Trinity in the Flesh...or Silicon

dr.kevin said:
who here reads korean ?

what's this saying ? especially that green man figure showing cpu and gpu of equal size ??

trinity_a10.jpg
Click to expand...

The 2 green guys are going out. :p - I think it signifies a CPU and GPU having a bromance and always working together.
As for the 2 blue guys. It looks like an adult has made a special friendship with a young man.
 
DJ Big T said:
Nice, looks to be a winner - big time.
Click to expand...

not really. win some, lose some.

that cinebench score is lower than the current A4-3310mx, which is clocked at 2.1ghz.


also, that 3dmark11 score is such bullshit. If only they could make their drivers work so that hybrid crossfire could net 70% graphics boost.
90% of games result in lower performance in hybrid xfire.
 
also, that 3dmark11 score is such bullshit. If only they could make their drivers work so that hybrid crossfire could net 70% graphics boost.
90% of games result in lower performance in hybrid xfire.
Click to expand...

I dunno where you get your information from, but first its not hybrid crossfire its called dual graphics. Dual graphics work when you combine a compatible discrete Gpu with The Amd APU. This setup is just like crossfire, except it only works in Direct X 11. Direct 9, and 10 games see no benefit from Dual graphics.

The 3dmark11 score is decent for the standalone notebook platform APU. The graph is of the a10-4600m, so not the top model.

Increases the CPU-NB clock to 3+GHz (if the L3 clock is still tied to the NB clock), or, tie the L3 clock to the 'core' clock to increase bandwidth
Click to expand...

A higher north bridge clock is not needed for more performance. The L3 is plenty fast, it just has horrible latency. Bulldozer works differently than Thuban or Phenom 2 and does not benefit from increasing the north bridge clocks.
 
ebduncan said:
I dunno where you get your information from, but first its not hybrid crossfire its called dual graphics.

The 3dmark11 score is decent for the standalone notebook platform APU. The graph is of the a10-4600m, so not the top model.

.
Click to expand...

same shit, different name. It used to be called hybrid crossfirex when pairing igp + dgpu

and the a10-4600 will be the top mobile model as far as i can see.
 
dr.kevin said:
same shit, different name. It used to be called hybrid crossfirex when pairing igp + dgpu

and the a10-4600 will be the top mobile model as far as i can see.
Click to expand...

I haven't really seen a list of there mobile parts but I assume at some point they'll release their "mx" branding and will continue on with that being the top mobile part, just like they're continuing with the K on the desktop version. Could be wrong and/or the "mx" might not make initial launch.
 
darkstarcow said:
I broke down and bought an A8 3870, depending on how trinity performs I will sell this in a couple months and upgrade.
Click to expand...

What MB did you pair the A8-3870 with? Most importantly how do you feel it compares to the 955 @ 4.0 ? I have a 1055 that I usually run @ 3.6 and have been thinking about dedicating that rig to VM's and building a new one just for daily driver.
 
I have an A8-3870k and was mostly disappointed. My disappointment comes specifically from the price tag. I bought it at release for $140 and it replaced an AMD 631 llano. the 631 would run 3.9ghz stable, this 3870k only runs 3.6ghz. I do like the on chip graphics overclock very nice and performs great. I currently run it at 800mhz for bitcoin mining and it adds an additional 80mh/sec roughly to my overall total while only consuming about 30 watts. I like that the CPU helps pull some weight in my mining rig, but its been really hard to recommend this CPU for people.

AMD 631 + used 6770 is almost the same price and a lot more performance for gaming.

Now with the price cuts, I think the a8-3870 is a really interested HTPC chip. Its powerful enough to play most new games on max settings 720p or older games like world of warcraft / call of duty MW2+ at 1080p.
 
dr.kevin said:
there better be some overclocking reviews for laptops
Click to expand...

Mmmmhmmm. If somebody reviews the mobile version without underclocks/overclocks then they missed a whole lot of Llano K10stat fun. The Llanos were amazing overclockers and I'm hoping Trinity chips will follow along those lines.
 
Just jumping in here, looking for news on reviews.

Anyone see any hints from the powers that be about having product in hand? Or is this just gonna be a paper launch type thing with product weeks down the road?
 
So far, it's looking like a mobile hardware launch only, with the desktop variety in August.
 
Battery life looks fine for me. I'd take one if the price is right.

GPU slightly slower than Intel HD 4000 graphics in one or two games. But let's be fair to AMD - they've got better drivers than at least Intel. Usually faster by a fair margin, but still could have expected more from the GPU.

Not mind-blowing, but overall a nice product if the manufacturers do their job and price it right. I don't believe in discrete GPUs for laptops.
 
I was expecting way more. Some games beat Llano by only 3%? And in some it was even beat by the 4000? Sad day.
 
Lower power + higher performance = higher price?

Needs more reviews to compare .... wonder if those are canned benchmarks they running...
Lacks that touch of [H],....
 
It's still a decent showing from their BD arch at least. There is some hope for AMD's CPU's, but I'm a tad disappointed myself, doesn't get anywhere near what they were claiming.
 
someone needs to run k10stat overclock, if it even works, and post results off the record on a message board somewhere.


A10 meets or beats Llano.
so at least it's an improvement. Single thread performance is faster when it has the fpu all to itself.
but then it lags in multicore.

trinity posts 15% faster single thread performance while running at 43% higher clock than llano A8-3520m.
bulldozer/piledriver IPC is still a bitch.

in terms of performance, this is like Intel tick+.

casual gamers will definitely enjoy this though, and AMD is surely betting on that.
Since college kids don't sit around all day running synthetic benchmarks, they will eat this up if AMD markets this correctly this summer.
 
Last edited:
jober1 said:
hmmm, that is much less impressive when phrased like that.
Click to expand...

trinity posts 15% faster single thread performance while running at 43% higher clock than llano A8-3520m AND gets longer battery life

A little extra context helps :p
 
dr.kevin said:


casual gamers will definitely enjoy this though, and AMD is surely betting on that.
Since college kids don't sit around all day running synthetic benchmarks, they will eat this up if AMD markets this correctly this summer.
Click to expand...


I tend to agree. I'm potentially in the market for two notebooks this summer and will consider a trinity notebook if they come in cheaper and are better quality (e.g. better screen, keyboard etc).
 
Im still excited for trinity for the improvement in the GPU. I really hope k10 stat still works... i have my a6-3400 overclocked to 2.4ghz and could go higher if i dealt with the heat.
 
sublimestuff said:
Tom's review has anti-aliasing benchmarks (but no HD4000, oddly). Pulls ahead of A8 by 40% in Skyrim with good visual settings. But in other games, mmmmm, its barely faster than Llano.
Click to expand...

This is probably driver related. It fairs really well in DX11 games with the exception of Batman + Metro2033, the latter which AMD generally does well in even with the Bulldozer/Piledriver CMT design. Considering integer performance is up per-core compared to Llano and the synthetic benchmarks show Trinity being a hefty bump over Llano I'd say the unimpressive DX11 gaming benchmarks are subject to change.

Edit - The Tom's benchmarks were run on DX9 because they were comparing the HD3000 which can't do DX11. The HD4000 comes closer and in some games beats the 7660G Trinity but I'm guessing this is likely due to driver issues. The only HD4000 Ivy chips that are available atm are the i7 ones and they are considerably more expensive so that's why they were omitted in the review. The VLIW4 architecture fairs far better in DX11 than it does in DX10 or DX9.
 
Last edited:
i am impressed however the sooner amd can get its post bulldozer arch out the door the better it will be for them
 
Hmm, maybe I will wait to update my HTPC in November/December time frame instead of now. Well, I will at least do something about the my storage problems. Too bad for AMD, but I am going to up date to Ivy Bridge on my gaming system in June/July. I will give them the time on the HTPC though.
 
Well I'm at work on my phone so its hard to peruse all the reviews but it looks like Trinity is 15-20% faster than Llano but 15-20% slower than Sandy Bridge which would likely translate to 20-25% slower than Ivy Bridge. That about right? Kinda disappointing if so. I was hoping to see AMD a little closer to 10% behind Intel.
 
Disappointing. Really looks like AMD rolled the wrong pair of dice with the BD arch.
 
http://www.anandtech.com/bench/Product/438?vs=600

^^ that really shows just how good Trinity is. It performs slightly better than a GT540m discrete card and does fairly well in CPU benchmarks as well. That's awesome, really. The only thing that can rain on their parade would be some Ivy i3's with HD4000 coming out soon and at equal or lower price ranges.
 
pelo said:
http://www.anandtech.com/bench/Product/438?vs=600

^^ that really shows just how good Trinity is. It performs slightly better than a GT540m discrete card and does fairly well in CPU benchmarks as well. That's awesome, really. The only thing that can rain on their parade would be some Ivy i3's with HD4000 coming out soon and at equal or lower price ranges.
Click to expand...

Shows that the GPU is strong, but from a CPU perspective still shows that at the same clock speed Sandy Bridge is over 50% faster, so it looks like the Trinity architecture has done little to nothing to improve the CPU performance of Bulldozer. :(
 
Sure it has, granted it looks to be mostly done via clock speed gains. I don't think they gained 10% IPC in most workloads. It's probably in the ~5% range? Either way the performance increased via clock speeds and the perf-per-watt increased dramatically. So much so that Trinity actually consumes less power than Ivy Bridge does at 22nm despite those crazy clock speeds... It's almost on-par with mobile Sandy Bridge as far as power consumption goes. Those are great signs, imo. Granted, ~10% IPC bump would have been ideal but Trinity is already at a disadvantage because it lacks L3 cache so in gaming that would roughly equal about 10% performance gain.

here
http://forums.anandtech.com/showthread.php?t=2245809

At first glance it can look somewhat unimpressive, but it's actually very good. I'd expect the games where it loses to the HD4000 to turn around in AMD's favor with drivers and the games where it does lead, which is all but 2, to look even better.

I think you're expecting SB-levels of single-threaded performance which was never going to happen. At least now the single-threaded workloads are respectable and actually improve upon Llano/Stars which BD failed to do. They're still off, by 20-30%? But considering the gap was 40-50% I think that's quite the achievement. Vishera won't have an on-die GPU and will likely clock very high, but how much that resonant clock mesh tech has helped Trinity and how much it will help Vishera is another matter. The clock mesh tech decreases power consumption by 10% or increases clock speeds by 10% at equal TDP. Those gains diminish as the clocks go past 4ghz.

err, in some places it's actually very good IPC gains. First pass x264 is a single-threaded benchmark so let's take that as an example:
49 for Trinity
53 for the i5 Sandy

The Trinity APU clocks up to 3.2ghz and the i5 clocks up to 2.9ghz, both include turbo.
Trinity gets roughly 1.53FPS per-mhz
the i5 Sandy gets roughly 1.83FPS per-mhz

That's only a 16% IPC deficit in that workload.

Here are the numbers from BD vs. Sandy in the same workload:
41696.png

i5 2500K at 3.7ghz w/ turbo
8150 at 4.2ghz w/ turbo

2.7FPS per-mhz for the 2500K
1.8FPS per-mhz for BD

That's roughly a third of the IPC deficit for that workload when comparing the 2500K to the BD, meaning AMD gained ~half of that in IPC and not just clock speeds.


Cinebench is only ~30% now from ~50% between an i5 Sandy and BD with roughly the same clock speed gap (2600K vs. 8150 Cinebench 11.5 single-threaded). It was never meant to catch up to Sandy with IPC but it looks like they may have gained a good bit of that 10% back (and then some in certain workloads), but it does bode well for Vishera as far as power consumption goes and at least showing some respectable performance, granted a year too late.

It won't be a sandy/IB killer but if they price it well nobody will complain about the power consumption, the heat and the performance won't be too bad either :p
 
Last edited:
I am not exactly going to try and make IPC comparisons between a mobile chip and a desktop chip.

Llano never was a cpu speedster, and it sold well, thanks to its good graphics. Trinity looks to be both faster in the GPU and CPU sides of things. Trinity's power draw is lower than of ivy bridge, is awesome esp when your comparing 32nm to 22nm.

I don't see how this chip is a fail? They are bringing discrete class mobile graphics to a embedded solution. Meaning to get similar gaming performance from a Ivy Bridge laptop they will need an expensive Gpu to keep up, no longer just a cheap mobile discrete card.

Cpu performance, well Amd LLano lacked severely. Most benchmarks show Trinity to be well over 20% faster than Llano. Granted Ivy Bridge is only 5% faster than sandy bridge Amd actually made a decent jump in cpu performance verses intel. Its not enough to make it faster than an I7 but it beats the i5's.

Either way you slice the bread, they are impressive. Amd made good strides in performance vs their competition in the markets they wish to compete in.

Pelo- interesting math, I won't dive into the math to figure out IPC improvements, but its certain that AMD made at least a 5% jump. Its hard to make a good comparison when one cpu is clocked differently, and has no l3, while the other has l3 and is clocked higher. They also have different methods of turbo core to further throw a wrench into things.
 
It wasn't a direct comparison between the desktop chips and the mobile ones but rather desktop vs desktop and mobile vs mobile. When you look at it that way you keep things fair and you can more accurately compare IPC in X workload. Either way it shows the IPC increases in those 2 workloads (which will be on every review site) in Piledriver when compared to Bulldozer and it roughly equals Husky mobile. If you take away the on-die GPU constraints and bump up the clock speeds (desktop trinity will reach over 4ghz with turbo and it too wastes 50% of its die area on the GPU) then you can see that they've made some significant improvements to the Bulldozer core.
 
Do you guys think that IPC could improve even further with Vishera since they wont have to be concentrating on energy efficiency and battery life with be desktop chips? Kinda like how Ford can put more power into their truck engines cause fuel efficiency wont be as big a concern as it would be in their small econo class cars? Or would that even affect IPC?
 
There would be some IPC improvements simply due to the presence of the L3 cache. However, AMD claims that they are using Trinity to test the Piledriver cores, and will improve the Piledriver cores used in Vishera, based on their roadmap. How significant the improvement is remains to be seen.
 
You must log in or register to reply here.
Back
Top