AMD Officially Denies Bulldozer Delay !

[EniGmA1987]

"...we indicated early summer for desktops and late summer for servers. And that's still where we're at."

Ummm....

I really dont know why everyone is throwing around all sorts of ideas and dates and everyone is saying all different things. The slides that leaked from under NDA back in January said launch or release info would be at E3 in early June. Since then nothing has been said for a change and now the PR department said it is still on schedule for that same release area as the slides from January said.

[Hedghawg71]

Denial is a river in Egypt. I'll have my BD chip by mid July Goddern it!

[audioave10]

I expect AMD to make a full anouncement at E3 in June. They have always said they would be there.

[jmm5351]

guys have a look at this. I have no clue if it is accurate but if it is then all I can say is yeah

http://translate.google.com/translat...130067349.html

Amd Zambezi scored over 80,000 points on 3DMark Vantage CPU Score

[chew*]

guys have a look at this. I have no clue if it is accurate but if it is then all I can say is yeah

http://translate.google.com/translat...130067349.html

Amd Zambezi scored over 80,000 points on 3DMark Vantage CPU Score

PHYSX enabled......

[DarthShader]

Old stuff... with PhysX on, a 2600K scores 70K or slightly more.

[Ccaution]

Just above Llano we will have the long awaited Bulldozer CPU. AMD originally wanted to launch Bulldozer at Computex but performance issues with its B0 and B1 stepping chips pushed back the launch. Now we're looking at a late July launch with B2 silicon, but performance today is a big unknown. Apparently the performance of B1 stepping silicon doesn't look too good.

http://www.anandtech.com/show/4365/c...8-motherboards








edit: http://www.xtremesystems.org/forums/...d.php?t=271578 ahaa, there was a thread! /nevermind

[zalbard]

Late July...

Apparently the performance of B1 stepping silicon doesn't look too good.

Fail.
And would explain the lack of leaked benchmarks.

[DarthShader]

And would explain the lack of leaked benchmarks.

But we did have B1 stepping leaked benchmarks...

[zalbard]

Photoshopped ones, you mean? lol. Poorly done ones at that...
There is almost no information available to make a reasonable guess about BD's performance atm.
Those that looked somewhat believable showed way too low performance to be true. Unless B1 (or was it B0?) is really bad and B2 is going to make a significant leap forward.

[bryanw1995]

Gosh when I was 16 I was enamoured with my P4 3.2C Northwood and 9800Pro. Oh how things have changed.

When I was 16 I was enamoured with my commodore 64!

[jmm5351]

Old stuff... with PhysX on, a 2600K scores 70K or slightly more.

The core I7 2600k scored 64146 in the test and if you read the article it says a gtx 580 was used for that test. The 8110 scored 81917 using only a gtx 560. If this is not real then it sucks but if it is real then that's amazing.

[informal]

Guys forget about those benchmarks,none of the are verifiable. Just wait ,it won't be long now.

[chew*]

The core I7 2600k scored 64146 in the test and if you read the article it says a gtx 580 was used for that test. The 8110 scored 81917 using only a gtx 560. If this is not real then it sucks but if it is real then that's amazing.

really it's not.

8 cores verus 4 cores plus 4 ht threads.......it's inline with what you would expect from an 8 core AMD cpu.....

It's a no brainer to calculate that amd will win in a thread to thread compare against intel, it's always been that way.

The problem is no one compares thread to thread, they go by core count which once again as I said before we will beat sandy in multithreaded, lose to gulfy in multithreaded.

Reviewers will however show this as an 8 core AMD cpu beating a 4 core intel but losing to a 6 core intel.......

[DarthShader]

Reviewers will however show this as an 8 core AMD cpu beating a 4 core intel but losing to a 6 core intel.......

Oh, you got your BD sample already?

[devguy]

Bulldozer listed already in China?

[chew*]

Oh, you got your BD sample already?

no, the calculations can be made using current cpu's. disable cores on thuban run 4 core, rerun with 6 cores, check variance in score, multiply by 2 you get a theroretical 8 core which will beat sandy but still be far enough from gulfy that even a cpc improvement would be tough to make up the difference.

The reality is 8 threads are handicapped to 12......

[Hondacity]

no, the calculations can be made using current cpu's. disable cores on thuban run 4 core, rerun with 6 cores, check variance in score, multiply by 2 you get a theroretical 8 core which will beat sandy but still be far enough from gulfy that even a cpc improvement would be tough to make up the difference.

The reality is 8 threads are handicapped to 12......

drats i was expecting this calculation...

will this be ok for 06/05? if no cb.... 6.4ghz x 8 core?

[chew*]

drats i was expecting this calculation...

will this be ok for 06/05? if no cb.... 6.4ghz x 8 core?

till we know what to expect clock wise from BD on ln2 I will go along with the prediction that 01/03/05 are still deneb territory for best 3d scores on AMD.

06 is up in the air as well as vantage and 3d 11.

*note* these are just educated guesses.

[informal]

no, the calculations can be made using current cpu's. disable cores on thuban run 4 core, rerun with 6 cores, check variance in score, multiply by 2 you get a theroretical 8 core which will beat sandy but still be far enough from gulfy that even a cpc improvement would be tough to make up the difference.

The reality is 8 threads are handicapped to 12......

One problem with theory above is that 1 Bulldozer core is not equal to 1 Phenom II core. It is supposed to be faster. As an example we have 50% throughput gain(in server workloads) with 33% more cores.Knowing nothing scales perfectly with cores,even server workloads,you end up with better cores than Deneb class.How much better and in what type of workloads(integer,fp) is unknown.

[chew*]

One problem with theory above is that 1 Bulldozer core is not equal to 1 Phenom II core. It is supposed to be faster.

sure but then the question remains....can bulldozer match deneb top end clocks.....thuban can't and has not.

Thuban has also not dethroned deneb in 01/03/05 or even 06 for that matter....

We know for a fact there are 6.6 05 deneb chips......the question we can't answer is how much faster per core are they and how high do they OC......

If we follow thuban trend adding 2 cores cost clocks.

[informal]

Well for one you have 32nm SOI highk/mg process.If it was just a shrink,without highk/mg it would bring a lot of headroom,but with highk/mg it brings a lot more to the table.Second,the way how cores are organized(modules),they draw less power than conventional cores,so you get more clock within the same power envelope.Third,the design has reduced FO4 which means a bit longer pipeline ,which in turns translates to around 20%+ potential clock raise (FO4 supposedly went from 22 to 17). Everything points out to higher clock potential of Zambezi,even with full 8 cores loaded.

[chew*]

I would be less inclined to make assumptions. I guess we shall see soon.

Sandy was a radical change to.....what it got was lesser clocks and wicked coldbugs. Without its IPC it would be a flop. It will also spawn a new commercial most likely ( so easy a dummy can do it )

The titantic looked good on paper and was unsinkable.

We all know how that ended.

I'm a peccimist, however I am one of those that are never dissapointed due to that fact

[FlanK3r]

dont fear about BD as Titanic . Better be a bit positive. I think, performance will be near 2600k in average results. For multithreading, think, can be 4 modules better than 4C/8T and in single thread a bit better for Intel CPU. With 32nm or step from higher to lower process is improve performance and OC potential ussualy.

[XRL8]

But we did have B1 stepping leaked benchmarks...

The only semi credible ones were with B0 stepping in cinebench using 890FX mainboard.