Hot Trinity CES Information

damn nifty trick by AMD. rumored numbers sound good though. lets hope it actually performs close to what they are claiming. 17w mobile processor with the same performance as the higher wattage liano's is going to be sweet when it comes to notebooks.
 
Let's hope they pull those performance numbers off. It'll also give us a sneak peak of what Piledriver can bring to the table.
 
Specifications of the Trinity APUs have not yet been released by AMD; however, if this leak holds true the Trinity APUs will have either two or four Piledriver CPU cores and TDP (thermal design power) of 65 W, 100 W, and 125 Watts (depending on particular chip). Clock speeds will further vary between 2.2 and 3.8 GHz at stock speeds (will run a bit faster with Turbo Core 3.0).
Click to expand...

AMD has stated Trinity will deliver a 25% increase in CPU performance and a 50% increase in GPU performance.
Click to expand...

God I hate marketing clowns. Are they saying Piledriver cores will be 25% faster than Stars at the same frequency? 25% faster at what?

Cool demo. Looks like a good product.
 
I quit believing in AMD performance estimates around when BD came out. Take them with a grain of salt or simply discard them altogether.

http://vr-zone.com/articles/amd-demonstrates-17w-trinity-gcn-hd-7000m-gpu/14502.html

When powering the 3 displays off the CPU alone it only reached around 60% off a 17W TDP chip. That's quite amazing. I'm really looking forward to Trinity, despite dumping the Stars architecture and adapting the new, and hopefully improved, BD design. It looks as if they've got some things figured out but it also seems they're looking forward to a first half of '12 release rather than Q1. If it performs as they say, I'd love a lappy or ultrabook powered by one of these :)
 
Also check out the article on HH showing the laptop reveal. All that action going on and it seems pretty fluid. I wonder if that is the 17W part or if it's higher TDP. If that's 17 that's extremely impressive.

Only one more thing I want to know: will the Trinity platform support Intel's advanced steering wheel interface? :D
 
I'm kinda disappointed in seeing more 100W & 125W APUs.

I'm actually more interested in Trinity's graphics performance, even more so from what that demo showed.
 
JMccovery said:
I'm kinda disappointed in seeing more 100W & 125W APUs.

I'm actually more interested in Trinity's graphics performance, even more so from what that demo showed.
Click to expand...

Honestly I'm not all that surprised. They want to give an option for high-end graphics performance, and having 125w parts allow much more flexibility and options than just having 65w parts.
 
JMccovery said:
I'm kinda disappointed in seeing more 100W & 125W APUs.

I'm actually more interested in Trinity's graphics performance, even more so from what that demo showed.
Click to expand...

why are you disapointed in seeing 100w/125w chips. hell i'm glad they have them. that means there will still be desktop performance processors and a usable gpu for more then browsing the web and playing video's of some one playing a game(yeah i had to put that in there, just to funny not to). right now the only thing the desktop market has is intel and their pos HD2000 and HD3000 igp's.
 
DeathFromBelow said:
God I hate marketing clowns. Are they saying Piledriver cores will be 25% faster than Stars at the same frequency? 25% faster at what?

Cool demo. Looks like a good product.
Click to expand...

25% faster then Llano at same TDP. Check the slides at the bottom.

Silent.Sin said:
Also check out the article on HH showing the laptop reveal. All that action going on and it seems pretty fluid. I wonder if that is the 17W part or if it's higher TDP. If that's 17 that's extremely impressive.
Click to expand...

It's the 17W part. Check the VR-Zone link. They do say whether it is quad or dual, but if you watch the HH video, the AMD guy says quad.
 
sirmonkey1985 said:
why are you disapointed in seeing 100w/125w chips. hell i'm glad they have them. that means there will still be desktop performance processors and a usable gpu for more then browsing the web and playing video's of some one playing a game(yeah i had to put that in there, just to funny not to). right now the only thing the desktop market has is intel and their pos HD2000 and HD3000 igp's.
Click to expand...

Well, not really disappointed, more like a 'meh'. Why? I would rather take a 6 or 8 core with graphics similar to the 35W mobile parts in a 125W TDP package rather than a higher-clocked quad with similar graphics to the 65W parts, just clocked higher.

It is like with the 2500(K)/2600(K), how many people that use the power of a higher-binned chip, make extended use of the IGP? If AMD came up with an 'Optimus'-like technology for desktops, where you connect the displays to just the motherboard's connections, and can make use of one or more discrete GPUs, I would be first in line for a higher TDP Trinity part.
 
Should have made that clear in your original post, otherwise people misunderstand (like me).

If AMD did do that, combined with the HD79xx zero idle technology or whatever that is called, that would be awesome for saving power in crossfire rigs. Although, the 79xx already use such little power, and people who are buying several of them aren't likely to be worried about power consumption anyways.
 
mastrdrver said:
It's the 17W part. Check the VR-Zone link. They do say whether it is quad or dual, but if you watch the HH video, the AMD guy says quad.
Click to expand...

I've only seen original articles from HH and vr-zone mention it was 17W and vr-zone's about as credible as the national inquirer so that's not much to go off of. It would have been easy to misconstrue AMD saying Trinity will go down to 17W in ultra thin platforms, not that the demo was running on a 17W part.

The other problem is I wouldn't trust any of the AMD marketeers at the moment considering the recent track record and company reorg's (BD is a 2 billion transistor chip huh?). Here's a video from AMD that implies the demo was done on the mainstream notebook (35W) platform and not the ultrathin parts which will scale to 17W on a BGA package: http://www.youtube.com/watch?v=agJxehoSBmY&feature=youtu.be

The S|A forum is diving into all the semantics about these videos too: http://www.semiaccurate.com/forums/showthread.php?t=5997&page=2

In the end it doesn't much matter, still an impressive demo but I wish companies were more forthcoming with specs at these events.
 
trinity looks good, particularly the 17W version.

anyone know yet how many shaders?

with VLIW4, and die-shots seeming to show six clusters, logic would argue for 384.

but die shots are never unadulterated, and there could easily be two more clusters offset from what appears to be the main stack.........
 
DeathFromBelow said:
God I hate marketing clowns. Are they saying Piledriver cores will be 25% faster than Stars at the same frequency? 25% faster at what?

Cool demo. Looks like a good product.
Click to expand...
AFAIK, 25% faster than current LLano parts. Not sure if they are comparing the chips at the same frequency to get that 25 % number or if they are getting a bit "creative" with it. Sorry if I wasn't more clear in stating they meant those numbers to be compared to Llano (though, again, not sure /how/ they compared them to llano parts or which llano APUs).

EDIT: article has been updated for clarity.
 
wolfofone said:
AFAIK, 25% faster than current LLano parts. Not sure if they are comparing the chips at the same frequency to get that 25 % number or if they are getting a bit "creative" with it. Sorry if I wasn't more clear in stating they meant those numbers to be compared to Llano (though, again, not sure /how/ they compared them to llano parts or which llano APUs).

EDIT: article has been updated for clarity.
Click to expand...

i believe they are clocked higher. clock for clock i don't think piledriver will be faster. but i believe the advantage of piledriver is that its more power efficient which allows them to increase the clocks while using the same amount of power thus gaining the 25%.

thats just based on the clock rumors for trinity. but who know's, like you said theres nothing really showing how/what they are using to compare. for all we know its 25% faster at multi-threaded apps but is on par with current llano when it comes to single threaded apps.
 
sirmonkey1985 said:
i believe they are clocked higher. clock for clock i don't think piledriver will be faster. but i believe the advantage of piledriver is that its more power efficient which allows them to increase the clocks while using the same amount of power thus gaining the 25%.

thats just based on the clock rumors for trinity. but who know's, like you said theres nothing really showing how/what they are using to compare. for all we know its 25% faster at multi-threaded apps but is on par with current llano when it comes to single threaded apps.
Click to expand...

Yeah, that's a good point. Once reviewers get their hands on them, I'm sure we'll know more about how much faster Trinity is :)
 
As gimped at bulldozer was.... I would expect at 15-18% performance increase over bulldozer. And a 25% performance increase compared to Llano in VERY few instances.
 
sirmonkey1985 said:
i believe they are clocked higher. clock for clock i don't think piledriver will be faster. but i believe the advantage of piledriver is that its more power efficient which allows them to increase the clocks while using the same amount of power thus gaining the 25%.

thats just based on the clock rumors for trinity. but who know's, like you said theres nothing really showing how/what they are using to compare. for all we know its 25% faster at multi-threaded apps but is on par with current llano when it comes to single threaded apps.
Click to expand...

I really hope that's not the case. Any time I hear chipmakers chasing clock speed I instinctively quiver. Bumping clock speed and ignoring the cache latency is a recipe for disaster. Thankfully it seems they've at least addressed the L1 size being too small

amd_15h_family_features.jpg


http://www.xbitlabs.com/news/cpu/di...ure_Peculiarities_to_Software_Developers.html

Not too hopeful, though. They've got a long road ahead of them and a massive 50% disadvantage in single-threaded performance. Opting to cure that with clock speed isn't going to work
 
R3MF said:
what have they done to the L1 cache?
Click to expand...

Based on the chart, they changed the size of L1 data TLB in the original Bulldozer design from 32 entries to 64 entries in the Piledriver cores. If that chart is accurate and the numbers mean what I think it means, they're changing the size from 32 kb to 64 kb.

Some people speculated that part of the poor performance (note, only part of the reason) is due to the really small L1 cache size. Another reason would be the much slower cache as compared to Sandy Bridge and other Intel architectures.
 
I thought I read BD's L2 was about as slow as SB's L3, which sounds pretty crappy to me.
 
trinity looks good. so far. I absolutely love my llano labtop.

I guess the real question with trinity is will it's "piledriver" cores make a large difference in cpu power. My guess based on current information with clock and scaling with bulldozer. Is trinity will be faster Via clock speed, not necessarily IPC. Might be a combination of both.

Either way, i will take look at them when they come out. I wouldn't mind spending 400-500$ on another labtop if indeed it had 50% more graphics performance,better battery life, and faster cpu.
 
jober1 said:
I thought I read BD's L2 was about as slow as SB's L3, which sounds pretty crappy to me.
Click to expand...

BD's L2 is nearly twice as slow as SB's L2. L3 is also significantly slower. In fact, BD's cache, L1 L2 and L3, were all slower than the Phenom/Thubans
 
what exactly is a "data tlb" ? 32 entries for a TLB seems quite small, unless they were constrained by high latencies.
 
Not sure, but it looks like they're doubling it with Piledriver cores.
 
You must log in or register to reply here.
Back
Top