AMD breaks 7.0ghz on its cpu

[Dominato3r]

http://www.youtube.com/watch?v=onYQxHx0tBQ

Was this really the world record before? i doubt that. Ive heard many rumors of people hitting like 10ghz. Dunno

 

[masteryoda34]

A few people hit 8.1GHz on NetBurst. I believe that is the highest clock recorded for a x86 CPU.

 

[alyarb]

nvm, but yeah. netburst.

 

[Idontcare]

Originally posted by: Dominato3r
Was this really the world record before?

If you have been following the saga in varying threads on XS you'll note AMD is putting a rather interesting spin on all the world records...they are world records on AMD hardware to date.

I.e. no one has clocked an AMD CPU to 7GHz before...so its a world record...on AMD hardware.

Same with the claimed 3Dmark06 WR, among a couple others.

For some reason all the world record claims fail to carry mention of this caveat until someone on a forum pushes back with the "what about so and so from 2 yrs ago?" data points.

Sad really, because if it weren't for the fine print on the WR claims we could ALL enjoy the coolness of the spectacle 100% instead of getting nit picky over what a world record means or is intended to convey.

 

[Tempered81]

world record on quad core too

 

[Sylvanas]

Does anyone really care though? I couldn't give a **** if a CPU hits a high frequency on Ln2- that is of no use to me as a prospective (high end) buyer, interesting that AMD (and many manufacturers, Giga, Asus, DFI) keep sponsoring these events- I would think the money could be better used elsewhere.

 

[masteryoda34]

Originally posted by: Idontcare

Originally posted by: Dominato3r
Was this really the world record before?

If you have been following the saga in varying threads on XS you'll note AMD is putting a rather interesting spin on all the world records...they are world records on AMD hardware to date.

I.e. no one has clocked an AMD CPU to 7GHz before...so its a world record...on AMD hardware.

Same with the claimed 3Dmark06 WR, among a couple others.

For some reason all the world record claims fail to carry mention of this caveat until someone on a forum pushes back with the "what about so and so from 2 yrs ago?" data points.

Sad really, because if it weren't for the fine print on the WR claims we could ALL enjoy the coolness of the spectacle 100% instead of getting nit picky over what a world record means or is intended to convey.

But it still is a world record on a quad. And NetBurst is no longer a relevant uarch.

 

[transitionality]

Originally posted by: Sylvanas
I would think the money could be better used elsewhere.

Oh no. This is very efficient in terms of marketing dollars spent. Most OEM component customers are idiotic little kids who don't know what the hell they're doing, so they buy based on what's cool. You see "AMD breaks world record" on an RSS headline or forum thread title, AMD suddenly becomes cool. They don't care about the details -- they probably wouldn't even understand them.

 

[Idontcare]

Originally posted by: masteryoda34
But it still is a world record on a quad.

I don't disagree, in fact I applaud team AMD on creating such a marvelously scaleable architecture.

An architecture that can scale from 3GHz target clocks to 7GHz is pretty darn impressive in my book.

Originally posted by: masteryoda34
And NetBurst is no longer a relevant uarch.

If China (or any other country) landed a man on the moon today, would it be cool for them to claim "first to land a man on the moon!" just because the Apollo technology is no longer relevant?

 

[Idontcare]

Originally posted by: transitionality

Originally posted by: Sylvanas
I would think the money could be better used elsewhere.

Oh no. This is very efficient in terms of marketing dollars spent. Most OEM component customers are idiotic little kids who don't know what the hell they're doing, so they buy based on what's cool. You see "AMD breaks world record" on an RSS headline or forum thread title, AMD suddenly becomes cool. They don't care about the details -- they probably wouldn't even understand them.

At least Sylvanas prefaced his statement with a comment to indicate he was just stating his opinion.

Do you have data to support your claims of knowledge regarding ROI for marketing budgets and the demographics of end-users cross analyzed wrt their comprehension abilities?

Or perhaps you meant to start your diatribe with "I would think" too?

 

[octopus41092]

Originally posted by: Idontcare

Originally posted by: masteryoda34
But it still is a world record on a quad.

I don't disagree, in fact I applaud team AMD on creating such a marvelously scaleable architecture.

An architecture that can scale from 3GHz target clocks to 7GHz is pretty darn impressive in my book.

Originally posted by: masteryoda34
And NetBurst is no longer a relevant uarch.

If China (or any other country) landed a man on the moon today, would it be cool for them to claim "first to land a man on the moon!" just because the Apollo technology is no longer relevant?

No, but if they said first Chinese man to land on the moon, it would be cool for a lot of people.

 

[VirtualLarry]

If you watch the video, it says they only got to 6.9, and they didn't run any benchmarks. They only clocked a single core that high, so it also doesn't really count as a true quad-core overclock (which I would think would require overclocking every core).

Meh.

 

[masteryoda34]

Originally posted by: VirtualLarry
They only clocked a single core that high, so it also doesn't really count as a true quad-core overclock (which I would think would require overclocking every core).

:thumbsdown:

 

[Cogman]

Originally posted by: VirtualLarry
If you watch the video, it says they only got to 6.9, and they didn't run any benchmarks. They only clocked a single core that high, so it also doesn't really count as a true quad-core overclock (which I would think would require overclocking every core).

Meh.

I couldn't watch the entire video. someone overclocking a CPU with large group of drunk people in the background shouting "OvverrCloCk!!!!1!".

What I did see is that they said they only went to 6.93 GHz and that they did in fact disable one of the cores.

 

[dmens]

Originally posted by: Idontcare
I don't disagree, in fact I applaud team AMD on creating such a marvelously scaleable architecture.

I disagree with your term scalable architecture. these records are achieved under conditions so extreme that no engineering work was done at those points. therefore any "record" is essentially a manufacturing oddity, not an intentional result of the architecture or design. those 8ghz p4's are literally freaks of process.

 

[Fox5]

Originally posted by: Idontcare

Originally posted by: masteryoda34
But it still is a world record on a quad.

I don't disagree, in fact I applaud team AMD on creating such a marvelously scaleable architecture.

An architecture that can scale from 3GHz target clocks to 7GHz is pretty darn impressive in my book.

Originally posted by: masteryoda34
And NetBurst is no longer a relevant uarch.

If China (or any other country) landed a man on the moon today, would it be cool for them to claim "first to land a man on the moon!" just because the Apollo technology is no longer relevant?

It's more impressive that AMD finally has a core that clocks well at all, under any conditions. I think the last time that happened was the pre-SOI 90nm Athlon 64s. Since they've implemented SOI, they've barely outdone launch speeds.

 

[Fox5]

Originally posted by: dmens

Originally posted by: Idontcare
I don't disagree, in fact I applaud team AMD on creating such a marvelously scaleable architecture.

I disagree with your term scalable architecture. these records are achieved under conditions so extreme that no engineering work was done at those points. therefore any "record" is essentially a manufacturing oddity, not an intentional result of the architecture or design. those 8ghz p4's are literally freaks of process.

The P4 architecture was made to scale though. IIRC, there were 6Ghz Prescotts on the roadmap before they realized it wasn't going to hit those speeds due to heat constraints.

 

[Idontcare]

Originally posted by: dmens

Originally posted by: Idontcare
I don't disagree, in fact I applaud team AMD on creating such a marvelously scaleable architecture.

I disagree with your term scalable architecture. these records are achieved under conditions so extreme that no engineering work was done at those points. therefore any "record" is essentially a manufacturing oddity, not an intentional result of the architecture or design. those 8ghz p4's are literally freaks of process.

You think it is because of a manufacturing oddity that the 45nm PhII's scale with cooling and voltage as they do while the 65nm Phenom's did not?

 

[alyarb]

while 8.1 GHz netburst chips are outliers, 7.6, 7.8 GHz overclocks were very achievable minimums with liquid nitrogen. they are not freaks of process. it's a similar bell curve that you get with normal cooling. AMD's architecture experiences this bell curve also. Most Deneb chips under LN2 don't quite make it to 7 GHz, but they are in the mid sixes. It's just like having a a Phenom II that does 4.0 GHz on air, while some only make it to 3.80 GHz.

AMD's architecture is surprisingly scalable because it is an evolution of the same architecture that they've had since 1999 with the K7. I think the pipeline went from 10 to 12 stages going from K7 to K8. I'm not exactly sure what has changed with K10, but I can tell you that the netburst pipeline evolved from 20 stages to nearly 40 stages over its life. Geometrically, things around the athlon pipeline have changed around as the memory controller was brought on die, caching has been completely rethought twice and new instructions are supported in the decode hardware, but they didn't have to keep piling on stage after stage because they were not locked in a suicidal frequency race with intel. They won the race to 1 GHz and they decided not to piss the queen off again.

AMD's really had a wonderful run with their architecture considering they haven't had to start from scratch in over ten years. Would it surprise you that a 130nm Athlon XP could do 4.0 GHz on LN2? Do you think a 45nm Athlon should do better than 7.0?

AMD was so badass with K7 and K8 it wasn't even funny. You had 2 GHz athlon's owning 3 GHz northwoods at 3d studio max. Anyone remember FlaskMPEG? what a miserable program, but I admit got the job done with SSE2. AMD really came through for us back then. From the get-go they gave us huge L1 cache and three FPU units per core. Did intel ever do that? Netburst had no raw muscle out of the box. It was all a gag and they banked.

But whatever, conroe and bloomfield have been great too. Both of them scale great with LN2.

 

[dmens]

Originally posted by: Idontcare
You think it is because of a manufacturing oddity that the 45nm PhII's scale with cooling and voltage as they do while the 65nm Phenom's did not?

im saying the world record runs are meaningless because they are freak occurrences and therefore not a good example to compare scalability, which is indeed better on 45nm than 65nm. guess i should have said, world records to not prove scalability.

 

[taltamir]

Originally posted by: Sylvanas
Does anyone really care though? I couldn't give a **** if a CPU hits a high frequency on Ln2- that is of no use to me as a prospective (high end) buyer, interesting that AMD (and many manufacturers, Giga, Asus, DFI) keep sponsoring these events- I would think the money could be better used elsewhere.

i have to agree... this isn't even on an expensive 700$ phase change cooler... this is a liquid nitrogen system that requires you to drive to the store and buy a new tank of nitrogen (assuming there is such a store near you) for every 3 hours your computer runs... I hate refueling my car at a gas station every other week, having to refuel my PC several times a day is unacceptable regardless of how rich I would be.

What I want to see is world records in PERFORMANCE (while overclocking) on a 100% stable system that requires only electricity from the outlet (no strange liquid consumeables)

 

[taltamir]

WTF, they are not even overclocking the whole chip... they are overclocking each individual core by itself and getting the highest a SINGLE CORE on a quad core CPU will go... that is 7ghz for ONE CORE ONLY.

 

[OCGuy]

Originally posted by: taltamir
WTF, they are not even overclocking the whole chip... they are overclocking each individual core by itself and getting the highest a SINGLE CORE on a quad core CPU will go... that is 7ghz for ONE CORE ONLY.

Yea that is major fail.


If the whole chip isnt overclocked, it is irrelevant.

 

[alyarb]

it's likely that all the other cores were crapping out between 6.8 and 7.0 ghz because that is where most others max out.

here's a quad at 6.89. Doesn't impress you?
http://hothardware.com/article...tem1324/amdoc_cpuz.jpg

i'm sure you'll be disappointed regardless though when you realize that they are barely stable enough to take screenshots when they do this. to get through a benchmark you'd have to go way down to like 6.4 or 6.5.

 

[Modular]

Man, I love liquid nitrogen.