AMD Zambezi news, info, fans !

[Opteron146]

...so it's th 19th of Sept, anything New?

That ...mber 19th is/was the joke of the year, just saw these slides:


Source: http://www.hwupgrade.it/forum/showpo...ostcount=22930
Thx@deadohiosky

comment:

Hi,

According to this page http://www.overclock.net/amd-motherb...tion-list.html

GA-970A-UD3 has low-rds VRMs too... plus heatsink

Think I'll get this cheap baby soon !

Why would you trust a 3rd party website? If Gigabyte writes "Ultra Durable 3 - classic":
619.jpg

Then it means that the board is only equipped with double CU layers and solid caps. No ferrite chokes or low-rds VRMs. The correct lable for these would be "ultra durable 3":

621.jpg

Believe whom you want, I believe Gigabyte ;-)

[EniGmA1987]

I love how everyone thinks they are so cool now being like "oh I didnt think it was September, see this slide here?" lol


I really wish it was today I need a new processor NOW, not later.

[Oliverda]

That ...mber 19th is/was the joke of the year, just saw these slides:

[Dirk Diggler]

That ...mber 19th is/was the joke of the year, just saw these slides:


Source: http://www.hwupgrade.it/forum/showpo...ostcount=22930
Thx@deadohiosky

comment:


Why would you trust a 3rd party website? If Gigabyte writes "Ultra Durable 3 - classic":
619.jpg

Then it means that the board is only equipped with double CU layers and solid caps. No ferrite chokes or low-rds VRMs. The correct lable for these would be "ultra durable 3":

621.jpg

Believe whom you want, I believe Gigabyte ;-)

Those are Phenom I slides

[aznsniper911]

http://www.pcgameshardware.de/aid,83...e/?iid=1555079

What do people make of that?

*Note i did not see this in the last couple of pages so I'm sorry if this is a really old repost*

[Mechanical Man]

SENSORED

What do people make of that?

The one we dont mention.

[Manicdan]

http://www.pcgameshardware.de/aid,83...e/?iid=1555079

What do people make of that?

*Note i did not see this in the last couple of pages so I'm sorry if this is a really old repost*

its in a language i dont read, but i see OBR, and the charts have a very familiar layout to the bad CB11.5 score we saw before. but some of these numbers look great, as if AMD improved massivly where they were weak with K10, and then surpassing intel in others. i would call such a product worth ~300$

[FlanK3r]

this all results are from him...so, nothing new

[Pestilence]

what?

http://www.pcgameshardware.de/aid,83...e/?iid=1555079

What do people make of that?

*Note i did not see this in the last couple of pages so I'm sorry if this is a really old repost*

The one who uses the dark side of the force

[imamage]

Those are Phenom I slides

Got me XD

[MaddMutt]

That ...mber 19th is/was the joke of the year, just saw these slides:


Source: http://www.hwupgrade.it/forum/showpo...ostcount=22930
Thx@deadohiosky

comment:


Why would you trust a 3rd party website? If Gigabyte writes "Ultra Durable 3 - classic":
619.jpg

Then it means that the board is only equipped with double CU layers and solid caps. No ferrite chokes or low-rds VRMs. The correct lable for these would be "ultra durable 3":

621.jpg

Believe whom you want, I believe Gigabyte ;-)

Thank you for making the previous posts understandable for us with less computer understanding

[dess]

SPi has never lost relevance Dave
SuperPi AMD vs AMD has always been comparable, Intel vs intel has been comparable.

I'm sorry, but I think it mostly lost it, as it's using x87 instructions primarily that modern CPUs aren't optimized for anymore. What if f.ex. Buldozer is even less optimized for x87 execution? In that case the new uarch would seem to be worse than K10, but then it will be worse only in a rather irrelevant (on the desktop and mobile) thing...

x87's peculiarity is that it supports Extended (80 bits) Precision which is needed almost only for some scientific algorithms. Otherwise it's rather limited. Most modern softwares use SSE2 instructions, instead, which support only Single (32 bits) and Double (64 bits) Precision, but is faster and more flexible. (Even if used for scalar computations, and indeed in SIMD mode, when several instructions are executed parallelly. Most modern SW use both.)

I disagree, FP is used extensively for gaming, and I'm not talking about 3D rendering (GPU has taken care of that for the last decade), rather I'm referring to physics. Physics accounts for a huge amount of FPU utilization in modern games. Physx would be the out here, however it's rarely used, and Radion has no support for it.

Aren't off-line 3D renderers use FP, as well? Tried to study them with PerfMonitor but unfortunately there are no separate meters for SSE and SSE2. (BTW, they use some x87, as well, but very-very little.) Of course, most people don't run such programs, but there are many hobbists that do. Not mentioning render-farms using PCs.

Although, there are initiatives to get FPUs more involved in this, as well, but it takes years.

[Dimitriman]

http://www.pcgameshardware.de/aid,83...e/?iid=1555079

What do people make of that?

*Note i did not see this in the last couple of pages so I'm sorry if this is a really old repost*

If those benchmarks are true at stock speeds, then Bulldozer is a success.
Keep in mind he has an early engineering sample so performance should improve on final silicon.

[cold2010]

Bulldozers to the FX-8120P [B2 latest version of the core differences between the measured Full CPU_NB]

http://forum.coolaler.com/showthread.php?t=273986

[Mats]

http://www.pcgameshardware.de/aid,83...e/?iid=1555079

What do people make of that?

*Note i did not see this in the last couple of pages so I'm sorry if this is a really old repost*

That's __°br__ . .
http://www.pcgameshardware.de/aid,83...e/?iid=1534720

Quelle: OBR

[danielkza]

Bulldozers to the FX-8120P [B2 latest version of the core differences between the measured Full CPU_NB]

http://forum.coolaler.com/showthread.php?t=273986

"[Quelle: OBR]"

Move along, nothing to see here.

[Dimitriman]

Bulldozers to the FX-8120P [B2 latest version of the core differences between the measured Full CPU_NB]

http://forum.coolaler.com/showthread.php?t=273986

Is he testing with DDR3 @ 1333 Mhz?

[Andi64]

Extremely poor L3 cache performance. That's why whe should take all these pre-release benchmarks with a grain of salt...

[JumpingJack]

given the ipc of those applications and the #fpu ratio. i don't see a big problem. Low ipc shows alot of dependant ops or lots of bubbles. So while it is ... <snip>

this code would run great on K8 architectures. (if they can predict and feed enough to their execution resources)

All good points but my intention was simply to point to data supporting the idea that there are cases where x87 is used extensively. Games running CPU code paths of Physx will evidently use x87 over SSE instructions.

[dess]

Well to be honest,they already said so @ Analyst Day presentations. They stated "80% of CMP" approach while having less die area and less power draw. So this 80% means 0.8x the performance of "native" X8 Bulldozer,done the old way of stacking cores next to each other.

Well, they've indicated 80% gain here, so +80%, therefore 0.9x of itself as a true 8-core.

13063_cmt.jpg

BTW, if it finally turns out the IPC is indeed lower than with K10, per core, I have an excuse for JF. At the beginning they've called a module a core (and a core an integer cluster), right? Well, perhaps an engineer referred to a module really when (s)he told JF that "IPC is higher, per core"...

[spursindonesia]

I simply expect 6/4 cores desktop BD (Zambezi) to be faster than 6 cores Thuban/4 cores Deneb @the same clock, ATLEAST on average. Anything else, then JF-AMD is a LIAR and Terrace/Savantu/Shintai/Any other Intel trolls were right on the money since the beginning.

[Dumo]

It is better to refrain from name calling

Imo, no one promise anything to anyone, especially for a cpu that still not available at retail channel.

Probly BD slower @ stock clock but faster than any cpu @ 7.5Ghz benching. Lol

[flyck]

Extremely poor L3 cache performance. That's why whe should take all these pre-release benchmarks with a grain of salt...

You cannot blame the l3 cache for that.... the l2 cache is even slower.. Actually if it didn't have a better latency than the main memory, they would be far better of removing those caches.

(so yeah i agree, these samples are crippled to the bone or something is terribly wrong...)

as xsecret said, the BD we have seen have serious cache issues. Even brazos has higher l2 bandwidth at half the core speed...?

[-Boris-]

And check out the L1 write. If those caches were where they should the performance would be much better.

[FlanK3r]

maybe some BIOS problem???