AMD Zambezi news, info, fans !

[Mats]

An norwegian E-tailer searchengine has slipped the price on FX 8150.

The 2600K is even cheaper than the FX in Sweden, but I guess those prices will change the next weeks.
Your USD prices are including 25 % Norwegian tax, right?

In Sweden:
http://www.prisjakt.nu/produkt.php?j...,995743,995734

[B.A.T]

All USD prices are including 25% Norwegian tax

FX 8120: 1630,- NOK = 281$

FX 6100: 1384,- NOK = 238$

[Mechanical Man]

You can wiggle a few months, but not 2-3 quarters. The customers aren't stupid, they've seen too the marketing slides from AMD, they've done their homework. If they were told Interlagos will be H1 ( Q2 ) , then rest assured they didn't expect to get it in early Q4. On HPC supercomputers, researchers apply for time slices, there is a lot of preparation going on even before the computer is installed. Imagine going back and saying " Hello everyone, we're delaying everything by 6 months ".

How likely are those customers to order again AMD based systems and even more interesting, how likely is Cray to bet high end deals on Opteron CPUs ? What about IBM, HP, Dell in the commercial sector ? Imagine production and banking systems. The factory is ready, the server is MIA or they're forced in the interim to use previous generation CPUs. Imagine the penalties and the extra downtime for upgrading once Interlagos is available. If you find the answer to these, AMD's server market share of <6% suddenly is explained.

If i recall correctly, server version was scheduled to h2 for long time now and i think there was info about it bd shipping to cray in summer. With Cray I think they needed hw not necessarily latest most performing version of bios & os scheduler guides. It might be that AMD delivered almost on time to them by using all they could get from 32nm process if those bad yield rumours are true. If B2F is latest version and it was B2 already before summer, it means they did not make any big modifications, only optimisations to get better yield ect. Ofc, we may never know.

[Mechanical Man]

Very interesting. There is a very clear difference between a recommendation and being forced to do something

I think I speak for most reviewers when not many, if any will divert from their regular testing methodology.

Well, it only means AMD would have come to its rivals level...

[Leeghoofd]

Well, it only means AMD would have come to its rivals level...

In some particular tests then... "optimised for" rings a bell ? Hence why I never run these tests and stick to my usual test suite, loved or hated by the fan boys...

[dess]

In some particular tests then... "optimised for" rings a bell ? Hence why I never run these tests and stick to my usual test suite, loved or hated by the fan boys...

Do you mean SuperPI?

[Leeghoofd]

absolutely pi has to stay in it... good indication of ram/bandwith scaling, wprime clearly shows the cores working...

[tiborrr]

There are many bios options that can effect the outcome of benches.

HPET is 1 for example, it stops the cpu from throttling back in mulithreaded apps.

Running pi on a cluster, versus a core ( 2 threads ) versus being able to disable a single cluster in a core ( which 99% boards/bios's do not have implemented so resources are not shared ) can all influence the results in single threaded.

Knowing all this tells you one thing for sure, you can make it look worse or make it look better all depending on your knowledge of the chip and or your intentions.

As far as PI it's an antiquated bench and has not been AMD's strong point for quite some time.

Granted some results shown tend to lead to the fact that 1m times are bad but looking at the bigger picture we also know that in many cases you can validate 1000mhz higher in many cases with BD, which would point to the fact that you can run 1m at alot faster speeds than current AMD tech.

Things that make you go hmm like what kind of times will we see at 8 gig or even comparing BD to deneb/thuban when same cooling is used.

This is why I like you so much. Major props.

[mav2000]

^^^ was that quote from chew* on another thread?

[dess]

absolutely pi has to stay in it... good indication of ram/bandwith scaling, wprime clearly shows the cores working...

Does it represent anything that is reflected in everydays performance?

[Mats]

^^^ was that quote from chew* on another thread?

Click the blue arrow in the quote and you'll see..

[Canis-X]

^This

^^^ was that quote from chew* on another thread?

Here is the link to it --> http://www.xtremesystems.org/forums/...60#post4952360

[radaja]

absolutely pi has to stay in it... good indication of ram/bandwith scaling, wprime clearly shows the cores working...

where is your review going to be posted at?are you still at madshrimps?looking forward to your review in particular,i have always liked you level headed thinking since the days of the asus P5Q-deluxe thread

[tifosi]

what i really dont understand is if the Bulldozer cpus are really that bad then why was Cray so eager to use these chips in its new supercomputers?

Erm, seeing specs for server chippery, one could note that BD came with more cores, higher clocks at same or lower TDP. It is a no brainer as it is drop in compatible. You don't have to change everything, but just upgrade bios and pop BD in. There's also support for DDR3 1600 Mhz on server side, i'm sure that will come handy too.

AMD marketing was doing damage control while they were busy tunning the ecosystem to get a few % more.

Launch is tomorrow. Who's fault is it if the whole chain isn't ready ? It's like buying the latest BMW and the sales guy will tell you " Well, we promised 300bhp and 40mpg, but since the latest software stack for the ECU isn't ready, you'll get only 150 bhp and 20 mpg until sometime next year when we'll fix it. Peace man." Imagine this to happen to you.

Yeah, who cares about x86 compatibility. Let's just ask all the SW vendors to recompile their apps for the new CPU with 10^-27 market share.

That to me equals sleazy tactics. "Configure the test properly" = choose the settings and scenarious which will put BD in the best light.
Rumours has it that AMD will force reviewers to use predefined SW and limit their resolution choices to very high details where CPU differences are masked by the GPU bottleneck.

You're average Joe user will make sure he has the latest silicon, latest microcode, latest BIOS, all the OS patches, drivers are latest, beta if possible, ask the ISV to recompile all his apps starting with Angry Birds and will read the 300 page BD optimization handbook that AMD will supply for free with each CPU. Only then, he'll start using his PC for his day to day activities and be completely satisfied.

Surely you don't think it is as scummy as paying off Dell etc to not use AMD chips, slow other chips in their compilers etc. Or for that matter, willy nilly deciding which ISA is supported or not (remember SSE5, FMA4). If you talk about X86 compatibility, then Intel has a lot more to answer than other two companies involved.

Also, wait till NDA lapses before you decry the chip. As it is, we all know how Nvidia screwed DX10. We also have recently learned that windows is more optimized for HT than actual cores, which is to be fixed. So either you could be sane, and rational, or just jump around tearing up your hair.

[ryboto]

And here's some stuff-
http://semiaccurate.com/2011/10/10/e...-architecture/

[Sparky]

Savantu is one of the resident AMD haters (and was banned from the news section for trolling the AMD threads I believe), so I do recommend taking anything he says regarding AMD with a fairly large amount of salt

[radaja]

And here's some stuff-
http://semiaccurate.com/2011/10/10/e...-architecture/

Exclusive,really charlie? kind of misleading

[mav2000]

I am just waiting around now, coming back to the same threads every half an hour or so, expecting some good news.....nothing yet..

[Leeghoofd]

anyione up to date on this ? My win7 and 8 were fully patched, no idea if this is included yes or no

"Actually, we already have such an issue known for Bulldozer, and NO bench-marked system has the patch installed!

The shared L1 cache is causing cross invalidations across threads so that the prefetch data is incorrect in too many cases and data must be fetched again. The fix is a "simple" memory alignment and (possible)tagging system in the kernel of Windows/Linux.

I reviewed the code for the Linux patch and was astonished by just how little I know of the Linux kernel... lol! In any event, it could easily cost 10% in terms of single threaded performance, possibly more than double that in multi-threaded loads on the same module due to the increased contention and randomness of accesses.

Not sure if ordained reviewers have been given access to the MS patch, but I'd imagine (and hope) so! Last I saw, the Linux kernel patch was still being worked on by AMD (publicly) and Linus was showing some distaste for the method used to address the issue""


And Tiborr, sorry but changing all sort of options in the bios is not done by most users, neither is setting affinity or co.. Either it works or it doesn't. Not everybody has Chews knowledge on the matter. Disabling PowerManageralready boosted multithreaded performance on my testbed...

[Oese]

Savantu is one of the resident AMD haters (and was banned from the news section for trolling the AMD threads I believe), so I do recommend taking anything he says regarding AMD with a fairly large amount of salt

The amount of salt needed around bulldozer is starting to damaging my health i believe

anyione up to date on this ? My win7 and 8 were fully patched, no idea if this is included yes or no

"Actually, we already have such an issue known for Bulldozer, and NO bench-marked system has the patch installed!

The shared L1 cache is causing cross invalidations across threads so that the prefetch data is incorrect in too many cases and data must be fetched again. The fix is a "simple" memory alignment and (possible)tagging system in the kernel of Windows/Linux.

I reviewed the code for the Linux patch and was astonished by just how little I know of the Linux kernel... lol! In any event, it could easily cost 10% in terms of single threaded performance, possibly more than double that in multi-threaded loads on the same module due to the increased contention and randomness of accesses.

Not sure if ordained reviewers have been given access to the MS patch, but I'd imagine (and hope) so! Last I saw, the Linux kernel patch was still being worked on by AMD (publicly) and Linus was showing some distaste for the method used to address the issue""


And Tiborr, sorry but changing all sort of options in the bios is not done by most users, neither is setting affinity or co.. Either it works or it doesn't. Not everybody has Chews knowledge on the matter. Disabling PowerManageralready boosted multithreaded performance on my testbed...

Thats great information. I doubt it is patched yet. And it would explain a LOT.

[AbortRetryFail?]

I swear, none of this still makes sense, particularly after reflecting upon these prices from a Microcenter employee:

http://www.overclock.net/amd-general...ate-specs.html

Assuming those listed specs are true, losing ECC RAM support in desktop BD is a bummer.

It is (was?) always nice to take an Asus desktop motherboard that had qualified ECC RAMs, and pop in a AMD desktop CPU for a low-budget workstation.

[Oese]

@ Leeghoofd: Where is this text from? Is it recently posted/mailed somewhere or quite long ago?

[Leeghoofd]

Talking to Maki in a moment, keep you lot updated ( if I am allowed )

[wez]

@ Leeghoofd: Where is this text from? Is it recently posted/mailed somewhere or quite long ago?

http://www.xtremesystems.org/forums/...=1#post4969164

[Warwian]

Talking to Maki in a moment, keep you lot updated ( if I am allowed )

Please do!