Trinity review

[Kevmanw430]

http://forums.anandtech.com/showpost.php?p=33460091&postcount=438

Edit: But feel free to bury your head and ignore things that don't meet your agenda.

THIS.

Look at other reviews, LolWut. Sure, Anand's put the HD4000 above the 7660G, but it's the only one I've read that had it this way.

You also seem to overlook the fact that AMD will have FAR superior drivers, image quality, and that all of these benches were not run on final drivers optimized for the Trinity platform. It is 100% possible another 15% performance can be squeezed out of Trinity, and even more in some specific programs.

 

[Kevmanw430]

So what you're saying is that Anandtech has an agenda? I didn't post those benchmarks, Jarred did. And Anandtech is a much more visited site than TechReport, and the journalistic integrity of Anandtech is much better than other sites.

I'll stick to the results Jarred got.

He didn't say Anandtech had an agenda, as we all know they don't. However, it does seem that YOU have an agenda.

Regardless, the majority of websites put that benchmark in the other order. I respect Anandtech's reviews, but sometimes there are anomolies.

 

[Khato]

35W Trinity is almost directly equivalent to a 25W intel, and not far at all from what's in this zenbook in terms of power consumption. By all rights, the A10-4600M is an ultrabook chip! According to the power consumption numbers they can stick this chip in that small a chassis and it will get no hotter than a 17W intel. Now will oem's actually do it? Probably not, because of the reputation AMD has of running so hot. Isnt it ironic that this reputation still persists even when intel 17W ULV chips pull nearly as much power as 35W A10?

Before getting all excited about the good performance of the Trinity prototype in the batter life tests, you may want to translate the data into a more common figure. (I do wonder why Anandtech normalizes into Min/Wh instead of simply translating into W.) Anyway, here's the reason why these battery life figures have pretty much zero indication of actual chip TDP - these are simply using the published battery size to translate the published runtime in minutes into watts:

Trinity A10-4600M Prototype:
Idle - 6.47W
Internet - 8.34W
H.264 - 15.07W

Asus NV56M i7-3720QM:
Idle - 10.09W
Internet - 10.94W
H.264 - 16.39W

Asus Zenbook Prime UX21A:
Idle - 4.47W
Internet - 7.98W
H.264 - 9.33W

From those, can you really draw any conclusions whatsoever about what the TDP values of the processors are? Especially considering that everything else in the platform has a marked effect upon batter life - despite the fact that Anandtech normalizes display brightness to 100 nits the differences in backlight efficiency will even come into play. The only real conclusion here is that the mobile quad core Ivy Bridge processors increased their idle power consumption which translates into higher figures for all the Anandtech tests. (Okay, you can't really draw that conclusion just from these tests as it could be an issue with that particular model, but it mirrors what notebookcheck observed in their review where they used the same platform and just swapped out the processor.)

 

[LOL_Wut_Axel]

No, Trinity is better in that benchmark. The Trinity battery is missing the 9W the i5 batter has, so that would look different. So he's right about the AMD Trinity A10 chips performing somewhere between the Intel ULV and 35W chips as far as battery life/power consumption is concerned. If you take into account perf-per-watt then things change in favor of i5 in heavy compute tasks and video (oddly enough) while AMD does better in gaming so it's more complicated than "this one is better." It's more like "this one is better at this task."

Yeah, it's better by a statistical error (0.27%).

There's a reason why it's called relative battery life. Yeah, the i5 laptop had a bigger battery, but it also delivered better battery life.

And it'd be rather ignorant to say that mobile Ivy Bridge won't be more efficient than mobile Sandy Bridge given on desktop we've already seen a very good improvement:

 

[LOL_Wut_Axel]

THIS.

Look at other reviews, LolWut. Sure, Anand's put the HD4000 above the 7660G, but it's the only one I've read that had it this way.

You also seem to overlook the fact that AMD will have FAR superior drivers, image quality, and that all of these benches were not run on final drivers optimized for the Trinity platform. It is 100% possible another 15% performance can be squeezed out of Trinity, and even more in some specific programs.

Care to explain how Intel wouldn't also be able to get a further 15% improvement in performance out of the HD 4000 with newer drivers? How is it that AMD is suddenly the only one able to do this? In other words, does AMD's driver department have secret magic pixie dust that Intel doesn't? That, and Intel has been much more aggressive about new driver releases than before.

 

[Arzachel]

Not sure if serious.

You realize that is an ULTRABOOK, right? The A10-4600M won't make it into AMD's ULTRA-THINS (or Sleekbooks, as HP calls them).

What will make it into Ultra-Thins is the A6-4455M, which is a single-module ("dual-core") chip clocked from 2.1-2.6GHz and with an HD 7500G (256 Radeon Cores) at 327-424MHz. For reference, the A10-4600M is a dual-module ("quad-core") clocked from 2.3-3.2GHz and with an HD 7660G (384 Radeon Cores) at 497-686MHz.

Sorry to destroy your now 45 minute dreams, but the APU in AMD's ultra-thins will have less than half the processing power of the A10 and only a bit more than half of the graphics processing power.

Also, funny that you say AMD is ahead in power consumption, yet you completely brushed over the fact that the one-year-old 35W i5-2410M has the same relative battery life as the 35W A10-4600M. You also brushed over the 17W Core i3-2367M, which has 18% higher relative battery life than the A10-4600M. Convenient, isn't it?

You know what's even more convenient? You totally ignoring the cute little number that represents battery capacity That said, this shows how well the chip is power gated. When it's running full bore, it probably pulls the same as an equivalent 35W Intel chip. They're not going to stick a 35W chip into a ultrabook, but they'll most likely have a 17W part with all the shaders enabled further down the road.

You don't seem to realize that the gpu is cut down in the other Trinity models. There is no evidence to this with ivy.

As far as people choosing this, the only selling point is price. People buying a gaming system certainly aren't going to use integrated graphics. You are implying that systems that use Intel cpu's are neither mobile nor suited for everyday computing. If that were the case, the market share numbers between the companies wouldn't be what they are today.

I'd get a decent 14/15" notebook with Trinity because I'd love to have something that doesn't weigh a ton and has good battery life without having completely anemic GPU performance.

 

[ninaholic37]

You realize that is an ULTRABOOK, right? The A10-4600M won't make it into AMD's ULTRA-THINS (or Sleekbooks, as HP calls them).

What will make it into Ultra-Thins is the A6-4455M, which is a single-module ("dual-core") chip clocked from 2.1-2.6GHz and with an HD 7500G (256 Radeon Cores) at 327-424MHz. For reference, the A10-4600M is a dual-module ("quad-core") clocked from 2.3-3.2GHz and with an HD 7660G (384 Radeon Cores) at 497-686MHz.

Sorry to destroy your now 45 minute dreams, but the APU in AMD's ultra-thins will have less than half the processing power of the A10 and only a bit more than half of the graphics processing power.

From what I can tell, there are two AMD chips planned for Ultrathins...

Has the A10-4655M been cancelled? Or am I missing something?

 

[pelov]

So what you're saying is that Anandtech has an agenda? I didn't post those benchmarks, Jarred did. And Anandtech is a much more visited site than TechReport, and the journalistic integrity of Anandtech is much better than other sites.

I'll stick to the results Jarred got.

Jarred fumbled that review. Honestly, it's one of the worst hardware reviews I've read on this site. He completely glossed over checking the IPC on a new architecture that's coming to the server/desktop. He also didn't bother to graph the FPS real-time so you get low FPS and high FPS as well as apples-to-apples which is how you properly review any gaming performance. He didn't bother with perf-per-watt improvements over Llano or even comparing them at all. That's not even ignored on the desktop yet it wasn't even mentioned in the Trinity laptop review... After Bulldozer sucked it up in power consumption in an epic fashion he practically ignores testing this in Trinity. Seriously? AVX performance was ignored completely. All he did was run the benchmarks, give a short little review of what the numbers were and did absolutely nothing to explain why he got the numbers he did.

Anand did far better with this ASUS Zenbook review, paying attention to a lot more. Granted, this is a finished product that is going into retail so it warrants a little prodding and poking. Jarred only had to test the CPU and GPU without the build quality of the platform yet he half-assed it.

Notebookcheck.com's review was more thorough. Though they didn't bother with IPC or other technical stuff, they at least included far more benchmarks so you get a better grip on the chip's performance.

TR's review was lacking a competing SB i5 retail chip to compare the Trinity/i7 IB chips to but they still did a more thorough review, paying attention not just to the numbers in the benchmarks but digging a bit deeper and attempting to explain it.

This is partly AMD's own fault. Instead of giving reviewers a finished product with proper working and mature drivers, they yet again opted to go with a paper launch and a product we may or may not see in retail outlets. The BIOS options were nonexistent and they haven't mentioned whether AMD will be providing the drivers or we'll have to rely on the OEMs yet again. They really need to fire their entire PR/marketing team. Apparently they did just that after the Bulldozer launch, it seems they haven't quite purged themselves of the same bad habits.

 

[Kevmanw430]

Care to explain how Intel wouldn't also be able to get a further 15% improvement in performance out of the HD 4000 with newer drivers? How is it that AMD is suddenly the only one able to do this? In other words, does AMD's driver department have secret magic pixie dust that Intel doesn't? That, and Intel has been much more aggressive about new driver releases than before.

Intel's history of performance increases through driver updates is spotty, to say the least. I don't remember the last time an Intel driver release mentioned a game improvment, but I'd love to be proved wrong, if you can provide proof.

AMD's driver team, on the other hand, offers regular driver updates targeted at games and other program improvements. They have had rough patches as well, but have a much better driver record.

It is much better to speculate on past history then on what YOU think is going to happen, as we are often proved wrong.

 

[Arzachel]

What version of Skyrim was tested in the Trinity review? If it was left unpatched, it probably accounts for the difference.

 

[Schmide]

So what you're saying is that Anandtech has an agenda?

No. Absolutely not. Ironically the guy who says the average is good refuses to average when it disfavors his fanboism.

I didn't post those benchmarks, Jarred did. And Anandtech is a much more visited site than TechReport, and the journalistic integrity of Anandtech is much better than other sites.

You dare question TechReport's journalistic integrity? I doubt anyone working at Anandtech would.

You show your true colors so easily.

I'll stick to the results Jarred got.

How does that sand taste?

 

[pelov]

What version of Skyrim was tested in the Trinity review? If it was left unpatched, it probably accounts for the difference.

It wasn't even mentioned what patch it was running. That's my point about Anandtech's reviews on anything gaming related. Great tech site and awesome reviews filled with detailed information (usually) but it seems like anything gaming related they just don't seem to understand as well as [H] or TR or a dozen or so other sites.

 

[Kevmanw430]

I would hope it was the patched version, didn't it imporve performance on AMD by like 50%?

 

[Joseph F]

Care to explain how Intel wouldn't also be able to get a further 15% improvement in performance out of the HD 4000 with newer drivers? How is it that AMD is suddenly the only one able to do this? In other words, does AMD's driver department have secret magic pixie dust that Intel doesn't? That, and Intel has been much more aggressive about new driver releases than before.

Because Intel's driver team is terrible.
Can you show us an example of an Intel GPU driver update actually improving gaming performance the way AMD and nV's updates do? (Aside from them fixing something that should have never been broken)
If not, do you have a good reason to suspect that Intel's driver team is going to put out at least semi-frequent gaming updates in the future?

Please forgive me if I am mistaken about Intel's driver team in recent times, it's been a while since I have used one of their GPUs.
(The last one I used was a GMA 950 in my old Thinkpad R60) D:

 

[LOL_Wut_Axel]

You know what's even more convenient? You totally ignoring the cute little number that represents battery capacity That said, this shows how well the chip is power gated. When it's running full bore, it probably pulls the same as an equivalent 35W Intel chip. They're not going to stick a 35W chip into a ultrabook, but they'll most likely have a 17W part with all the shaders enabled further down the road.



I'd get a decent 14/15" notebook with Trinity because I'd love to have something that doesn't weigh a ton and has good battery life without having completely anemic GPU performance.

The Zenbook UX21A has a 32Wh battery and the Trinity Prototype has a 56Wh battery. The Zenbook posted an internet-browsing battery life of 263 minutes, and the prototype posted 403 minutes. That means the Trinity Prototype had a 75% bigger battery yet posted 53% higher battery runtime. In other words, the i5 or i7 17W ULV Ivy Bridge had 22% higher relative battery life than the 35W A10.

Argument not found.

 

[Schmide]

(snip)

Argument not found.

It's a freaking prototype?

Edit: Nah there is no room for improvement anyways.

 

[Kevmanw430]

http://techreport.com/articles.x/22932/10

Right here, does Trinity significantly outperform the 3720QM. However, I think the more important metric the the frame spikes. Trinity is not only faster, but has a much more even frame rate, giving you the impression it's running miles faster than the 3720QM.

Another thing to keep in mind is that the A10-4600M isn't supposed to compete against the 3720QM, but the weaker, likely comparitively priced i5 models.

 

[Arzachel]

The Zenbook UX21A has a 32Wh battery and the Trinity Prototype has a 56Wh battery. The Zenbook posted an internet-browsing battery life of 263 minutes, and the prototype posted 403 minutes. That means the Trinity Prototype had a 75% bigger battery yet posted 53% higher battery runtime. In other words, the i5 or i7 17W ULV Ivy Bridge had 22% higher relative battery life than the 35W A10.

Argument not found.

I even said that no one will be putting an 35W CPU into ultrabooks but you went and strawmanned anyways. I was pointing out that the comparison to the i5-2410m wasn't apples to apples. That said, you linked to the normalized chart in a later post I guess.

 

[pelov]

Please forgive me if I am mistaken about Intel's driver team in recent times, it's been a while since I have used one of their GPUs.
(The last one I used was a GMA 950 in my old Thinkpad R60)

They're improving but it mainly revolves around fixing issues with games rather than squeezing out more performance.

http://downloadmirror.intel.com/21231/eng/ReleaseNotes_GFX_64.pdf
http://downloadmirror.intel.com/21034/eng/relnotes_gfx.pdf

So they're getting around to some of the known issues and annoying bugs but I wouldn't expect any performance gains. Roughly speaking, nVidia and AMD get up to ~10% more performance through drivers over the course of a year.

 

[LOL_Wut_Axel]

Intel's history of performance increases through driver updates is spotty, to say the least. I don't remember the last time an Intel driver release mentioned a game improvment, but I'd love to be proved wrong, if you can provide proof.

AMD's driver team, on the other hand, offers regular driver updates targeted at games and other program improvements. They have had rough patches as well, but have a much better driver record.

It is much better to speculate on past history then on what YOU think is going to happen, as we are often proved wrong.

Because Intel's driver team is terrible.
Can you show us an example of an Intel GPU driver update actually improving gaming performance the way AMD and nV's updates do? (Aside from them fixing something that should have never been broken)
If not, do you have a good reason to suspect that Intel's driver team is going to put out at least semi-frequent gaming updates in the future?

Please forgive me if I am mistaken about Intel's driver team in recent times, it's been a while since I have used one of their GPUs.
(The last one I used was a GMA 950 in my old Thinkpad R60) D:

LOL at people praising AMD's driver team when you have threads here with people complaining about issues with the HD 7000 series.

You do realize in the Anandtech review they did not mention having any issues with the drivers, right?

And you do realize the GMA 950 is a 6-year-old IGP, right? If we were to make arguments here making reference to issues with 6-year-old products everyone here would be complaining about the horrible driver support for the Radeon HD 2000 and 3000 series and how that means AMD has horrible driver support.

 

[wlee15]

There's also an 19W embedded variant called R-452L which is a quad-core 1.6ghz/2.4ghz, 256 shader 327mhz/424mhz. I expect a non-embedded version will make into ultrathins as well.

 

[Kevmanw430]

The Zenbook UX21A has a 32Wh battery and the Trinity Prototype has a 56Wh battery. The Zenbook posted an internet-browsing battery life of 263 minutes, and the prototype posted 403 minutes. That means the Trinity Prototype had a 75% bigger battery yet posted 53% higher battery runtime. In other words, the i5 or i7 17W ULV Ivy Bridge had 22% higher relative battery life than the 35W A10.

Argument not found.

Obviously, a 17W chip would get better relative battery life, because it has a whole 18W of TDP less to use.

Interesting that while the A10 had more than 100% more TDP available than the ULV IVB chip, but its relative battery life was only no where near 100% worse.

I know that is not how TDP is quantified, but you get my point.

 

[hans030390]

But they didn't. HD 7660G isn't enough for playing most games at High settings.

WTF. No one is arguing that it can play most games at high settings. No one has argued that it would be able to, either. There is no point in mentioning this.

As for your comment regarding the A10:

Which there's not, because laptops featuring the A10 will cost more on average than laptops featuring the Core i3.

And, in gaming, they A10 will perform around 20% better on average than the i3 (according to benchmarks). So, yeah...kind of makes sense that the A10 will cost more than the i3.

You're aware that from a pure price-to-performance ratio, Trinity is overall a better deal than IB will be when it comes to budget gaming, right? I mean, you keep posting graphs/benchmarks and mentioning prices, so I'm assuming you know how to properly read those and correlate the data with pricing.

We're not arguing what is "good enough." Going strictly from the numbers, Trinity wins. You're not going to be able to get an Intel laptop for $600 that competes with Trinity in that regard (outside of special deals, which you can't really factor into arguments like these). This is assuming the A10 is put in laptops around that price. Details are pretty sparse there.

Battery life is kind of a moot point. OH NO, THIS LAPTOP GETS 7 HOUR AND THIS ONE GETS 8! WOE IS ME! Both are fantastic for battery life. Both will allow for sleek, very portable machines. If someone needs a larger focus on CPU performance, go with Intel. If someone is more focused on budget gaming, Trinity wins out on average in similar price ranges. Easy, easy, easy.

It's also very strange how you seem to ignore conflicting benchmarks. Haven't some shown Trinity beating the HD4000 in Skyrim using similar settings compared to other sites?

On top of that, you definitely don't factor in driver support for games (AMD wins here, hands down). Same for image quality. AMD will do well in these areas. Intel? Eh.

You can call people fanboys (or equivalent terms) all you want, but that's usually just a sign of someone who is unwilling to back down from their position no matter what they're up against (aka, a fanboy). And, for future reference, it makes one look more childish to argue non-stop than to take a step back and actually consider someone else's position. It takes a real man to say, "Hey, you know, you have a point. I might even be wrong in this regard."

I always go for hardware that will give me the best performance for my needs in my budget. That is why I'll be picking up a Trinity laptop and not an Intel one, unless I can find a sleek, portable Intel laptop with a decent iGPU and great battery life for around $600 (not counting refurbished models or special deals, again). Unless, of course, AMD really screws up and puts A10 in the $700-800 price range.

 

[happysmiles]

lol_wut vs the forum.

 

[hans030390]

LOL at people praising AMD's driver team when you have threads here with people complaining about issues with the HD 7000 series.

You do realize in the Anandtech review they did not mention having any issues with the drivers, right?

And you do realize the GMA 950 is a 6-year-old IGP, right? If we were to make arguments here making reference to issues with 6-year-old products everyone here would be complaining about the horrible driver support for the Radeon HD 2000 and 3000 series and how that means AMD has horrible driver support.

People bitch about any drivers. Period. The end. Your argument is invalid, because it can apply to EVERY PIECE OF HARDWARE EVER MADE. I've had NO issues with drivers from Nvidia or AMD, just so you know. Does that mean everyone should have no issues? Nope. It can happen to anyone with any hardware.

You do realize that Anandtech doesn't always cover the same topics as everyone else, right? This can include drivers and image quality. I'm not saying they never touch on these topics, but they don't for every article and review.

You're still ignoring that Intel's HD4000 still has questionable image quality for games (based on some things people have pointed out on here). That's a legitimate concern, I'd say. You're also ignoring frame spikes that were mentioned before. BUT ANANDTECH DIDN'T MENTION THESE SO IT PROBABLY ISN'T A BIG DEAL EVEN THOUGH OTHER SITES HAVE AND IT IS APPARENTLY WORTH MENTIONING BUT I DON'T KNOW I'M JUST GOING TO KEEP ARGUING WHATEVER I FEEL LIKE IN MY GUT BECAUSE I KNOW I MUST BE RIGHT IN MY HEART GO AMERICA!!!