Tuesday, September 13th 2011

AMD FX Sets Guinness Record for Clock Speed

by
btarunr
 Discuss (225 Comments)
Weeks ahead of its market launch, AMD pulled off a nice PR feat by setting making its trusty squad of overclockers, Sami Mäkinen, Brian Mclachlan, Pete Hardman, and Aaron Schradin set a new clock speed world record (as in Guinness World Record). With just one of its four modules enabled, the eight-core FX-8150 engineering sample was overclocked to a stunning 8429.38 MHz. The chip was able to tolerate a brutal core voltage of 2.016V. Even for a one-in-a-million cherry-picked chip, those are staggering numbers.

8429.38 MHz was achieved using a base clock of 271.92 MHz, with 31.0X multiplier. The memory used was a Corsair Dominator GT single module, which apparently tolerated 3:10 DRAM ratio and timings of 2-16-2-22. That's right, 2-16-2-22. ASUS Crosshair V Formula seated the platform. Cooling was care of a custom liquid-nitrogen evaporator setup. The team used liquid nitrogen as its cooling medium, and switched to liquid helium halfway, which has a lower boiling point. The team cherry-picked chips from the best lots on-site.
A video of the feat follows.


This feat was more of a hit-and-run, in which the system could run at the desired frequency stable enough to make a CPU-Z validation, no proper stability testing was done. AMD claims that frequencies over 5.00 GHz were possible using sub-$100 cooling solutions (now that can be anything between a high-end heatsink and a cheap closed-loop liquid cooler). AMD did a similar overclocking feat ahead of its Phenom II processor launch.
Source: Overclockers.com

Related News

Add your own comment

225 Comments on AMD FX Sets Guinness Record for Clock Speed

#201
HTC
EarthDogNow? Its always been that way (well since multi cores were out). When shooting for highest clocks you should always disable cores.
Let me quote hipro5 from XS:
Well yes - you read it well!...... :D

An EASY way to KILL your i7 980X cpu is VERY SIMPLE... :D and :(


IF your cpu sucks under LN2 or so, an EASY way to kill it is the following:

Power up you mobo (it doesn't matter what mobo you have)
Go into cpu features
Disable Hyper Threading.
Select 2 Cores for slow die
OR
Select 1 Core for FAST die
Save and exit and.........you're ready - No overclock no nothing - :mad:


Here's my story:
I had a decent i7 980 Cpu, benching Vantage ~5900MHz
I putted on the GIGABYTE UD7 mobo
I was playing OK UNTIL I tested the Super-Pi 1M
Disable HT
Test 1M
Disable 4 cores and I left only the 2 of them
Playing 1M OK
Playing and fooling around with 2 cores for a couple of sessions
OK
ENABLE ALL cores and HT
NO Windows at 5800MHz which was EASY before
NO Windows at 5600MHz
NO Windows at 5400MHz
NO Windows at 4000MHz
Booted into Windows with full cores and HT at DEFAULTS = NO GO

I said f(l)uck!...The cpu degraded THAT FAST.... :(


NO

Another Cpu on EVGA
Decent one too

Playing, testing it and fooling around with full cores and HT and LN2 = OK
Next day I had no LN2 to play with so I did some testing on air with DEFAULT EVERYTHING
Testing Super-Pi with 2 cores = OK
I then go into bios and pick ONLY ONE CPU Core with No HT.
Save and exit.
Boots and freezes
Pressing Reset
NO BOOT

CPU = DEAD!.... :mad:


I then call to a friend of mine who is on air and he doesn't Overclock at all.
I go to him
Get into bios and disable HT and pick one core
Did you do it?
Yes?
OK - Now save and exit -
OK

........

........

George it doesn't booting!.... :eek:

I go - clear c-mos and retry.

George no go, what happened? you killed my cpu?

Am I at your home? How could I kill your cpu?..... :D

SHI(f)t!!!


So if Intel hasn't predicted such behaviours, RMA THE BIATCHES guys..... :mad: ;)

DEFAULT with ONE core enable and no HT = DEAD.....

RMA - Period



I suggest to the motherboard manufactures to leave into their bioses ONLY the option of the Hyper Threading and REMOVE the option form the core's selectance with a bios update. :(



.
And this is why they stopped searching single thread frequency records for 980x and such.


On topic: congratz AMD for a new frequency record :toast:

Here's to hoping performance comes along with frequency: frequency without performance isn't much to hope for, IMO.
Posted on Reply
#202
EarthDog
Way to find the exception to the rule! That was a known problem with gulftowns. Too bad nobody in their right mind would shoot for a CPUz record with one of those in the first place. ;)
Posted on Reply
#203
cadaveca
My name is Dave
EarthDogWay to find the exception to the rule! That was a known problem with gulftowns. Too bad nobody in their right mind would shoot for a CPUz record with one of those in the first place. ;)
Right, that's why HiPro and others were killing them.:laugh:


:slap:

I stayed far away from 6-core iNTEL CPUs because of that thread...
Posted on Reply
#204
EarthDog
Well, as you should have been able to see from that thread, they were trying to rock single threaded benchmarks, not CPUz...and of course after this was confirmed that all stopped...though some of those were good for 7Ghz+ which is more than we can say for a lot of CPU's. :p
Posted on Reply
#205
cadaveca
My name is Dave
Sure. But it does say something very interesting about how AMD deals with disabled cores, if they are pushing so much wattage through them.
Posted on Reply
#206
EarthDog
That, my mousey friend (your avi), is well above my head on how the architecture works inside the silicon. But you are correct. :)

I just wonder (and this is bet for another thread at possibly another site) why that only happened with Gulftown... 32nm process? Not sure thats it as my 25/2600k works fine down to 1c/2c...
Posted on Reply
#207
cadaveca
My name is Dave
6-core INtel chips are 32nm, like SB. I seem to recall something about current run-off.

Can the 6-cores adjust core speed independantly? Pretty AMD's can.

I mean, as already stated, keeping that 8 GHz+ frequency stable is a big thing, just kind of a useless thing for me personally. My reviews won't cover stuff like that.
Posted on Reply
#208
cdawall
where the hell are my stars
tricksonOk lets start this then .
1. Higher IPC + We do not know how this performs in real world applications .
2. More cores , Ok it has 8 cores ( That is a start . ) .
3. It clocks like a mother Trucker . On One core ( or was it 2 cores ) Never the less NOT on ALL 8 cores for sure !
4. Just how is this going to help me in making a logical sensible buy on a product ?
5. They broke the world record . That is awesome ! Great job .

Am I flame baiting or thread crapping ? Or are you being a tad bit too over the top ? I want more information but it just ekks out of AMD . And I think that all the Fluff is not going to make me change my mind about this BD chip . I want more than a world record holding over clock chip .
Actually we do know how higer IPC works in the real world. Conroe vs K8, conroe had a higher IPC and outperformed at lower clocks, K8 vs netburst opposite of K8 vs conroe, netburst vs mobile P3 based P4 the mobiles performed better with a higher IPC, K10 vs sandybride again higher IPC lower clocks better performance. So if we were to extrapolate said common knowledge hmmm I wonder if higher IPC makes a difference. There are plenty of clockers that disable cores including all of the high amd clocks as it stands the most anyone has really seen with all 6 or all 4 cores going on ANY manuf's chip is in the mid/high 6ghz. this is not designed to make YOU buy this chip it is to increase awarness and basically be a PR stunt before release.
Posted on Reply
#209
EarthDog
cadaveca6-core INtel chips are 32nm, like SB. I seem to recall something about current run-off.

Can the 6-cores adjust core speed independantly? Pretty AMD's can.

I mean, as already stated, keeping that 8 GHz+ frequency stable is a big thing, just kind of a useless thing for me personally. My reviews won't cover stuff like that.
Right, but SB doesnt have this issue, so its not the process... I do not believe gulftown can do that, no.

Our reviews do... sort of. We arent as standardized as you all are, so it depends on the reviewer. We ALL overclock, some brink overclock on water, others brink overclock on LN2 for theirs. But again our user/reader base is a bit more geared towards overclocking.
Posted on Reply
#210
trickson
OH, I have such a headache
cdawallActually we do know how higer IPC works in the real world. Conroe vs K8, conroe had a higher IPC and outperformed at lower clocks, K8 vs netburst opposite of K8 vs conroe, netburst vs mobile P3 based P4 the mobiles performed better with a higher IPC, K10 vs sandybride again higher IPC lower clocks better performance. So if we were to extrapolate said common knowledge hmmm I wonder if higher IPC makes a difference. There are plenty of clockers that disable cores including all of the high amd clocks as it stands the most anyone has really seen with all 6 or all 4 cores going on ANY manuf's chip is in the mid/high 6ghz. this is not designed to make YOU buy this chip it is to increase awarness and basically be a PR stunt before release.
I get the ICP stuff I know that with more there will be more performance , Thing is Just how MUCH more will it have over the SB ? THAT IS THE QUESTION ! All AMD can do is stupid stunts and give out useless crap on there chips at this time . This is all I am saying . :shadedshu
Posted on Reply
#211
cdawall
where the hell are my stars
tricksonI get the ICP stuff I know that with more there will be more performance , Thing is Just how MUCH more will it have over the SB ? THAT IS THE QUESTION ! All AMD can do is stupid stunts and give out useless crap on there chips at this time . This is all I am saying . :shadedshu
How is it a stupid stunt they took some of the best AMD clockers handed them a handful of new cpu's and they set a WR. Thats not a stunt thats a PR win.
Posted on Reply
#212
trickson
OH, I have such a headache
cdawallHow is it a stupid stunt they took some of the best AMD clockers handed them a handful of new cpu's and they set a WR. Thats not a stunt thats a PR win.
I guess your right , It will go good with this . And means about as much . But it is a great thing to see done . Congratulations AMD ! A job well done indeed . :rockout:
Posted on Reply
#213
cdawall
where the hell are my stars
tricksonI guess your right , It will go good with this . And means about as much . But it is a great thing to see done . Congratulations AMD ! A job well done indeed . :rockout:
I mean its not like its the first high end cpu to break the top 10 fastest chips or anything :laugh:

Posted on Reply
#214
trickson
OH, I have such a headache
cdawallI mean its not like its the first high end cpu to break the top 10 fastest chips or anything :laugh:

img.techpowerup.org/110915/Capture134.jpg
So this looks to be the first AMD record for over clocking . :respect::rockout:
Posted on Reply
#215
a_ump
i think its awesome. its the beginning of a new architecture. Few more family releases of it and i wonder how far it'll take AMD. My hopes and expectations are high.
Posted on Reply
#217
Steevo
Interesting that chew says its a 4 core 8 threaded CPU. If that truly is the case (or at least how AMD plays it if the market allows) they will be killing Intel on IPC.
Posted on Reply
#218
cadaveca
My name is Dave
SteevoInteresting that chew says its a 4 core 8 threaded CPU. If that truly is the case (or at least how AMD plays it if the market allows) they will be killing Intel on IPC.
In some situations, yes, it is just a quadcore.

:banghead:


In fact, it IS just four MODULES. But the OS recognizes each module as 2 cores, like Intel's HT, but it's so much more than just HT, they have to make that distinction somehow.

:ohwell:

For most instances, I think, for most users, it'll effectively be a 8-core CPU.

But it's still just a quad. Module. But quad.:wtf:

:D
Posted on Reply
#219
Goodman
Damn_SmoothDon't think. Do.
You didn't notice... i was sarcastic ;)
EarthDogNow? Its always been that way (well since multi cores were out). When shooting for highest clocks you should always disable cores.
Well all CPU-z scores on HWBot are all full cores enable except for Celeron & P4 which are single core CPU , anyway Celeron still/was single core cpu... as far as i know
cadavecaIn some situations, yes, it is just a quadcore.

:banghead:


In fact, it IS just four MODULES. But the OS recognizes each module as 2 cores, like Intel's HT, but it's so much more than just HT, they have to make that distinction somehow.

:ohwell:

For most instances, I think, for most users, it'll effectively be a 8-core CPU.

But it's still just a quad. Module. But quad.:wtf:

:D
You'll still get 8 threads ;)

Also i wonder if we're going to see some 4-6 cores FX unlock in to 8 cores FX?
That would be so cool...:rockout:
Posted on Reply
#220
EarthDog
Well all CPU-z scores on HWBot are all full cores enable except for Celeron & P4 which are single core CPU , anyway Celeron still/was single core cpu... as far as i know
All. Hardly.

Most any CPU with HT, at minimum HT is disabled, in a lot, cores are disabled. This is the best way to achieve the highest clocks. If someone didnt disable HT and some cores, they likely havent maxed out their CPU. Its the way it is and there isnt any getting around that.

2600k - hwbot.org/hardware/processor/core_i7_2600k/
i7 920 - hwbot.org/hardware/processor/core_i7_920/

I can go on...and on, but hopefully you get my point.
Posted on Reply
#221
cdawall
where the hell are my stars
GoodmanWell all CPU-z scores on HWBot are all full cores enable except for Celeron & P4 which are single core CPU , anyway Celeron still/was single core cpu... as far as i know
well all of the P4's on the list have HT disabled since most are P4 631's. the P4 631's on the >8ghz list are very very single core single thread. As is seen here and here.
Posted on Reply
#222
Damn_Smooth
GoodmanYou didn't notice... i was sarcastic ;)
I actually think you might be able to pull it off. I wasn't trying to be a dick or anything, I just wanted to see how close you came.
Posted on Reply
#223
twicksisted
Damn_SmoothI actually think you might be able to pull it off. I wasn't trying to be a dick or anything, I just wanted to see how close you came.
I nearly came!
Posted on Reply
#225
Goodman
EarthDogAll. Hardly.

Most any CPU with HT, at minimum HT is disabled, in a lot, cores are disabled. This is the best way to achieve the highest clocks. If someone didnt disable HT and some cores, they likely havent maxed out their CPU. Its the way it is and there isnt any getting around that.

2600k - hwbot.org/hardware/processor/core_i7_2600k/
i7 920 - hwbot.org/hardware/processor/core_i7_920/

I can go on...and on, but hopefully you get my point.
I didn't look at every CPU-z shot on HWbot just the highest ones ~40 , but both #1 in the links you posted shows 4 cores 4 threads on CPU-z
Anyhow i always thought that to be able to have your overclock record with CPU-z that you must have all core enable otherwise it is rejected , i guess i was wrong...

Still wonder why so many people , most of them (from what i see so far) try the highest score with all cores enable , when much simpler with only 1-2 cores?
maybe most of them like me don't know that you don't have to have all cores enable?

Anyways case close & watch out for next CPU-z shot...:laugh:;)
Damn_SmoothI actually think you might be able to pull it off. I wasn't trying to be a dick or anything, I just wanted to see how close you came.
twicksistedI nearly came!
I'll let you two discuss that in PM , i don't wont to know anything about it.... :roll:
Posted on Reply
Add your own comment
Apr 25th, 2025 09:25 EDT change timezone

Latest GPU Drivers

New Forum Posts

Popular Reviews

Controversial News Posts