AMD Piledriver rumours ... and expert conjecture

Page 26 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.
We have had several requests for a sticky on AMD's yet to be released Piledriver architecture ... so here it is.

I want to make a few things clear though.

Post a question relevant to the topic, or information about the topic, or it will be deleted.

Post any negative personal comments about another user ... and they will be deleted.

Post flame baiting comments about the blue, red and green team and they will be deleted.

Enjoy ...
 
if its the case
then they are spoiling the awesome phenom 2 series

but if it is a pair of phenom 2 x4 bridged together and in 32nm so as to keep its tdp in limit, then we can compare it to i7-2700 in multithreaded application.
i would love it to be k10.5 based (and if it is at comparable price to a fx 8120 im buying one in an instant), but those rumours of bad yields at glofo indicate that a lot of bad dies are floating around, and this is amd trying to dispose of them
 
i think
i have another thing to share
todays apps are note ready for 8 core (8 core is a future design, it can make some sense in future but is not needed now)
instead of p2 x8 with lower clock a p2 x4 with a higher clock (use 32nm fabrication and more transister to achieve higher clock with low tdp) and more cache will be a hit, and instead of sse5 if they can use something like avx (which is far more better and future proof than sse5) then they can beat any current intel cpu.
 
http://www.xbitlabs.com/news/cpu/display/20111222231004_AMD_May_Be_Preparing_Phenom_II_X8_Microprocessors.html

For real? XD

And trinity wasn't gonna use GCN IIRC; it was gonna use the VLIW4 seen in the 69xx series. Lemme go get the link to confirm that.

Cheers!

EDIT: I've only found rumors and conjectures, bur nothing confirmed about the 7k series in Trinity xP

The only thing we have right now is info on the HD7970, and probably the HD7950. Everything else below it will probably be a die shrink of the current HD6K series and thats how they will do it with Trinity. It will be like the low end HD7Ks but still use the VLIW4 instead of GCN.

The only discrete GPUs I see using GCN as well is the HD79XX series, everything else below will probably still use the VLIW4.

Phenom II X8 according to XbitLabs

http://www.xbitlabs.com/news/cpu/display/20111222231004_AMD_May_Be_Preparing_Phenom_II_X8_Microprocessors.html

the thing that makes me think is the ClockSpeeds 2.8 Gzh max jezz i don't know what to think...maybe they're Failed FX with locked Cores..or as the're saying in AnandTech...Fx with L3 Cache Deactivated.

Would be pretty pointless. And if anything it would be Phenom III, not II since it would be the third arch with the name unless they are actually just shrinking K10.5. But even if I would expect higher clocks from it with a die shrink, unless K10.5 doesn't like 32nm much like K10 wasn't a fan of 65nm.
 
p ahenom 2 x4 would have to clock to >4.5ghz to beat a stock 2500k, if you oc the 2500k the x4 would need to reliably hit >6ghz to be a better overall chip

Well, that's not really true. Using SuperPI as example the Phenom II X4 955 BE overclocked at 7.2GHz takes 9.578 seconds to finish SP 1M while the i5-2500k takes 6.203 seconds only overclocked at 5.9GHz.

In a daily use, the 4.5GHz are impossible on a Phenom II Deneb or Thuban core
 
Well, that's not really true. Using SuperPI as example the Phenom II X4 955 BE overclocked at 7.2GHz takes 9.578 seconds to finish SP 1M while the i5-2500k takes 6.203 seconds only overclocked at 5.9GHz.

In a daily use, the 4.5GHz are impossible on a Phenom II Deneb or Thuban core

well 980 got 4.3 on air so i think it can go beyond 4.5 with water cooling
www.overclockersclub.com/reviews/phenom_2_x4_980/3.htm
and they can hit more than 4.6 (but stablity is not confirmed), your own tables contains the proof
 
well 980 got 4.3 on air so i think it can go beyond 4.5 with water cooling
www.overclockersclub.com/reviews/phenom_2_x4_980/3.htm
and they can hit more than 4.6 (but stablity is not confirmed), your own tables contains the proof

Try to get those 4.5GHz and let me know if is successfully.


i think toms is the first site to discuss that p2 x8

That sounds more interesting, the Phenom II line is very solid for budget rigs, even, I still have a X6 1090T working very good on my house.
 
Try to get those 4.5GHz and let me know if is successfully.

unfortunately i cann't do that, because i am tight on budget and cannot afford that 250doller upgrade.

But i will be posting my overclocking soon, and which test (small ftt, inplace large ftt or blend test) is needed to confirm stability and for how much duration(minimum)
 
The only thing we have right now is info on the HD7970, and probably the HD7950. Everything else below it will probably be a die shrink of the current HD6K series and thats how they will do it with Trinity. It will be like the low end HD7Ks but still use the VLIW4 instead of GCN.

The only discrete GPUs I see using GCN as well is the HD79XX series, everything else below will probably still use the VLIW4.



.

at the beginning of the 7970 review it says that all cards will be using gcn


 
yeah, i noticed that too.
the preview says cards from 7700 to 79xx will use gcn. it also says that there might be some rebranding/rebadging. i guess amd will use 75xx and 74xx cards for rebadging.... may be to maintain hybrid cfx compatibility with trinity apus? i am only guessing in case trinity does support hybrid cfx like llano does now.
 
"Meanwhile for integrated GPUs AMD’s Trinity will be using a VLIW4 design derived from AMD’s 6900 series Cayman GPU, and at the same time it stands to reason that at least some of AMD’s 7000 series will be VLIW4 in order to have something to CrossFire with Trinity"

http://www.anandtech.com/show/5200/introducing-amds-radeon-7000m-and-nvidias-geforce-600m-mobile-gpus

That's were I read about Trinity using VLIW4.

It doesn't say anything about the full batch of cards though, any1 with a link?

And as a bonus track:

http://www.phoronix.com/scan.php?page=article&item=amd_a8_3870k&num=1

Look where the FX stands in those tests. If you ask me, it kinda looks bad for Trinity CPU wise, lol. And that's Linux for you; in windows there's a shorter gap prolly.

Cheers!
 
Llano is a mainstream part that doesn't generate that much heat IMO. That's why it's prolly less fun to explore it 😛

It's nice to see how it performs though. The GPU couldn't OC because of a certain bug they found, but once they do, I'm really curious to see how it performs. Since at stock can drive some games at 1080p with medium setting, OC'ing it could be a really nice surprise 😛

Cheers!

EDIT: "Heat" as in "page traffic", or whatever is called, haha
 
Now that you mention it, yes it should.

They're actually really close in price and the A8 has a GPU that doesn't suck. There are also very nice MoBos for it.

But the more spectacular the A8 performs, the more fear I get with Trinity around the corner... xP

Besides, the A8/A6 is the closest thing we'll get to a 32nm Phenom II X4, haha.

Cheers!
 
now i have another thought
bulldozer owner can try to do overclock in only few core, it will keep the total tdp low, temp low, and more headroom for overclocking.

Why partial overclock?
Because 2-3 cores are enough for games and you can assign the overclocked cores to game so as to run a game more faster by providing them more speedy core.

E.g 8150@3.6
we can overclock it like this
c0@5.5
c1@5.5
c2@5.5
other cores at normal
and assign these cores to the game and thus we can see some performance boost in low threaded task over complete overclocking
instead of
8150@4.3 (which is equal to partial overclock (in tdp))
and to finalize this amd can develop an overclocking software to do this work of partial overclock and core assignment.
 
IIRC, when you OC without other components "turned on", you still have current leakage. So anyway yo do it, it's not a linear gain in max TDP. Anyway, your theory is correct, you get most of your OC when you have the least resources producing heat, but the 2M/4T FX-4100 would have to be over 4.6Ghz to beat steadily a PhII X4 @4Ghz.

And they actually do this already with the VGA component in the Llano series. They can turn of almost completely the modular components, leaving more room for CPU heat. I don't know if once Windows/Linux knows there is a 4/6/8 core in it, it can turn it off like in the UEFI/BIOS settings.

Cheers!
 
Try to get those 4.5GHz and let me know if is successfully.




That sounds more interesting, the Phenom II line is very solid for budget rigs, even, I still have a X6 1090T working very good on my house.

I've hit 4.5Ghz with water on my 970BE. Eventually had to back it down to 4.2 as 4.5 isn't stable for very long, not without over-volting it to a point that I'm not comfortable with.
 
I've hit 4.5Ghz with water on my 970BE. Eventually had to back it down to 4.2 as 4.5 isn't stable for very long, not without over-volting it to a point that I'm not comfortable with.
got an x2 560 (i think) oc'd to 4.2 on air with 1.5 volts, didnt get very hot either only needed a freezer 13 pro to keep it below 50c
 
Now get all those 4.5GHz stable 24/7/365, that's the real challenge.

It's can be done, just need to be willing to volt higher then 1.5, something I'm not willing to do as it can dramatically shorten the life of the CPU.

Also it's easier to OC a two or three core CPU then a four and even harder for a six / eight. The less processor resources you have running the less heat is generated at those resources.

Anyhow, 4.2 is absolutely fine with me. Going to water was about reducing ambient noise from the system rather then overclocking. The two EVGA GTX580 Hydro Coppers I have in the same loop as that 970 are the real heat generators.
 
Status
Not open for further replies.