Sunday, September 18th 2011

AMD's Bulldozer 8.4GHz+ OC Achievement: Cooled to Near-Absolute Zero

AMD's Bulldozer 8.4 GHz+ OC Achievement: Cooled to Near-Absolute Zero

TechPowerUp recently brought you news on AMDs fantastic overclocking achievement with their new processors. Now we can tell you how it was done: cherry-picking the chips and slapping on some water cooling isn’t quite enough. AMDs new processors can operate at much lower temperatures without displaying the “cold bug” - where it just gives up and goes home - and performance scales very well at super-low temperatures. The problem is that the cold affects lots of things such as timing, but more importantly, power circuits, which stop switching and just fry everything in sight - surely one to avoid. AMD senior manager of social media, Simon Solotko explains in detail how it was all done, using both liquid helium and liquid nitrogen to make the poor processor really cold. The new processor had these great qualities, according to Solotko:
It was able to take a lot of voltage, extremely low temperatures, extremely high frequencies," he said. "It was very durable under extreme overclocking. So that was awesome. So it worked well, it scaled well, it responded to cold well – all the right variables.
This overclock is an impressive feat and it will be interesting to see if Intel can match it.Source: The Register
Add your own comment

116 Comments on AMD's Bulldozer 8.4GHz+ OC Achievement: Cooled to Near-Absolute Zero

#1
trickson
OH, I have such a headache
cadaveca said:
I think part of the problem is that, for example, with the phenom I launch, we were told by Leslie Sobon that 3GHz cpus were coming. They did, but not with Phenom I, at least, not for sale @ 3 GHz.

However, AMD did show CPUs running @ 3 GHz.

So, now we got one CPU @ 8.4 GHz. Going by the past, this may very well be the only chip that ever does that, and many are a bit hesitant to expect much from AMD, as they have in the past made promises, just to fail to deliver.

I don't think anyone really wants to ruin anyone else's fun, but it will be quite disappointing for many if they set their expectations to high, so many are quick to quash any ideas that seem overly excited.

AMD can only blame itself for customer reactions, after all.
Best statement in the entire thread ! :respect:
Posted on Reply
#2
Steven B
cadaveca said:
Um...there was more than VID binning, IMHO, just FYI. I see more than just VID listings there:
http://www.techpowerup.com/forums/attachment.php?attachmentid=43658&stc=1&d=1316449766




Of course, personally, i think pre-binning for something like this is not really important(and expected), but it does show that not all chips are able to do this, which does raise question to just how important, or not, events like this are. At least AMD isn't claiming that all chips can do this, but they DID SAY "expect higher clocks soon".
For a public demo, i don't think there is any public display of OC like this that use non binned CPUs. Its embarrassing if one CPu can't even do 6ghz, and it takes too much time otherwise.

Yea well technically when you go for an extremely high frequency you want ES processors, whether its from intel or AMD, because usually they have unlocked/no set TDPs and thus can have very leaky transistors, high leakage = higher frequency potential. That is one reason you see Intel ES CPUs OC higher than retail, with Sandy bridge CPu PLl overvoltage changed that, but none the less. These frequencies are amazing, the only down part is that it doesn't take a nice CPU to hit those speeds, I think celerons owned the record before these buldozers. i just hope they can get to a moderate high frequency to beat a 6ghz SB-E, if we can see them benching at 1ghz higher AMD might be in luck!
Posted on Reply
#3
cadaveca
My name is Dave
Yes, I do expect hard bininng for demo like this.


I DO NOT expect anyone to claim otherwise. I wasn't evne going ot post in this thread; I just wanted to get that picture in there. I probably shouldn't have even bothered.

I've said before, it's cool to see, and a bit exciting, but I'll feel much better with chips in my hands to play with. Doesn't matter what ANYONE says, I refuse to make any judgments on Bulldozer until I do it myself, as many sources out in the wild are suspect, and this whole Bulldozer story has caught out quite a few people, for sure.

And myself, I'm fully aware of what chips were doing what. I still have a 347 cellery sitting here on my desk. Didn't keep it for no reasion. ;)
Posted on Reply
#4
EarthDog
cadaveca said:
I think part of the problem is that, for example, with the phenom I launch, we were told by Leslie Sobon that 3GHz cpus were coming. They did, but not with Phenom I, at least, not for sale @ 3 GHz.

However, AMD did show CPUs running @ 3 GHz.

So, now we got one CPU @ 8.4 GHz. Going by the past, this may very well be the only chip that ever does that, and many are a bit hesitant to expect much from AMD, as they have in the past made promises, just to fail to deliver.

I don't think anyone really wants to ruin anyone else's fun, but it will be quite disappointing for many if they set their expectations to high, so many are quick to quash any ideas that seem overly excited.

AMD can only blame itself for customer reactions, after all.
I have to disagree a bit with a couple of things...

1. You saw yourself from the picture in the news thread how many were labeled 8Ghz and how many were close. At least a few. Chew has also told us his method of selecting the chips and it wasnt nearly as in depth/binned as you think it was (quote to follow)...

2. If people here are setting their expectations that high, they just are clueless about how it was achieved.

2a. This press release wasnt really meant for average joe TPU so much as it was for others that have an idea about extreme overclocking and what that means. Generally, and there are many exceptions, the higher a CPU can clock like this, the higher it can go with normal cooling...and that it scales quite well under cold and voltage.
Posted on Reply
#5
digibucc
cadaveca said:
but it will be quite disappointing for many if they set their expectations to high, so many are quick to quash any ideas that seem overly excited.

AMD can only blame itself for customer reactions, after all.
i don't mean to be rude, but obviously. anyone who gets their hopes up because of this can blame themselves though.

my point is simply take it by itself. a silicon chip can now do instructions at 8.4ghz. that is technologically awesome and i don't think anyone saying so is expecting their retail chip to do anything close.
Posted on Reply
#6
cadaveca
My name is Dave
Oh, I agree, 100%(both with you, EarthDog, and you, Digibucc). This means much different things to the extreme user. My problem is that the extreme user, those that go to LN2 and better, are a dying breed, even in enthusiast circles.


Which is why I question how this event was marketed in the first place.


You get what I mean? I'm not trying to downplay how cool this is...but the fact remains it's only really cool to a few people, globally. like less than 0.000001% of the world's population cares that this CPU design can set a world record, and even less care that they can do it ONLY near absolute zero.

I mean, hasn't Intel shown 100GHz tansistors, or some such nonsense? How about a 100-core CPU? This is just as unimportant to daily life, and should not, really, be hyped as any real technological feat.

At least, not so early before CPUs are in retail. If this was done on launch day, I wouldn't have taken this side of the argument. I'd be busy trying to get chips to get these clocks myself.

Except reality has me sitting here analyzing it instead.
Posted on Reply
#7
trickson
OH, I have such a headache
cadaveca said:
Oh, I agree, 100%(both with you, EarthDog, and you, Digibucc). This means much different things to the extreme user. My problem is that the extreme user, thos ethat go to LN2 and better, are a dying breed, even in enthusiast circles.


Which is why I question how this event was marketed in the first place.


You get what I mean? I'm not trying to downplay how cool htis is...but the fact remains it's only really cool to a few people, globally. like less than 0.000001% of the world's population cares that this CPU design can set a world record, and even less care that they can do it near absolute zero.
Agreed . Nothing here has been shown to the MASSES at to what BD will be offering up performance wise . Most people could careless if a chip can clock to 8+ GHz if the performance is not there . It is a PR stunt and I hope to GOD it doesn't back fire on them .
Posted on Reply
#8
EarthDog
To quote a song from the 90's: This is ponderous man...really, ponderous.
Posted on Reply
#9
digibucc
cadaveca said:
You get what I mean? I'm not trying to downplay how cool htis is...but the fact remains it's only really cool to a few people, globally. like less than 0.000001% of the world's population cares that this CPU design can set a world record, and even less care that they can do it near absolute zero.
i get it.

still though, i am not an overclocker. my hardware is perfect for it, but i just don't have an interest. i don't attend any shows, i don't do any of that. i just think world speed cpu records are cool, and always have.

i think more people appreciate it just as a technological advancement. than you seem to think. they don't have to understand the architecture to know 8.4ghz is a high number, a world record is a cool thing (imo) and a jump in speed (even under such conditions) is always welcome.

something of course can be said for how they released it, but i really just think people are sick of waiting and so are being cynical. i choose to take it on it's own.
Posted on Reply
#10
EarthDog
digibucc said:
just as technological advancement than you think
:toast:
Posted on Reply
#11
eidairaman1
The Exiled Airman
heky said:
Who are you to tell me i dont have to stay. I have the right to express my opinion. And for your information, i am not an intel fanboy, i have many amd rigs at home. For my main rig i am currently using a intel i7 2600k, since it just brigns the best performance for money atm.
they way your portraying your attitude in every AMD thread proves that you are being one.
Posted on Reply
#12
erocker
Senior Moderator
Keep the conversations on topic and not pointed towards others.

Thank you.
Posted on Reply
#13
lashton
cdawall said:
this record is for silicone based cpu. this record was also done on LHe not LN2...
Considering you are correcting everyone else, this was done on LN2 AND LHe, you may want to read the original article again!
Posted on Reply
#14
lashton
this OC was done because people said that the Bulldozer would not be overclocked (also as a PR stunt) well they were proved wrong
Posted on Reply
#15
Steven B
i don't think anyone said bulldozer wouldn't OC well, a 4.8ghz or so turbo(i am unsure of the actual turbo number but i think it was 4.8 or 4.4ghz) shows that a CPU can do 4.8 or 4.4*ghz under manufacturer specs means its going to OC very high, and AMD not really having issues with cold means that you can expect some very high frequencies.

I just hope that these can bench at 7ghz consistently(also AMD later batches tend to be better than ES, just because the processing tech gets better(llano for example)), and hopefully be able to beat a 6ghz SB-E, if AMD can pull that off, 1ghz higher, same core numbers or even higher. If they can pull that off, average joes will buy them up and OC them to hell and not spend a fraction of what they would spend on a SB-E system. that is just the truth, AMD has beaten Intel before, if anyone here is old enough to remember the P4 days, AMD rocked, Then Intel gained room. I left the OC sense when Intel sucked, and my P4 561 held its own WR(still does), i come back and Intel is way ahead, now maybe the tides will turn. From what I have seen ambient temps and so Intel is far ahead, even with bulldozer, but this gives AMD a chance to win over the extreme overclockers.
Posted on Reply
#16
trickson
OH, I have such a headache
Originally Posted by cdawall
this record is for silicone based cpu. this record was also done on LHe not LN2...
Originally Posted by lashton
Considering you are correcting everyone else, this was done on LN2 AND LHe, you may want to read the original article again!
You just got SCHOOLED ! :laugh: :roll:
Posted on Reply
#17
Steven B
i have a question how do you use LN2 and LHe?
Posted on Reply
#18
Fx
DrPepper said:
Generally the cpu's which have achieved world record oc's can achieve very high speeds on air and water.
+1

I agree, I dont try to do any extreme overclocking. this year is when I finally tried out water-cooling for my cpu

I dont ever overclock to any speed that is a pain to stabilize so I appreciate when I can take a cpu and easily overclock for a major bonus
Posted on Reply
#19
qubit
Overclocked quantum bit
Fx said:
+1

I agree, I dont try to do any extreme overclocking. this year is when I finally tried out water-cooling for my cpu

I dont ever overclock to any speed that is a pain to stabilize so I appreciate when I can take a cpu and easily overclock for a major bonus
+1 again. I got my E8500 CPU because I knew there was a very good chance I'd get a gig overclock out of it - and I did. All this with a few simple tweaks to the BIOS and good quality air cooling. Heck, most of the settings I just left on Auto! To get much more out of it though, I have to start getting a better cooler and fine tuning all the settings to make it stable, so I'm happy with the 4.11GHz I've got out of it. ;)
Posted on Reply
#20
cadaveca
My name is Dave
Steven B said:
i have a question how do you use LN2 and LHe?
It's in that video, no? ;)
Posted on Reply
#22
NirXY
lashton said:
Considering you are correcting everyone else, this was done on LN2 AND LHe, you may want to read the original article again!
I suggest you go and read it yourself first..
LHe was used for the 8.4 record, LN2 (cheaper) was used to bin the chips (lower o/c).

not only they didn't use it together, it was a complete process to completely remove the LN2 residues before using LHe..

on topic, too sad they disabled 6 cores, i'm curious to max o/c on all 8 cores..
I really hate PR stunts :banghead:
Posted on Reply
#23
cadaveca
My name is Dave
Steven B said:
ahh i didn't watch it
They used LN2 to cool the CPU down as low as they could, then swapped to LHe. They did the same @ the blackops event(although they screwed up at the blackops event and didn't hapve proper fittings, and wasted tonnes of LHe waiting for it to build up ice to seal. was very scar to watch them panic when the LHe leaked all over the garage on the streamcast).
Posted on Reply
#24
EarthDog
Those hose popped out at this event too, was quite funny. :p
Posted on Reply
#25
cadaveca
My name is Dave
EarthDog said:
Those hose popped out at this event too, was quite funny. :p
it didn't pop out at the blackops, event, it was never in!

And yeah, i saw it pop. Crazy stuff done by crazy guys, setting a crazy record. Seems fitting. :laugh:
Posted on Reply
Add your own comment