AMD Zambezi news, info, fans !

[Apokalipse]

but it is only 20c warm there?

under 20 most of the time.
Google tells me it's 14c right now

[undone]

OBR blog updated.

Googlish:
For kids in the forum: I'm not saying there's always something else from the beginning, I argue that performance is poverty and IPC rose only moments (units per cent), that in my posts that you read your own performance projections, and God knows what, is your business. My general opinion of the performance has not changed, find the first and last post. The fact itself is unchanged at 4.6 GHz Ocing stabiulní for Spi ...

[informal]

So he admits BIOSes are crippling the FX performance but "his opinion has not changed"?! That guy has no idea what he wants...
Until FX launches and we have a review from some relevant hardware website,this guy is blowing smoke out of his ass with those BS ES numbers.He is just seeking attention,that's all.
Not to mention he sold an BD ES over the web,an illegal thing to do.

[XRL8]

to anyone wondering, he "whited out" all the results ,even the last one is partially whited out.

@informal

He said bios works badly on gigabyte, on msi it just doesnt work at all.And on Asus it works perfectly...
However this shot could be from gigabyte.He didnt say.

[undone]

Not to mention he sold an BD ES over the web,an illegal thing to do.

One guy that update his blog frequently, to 'leak' secret, buy things from others, make flamebait in the message, everything indirectly prove that the guy doesn't have any trustable clue.

EDIT

to anyone wondering, he "whited out" all the results ,even the last one is partially whited out..

B0 chip with no doubt.

[XRL8]

I looked at multithreaded results at anandtech.If 4 BD modules behave in multithreaded situations similarly to two PHII cores.BD would win pretty much all scores with 2600K,sometimes with a big margin.
Thats just PHII cores at 1333mhz mem.If you add faster memory,new instructions, you have somewhat more boost.That leaves us with a cpu that is faster than current sandybridge in multithreaded situations.Just as expected.SuperPi with a statically set cpu frequency would look bad in that case, cause it doesnt use neither MT in any way, nor new instructions.But its useless as a real world scenario thing.

[LowRun]

Could you please stop bringing some OBR stuff in here? The guy has always been a total joke like when he came and said the retail Q9600 G0 was a crappy OCer compared to the ES and that he tested two of them on two different mobos, then later on he came and said his bad results were due to faulty ram sticks (he edited out the faulty ram part afterward, probably realizing how dumb it did make him look). Who needs computer hardware lecture from someone who can't even spot faulty ram sticks?

[AKM]

http://www.brightsideofnews.com/news....aspx?pageid=1

AMD Insiders Speak Out: BAPCo Exit is An Excuse for Poor Bulldozer Performance

When asked about core performance, surprising information was that a Bulldozer core versus the existing cores in Llano will result in minimal improvements overall.

[ryboto]

http://www.brightsideofnews.com/news....aspx?pageid=1

AMD Insiders Speak Out: BAPCo Exit is An Excuse for Poor Bulldozer Performance

Well...that's disappointing.

[informal]

LoL the same Theo Valich who "brought" us Reverse Hyperthreading? World has not been the same ever since.
Guy is full of it and HE knows it.Unsuspecting masses don't,unfortunately.

BTW He "saw" the Bulldozer Vs Sandybridge "integer" scores from "internal" AMD document in which BD supposedly destroys SB at the same clock,but fails in floating point.Another BS article written by that guy.

[Nintendork]

So he admits BIOSes are crippling the FX performance but "his opinion has not changed"?! That guy has no idea what he wants...
Until FX launches and we have a review from some relevant hardware website,this guy is blowing smoke out of his ass with those BS ES numbers.He is just seeking attention,that's all.
Not to mention he sold an BD ES over the web,an illegal thing to do.

XS need to stop posting vaporware from this guy.

In some other forums people(haters/negative guys/need a excuse to justify SB) take their cinebench numbers like the bible wich means that top FX 8000 cpu performs worse than my AII X4 620, even when costing 5x more

[dave_graham]

ah Theo Valich...I remember that name from somewhere in the "slag AMD" annals of time. whatever. my focus isn't on desktop but on server. consequently, I could really care less about Zambezi as a consumer exercise.

[XRL8]

XS need to stop posting vaporware from this guy.

Hate to be the devils advocate.
but the numbers hes posting are the best of all the other leaks.He also confirms that most mainboards have bios issues with BD (which i find a bit worrying).
He also shows that even that early BD sample overclocks very well.
As for the single thread ,B0 may have low clocked uncore.Which can give for example weak gaming scores.Just a theory.
Anyhow, its not like we have many leaks to choose from, the other ones posted crappy numbers and silenced.

[Formula350]

YAY Another long post due to not reading the thread for 2 pages lol [/sorry]

As far as core scaling goes, when both BD cores in each module are being used, AMD themselves said multi-threaded performance would increase by 80% (remember 80% more than 100% is 180%)
And that's not compared to K10. That's compared to doing one thread on one BD core.

AMD also said they got "50% more performance from 33% more cores"
Of course they didn't say what type of workload. But in that context I'd assume it means something that uses >=8 threads (still not THAT specific).

I think the biggest thing you're forgetting is that those are all G34 Server related performance numbers. I've yet to see anything from AMD on Bulldozer talk about it in context of the consumer desktop chips. While yes, it's the same architecture between them, I have a feeling that it won't run on HT3.1 specs for 990FX. We all know what going from just 2000MHz to 2600 or 2800MHz can do for performance gains (I've been doing a lot of AIDA benching with my 1090T if you've not seen anyone show the difference). So not only is that something to consider, but also that those numbers might actually be in relation to server-oriented tasks. 50% more performance for serving content over the web or crunching simulations for what-have-you, doesn't necessarily equate to 50% more performance for gaming or raising your EP [e-peen] level when playing MMOBUC [benchmark, upload, compar] on HWBot lol Seriously though, who knows how it will pan out for us :\ OBR's memory scores in AIDA are rather odd (see: low), especially considering it's at 1866 :\ The L1 comparatively seems to be rather low, L2 is alright and parts of L3 seems quite low. Mem write is bad, with read being the only significant gain. Though I've gotten close to that with 1090T @ 3.2/3.6T with NB@2800 with 4x2GB DDR3-1600 8-8-8-20 1T 26 5-5-5-8-7-6-2-4-4 110ns (yea, those sub timings aren't JEDEC but I can translate if anyone cares).

However the guy is known for having its connections and getting ES stuff, while he is disliked by most guys here, no one "in the know" dismissed tha fact that he gets samples.They are even irritated because he breaks NDA`s and still gets samples.
Guy has pictures of samples, guy has sold one , he provided many screenshots.What more do you want ?

UGH, tell me about it But, I guess that's the benefit of being a valued tester (it's the only thing that makes sense given what's happened...)

As far as i remember there is only one price leak, and its a simple excel chart,and not confirmed in any way at that...

Well I had asked my friend who has the dual Interlagos if he knew of what the 8100 FX models would run and that was before the leaked prices came out, which he had said "they're saying* low $300-400" [*general corporate 'campfire talk' from people privileged to know/access that info, I think he means]. So while both my info and the original can't be confirmed, I think it's a fair bet we'll see them arrive at those prices.

One guy that update his blog frequently, to 'leak' secret, buy things from others, make flamebait in the message, everything indirectly prove that the guy doesn't have any trustable clue.

I think you're talking about just OBR in that msg, and if so then I think that's probably rather accurate lol Who knows, maybe he's something like the lead QC person at one of the big mobo makers and since AMD has gone through 2 steppings now, working on their third, those are the old test samples. BUT perhaps he is also and INTELamer and is 'sabotaging' the results lol [Disclaimer: OBR, if you read this, I'm speculatively joking, so don't get all bent out of shape lol]

Hate to be the devils advocate.
but the numbers hes posting are the best of all the other leaks. He also confirms that most mainboards have bios issues with BD (which i find a bit worrying).
He also shows that even that early BD sample overclocks very well.
As for the single thread ,B0 may have low clocked uncore. Which can give for example weak gaming scores. Just a theory.
Anyhow, its not like we have many leaks to choose from, the other ones posted crappy numbers and silenced.

BIOS issues on 990FX boards I assume was/is implied? I mean, I know it's the same chips and all, but some boards ARE new designs and so more than likely a fresh BIOS. Especially considering they are UEFI now too. Boards like the 890FX Extreme4 and 890FXA-GD65 though I'm quite certain are carbon copies of the 890FX variants. Doing a back and forth between the GD65 shows that the only difference layout-wise is no Clr_CMOS jumer installed on the 990FX model, some adjustments/additions to silkscreened labels and the change on the heatsinks from Blue to Black (though it does look like the 990's heatsink is a hair smaller). Then again these boards are likely to be simply ones that were already made for BD, but when things originally got delayed they just released them as a silly stop-gap under the 890 branding. Because the board model on the GD65 both say "MS-7640 VER: 3.0" heh Wonder if flashing the 990's BIOS (with it's 990 model number strings) on the "890" version will soft mod them to a 990, giving those owners SLI capability...?


STOP! NOBODY POST FOR A SECOND SO THIS REPLY IS CURRENT!! lol

[Apokalipse]

http://www.brightsideofnews.com/news....aspx?pageid=1

AMD Insiders Speak Out: BAPCo Exit is An Excuse for Poor Bulldozer Performance

Because let's just ignore the fact that Nvidia and Via also decided to quit.
Is Nvidia "angry about their GPU's being sub-par to Intel"? lol......

[flippin_waffles]

ah Theo Valich...I remember that name from somewhere in the "slag AMD" annals of time. whatever. my focus isn't on desktop but on server. consequently, I could really care less about Zambezi as a consumer exercise.

Will be interested to see your opinion once/if you get one, and when you can talk about it.

[AAbenson]

a 32nm part running at 1.5v...gimme a break

[PatRaceTin]

4.6xxxGHz @ 1.5xxxxV not look good

[Dresdenboy]

4.6xxxGHz @ 1.5xxxxV not look good

I assume that it has been read out correctly. If that is a modified base clock (original should be 3.2GHz) I'd like to know if any turbo modes had been switched off.

[FlanK3r]

I think yes...but
1)u can nnot compare 32nm INtel and 32nm AMD. Look at AMD history...65nm Phenoms has about 1.45V as safe voltage, but Phenoms II at 45nm 1.55V! BD is still SOI with HK!
2) its ES

[undone]

Few months have past, still no meaningful info, sigh.

[rintamarotta]

Dont worry, litle baby-monster will come out of AMD in few months

[Formula350]

The ONLY thing that I've got going for me with the whole Bulldozer thing, is I'll be on Vacation here from Jul 8 to Aug 5, so when I get back then Dozer will hopefully be out in the next couple days!

GOD I hate this, but it better be for a damn good reason AMD! If you make it worth our while then I think you'll be unanimously forgiven -_-

[AAbenson]

even for the current AMD cpus 1.55v is borderline degrading.
with or without SOI no way the new 32nm generation AMD cpus wont degrade at 1.55v...thats prolly just a suicide shot just to show that 4.6ghz is doable on the Bulldozer

[tifosi]

Screenshot still shows 186W as TDP Which means that the chips would be like Power 7 chips? 5 Ghz+ at launch?

http://www.brightsideofnews.com/news....aspx?pageid=0

I think Charlie should may be say something about this... could it be that Intel is paying bsn, or someone else is misleading them, plain and simple. Think about it this way... Intel is the only chip manufacturer there on BAPCo, rest all have left. If you take out credibility of someone, you don't have to discredit what they're saying.

We have on one hand, a guy with SR2 rig, 'Rintamarotta!!!' ding ding ding... who's claimed in an earlier post that it would easily best 990X. On the other hand we have people who're saying the opposite. The again we have OBR who's just being a total tool... as he could share some information as long as he hasn't signed a NDA(even a gentleman's agreement), i think that's fairly ok.