AMD Zambezi news, info, fans !

[undone]

I just find it odd that it's different to the other ES AMD chips I have. COMPLETELY different

You mean you have several bulldozer ES?

[Mats]

I just find it odd that it's different to the other ES AMD chips I have. COMPLETELY different

Yeah, the specifiaction in CPU-Z usually never include that much codes.

[[XC] Oj101]

You mean you have several bulldozer ES?

I wish

[Kobaltrock]

I fell off my chair laughing at that too mate:
ROFL @ AsRock

Ok, So AsRock confirms it, BD can work with AM3 boards.

That leaves BioStar the only one left that hasn't said anything about what they are going to have out...

Though, I guess we can hold a contest to see which AM3 board goes up in flames the fastest with a BD installed.

[FlanK3r]

Kobaltrock:not with all AM3 boards! Its important! There is 2 conditions:
1)16MB big ROM flash
2)all mobo contacts for pin must be electricaly active

And there is not AMD support! I think, not only for me will be better new AM3+ mobo.

[Formula350]

I don't have time to read the next 2 pages at the moment, but all this talk of "really, no 'Dozer bench leaks yet?!" makes me assume that no one has seen these? o_0 Which if not, is surprising I think >_>
http://www.phoronix.com/scan.php?pag...er_linux&num=1

Granted, it's not the desktop variant, but from my understanding the 'Dozer CPU modules are going to be the same between everything. It'll just be small tweaks to certain areas (likely the HT Link for the SMP cross-talking [if you will]), but the only big big difference is the amount of cores the C32 and G34 have since they both have got over 1.25x and 2x the amount of pins available, respectively. Valencia might be able to run 10cores, but AMD probably figures they should keep it down to 8C(max) to make buying the G34 worth the investment :P Because we're getting 8C on 942pins with DDR3-1866 Dual Channel, and C32 is 1207 with the same :\ Those ~260 pins really all for the SMP Hyper Transport?

Speaking of HT quick... Why the hell is AMD upping the Link speed to gain a higher MT/s, instead of just friggen just using the 32bit that all our NB and HT have supported since it started getting used, ditching the damn 16bit?! *sigh*
From Wiki:

Sucks when there is an available ability that doesn't get utilized! Th... that'd be like, disabling perfectly good cpu cores and selling it as a down-cored model of a the same CPU!! Oh... wait... NM :P lol But seriously, I don't get why they keep with 16bit when 32 is right there :|


Read this quick while posting:

Seems CPU-Z doesn't work properly with BD

My friend said his reads as a Phenom II FX something, which is in line with what the releases are going to be, even though his is an Opteron.

All I know is that the CPUID reads fine for this E350 Zacate on the Specifications line in CPU-Z 1.37 (rrrreally old), and a program called CPUMSR (even older). So CPU-Z doesn't care too much about that, but where my friend is getting that FX reading from, I'm not 100% sure. I'm trying to squeeze a CPU-Z SS from him

[Mats]

I don't have time to read the next 2 pages at the moment, but all this talk of "really, no 'Dozer bench leaks yet?!" makes me assume that no one has seen these? o_0

I'm not sure, but I think this was posted in the news last month, and the conclusion was that the systems had too different specs to be of any use.
Something like one was running at 640x480, the other at 1920. One had twice as much RAM as the other etc.
Can't find the link..

[[XC] Oj101]

Yes, the graphics cards and storage were also different. It was a useless bit of info.

[BeepBeep2]

Sure you guys have seen this:
http://www.youtube.com/watch?v=mr7kr4kimeM

If you havent, some nice short videos restating things we already know

[Formula350]

So I was poking around the ASRock eXtreme Tuning Utility, hoping to maybe figure out a way to edit it or add in an entry to get it to work with the E350M-1 board, first file I've looked in though turned up something of relevance to the thread...

Everything runs off XML files, which are for basically each individual board to state limits, MOSFET controllers (or whatever) and the sensor IO, among other things. For the 890FX-Deluxe3 that just came out, with AM3+ support, it says:

<Item Conf="890FXD5.xml" Name="890FX Deluxe5" Index="21" IesConf="L6717.xml"/>

Looking down the list more and I see...

<Item Conf="890FXD5.xml" Name="990FX Deluxe4" Index="45" IesConf="L6717.xml"/>

So apparently the 990FX will either be such a minimally tweaked chip that it is, as some of the past SB have, just a drop-in for motherboard manufacturers. Or not at all and just rebranded as is what has basically been said/hinted at. What that means is, it would seem at least, is FlanK3r doesn't have to worry about any performance issues with the AM3+ boards that are out there right now. I mean, it's just speculation right now, but appearss as though at least one of ASRock's boards will be mostly the same as far as the finer details are concerned, since that profile to overclock everything points back to the 890FX D5. Likely will be layout and heatsink changes, just to mix things up. Especially considering the Deluxe4 and 5 show very litter difference in terms of component locations across the board, which only shows minor layout difference versus the Deluxe3. Biggest one is the change from VIA audio to Realtek :\ The only difference in the profile that I don't know what is for, is the "Index=##" part, which changes from 21 (890FX) to 45 (990FX)

Anywho... While I'm on the topic, I also came across some info on undocumented APU graphics models, their designation, Crossfire compatibility with a discrete card, and what Radeon model that APU+GPU Crossfire will be redesignated That was in one of AMD's programs ('spose I should type up that info I was going to and publish it... Just cuz I wanna feel 'spechul' and get it online before anyone else! lol)

EDIT: Can't say they're not coming out soon They've apparently gotten into the hands of someone who is providing the AIDA team with info, I suppose for the sensors and the like.

AIDA64 Extreme Edition v1.60.1369 Beta

preliminary support for AMD K12 (aka Llano) processors
preliminary support for AMD K15 (aka Bulldozer) processors

[BeepBeep2]

I can pretend...
This is a fake image btw.

[FlanK3r]

beep: why do u think (1.57.3, myabe better support)?

[BeepBeep2]

beep: why do u think (1.57.3, myabe better support)?

You should already know why

[LightSpeed]

I so hope to god that Bulldozer performs remarkably well and go head to head with intel again, it would be such a welcome change in the microprocessor world. ~15% faster per core than K10.5 would make the 8 core BD very competitive indeed. Here's hoping its faster still

[FlanK3r]

4x 2048???etc? Others, I dont really know

[BeepBeep2]

4x 2048???etc? Others, I dont really know

Nope, thats the correct cache structure, at least I think so, so keep guessing

Where do you think I got that image from?

[Lightman]

Nope, thats the correct cache structure, at least I think so, so keep guessing

Where do you think I got that image from?

Eng sample name suggests it's 2.8GHz [3.6GHz Turbo] and faked pic has 3.2GHz

Anyway I'm looking forward to next couple of weeks, it will be fun sorting fake info from somewhat legit one

[Formula350]

Eng sample name suggests it's 2.8GHz [3.6GHz Turbo] and faked pic has 3.2GHz

That doesn't mean anything. It doesn't display the Multi or HTT, so it could be overclocked for all we know But 3220, according to what I'm seeing, is correct for 3.2GHz; 2.8GHz would be 2820.

Nope, thats the correct cache structure, at least I think so, so keep guessing

My guess is due to it not reporting how many multiples for the cache there are, IE:
8 x 16KB
4 x 64KB
4 x 2048KB

But, I'm just basing that off what it reports for my Bobcat and in all likelihood isn't right :\

Unless it's that engineering samples show up as single core, and that is why it doesn't indicate their cache multiples, thus 8 then being incorrect...
(doubt the Processor #1 being shaded is your reason, as again my Bobcat's is and it's dual-core)

[BeepBeep2]

Eng sample name suggests it's 2.8GHz [3.6GHz Turbo] and faked pic has 3.2GHz

Anyway I'm looking forward to next couple of weeks, it will be fun sorting fake info from somewhat legit one

No, my pic shows a 3220 model number tech docs state that as a 3.2 Ghz CPU with 3.7 Ghz Turbo

[Formula350]

Here's that APU Crossfire and 990FX data I stumbled across that I said I would do a writeup about
http://pro-clockers.com/industryrevi...ain-sight.html

[[XC] Oj101]

The specification really throws me off here, none of my AMD ES chips have anything which remotely resembles that. I look forward to seeing TRUE ES pictures

[Lightman]

That doesn't mean anything. It doesn't display the Multi or HTT, so it could be overclocked for all we know But 3220, according to what I'm seeing, is correct for 3.2GHz; 2.8GHz would be 2820.

...

True, but lets assume current beta BIOSes are not allowing OC.

No, my pic shows a 3220 model number tech docs state that as a 3.2 Ghz CPU with 3.7 Ghz Turbo

I was talking about this section of CPU name:

ZD322046W8K44_36/28/20_2/8 A

This is my speculation, but to me it reads 3.6GHz Turbo. 2.8GHz Stock. I'm not sure how to interpret 20_2 and 8 A.
Once again my speculation

[Apokalipse]

The specification really throws me off here, none of my AMD ES chips have anything which remotely resembles that. I look forward to seeing TRUE ES pictures

What sort of specifications are you referring to? (if you're allowed to say)

Is it something that AMD can change in different ES chips so they can trace the source of leaks?

True, but lets assume current beta BIOSes are not allowing OC.



I was talking about this section of CPU name:

ZD322046W8K44_36/28/20_2/8 A

This is my speculation, but to me it reads 3.6GHz Turbo. 2.8GHz Stock. I'm not sure how to interpret 20_2 and 8 A.
Once again my speculation

Week and batch maybe?

[2good4you]

much info here also:

http://translate.google.com/translat...%2Findex6.html

[Olivon]

Nice 2good4you Thanks