AMD Bulldozer reaches 8.43 Ghz on liquid Helium

[FlanK3r]

OMG, that is a 2! Look at the actual 7 right above the 2 and compare! Whats this mean chew*? Default vcore?
Sorry if I'm behind the time with this, but just noticed and haven't yet read all 6 pages!

Attachment 120151

I think, its VID of CPUs

good to see ya around! i made a similar post the other day regarding process improvements & if we see similar improvements on 32nm as we did on 45nm, who knows, why stop at 9ghz, shoot for the stars, maybe 10ghz will be taken w/ 32nm! 45nm went from ~6ghz -> 7.3ghz+ so if 32nm is already at 8.4ghz

Maybe with Piledriver?

[M.Beier]

59 with the gigaybte "special " bios ....

Disabling cores, or turbo multiplier^^

[n!Cola]

http://www.geek.com/articles/chips/i...2011-20000726/

Some of the comments are very amusing

[Lanek]

Disabling cores, or turbo multiplier^^

well i was not clear.. just this specific bios ( posted by dino ) was allowing a max multi of 59x in bios instead of 57x .. does it have work ? i don't remember on the UD7 section if someone have success .. my 2600K will not boot anyway at 5.5ghz+, ( anyway not under water )

[Snow Crash]

Conratulations on a job well done, I hope you all jump into the fray to beat your own incredible record
Thinking of all the accumulated time, experience, knowledge, and research involved to acheive this goal fills me with serious respect for the people involved, not only for the record itself (did I say I think that what you've done is freaking amazing???) but also for setting the new bar for xtreme overclockers to aim for.

[CaptMorgan]

Snow Crash does make the point I believe "I" myself did not say- Congratulations on an Amazing Achievement!!!!

[Pestilence]

http://www.shopblt.com/cgi-bin/shop/...er_id=!ORDERID!

Have an in stock date of next Thursday for the 8150/8120/6100

[Apokalipse]

I.... wouldn't interpret it that way

It took 6 years + liquid Helium to beat the LN2 MHz of a good Celeron by 2%.

Well, Netburst was designed for frequency and only frequency.
Today's CPU's are designed with much higher IPC, core count, and do more types of instructions etc, so they do a much wider range of things

[bamtan2]

http://www.shopblt.com/cgi-bin/shop/...er_id=!ORDERID!

Have an in stock date of next Thursday for the 8150/8120/6100

22nd. interesting. $35 extra and $45 extra between them. not much difference. and if they're unlocked, I don't understand the diff between the two top ones. hope it is an interesting week coming up.

[l0ud_sil3nc3]

Then as your rebuttle says, it should be NO issue to throw these types of "numbers" down with a board with MAX memory then correct??? (Depending on "insulation" of course)???

These type's of bench's are missleading to the "general" enthusiast community, 1 core, 2 core, on a 8 core proc???

Not too impressed IMHO, when this can be done with a full loadout of max board memory, and 2+ pci-e vid cards, then I will give props.... Kinda like the guys the "build" open-air rigs, that bench high numbers, but couldnt keep em "stable" in a case....

troll comment

this a sick achievement, congrats guys you have sold me as long as retails are somewhat similar

[Movieman]

troll comment

this a sick achievement, congrats guys you have sold me as long as retails are somewhat similar

My sentiments exactly and what he really missed is that the 8307 was on a single core by Intel and AMD did the 8429 on TWO cores being used..

[Embalmed]

My sentiments exactly and what he really missed is that the 8307 was on a single core by Intel and AMD did the 8429 on TWO cores being used..

I hope he was using that time at 8429 to run BOINC

[64NOMIS]

Thank You.

We are very fortunate to do this kind of work and everyone is thrilled we could make the record happen and share it with you.

Personally I appreciate the encouragement and collaboration. We have done a number of small scale experiments which have been informative and fun.

This one was a whopper. I remain shocked that we were able to break the record in a small span of time. As those we have worked with in the past will readily acknowledge, Helium rarely behaves on cue.

I appreciate the work of Sampsa, SF3D, K|ngp|n, Elmor, giorgioprimo, and many others leading up to this event. Special thanks to HWbot.org, CanardPC, CPU-Z, and of course, XS.

I also want to thank runmc who built us one hell of a phase unit. More on that later.

64NOMIS

[Stewie007]

Question is, how high can it go on all 8 cores?

[Movieman]

Question is, how high can it go on all 8 cores?

We'll see that soon won't we!

[Stewie007]

Can't wait.

[\/icious]

Freakin awesome. Congrats guys.

[halofanman]

Some of the comments are very amusing

lol

Moore's Law:

early 2001: 1 GHz
mid 2002: 2 GHz
early 2004: 4 GHz
mid 2005: 8 GHz
early 2007: 16 GHz
mid 2008: 32 GHz
early 2010: 64 GHz
mid 2011: 128 GHz

lol not one person in the comments from back in 2000 mentioned a processor with multiple cores, everybody just assumed GHz would go up
even with all 8 of Bulldozers cores running at 6GHz 6*8=48 according to their predictions that would put it somewhere in 2009..

[JaD]

Moore's law has nothing to do with CPU frequency.

[-Boris-]

lol


lol not one person in the comments from back in 2000 mentioned a processor with multiple cores, everybody just assumed GHz would go up
even with all 8 of Bulldozers cores running at 6GHz 6*8=48 according to their predictions that would put it somewhere in 2009..

Moores Law says that the complexity doubles roughly every two years. That means transistor count.

Let's check it out, based on K7 with 22m trannies, what we got is inside brackets:

1999 - 22m (Athlon K7 22m 250nm)
2001 - 44m (Athlon XP 37.5m 180nm)
2003 - 88m (Athlon 64 103m 130nm)
2005 - 176m (Athlon X2 243m 90nm)
2007 - 352m (Phenom 452m 2008 65nm)
2009 - 704m (Phenom II 702m 45nm)
2011 - 1408m (Llano 1450m 32nm)

Looks like it's still in action? You can also look at the nodes, every new processing node roughly doubles the transistor count, as you can see from the list it follows the law as well.

[R101]

Well, Netburst was designed for frequency and only frequency.
Today's CPU's are designed with much higher IPC, core count, and do more types of instructions etc, so they do a much wider range of things

If I remember correctly, integer units in netburst ran at 2x the frequency, so it's a bit tricky since in case of this BD it is not a whole chip that ran at 8.xx. And if we consider it a free interpretation - this Cellys ALUs were screaming at 16 GHz..