Power is still estimated based on workload, which AMD claims has less than a 1% error rate, but the new model gets accurate temperatures from those estimations. The thermal model delivers accuracy at or below 2C, in real time. Having more accurate thermal data allows the turbo microcontroller to respond quicker, which should allow for frequencies to scale up and down more effectively.

At the end of the day this should improve performance, although it's difficult to compare directly to Llano since so much has changed between the two APUs. Just as with Llano, AMD specifies nominal and max turbo frequencies for the Trinity CPU/GPU.

also

SB and IB use the L3 (LLC) for the iGPU and the CPU. This is one of the reasons the iGPU in Core i7 (8MB) perform higher than the Core i5 (3MB).

Originally Posted by IntelUser2000

http://www.anandtech.com/bench/Product/348?vs=327

Actually, in case of Sandy Bridge, there's usually only 10-20% difference in games between the 45W quad core and 35W dual core.

 

[Gigantopithecus]

The bottom graph is from PCPer.

How does such a wild discrepancy happen?

 

[Silverforce11]

You guys need to lay off the crackpot if you think ppl don't game on laptops.. it's utter rubbish.

I just got myself a dirt cheap dv6 with 6770M and here I am playing Diablo 3 at max settings for the past few hrs, its running at 60 fps vsynced. It also handles BF3 at medium (which looks heaps better than consoles already) in MP without dipping < 30 fps.

Trinity looks to be onpar with the 6770M discrete, that's in a package with great power efficiency and functional without plugging in for a few hrs of gaming. I would not hesitate to get a ultrabook with trinity in it at all.

Also, CPU matters very little in laptops, what is the most common task on this platform? Surfing (flash), skyping, viewing HD media, doing office work and gaming. Which ones of these task need more than Trinity has in terms of CPU power? None. Which ones work better on the iGPU? You can figure it out. Unless you buy a laptop specifically to encode videos, paying more for sb/ivb cpus is an utter waste of $$.

The CPU hasnt been a bottleneck in PC experience for a long long time.

 

[Joseph F]

IMO, what Trinity needs (besides better CPU performance) is a 32MB ZRAM GPU cache.

You guys need to lay off the crackpot if you think ppl don't game on laptops.. it's utter rubbish.

Indeed. Back when I had a laptop, (Lenovo Thinkpad R60) I would play lots of games on it, despite the agonizingly slow Intel GMA 950. D:

 

[pelov]

Could be AA cranked up or even other BIOS settings. Plugged in versus on battery? That's why it's important to read a variety of reviews That's also why I'm convinced the drivers are lackluster and the GPU performance should see some good improvements in certain games where it's rather obvious (Metro 2033 and Batman Arkham).

The price is apparently ~$600 without the SSD, so I guess that isn't that bad. The other reviews I read stated it was $700 without it (VR-zone. It was a really really shitty review that got quite a bit flat-out wrong so don't bother reading that). So maybe the price wasn't as bad as I originally thought it was. ~$500 something in standard notebook form 15'6" form? Discrete GPU + 1080p screen would cost roughly $700-800 on HP if my assumptions regarding the prices are correct. That'd be a pretty sweet deal. Add another $50-75 for a discrete GPU and hope to god they're actually taking mobile crossfire seriously or crossfire in general, really.

 

[ShintaiDK]

The bottom graph is from PCPer.

How does such a wild discrepancy happen?

When you cripple one system on purpose.