PCPER.com: 5 GHZ+ Phenom II Overclock on Dry Ice, 6Ghz on LN2

[Boissez]

Nice to see AMD coming back BUTBUTBUT... Its still going to be a tough sell.
The way I see the market from Q1 2009:
Server: Intel win across the board
High-end desktop: Intel win
Value desktop: AMD win
High-end mobile: Intel win
Value mobile: Intel win

Still not a very good position by all accounts. IMO they should churn out some mobile parts ASAP.

A mobile X3 deneb with an undervolted RV730/RV770 would be a sweet sweet platform.

[D-Cyph3r]

Nice to see AMD coming back BUTBUTBUT... Its still going to be a tough sell.
The way I see the market from Q1 2009:
Server: Intel win across the board
High-end desktop: Intel win
Value desktop: AMD win
High-end mobile: Intel win
Value mobile: Intel win

Still not a very good position by all accounts. IMO they should churn out some mobile parts ASAP.

A mobile X3 deneb with an undervolted RV730/RV770 would be a sweet sweet platform.

I'm pretty sure the "Value Desktop" market is larger than all the others put together though (excluding server of course).

[Boissez]

I'm pretty sure the "Value Desktop" market is larger than all the others put together though (excluding server of course).

IIRC The mobile market is the largest in terms of volume and the server market is the largest in terms of value.

[stangracin2]

Nice to see AMD coming back BUTBUTBUT... Its still going to be a tough sell.
The way I see the market from Q1 2009:
Server: Intel win across the board
High-end desktop: Intel win
Value desktop: AMD win
High-end mobile: Intel win
Value mobile: Intel win

Still not a very good position by all accounts. IMO they should churn out some mobile parts ASAP.

A mobile X3 deneb with an undervolted RV730/RV770 would be a sweet sweet platform.

amd is gonna do good in the servers because its just a drop in chip
instead of new board/ram/cpu's

[Xel'Naga]

amd is gonna do good in the servers because its just a drop in chip
instead of new board/ram/cpu's

Also, in servers performance per watt is the most important. And although Intel's platform will be faster, the CPUs consume a little more and FBDIMM makes it a lot worse

[JumpingJack]

Also, in servers performance per watt is the most important. And although Intel's platform will be faster, the CPUs consume a little more and FBDIMM makes it a lot worse

AMD has some good marketing advantages over Nehalem and server, the drop in compatibility is a huge win for companies in this environment.

However, Nehalem 2P servers will not use FBDIMMs they will be using low voltage DDR3.

Jack

[Miss Banana]

Fortunately for AMD, there are more segments than just "value" and "high end". What about enthusiasts that want the best bang for the buck?
What about students that need all the horsepower they can for audio/video tasks, without wanting to pay the jackpot?

To me it seems the highest performing Deneb will come very close in performance to the highest performing Nehalem.
Add to this Deneb's superior performance per watt, and the fact that MBs that support Deneb will be much cheaper than MBs that support nehalem. (especially the AMD boards that still use the old southbridge)

About time though, for a long time it made no sense to buy AMD for whatever purpose besides wanting to go cheap. (server market excluded)
Glad we're past that phase.

[informal]

People shouldn't underestimate mobile market since it's growing insanely fast.That's why AMD preps the Deneb core based low power models for notebooks.Netbooks should get a single core variant of Deneb too,only cache stripped and running at low volts.

[Hornet331]

People shouldn't underestimate mobile market since it's growing insanely fast.That's why AMD preps the Deneb core based low power models for notebooks.Netbooks should get a single core variant of Deneb too,only cache stripped and running at low volts.

Whats the point of that, singelcores are dead, even for notebooks (beside for UMPCs, but even there dualcores take over)?

They need a chip that is dirt cheap and consumes next to nothing, and perform a bit better then atom -> zing amd would own the whole sub 400€ market.

Just look at the success of atom. Even if you strip down phenom 2 off everything it wont come close the the power consumption of atom.

Amd already had massmarket SoC way befor intel (remeber cyrix? ), but they dumped it... if they would have forced the development based on MediaGX core they would have a very competentive product against Atom.

As for a competition against the centrieno platform, that depends how much impact the cache strip has on phenom phenoms performance.

[Stukov]

Whats the point of that, singelcores are dead, even for notebooks (beside for UMPCs, but even there dualcores take over)?

They need a chip that is dirt cheap and consumes next to nothing, and perform a bit better then atom -> zing amd would own the whole sub 400€ market.

Just look at the success of atom. Even if you strip down phenom 2 off everything it wont come close the the power consumption of atom.

Amd already had massmarket SoC way befor intel (remeber cyrix? ), but they dumped it... if they would have forced the development based on MediaGX core they would have a very competentive product against Atom.

As for a competition against the centrieno platform, that depends how much impact the cache strip has on phenom phenoms performance.

And how are the margins in the sub $400 computers components?

[Mescalamba]

Well, maybe someone will think that Im stupid (maybe most of you), but how you comed to thinking that Phenom II wont be comparable to i7?

i7 has integrated memory bus, guess what.. Phenom II has integrated memory bus too. Both will use DDR3 (when AMD release AM3 platform). Phenom II doesnt have any memory clock/voltage limitations. Only thing it lacks really is Hyper-Threading and triple channel. But as much as I know, there is not big difference between dual/triple channel if you have integrated memory bus in CPU. i7 is just re-designed Core 2, nothing more. I think that if Phenom II will be able to do 4GHz at air, it can actually beat i7 (just imagine Phenom II under watercooling - 4,5GHz + some 2+GHz rams). Even with my poor Athlon X2 on 3GHz and Cellshocks at 1000MHz CL4.. I have over 9k bandwith.. if I imagine having some Phenom II on 4GHz.. heh bandwith over 15k? Maybe.. cause I can clock Cellshocks higher, just FSB cant take it).

If these OC reports are true, I believe that Phenom II will be much more than just comparable to i7. (at least much more friendly to OCers.. dont liek that idea of dead CPU if my memory voltage is too high).

[Kuemmel]

...
i7 has integrated memory bus, guess what.. Phenom II has integrated memory bus too. Both will use DDR3 (when AMD release AM3 platform). Phenom II doesnt have any memory clock/voltage limitations. Only thing it lacks really is Hyper-Threading and triple channel...snip...If these OC reports are true, I believe that Phenom II will be much more than just comparable to i7. (at least much more friendly to OCers.. dont liek that idea of dead CPU if my memory voltage is too high).

You are missing some things here, it's not just about HT, etc. The problem is that AMD didn't optimize the execution units on Phenom / Phenom II as good as Intel did on Core 2 / i7.

The execution units of the Intel CPU's are much more flexibel and therefore depending on the application can excecute more instructions compared to AMD at the same time. In my on fractal assembler code I see that even Core 2 is about 30 to 40 percent faster at the same clock speed than a Phenom, when it comes to FPU or SSE2 performance.

If you wanna know more detail, take a weekend off ;-) and fight your way through the Agner Fog's nice guides:
http://www.agner.org/optimize/microarchitecture.pdf

[Zucker2k]

You are missing some things here, it's not just about HT, etc. The problem is that AMD didn't optimize the execution units on Phenom / Phenom II as good as Intel did on Core 2 / i7.

The execution units of the Intel CPU's are much more flexibel and therefore depending on the application can excecute more instructions compared to AMD at the same time. In my on fractal assembler code I see that even Core 2 is about 30 to 40 percent faster at the same clock speed than a Phenom, when it comes to FPU or SSE2 performance.

If you wanna know more detail, take a weekend off ;-) and fight your way through the Agner Fog's nice guides:
http://www.agner.org/optimize/microarchitecture.pdf

Interesting, I look forward to seeing some head to head comparos between Deneb and Ci7; it'll be a shame if 4Ghz Deneb = 3.6Ghz Ci7. If so, AMD is will be 18 months late to the market, again.

[stangracin2]

Interesting, I look forward to seeing some head to head comparos between Deneb and Ci7; it'll be a shame if 4Ghz Deneb = 3.6Ghz Ci7. If so, AMD is will be 18 months late to the market, again.

it will prolly be like that because of the extra 4 threads on i7

[Calmatory]

it will prolly be like that because of the extra 4 threads on i7

...which do not usually help that much under heavy load. HT is not the main concern, IPC is.

[demonkevy666]

You are missing some things here, it's not just about HT, etc. The problem is that AMD didn't optimize the execution units on Phenom / Phenom II as good as Intel did on Core 2 / i7.

The execution units of the Intel CPU's are much more flexibel and therefore depending on the application can excecute more instructions compared to AMD at the same time. In my on fractal assembler code I see that even Core 2 is about 30 to 40 percent faster at the same clock speed than a Phenom, when it comes to FPU or SSE2 performance.

If you wanna know more detail, take a weekend off ;-) and fight your way through the Agner Fog's nice guides:
http://www.agner.org/optimize/microarchitecture.pdf

you have a point there the execution unit is tweaked. I remember reading somewhere that it's basily makes use of both cores from a single thread.

[Nedjo]

I remember reading somewhere that it's basily makes use of both cores from a single thread.

pfff...What?!

[dinos22]

He was the first person to ever get one I think.
If Macci is hitting those kinds of clocks on LN2 with Dragon Evo...wonder what he would do with a F1 extreme...Sami PM me your info again, time for an upgrade
Looks like AMD is back in the LN2 game after all this time. Pretty cool
Nehalem kind of sucks for extreme overclocking on LN2 to be honest. The LN2 party since dothan had to end some time I guess lol.

ho ho ho.....what are you trying to do....allow AMD to hit 7 GHz mate

The only downside I see here, is the amount of LN2 we will have to spend running the CPU at -190C all day long

Nice to see that AMD is back in business. Now where is my AM2 board again? After CES things should get interesting again =) Finally a change in the benching scene. Benching C2Ds all the time is a little boring and Nehalem is not exactly much fun under LN2 as of right now =/

Fingers crossed, AMD is right and that CPU is not cherry picked or special in any form.

the beauty of using F1EE is that it is use FAR less LN2 than any other pot out there so to maintain and be at those temps this is THE pot to use...............really glad KP will hook up macci

I ll try with and without couple of diodes too ;-)
Get ready for a trauma ...

why not just rip them off for good.....if a couple of diodes are making Nehalem run like a dog on LN2 get rid of them i say

I usually do not organize what I call "freak of nature" demo, what I mean by this is that I am always very careful to use statistically representative parts to avoid misleading anybody.

This time, I got a Saturday FedEx delivery ...

http://www.v12extreme.com/post/IMG_6158.JPG

This is from my Buddies Milke M and Mike H (thanks for going the extra miles), this is a freak of statistic, this part is not in the middle distribution of the curve, it actually leak some serious current ... but it does OC, really, like hell.
6Ghz, i know i can do it already, let s see what is the next Digit , and to make sure everything is fair, I ll be joining a master of OC, he knows better than I do about LN2

I am just going to be doing the same as our green buddies, dissable the thermal control, and let s see

I ll drop the result here before thanks giving week end!

so this CPU will never actually be a retail product you would sell publically???

Make sure the oc master is a member of this community or it won't do you much good

Why don't you fly to vegas and ask FUGGER himself to bench it with you on ln2. And make sure to videotape it

I want vid's or it didn't happen

it's going to be Charles dude lol

Every body take it easy ... The Demo is coming, and the Phenoms II will be in store soon ...
And for the demo, we will do on CAM, with and without tricks, I need the community to understand that Manufacturer demo is kind of not appropriate for OC, it is why i always let Charles and Co to do it.
So, we will know soon.

you are operating on assumption that AMD cheated by turning off a few parts of the chip and running those demos and telling ppl they are normal run of the mill CPUs. what if you are wrong? Will Intel look like an arse trying to do something AMD didnt do....maybe it's a bad idea to do this sort of demo i donno Francois.

if you say that Intel has always tried to have a representative CPU you should stick to that. Long term it's a smarter attitude to have. You should let XS users smell a rat rather than you try to show us imo. We're are pretty good at it too you know and consequences will be bad for AMD as they will be if Intel starts playing trickery with people.........

what is the point of showing us a 6GHz or even 7GHz nehalem when average benching stable clocks on nehalem subzero are below 5GHz atm, even below 4.8GHz most likely. You are only going to exacerbate people's dissatisfaction with spending crazy amounts of money only to get CPUs that run like dogs on LN2 ..........

what we really need now is for you guys to sit down and fix cold bug problems and release retail chips........so that we can all reach monster clocks without temps affecting the OCs. Shamino thinks that there is a high vcore cold bug on these CPUs...He's very experienced and would have tested a fair few CPUs to form an opinion like that. Time to get ot work and make us ALL happy heheheh

Don't worry, F1 EE will be used very soon, date is already set

[Marvin_The_Martian]

it's going to be Charles dude lol

I know he told me minutes later

I agree btw on the not accusing as well, think it more classy to just show us lol

[Drwho?]

Guys, if you concidere yourself a Blue team member , please let's the green team guys have their fun, it is nice to see them reviving a little, give them some space.
Instead, get back to work, and break the 7Ghz on Core i7
Let s squeeze the best of both platform.

Don't be too proud of this technological terror you've constructed. The ability to destroy a planet is insignificant next to the power of the Core i7

Darth
http://www.youtube.com/watch?v=3MaJy7LH5js

[Blaber]

Guys, if you concidere yourself a Blue team member , please let's the green team guys have their fun, it is nice to see them reviving a little, give them some space.
Instead, get back to work, and break the 7Ghz on Core i7
Let s squeeze the best of both platform.

Don't be too proud of this technological terror you've constructed. The ability to destroy a planet is insignificant next to the power of the Core i7

Darth
http://www.youtube.com/watch?v=3MaJy7LH5js

Please Stop dividing enthusiast into BLUE or GREEN team members or communitys.

Enthusiast picks hardware that does best for him.Not depending on any blue or green team.

And all of that with Olympic spirit in mind

[massman]

Let's start with breaking 6G before we rush to 7G . No dirty tricks allowed.

[Drwho?]

Let's start with breaking 6G before we rush to 7G . No dirty tricks allowed.

answer in video: http://www.youtube.com/watch?v=t3dI-ghYimg

[STaRGaZeR]

answer in video: http://www.youtube.com/watch?v=t3dI-ghYimg

Oh man such good memories of JK2 and JKA

I'm going to play some multiplayer now

[Drwho?]

Oh man such good memories of JK2 and JKA

I'm going to play some multiplayer now

this one is awesome: http://www.youtube.com/watch?v=3zYOw7v6TFE