Trinity’s GPU Performance
AMD has allowed a quick preview of their new APU
Editor's Note: Right before the release of this story some discussion has been ongoing at other hardware sites about the methods AMD employed with this NDA and release of information. Essentially, AMD allowed us to write about only the gaming benchmarks and specifications for the Trinity APU, rather than allowing the full gamut of results including CPU tests, power consumption, etc. Why? Obviously AMD wants to see a good message be released about their product; by release info in stages they can at least allow a brief window for that.
Does it suck that they did this? Yes. Do I feel like we should have NOT published this because of those circumstances? Not at all. Information is information and we felt that getting it to you as soon as possible was beneficial. Also, because the parts are not on sale today we are not risking adversely affecting your purchasing decision with these limited benchmarks. When the parts DO go on sale, you will have our full review with all the positives and negatives laid out before you, in the open.
This kind of stuff happens often in our world – NVIDIA sent out GTX 660 cards but not GTX 650s because of lack luster performance for example – and we balance it and judge it on a case by case basis. I don't think anyone looking at this story sees a "full review" and would think to make a final decision about ANY product from it. That's not the goal. But just as we sometimes show you rumored specs and performance numbers on upcoming parts before the NDAs expire, we did this today with Trinity – it just so happens it was with AMD's blessing.
AMD has graciously allowed us the chance to give readers a small glimpse at the performance of the upcoming A series APUs based on the Trinity processor. Today we are covering the SKUs that will be released, general gaming performance, and what kind of power consumption we are seeing as compared to the previous Llano processor and any Intel processor we can lay hands upon.
Trinity is based on the updated Piledriver architecture, which is an update to Bulldozer. Piledriver improves upon IPC by a small amount over Bulldozer, but the biggest impact is that of power consumption and higher clockspeeds. It was pretty well known that Bulldozer did not hit the performance expectations of both AMD and consumers. Part of this was due to the design pulling more power at the target clockspeeds than was expected. To remedy this, AMD lowered clockspeeds. Piledriver fixes most of those power issues, as well as sprinkles some extra efficiency into the design, so that clockspeeds can scale to speeds that will make these products more competitive with current Intel offerings.
The top end model that AMD will be offering of the socket FM2 processors (for the time being) is the A10 5800K. This little number is a dual module/quad core processor running at 3.8 GHz with a turbo speed of 4.2 GHz. We see below the exact model range of products that AMD will be offering. This does not include the rumored Athlon II editions that will have a disabled GPU onboard. Each module features 2 MB of L2 cache, for a total of 4 MB on the processor. The A10 series does not feature a dedicated L3 cache as the FX processors do. This particular part is unlocked as well, so expect some decent overclocking right off the bat.
The A10 5800K features the VLIW 4 based graphics portion, which is significantly more efficient than the previous VLIW 5 based unit in Llano (A8 3870K and brethren). Even though it features the same number of stream processors as the 3870K, AMD is confident that this particular unit is upwards of 20% faster than the previous model. This GPU portion is running at a brisk 800 MHz. The GPU core is also unlocked, so expect some significant leaps in that piece of the puzzle as well.
That is about all I can give out at this time, since this is primarily based on what we see in the diagram and what we have learned from the previous Trinity release (for notebooks).
Click to read the entire post here.
We did not have access to an Intel processor with a 4000 series graphics unit at this time, but I will be including results from such a product in the full review. I used the previous A8 3870K and the Intel i3 2105 (which features the 3000 series graphics unit). I took these products for a spin in four of the newer, more popular games out there. I was not able to enable full resolution and quality selections, but the results still looked pretty good considering the performance.
The Intel part is not up to the task obviously. Not surprising, as it is a last gen part with inferior graphics as compared to the 4000 series with the latest Ivy Bridge processors. What is more interesting is the significant increase in performance from the 3870K to the 5800K. We are looking at nearly a 25% increase, and overall smoothness and playability was increased dramatically.
I selected the High preset and put post processing to medium. 4X AA was enabled as well. The built in benchmark was used.
Yet again the Intel 3000 is chunking. Hard. We see a very solid 20% increase in performance for the 5800K over the 3870K. I am still amazed that you can get this kind of performance and quality out of an integrated part.
Settings were placed at Medium presets with a 1280×960 resolution. I did a manual runthrough of the first mission up until bad people start really shooting at me. I don’t like it when people shoot at me. Much.
We again see a nice a bump in performance going from the 3870K to 5800K. Overall smoothness in the title itself is much improved. It isn’t buttery smooth, but it is more than playable in single player mode. The Intel part continued to crash throughout testing and we were unable to get any data out of this run.
The new hotness got its moment in the sun for this review. I hand tuned the settings with AO on, DoF off, AF off, far view distance, and FXAA off. I ran around the first mission and killed some poor creatures, all the while capturing it with FRAPS.
We see a little less than 20% scaling, but with a brand new title on the market it is not surprising. We do see a lot more performance, and gameplay was much smoother. The 5800K does seem to deliver on promises about how much better the graphics are for this particular product. The Intel i3-2105 also continued to crash throughout testing, so we were unable to get any results.
The A10 was very efficient at idle, consuming only 44 watts at the wall (this was the entire system). Llano did a little worse, but not bad. The older Intel model had much worse idle performance, but anything under 60 watts is really good. Once we get into Load then we see that the A10 is the hotter processor. The Llano is quite a bit more efficient, but it is also a lot slower. The Intel i3… that is very efficient, but when we look at performance we see that it is lacking.
The graphics performance of Trinity is second to none when we are looking at that particular piece. But that is not the whole of the product. How does Piledriver perform? Are the improvements at the full 3.8 GHz enough to get it over the hump and beat Intel at that price point? Do GPGPU type applications really see a speed up in this product? These are all good questions, and they go unanswered with this preview. You will need to read what will come next week to see the entire product. We showed you the good parts, but the warts are going to be exposed. Perfect product? Not really, but at least as far as we can see a step in the right direction. Stay tuned for next week as we show off the full monty.
Uhh, you got a credibility
Uhh, you got a credibility problem here.
I really don’t see the
I really don’t see the problem here. companys do that all the time. someone is just cranky. hence why im not on that particular sight and here reading a preview of trinity.
I want to read it, too. I
I want to read it, too. I just think pcper could have done better to inform readers of the situation so that we can judge ourselves.
Btw, I am not saying you
Btw, I am not saying you can’t publish this preview. But it’s better to make it absolutely clear that this preview is limited by what AMD allows you to do. At least, please let people know what’s going on. Anand dedicated a full section to explain how their preview is limited by NDA.
Fair feedback, but knocking
Fair feedback, but knocking our credibility is kind of a crock. Limited releases of information happen all the time either by hardware hold backs, driver hold backs, etc. this just happen to be via email agreement.
Updating to point that out should be done though.
I apologize for my original
I apologize for my original offensive wording. I followed you for quite a long time now, and I was quite disappointed to see that you didn’t mention AMD’s NDA at all. It will make your article look like a deliberate choice to please AMD, instead of a limited preview due to legal reasons.
Btw, I am aware of the last paragraph, but I can’t tell if you are willingfully postponing it, or being forced to do so. That makes a difference in my opinion, and thus my first comment is born. Thank you for the clarification.
Just added an editor’s note
Just added an editor's note at the top.
Thanks, that helps. I’d
Thanks, that helps. I’d still say that NVIDIA sending 660’s for review and not 650’s isn’t the same thing. The same thing would be saying you can only test the 660’s using the following games, BL2, BF3 and you can release full benchmarks in a few weeks. Am I missing something?
Still good to see information I guess. Just looking at it with a HUGE asterix next to it 🙂
Thank you. I agree data is
Thank you. I agree data is data, and the updated note should solve any potential confusion of the situation.
Between your stance on
Between your stance on the 7950 ‘now with boost’ and your stance on this issue, I’m suddenly noticing that pcper is starting to go down in quality. And I don’t blame your guys Ryan, I blame you because they all march to the beat of your drum. Maybe it’s been too long since you we’re just a regular guy with a blog, and you’ve forgotten what it’s like not to be a fancy always-traveling super-journalist.
You run your own successful business and I have enormous respect for that. You’ve been producing very interesting, high quality and often very original content for quite a while now. Your podcast is just about the best one out there. But you’re risking it in the long term if you don’t remember who your customers are. We are your customers Ryan, your readers, listeners and fans.
For better or for worse, when
For better or for worse, when I was approached to do this preview, I really did not hesitate to do it. It was not a money grab for me. I really didn't worry about page views either. My first thought was, "Hey, this is cool that we can show what the integrated graphics portion looks like in games, which is what people are really curious about!" The enthusiast in me jumped for it. We get to show off a major piece of the puzzle that is Trinity.
Throughout writing, I made sure that it was clearly labeled "preview". I even put in the somewhat sarcastic "graciously" part, though I probably should have been a bit more obvious with how presenting it. We figured that AMD was attempting steer reviews into showing the strength of the product rather than dwelling on the weaknesses. While some may look at this and say, "Oh, you clever AMD…" I think that what they did here is shortsighted. The ethics here are really not all that bent. I really don't have a problem with what AMD had, because we as journalists have a job to get to the bottom of any product. Once these puppies are out in the wild, and we have the opportunity to report fully on this part, it will be done with an eye on testing every aspect of the product. The bomb for AMD is that people see the preview, they get rather excited about what else this mysterious little bit of technology can offer, and they get their hopes up. Once the full review hits and people see all aspects of performance, I think the backlash is going to be much more severe because of the greater expectations that were lead on by those GPU results.
I think AMD will certainly live and learn from this example. But don't blame Ryan, or think that he tarnished his integrity. We wanted to get readers (and consumers) information about this product ASAP. If a user wants an APU and wants to be able to play these games on them, then they know what kind of performance they will get. If a user wants something that performs well all around, they have to wait a few days to see that, as this preview did nothing to satisfy them. All questions by these folks will then be answered, and they will have the ability to decide for themselves what is best for their usage models.
Look at it from our point of
Look at it from our point of view. We have a super annoying Sleeping Dogs ad on this site that’s way too easy to accidentally click on, and it’s now 2 occasions in the last few months where AMD showed its customers the middle finger and the official pcper stance has been: hey, this kind of thing happens all the time, it’s not that bad.
For what it’s worth, I believe that you believe everything you said. And I honestly appreciate the sentiment.
Pcper should have stood in solidarity with techreport on this issue. If all the noteworthy tech sites did that, we could collectively slap AMD across the face like a naughty puppy.
As an aside, I want you to consider this. There is not enough money at AMD to be spent on serious R&D, or hell, even on their driver engineers, but they can pay a bunch of marketing assholes to act like intel in 2004.
This is an interesting
This is an interesting situation. While pre-release’s do happen, this does seem like an attempt by AMD to use “first day” news to get all the kudos they can. I’m not sure I agree with the “happens all the time” though Ryan. If a company has hardware holdbacks or the product really is a “preview”, the company tells you that and you call it out.
What is AMD’s reason for waiting until some time in October to release the rest of the information? Is there some missing hardware, new code, driver, did they forget a pin? Seems to be just a timing thing which is what makes it smell funny.
Without being on the inside, I can only form opinions based on what I have access to. I will say I don’t think it changes your credibility but I would agree that a blog post or something about the situation and your take on it and why you are posting “preview” info would be helpful for your readers..
my two cents 🙂
i really hope you talk about
i really hope you talk about this on your next podcast!
i somewhat understand the pros and cons of this, but it seems like tech reports stance on this is a little backwards.
where was the data? instead i got something i didnt wanna read about and ended up going elsewhere.
i understand Amd and tech websites wanna generate buzz and views/hits and inform their readers but TR did it in the wrong way and made something out of nothing.
data is data and people wanna know
hence why im here
Ooooh, how exciting! Our
Ooooh, how exciting! Our first big controversy of the Fall season! Truth be told, I was not entirely pleased with the preview, because I had not gotten all the data I wanted. However, the first actual results are interesting and people want to know how it performs. We have a limited view of performance (to the GPU), but it is a start. Hopefully it will cause people to lick their lips and want to dig in more. That is what the full release will be about. Trinity is an interesting processor, but what great novel have you ever read without both agony and ecstasy involved?
AMD explained a few times why
AMD explained a few times why they haven’t said much about Trinity. They have been trying to sell all of the Llanos they can before Trinity comes out. There have been trinity desktop chips being sold by certain OEM’s for some time so there should already be plenty of information floating around about them. If readers are really concerned about what Trinity can do, you can look it up from a few months ago. This is just publicity. All of the relevant information was released during the summer.
two questions in the spirit
two questions in the spirit of fairness and reproducibility:
1) what drivers did you use for the Intel parts?
2) why did you choose to compare against Intel HD 3000 vs the Intel HD 4000 which is available at a similar price point (Core i3-3225)?
3) AMD stipulated that the review had to be about gaming experience. How about a $60 Sandybridge Pentium or Celeron + a $60 discrete card?
I thought there was only two
I thought there was only two questions?
We used the latest Intel graphics drivers for the i3 2105. There was no skimping there.
Intel only provides a certain number of CPUs for any launch, and those are mainly reserved for the biggest/greatest. So, on such short notice, we did not have the best potential candidates for comparison. Budgets are limited, and we cannot purchase every piece of technology ourselves without going flat broke. As such, we did purchase a 4000 series based processor, but it did not arrive in time for testing. Plus, there is that issue of price point for that particular processor.
Results of a $60 CPU plus $60 graphics card will be forthcoming. You can even get a glimpse of performance if you look at GTX 630 based review around the web being run on high end processors…
Thank you for making it known
Thank you for making it known that *everyone* does this. Nvidia and Intel.
I think this whole thing just needs to get blown way out of control so people can finally wake up and realize that all of these reviews come with guides that tell you how you’re supposed to review the product to make it look the best.
AMD has always sucked at this and they’re finally doing what Intel and Nvidia do, and everyone is throwing a pissy fit. If you don’t like what AMD is doing, bitch at everyone instead of just picking on AMD. They guidelines of making this sort of behavior have been established long ago.
Thanks for the post Josh,
Thanks for the post Josh, Trinity is showing some real promise in what you have shown us so far, looking forward to your full review with positive anticipation. Do you know when we can expect to see some AMD FM2 socket motherboards??
While FM2 boards are not
While FM2 boards are not available at retail yet, most of the mobo guys have FM2 parts listed on their websites. We will probably see some models leak out this weekend in anticipation of the early October release.
Well I understand that you
Well I understand that you guys are just humoring them with this post. I don’t think it undermines ANY of PcPers credibility.
Just a bunch of whiners on here today i guess.
I understand that you don’t want to piss off someone who gives you guys parts to review. Its obviously the case with anybody else who posts a similar review.
I also believe that you are being truthful, and appreciate any information that you guys are allowed to give us.
I just read a supposed copy of AMDs e-mail.
Some if it kinda made me mad, how they try and manipulate you guys with the clever wording.
“We are allowing limited previews of the embargoed information to generate additional traffic for your site”
Somebody kick that guy in the nuts, that’s just plain rude.
Are reviews referred to in this fashion often?
As a dedicated reader, I want to know about it.
“We believe there are an infinite number of interesting angles available for these preview articles within this framework.”
I guess you guys wouldn’t know what to write, apparently they dont think you guys do this for a living?
“In previewing x86 applications, without providing hard numbers until October, we are hoping that you will be able to convey what is most important to the end-user which is what the experience of using the system is like. As one of the foremost evaluators of technology, you are in a unique position to draw educated comparisons and conclusions based on real-world experience with the platform.”
Last i checked, anybody that gives a crap about new processors are the same people that want hard numbers. No?
Any normal users I know dont even know there are 2 processor manufacturers, they definitely dont care about benchmarks.
Well after reading that e-mail, i would really appreciate it if you guys would bring up and openly talk about these kinds of things.
I don’t like that you guys are referred to like that, and i don’t like that readers are referred to like that either.
That email just lost AMD a customer.
I’m sure that Intel and NVIDIA aren’t any better, but AMD isn’t in any position to be cocky about anything.
“”We believe there are an
""We believe there are an infinite number of interesting angles available for these preview articles within this framework."
I guess you guys wouldn't know what to write, apparently they dont think you guys do this for a living?"
lol, good point :p. I can't really comment on whether this is common or not as I don't handle that sort of back end stuff, but we tried to find a good mix of graphical benchmarks to compare against the Intel chip at least. The full review will have a lot more information, and hopefully an Intel chip with HD4000 graphics for sake of comparison.
Don’t get mad, bro. All
Don't get mad, bro. All manufacturers do this. I guess his email sorta directing us didn't even really phase me. All PR people send out review guides, list out "best practices", and in situations like this tell us what we can and can't release. There has obviously been a lot of turnover at AMD, and we are dealing with newer people who may not have experienced the pitfals that Peter obviously walked into. I would not go as far as Scott in publishing the email, but the overall tone and topics covered is really nothing new. I think a lot of PR folks will probably take this to heart and really watch what they post out. It is something of a gentleman's agreement that most correspondence between reviewers and PR people are meant to be kept out of the public spotlight.
Also, we are all big boys here. I have been doing this for some 15+ years now, Ryan has been running this site for 11 or 12 years as well. We have our ways of testing things, and while we listen to the PR guys, at the end of the day we decide how to review a product, and what properties we find most important. We all understand that an uneasy truce that lies between PR and press. PR people have a job to promote their product and help to sell it… so that they continue to find meaningful employment. Press has a job to dig deep into these products and expose the positives and negatives in an even handed manner so that they continue to have an active readership, and continue to find meaningful employment.
Thanks for the replys guys.
Thanks for the replys guys.
I was just venting, sorry about that.
I dont take it back though, i hope the new guy is reading all of this hate from the hardcore, and learns a lesson.
I’m glad to hear that this doesn’t phase you, and i know you are a big boy. However its really shady.
I still would like these kind of back room dealings to be made more public. Weither its before of after it happens. I understand that you sign NDAs. However this stuff needs to be discouraged and looked down upon, for fear of word getting out.
I’m sure NVIDIA encouraged everyone to not do non gaming benchmarks for the 600 series (they seem to be hard to find). Even that is not ok. No, i personally dont care about non gaming benchmarks for GPUs (I game on my cards), but im sure plenty of other people do.
I will be more than happy to buy a competitors part if somebody wants to be a jerk about shaping the press.
Will AM3+ version also be
Will AM3+ version also be released along the FM2 version? Or do we have to wait longer?
Still, so far so good. Can’t wait for full review next week 🙂
AM3+ gets Vishera next month.
AM3+ gets Vishera next month. Apparently AMD will port Steamroller to AM3+ as well, but that is next year and things can change. The latest leak does show that AMD is wanting to consolidate the socket infrastructure for the desktop, so I would imagine that we will see more of a move to FM2 for most enthusiast level parts by next year at the earliest.
Very shady to play along
Very shady to play along under AMDs dictatorship of releasing information. We can all play dumb but we know full well that by allowing only the good benchmarks first and releasing all the poor ones later is no different than making an agreement with AMD to not publish the bad benchmarks at all. We’re all smart enough to know that ‘real’ reviews posted several weeks later aren’t payed attention to. By that point the damage has been done and everyone views the product as some great hardware that it really isn’t.
Shame on AMD.
I will disagree with you
I will disagree with you there. I think in fact AMD's handling of this will backfire on them, but not for the reasons you state. By allowing a preview that contains only the good stuff, most users' curiosity will be piqued. They will want to learn more. They will be hopefull of this new architecture and what it brings. When they see the final product in a few days, I am wondering if that initial positivism will be dashed. And in the end, users will look at it in a much more negative light because their initial impression was so positive, and the swing towards disappointment will be much more severe? I guess we will find out next week!
One aspect I really don’t
One aspect I really don’t like about this is that it puts the reviewer in a no-win situation where you either piss off the manufacturer or piss off your audience. As a viewer of this webpage (Or as a consumer in a consumer based society in general) I think that we, the masses, play an important role in shaping the boundary of ‘what is right’ and what is not. Please don’t confuse our displeasure with disrespect for the work that you do. We understand that if you don’t publish, many other sites will. While I respect TR for their choice I also understand that not every reviewer will feel the same way. That is why as a ‘customer’ of this website AS WELL AS AMD I can act as a positive force to push AMD to behave ethically (as I see it) by making my thoughts feel known in the comment sections of articles such as this.
The only way that we viewers can keep honest review sites running is by standing up for sites like PCPer. An honest PCPer cannot exist if it’s viewer base is not honest itself.
I forgot to mention earlier
I forgot to mention earlier (yes Josh, I’m still mad at AMD)
To whoever is bitching about the vs HD 4000 graphics.
Is just looking for things to whine about.
Seriously who needs a benchmark to show them that Intel’s GPUs are still bad?
Ill give you a sneak preview right now of the outcome of that test. Intel graphics still suck. Shocking, i know.
Who would have thunk it?
When Intel actually cares about supporting their GPUs then benchmarks will actually matter.
LOL, what Intel drivers did you use.
I’m gunna go out on a limb here and say that there hasn’t been a new driver since launch.
Anyway, i’m excited to read the full review!
And you would be wrong.
And you would be wrong. Sandybridge launched Jan2011. There were at least 5 driver updates last year and several this year including drivers that support Sandybridge and Ivybridge on Win7 and Win8 in July; expect another update in the next couple weeks for win8 launch.
Hd4000 may not be as fast an iGpu as these top two desktop trinity skus, but at least it wouldn’t be a lopsided/unfair comparison. Can you imagine the howling if someone did a review comparing intel hd4000 against an AMD 785 igp?
Would you say it’s similar or
Would you say it’s similar or identical to comparing intel CPUs at twice, or more, the price range of AMD APUs in a review, knowing that APUs aren’t going to be great in x86 performance?
If you’re interested in graphics performance, maybe reviews like legitreview did with an i7-3920XM, which introed for over 1000 dollars, vs an A10-4600M more to your liking?
Did you know that intel compiled benchmarks run better on intel systems?
Can you imagine the howling?
We have these in stock at
We have these in stock at Micro Center. No idea when compatible boards will arrive..
To be clear.. They don’t
To be clear.. They don’t exist in our internal or external sites, but I’ve had these in my hands.
To be clear.. They don’t
To be clear.. They don’t exist in our internal or external sites, but I’ve had these in my hands.
there is a lot of good info
there is a lot of good info in there … will be building a new windows 8 machine soon and was curious about the FM2 CPUs and what GPU they would have built in … i use my machine for writing, a bit of web design, and some occasional WoW … looks like A6 is my best bet … i try not to give Intel any money if I can.