AMD Zambezi news, info, fans !

[Smartidiot89]

Things that can affect CPU performance include:
* BIOS settings/features
* microcode
* CPU thread scheduler (OS), or various patches for the OS
* program compiler, and what instruction sets are used
* CPU revision/stepping

Some things are known, but some things aren't.

I was oblivious to this please tell me more Isn't it reasonable to believe that AMD sent out the right stuff to the media, no?

[Apokalipse]

I was oblivious to this please tell me more Isn't it reasonable to believe that AMD sent out the right stuff to the media, no?

Not if they don't want it's performance to be revealed until the actual day the NDA expires.

[Smartidiot89]

Not if they don't want it's performance to be revealed until the actual day the NDA expires.

So your saying they are tricking the press, media and OEM's that their product is bad then when all the reviews globally are released they will drop the bomb saying this is the real deal? Sounds like a perfect way to ruin a company, brand and their stock to me.

[TESKATLIPOKA]

xdan

Yes the arhitecture which was wrong from the begining.
What if the penalty from sharing resources between cores is biger than AMD was saying.
Finally three more days to the complete truth.

It seems you forgot about the barcelona launch that cpu would be another fail for you yet the successor without any bug and more mature process was trading blows with Core 2 duo.
The problem in BD is not sharing resources but the performance in single thread when nothing is shared.
I would wait for Trinity and compare it to FX4170 clock to clock and see how much IPC increases, instead of talking how wrong the architecture is when you don't even know what is wrong and what can be corrected in the next design.

[xdan]

Barcelona was not a fail compared to previous generations AMD cpu's but with Intel quads from that time.
It was not a fail compared to Athlon 64 X2 5000+.
But BD due the higher clocks than Phenom II, larger cache, so called "8 cores" it's a fail to Phenom II.
FX 4170 will offer a better picture of this compared wit X4 955.
tmpgenc.png
http://www.neoseeker.com/Articles/Ha...om_9600/9.html

[zalbard]

Does Windows need a patch to make the AMD processors work correctly AGAIN?

That's possible.
Then again, they must've been doing some in-house testing... Hard to get sensible performance data when it requires a software patch and there is none. And what about reviewers? Where is the said patch?

[zagortenay]

I can understand AMD not keeping up with the competition on the manufacturing front, all the yield issues and related delays, but come on now! Did AMD forget about all the expertise they had on the x86 cpu design? How can they possibly design a chip after 5 years of research, that fails to deliver the performance of their former cpu? What is the point in launching such a miserable chip? Why BD chip is so big in the first place?
It just doesn't make sense... especially when AMD graphics department is doing so well.

[haylui]

isn't the AMD presentation already tell us that
some benchmarks are close to i7 2600k while others like wprime lag behind?
look at Handbrake and wprime

Presentation from AMD (assume)


Handbrake by Lab501


wprime by Lab501

[Apokalipse]

So your saying they are tricking the press, media and OEM's that their product is bad

I think it's not exactly a case of AMD deliberately trying to make it look bad. They don't want the people with access to the chips leaking at all.
It's just extremely difficult to stop people from leaking, so they only give enough out to show it's working, but not how fast. Looking slower than it actually is is not the goal, but the side effect.

then when all the reviews globally are released they will drop the bomb saying this is the real deal? Sounds like a perfect way to ruin a company, brand and their stock to me.

How exactly would having higher performance than expected ruin a company?

[informal]

Guys,the reviewers already have all they need to test the platform correctly. AMD is not going to supply them with a last minute magic BIOS/OS workaround/whatever ,that is not how they operate. You cannot do that so close to product launch,it would invalidate all the work reviewers have done.And to make a decent review it takes a lot of time and testing.
This is it when it comes to performance. How is AMD going to spin this is unknown. I await for explanations why single thread IPC is not up like we were promised,why legacy SIMD(SSE) performance is slower than X6/X4,why multicore scaling is so poor in some workloads etc.

[Apokalipse]

Guys,the reviewers already have all they need to test the platform correctly.

How could you possibly know this?

[TESKATLIPOKA]

xdan I was referring to your statement about how the architecture was a fail from the beginning and thats why I said lets wait for Trinity to see if true or not, because the only problem is low IPC.

FX 4170 will offer a better picture of this compared wit X4 955.

they will be at best comparable because of lower IPC and its still a comparison between 2 module and 4 core, I would say in single it will be faster but in multi slower thanks to higher clock.

[informal]

How could you possibly know this?

Ok 3 more days.
If you really think AMD is going to release a workaround at the launch then there will be many websites,including the biggest ones,which were scr*wed over by AMD since they tested the incomplete platform. This won't happen of course.

In order to get to level projected from Nov 2010 slide that was leaked,performance needs to jump up dramatically,think 40+% over this(yes this much). Not going to happen. There are only few things in a chip that can lower performance this much and still not be a bug (one of them are 2 new prefetch mechanisms AMD stated were used to offset the IPC impact of deeper pipeline;but I doubt anything is wrong or disabled,this is the final chip).

[Apokalipse]

Having something like prefetching disabled can lower performance by about half; microcode in the BIOS can do this, for example.

If leaked performance really is final performance, then I would say the entire management at AMD would need a visit by the nice men in white coats for actually releasing something worse than K10.5
Because if it is actually worse, then it makes no sense at all to release it.

Now I won't claim to know if it will beat SB. For all I know SB could be faster. But to be slower than K10.5 would be utterly retarded.

[-Boris-]

Having something like prefetching disabled can lower performance by about half; microcode in the BIOS can do this, for example.

If leaked performance really is final performance, then I would say the entire management at AMD would need a visit by the nice men in white coats for actually releasing something worse than K10.5
Because if it is actually worse, then it makes no sense at all to release it.

Now I won't claim to know if it will beat SB. For all I know SB could be faster. But to be slower than K10.5 would be utterly retarded.

And that's what the fuzz is all about. If the reviewers should be able to post reviews this Wednesday then they must have the final chip with final bios and final configuration already. So if reviewers says it sucks, when it probably will suck at this point. We aren't talking about leaks of engineering samples with features disabled and beta bioses. This is the real deal.

[Apokalipse]

And that's what the fuzz is all about. If the reviewers should be able to post reviews this Wednesday then they must have the final chip with final bios and final configuration already. So if reviewers say it's suck, when it probably will suck at this point. We aren't talking about leaks of engineering samples with features disabled and beta bioses. This is the real deal.

Final silicon and hardware probably. That doesn't mean everything is final.

If it's slower than K10.5, they shouldn't even bother to release it. Simple as that.

[Andrea deluxe]

from 2005 amd is incapable to produce a cpu with superior or equal capabilities.....

incredible.

my latest amd cpu..... athlon 64 X2 3800+ overclocked @ second boot.... (good times....)

BULLDOZER

[Oese]

Ok 3 more days.
If you really think AMD is going to release a workaround at the launch then there will be many websites,including the biggest ones,which were scr*wed over by AMD since they tested the incomplete platform. This won't happen of course.

In order to get to level projected from Nov 2010 slide that was leaked,performance needs to jump up dramatically,think 40+% over this(yes this much). Not going to happen. There are only few things in a chip that can lower performance this much and still not be a bug (one of them are 2 new prefetch mechanisms AMD stated were used to offset the IPC impact of deeper pipeline;but I doubt anything is wrong or disabled,this is the final chip).

i can believe the prefetch mechanism has flaws. That would account for a bad power consumption without actual performance.

not amused at all by what can be seen by now..

[dess]

nightmare, but not unlikely

I hope we will all be waken up 3 days later with a big "Phew!"...

Do you have access to the benches under NDA yet?

Yes?

...kind of anti-competitive don't you think? A bit of a Monopoly, don't you think?

Definitely. And there is more to it: it's againt the interests of the whole industry.
That's why I for one won't go for SB even if BD is worse.

Can I sincerely ask how are these (i7 numbers) possible with stock ram frequency? I mean, top bandwidth for DDR3-1600 (the max. that SB officially supports) 12800 MB/s... Even with DDR3-1866 you get 14933...
http://en.wikipedia.org/wiki/DDR3

AMD wants to keep Intel in the dark, and have shown they're capable of keeping a secret (Eyefinity, RV770 shader count)

It was surely the case until now. But, if there is anything that can help BD, I don't think they're holding it back intentionally, anymore. Look what I've wrote about it earlier... Perhaps they're still fiddling with AGESA to find some nasty bugs, or something... But, the management said no more delays, no matter if they're able to find a solution before lanuchday or not. I don't think it will come in time, but perhaps somewhat later.

Why BD chip is so big in the first place?

BD surely should be capable of better performance to justify it's size.
Although, cachesizes also have a role:
SB: 1 MB L2 + 8 MB L3 = 9 MB total
BD: 8 MB L2 + 8 MB L3 = 16 MB total

Because if it is actually worse, then it makes no sense at all to release it.

O.C. could justify it... That way you can top K10.x performance.
Don't forget FX is not a mainstream productline... It's Black Edition, meant for O.C.
Probably it wasn't the original plan, though.

[xdan]

Memoriile folosite pentru ambele platforme au fost A-Data 2 x 2GB XPG Plus 2200MHz CL8 echipate cu chipuri Elpida Hyper. Acestea au fost setate la 1866MHz cu latentele 8-8-8-24 1T in cazul ambelor platforme.

The memory was set at 1866 CL8, and may be with Hypers even some more latency where lowered.

Can I sincerely ask how are these (i7 numbers) possible with stock ram frequency? I mean, top bandwidth for DDR3-1600 (the max. that SB officially supports) 12800 MB/s... Even with DDR3-1866 you get 14933...

14933 at 1866CL9 may be with AMD...

http://www.xtremesystems.org/forums/...xtreme4-Review

[maltrabob]

@Apokalipse: I really do admire your bravery to claim such opinions publicly :-). Personally, I think there is a possibility that you are right but common sense says otherwise. Either way, I cannot wait to read first reviews on Wednesday.

[slaveondope]

Was it not said that the FPU and integer cores operate at different clocks? I'd still like to see some higher base clock speeds and NB clocks. 501 only took the fsb up to 202 from 200 and didn't touch the NB. Which would explain the difference in memory bandwidth to SB as I recall adjust it's IMC speed automatically when memory divider is adjusted correct?

[zalbard]

One thing that may happen is the same as with 69x0 cards, when they said (or people said?) that better drivers (so software / BIOS patches for BD) would improve the performance dramatically.
Well, the performance did improve, but so did the competition...

[-Boris-]

Final silicon and hardware probably. That doesn't mean everything is final.

If it's slower than K10.5, they shouldn't even bother to release it. Simple as that.

They don't send out beta bioses or anything beta to reviewers. They don't want crippled performance in the reviews people are judging their entire new lineup from. And it might not be slower than Phenom II, but it may be only a small performance boost. Remember prescott? Slower, hotter and more complex than Northwood. Still it was released. Remember Willamette where they crippled P3s so they wouldn't compete with their new P4?

[Smartidiot89]

I think it's not exactly a case of AMD deliberately trying to make it look bad. They don't want the people with access to the chips leaking at all.
It's just extremely difficult to stop people from leaking, so they only give enough out to show it's working, but not how fast. Looking slower than it actually is is not the goal, but the side effect.

How exactly would having higher performance than expected ruin a company?

Reviewers all over the globe report the same performance. If AMD release this magic/fairy dust patch when NDA lifts do you really think reviewers will have time to retest the ship and rewrite the review. There will be reviews saying how much is sucks, and yeah that would hurt the company, and their FX-brand. What you are portraying won't happen. I have very high confidence what AMD sent out is "final everything" for several reasons, not only what has been spoonfeed by AMD.

Informal and -Boris- seems to be the most down to earth people here.