http://translate.google.com/translat...y-bulldozer%2F
http://www.imagengratis.org/images/trinity.jpg
http://www.imagengratis.org/images/trinitycore.jpg
Printable View
Translation is messed up. It's translating spanish text from english to spanish. Fix that. Otherwise you have to select ingles and espanol by hand (right the other way its selelcted by default).
fixed up!
This looks suspiciously like the existing Photoshopped Bulldozer die with the L3 part being blacked out and called the GPU.
lol first let them bring out bulldozer forever (play on duke nukem forever) before they start wasting their damn time with trinity.
You surely dont think that.Quote:
Originally Posted by god_43
For trinity to be available next year, they already have to have first silicon(could be not working tho).Planning stages had to be done at least year ago.
Introducing new cpu is a very lengthy process, and thats a new cpu architecture with a recent gpu part.
4 BD cores with a gpu ~6850 in 220mm2 sounds like a good chip to me. although i wonder if they will ever try to implement MCM so they can double everything while still having good yields.
I'm not sure if two modules is enough, sure it'll make a great laptop but on a desktop it might be crippled.
What like Crossfire the GPU at socket level?Quote:
Originally Posted by Manicdan
2B transistor = 400mm^2 die size?
This is not economical for mainstream uses.
considering how many games uses less then 4 cores, and how important clock speeds seem, 4 cores is about the best perf for the price and size for any gamer now, and probably years to come. a 2 core chip is really awesome for gaming considering its low price, but were not really at a clock limit on 2 cores anymore, so its not going to be any faster in specific cases like it use to be.Quote:
Originally Posted by Iconyu
basically. if they make it a smaller chip, then using 2 will make it cover the needs of a desktop without the risk of really crappy yields. where amd is constantly behind is the perf/mm2 of the whole package, so i think they need to focus on that a little more so costs are down and supply can be higherQuote:
Originally Posted by Motiv
i think that BD has been delayed and reworked long enough. they said bd will be here in q2....then bring the :banana::banana::banana::banana:ing thing out in q2! idc what the issues are, if charlie is right and they just moved it back for strategic reasons, then that is unacceptable. its been in development for 7+ :banana::banana::banana::banana:ing years man! i am just tired of trying to understand amd's plight, my next build will be an ivy-bridge system, i am done with amd.Quote:
Originally Posted by XRL8
Those guys are just guessing,it's obvious.
do you know how long it take for intel to make fully monolithe quad with imc onboard? 6 years. do you know what chip i am talking about?Quote:
Originally Posted by god_43
it was bloomfield.
AMD wanted to catch up with intel (not with monolithe quad. with the performance), they just said about BD to early.
Are you sure it'll have a low price? Because it sounds like it'll be the biggest APU on the block.Quote:
Originally Posted by Manicdan
Also loads of games use four cores, almost every new game will use four threads even if frame rates are 'happy' on two cores any hard drive access will cause a hitch. I really doubt many of them will really benefit from having four true FPU/SSE units to abuse, I guess it really depends on the game.
When software starts really abusing an APU in a way a CPU and desecrate card can't match, I'll jump on board, until then I'll just watch and see what developers do with the guaranteed level of GPU power these APU's will provide.
@prznar1: The real difference is that Intel took a few steps introducing new CPU's until Bloomfield, the march is unceasing and Intel are happy to ditch an entire platform to enable new features. I hope Bulldozer can live up to our hopes on this, because the bar is set pretty high.
im not making assumptions on price for this specific chip, i was just saying that a duel core cpu is just fine for many games out there, and ridiculously cheap.Quote:
Originally Posted by Iconyu
also many games that support more than 2 cores, just give the appearance that they do because of the thread scheduler. there are tasks in games that cannot be split into multiple threads and never will be. note how few games out there push 70% cpu usage on a quad, and i dont think this will change until the new consoles come out, and then give it about 2 more years after that. with e3 right around the corner, id like to see a new review with how much of a perf loss there is between the same clocked quad/tri/duel cpu
if i had to pick, i would still go with a fast duel core with a decent gpu in my next laptop. at work we switched from a 2.8ghz C2D to an i7 that turbos up to 2.8ghz, the only thing i noticed was having 1/3 the battery life. these APUs require very strict balancing so that people get just enough of what they need, without draining all the battery power. then theres the transition to opencl and who knows how much the cpu side will matter then. no point having too many cores when your apps never use them.
keep in mind that alot my arguments are based on buying a product to keep for about 3 or more years. for those who upgrade every 18 months or so, then its ok to get what works best now
Relax man.BD project was stopped in the meantime (hector ruiz rulz), then it was started again with different approach, first one was scrapped.The cpu you are talking about, is probably in the pipeline for 2-3 years.And it has beed delayed for 2-3 months.Whoopsie doo, intel had 3 month slip THIS year due to the faulty chipset.And thats intel, they are like F-in BORG.Quote:
Originally Posted by god_43
What im getting at, they have to do multiple projects at once now, while being limited on engineers and money at the same time.
Bobcat
Llano
BD
Cayman
Barts
->trinity etc.
Theire doing :banana::banana::banana::banana:load of things for being relatively small company.And not long ago there was talks of AMD just disappearing from the face of the planet...
http://www.pcmasters.de/fileadmin/ne...r_die_shot.jpg
Just everything below the first two modules is blacked out. The Hd6770 is not VLIW4, it's unknown if they go for VLIW4 already in Trinity. The 220mm^2 guess sounds too optimistic, given the other spec... why did someone's guesses make news? :confused:
Please... this is a pointless thread about Trinity, so why make it more pointless by posting a pointless rant about something unrelated to the topic? If you want to vent about how enthusiasts aren't the top priority for AMD right now, due the situation they are in, do it elsewhere.Quote:
Originally Posted by god_43
One Llano core takes 16mm2, one bulldozer mudule takes 30.9mm2, so a dual module bulldozer may smaller than quad Llano, about 10-15%, including some blank between each of the Llano core.
30.9 mm^2 includes only 1 module and 2MB L2 cache.Quote:
Originally Posted by undone
If you expand to 4MB L2 cache as in post #1, you will need more die space.
And in post #1, only 40 mm^2 of die space is for I/O, I doubt it if you need to include 4 PCI-E links and a dual-channel MC.
i dont think module need 4mb L2, 2mb is enough. As for the I/O part, they should take Llano for reference, not bulldozer.Quote:
Originally Posted by qcmadness
edit: btw if amd replace vliw5 with vliw4, it can do an HD6670(turks. 480sp) on die with almost same space. If amd dont, those space also can turn into cache, but I doubt amd will do it for vliw4 architech, i'd rather think HD7000 would have massive cache, first.
BD has not been in developement for so long at all. AMD had at least two projects going before they started BD. They saw at an early stage that the chips wouldn't do so they started from scratch to avoid a Pentium 4 of their own. K9 and K10 was both scrapped. (No Phenom is not K10)Quote:
Originally Posted by god_43
AMD don't have 10% of the resources Intel has. And in order to survive they have been forced to have several projects running at the same time. Bobcat, Llano and BD is three different large projects, that's three times more than anything they have done before. If people actually bought what was best between 1999 and 2006 AMD might had more R&D capacity today.
If you switch to 2 modules with 2Mb cache per module you will have 2x 30.9 = 61.8mm for the cores and cache in a quadcore. That's still less than 4x 16 = 64mm for Llano. So if you just switch Llano cores with modules you will save space. With a bit more shaders Trinity will possibly be larger than Llano, but the guys who did the calculations in OP post are idiots.Quote:
Originally Posted by qcmadness
who cares about adding graphics to bulldozer? I don't get it. Graphics are a waste of silicon on high end workstation/server chips if anybody asks me.
I don't think Trinity is a high end chip…Quote:
Originally Posted by bamtan2
Graphics are not the only reason for the addition. Read up on AMD's vision of how these APU's will be used in the future.Quote:
Originally Posted by bamtan2
Here is AMD doc. given to investors with allot of APU info .... http://phx.corporate-ir.net/External...xUeXBlPTM=&t=1
Yeah I totally agree, Parallel processing is such a waste of time, alongside OpenCL, adding parallel processing to a cpu... what a waste of time... in a very cynical way...:shakes:Quote:
Originally Posted by bamtan2
Yeah and the 486DX was such a waste of time compared to the 486SX.
People buy what they are comfortable with, if people used the "buy whats best" ideology things might have turned out much worse for AMD in the end.Quote:
Originally Posted by -Boris-
It all comes down to the Pentium. Not the actual product, but the name. Before then it was a bunch of numbers. By the time AMD was something to look at, they had lost the marketing. Most of this technical stuff is just way above the head of the average user. They don't understand MHz, GHz, i7 or any of the other silly numbers that they use. Thats why Apple is so successful.
I am pretty sure they would be far more competitive today had Intel not used some of the practices they did. But I think Intel would still be leading by a healthy margin.