Brisbane Testing in progress...

[alucasa]

If it's NN's avatar, he's a player. (It changes frequently.)

Ryan

Gets an early chip and , at the same time, a player? Can't get better in life, eh.

Though I'd say Arisythila's cat's cuter.

[nn_step]

Gets an early chip and , at the same time, a player? Can't get better in life, eh.

Though I'd say Arisythila's cat's cuter.

You into beastuality or something?

[Lightman]

What I will post now is a pure unconfirmed information/speculation or so called FUD.
This is quote taken from techpowerup forum.
AshenSugar wrote:

ok first off, 65nm will come first to almost unaltered k8 cores, should be seeing them poping up anytime, then mid next year k8L will show up, the one advantage of the new 65nm chips will be higher clocks with lower temps/volts, this should be a very exciting time for amd users who already have am2, if not, well u got what you got.

i was just told by a buddy at intel test labs that they aquiered 2 65nm chips, he dosnt know where they got them(nobodys talking) but they are easly clocking to the max the boards can handel at 1.4v, he said one of them is running at 355x12 (chipset/board can do more then 355 stable) thats 4260 at 1.4v they didnt try lower volts from what he said, hes gonna try and test it come more himself over the weekend(he can work some overtime and play with some of the new toys without alot of other people getting in the way...)

oh and temps where maxed at 35c with stock cooling from a sempy64(a chip they fryed by giving it 2.7volts ROFL) thats crazy, little cooler+slow fan+high overclock=low temp........

im trying to get him to take me with him this weekend, i just wana see these chips, play with them, see if im gonna be forced to buy one when they come out :P

I bolded interesting bit. If what he says is true or at least close to true then I think Intel get hold on 65nm K8 SiGe based core or basically revG2. Everything is possible. If you read my previous post I approximated date of release for revG2 and who if not Intel can get those so fast??

Just a theory and high hopes .

[K404]

nn...you KNOW that if ur gonna change ur avatar all the time, ya gotta keep us up to date with who she is, plus the important stuff like phone number....


@Lightmans post....I straight up dont believe what that guy says.

[stealth17]

What I will post now is a pure unconfirmed information/speculation or so called FUD.
This is quote taken from techpowerup forum.
AshenSugar wrote:


I bolded interesting bit. If what he says is true or at least close to true then I think Intel get hold on 65nm K8 SiGe based core or basically revG2. Everything is possible. If you read my previous post I approximated date of release for revG2 and who if not Intel can get those so fast??

Just a theory and high hopes .

Well considering that is so far out of line from current chips, I will call bogus on this for now. They old saying "if it sounds too good to be true, it usually is" has a lot of truth to it.

I'm not saying it's impossible and I'm not calling anyone a lier. I'm just stating that with the given proof we have now, I will only dream that 4.2ghz on 1.4v and stock cooler is true. In reality, fat chance.

[XFxGeforced]

@Lightmans post....I straight up dont believe what that guy says.

was that about the first gen 65nm cpus we're about to see?! that's way too insane to believe I have to side with stealth on this one ^

[biohead]

why do all the people getting early 65nm samples must be special enough not to have the right hardware to actually TEST it the way it's meant to be tested... huh? it's either crappy cooling, crappy mainboard, etc, whatever. how hard can it be? [/rant]

Lightman that part you quoted... 'if it sounds too good to be true, it usually is' <- LOL stealth beat me to it, read it afterwards.

[nn_step]

lets do some math here people, There are two ways in increase clock speed.
1) Have more Stages
or/AND
2) Smaller process.

Given the fact that conroe has 2 more stages and is on the same process. One can assume on average it'll be closer to Conroe's 3.2-3.5 average overclock than 4Ghz unless they did something that Intel couldn't even with Billion and billions of research dollars. In which case I would laugh my ass for the next decade

[Lightman]

From the other posts from quoted guy it seems that he was in this Intel Lab and saw some older experiments.
Here is link for whole thread:

http://forums.techpowerup.com/showthread.php?t=21757

I'm simply want to see 65nm chip from 0651 onwards.


PS. Anyone know what is the meaning of "S3'" mark on Brisbane CPU??

[K404]

I just read that thread... little discussion on the 65nm AMD, goes off at a tangent very quickly.

I saw the graphs tho, definately a catch up, but i`d like to know what AMD have done to improve benches.

[Lightman]

At least we are back on topic ---> 65nm K8!

[awdrifter]

I think Nedjo's 3.1ghz is very reasonable. This is AMD's first batch of 65nm, so of course it's not going to be that much faster than the current optimized 90nm process. I think in a few month the new 65nm chips will hit maybe 3.3-3.4ghz, just like the 90nm chips gained a few hundred mhz after it was in production for a while.

[Fred_Pohl]

Everything is possible.

Perhaps so but some things are so highly improbable as to be laughable. IMHO a Brisbane clocking to 4.26GHz @1.4V and running at 35C with a Sempron HSF is a great example.

I really don't see what all the fuss is about. Nedjo's results are probably very typical for the higher end Brisbanes at this time. 3.1GHz @1.5V running >50C under load with relatively high-end air cooling and exactly the same performance as Windsor is hardly unexpected. K8 simply isn't designed to ever reach 4GHz and AMD didn't work any miracles with the shrink to 65nm.

[pauldovi]

What I will post now is a pure unconfirmed information/speculation or so called FUD.
This is quote taken from techpowerup forum.
AshenSugar wrote:


I bolded interesting bit. If what he says is true or at least close to true then I think Intel get hold on 65nm K8 SiGe based core or basically revG2. Everything is possible. If you read my previous post I approximated date of release for revG2 and who if not Intel can get those so fast??

Just a theory and high hopes .

I highly doubt that a die shrinkage will result in a X2 doing 4.2Ghz "easy" at 1.4V and having max loads of 35C on the stock cooler. This is actually quite humorous. How could you believe this! Why doesn't Anandtech share similar results? Why does Anandtech get 51C at stock clocks? In other words, that post is "FUD."

[Lightman]

I highly doubt that a die shrinkage will result in a X2 doing 4.2Ghz "easy" at 1.4V and having max loads of 35C on the stock cooler. This is actually quite humorous. How could you believe this! Why doesn't Anandtech share similar results? Why does Anandtech get 51C at stock clocks? In other words, that post is "FUD."

If they used CoreTemp as Nedjo to read temps. then Intel results are in line witch those from Nedjo.

[duploxxx]

I highly doubt that a die shrinkage will result in a X2 doing 4.2Ghz "easy" at 1.4V and having max loads of 35C on the stock cooler. This is actually quite humorous. How could you believe this! Why doesn't Anandtech share similar results? Why does Anandtech get 51C at stock clocks? In other words, that post is "FUD."

yeah right our dear anand.... did you see the screenshot? htt link is on 1125... and you question why they reached a wall? temprature is indeed in normal range

[gOJDO]

lets do some math here people, There are two ways in increase clock speed.
1) Have more Stages
or/AND
2) Smaller process.

Given the fact that conroe has 2 more stages and is on the same process. One can assume on average it'll be closer to Conroe's 3.2-3.5 average overclock than 4Ghz unless they did something that Intel couldn't even with Billion and billions of research dollars. In which case I would laugh my ass for the next decade

There are more ways to increase the CPU freqfency:
1) optimize stages, so all stage will have roughly equal propogated delay.
2) increase the number of stages, but reduce the propogated delay per stage. that means reducing the logic done per stage
3) increase CPU VID. that means that the energy efficiency will be lower.
4) make CPUs on process with faster switching transistors and better transistor interconnections. That does not mean that the process needs to be smaller. In theory smaller transistor means faster, but when scalled proportional. The problem is that the gate oxide allready reached its scalling limit and is same on both 90nm and 65nm, Intel "bulk" and AMD SOI-2/SOI-3 processes. Also the same materials with same parameters are used for transistor interconnetions on the mentioned processes.

AMD better be happy if Intel would not release Penryn before K8L. My sources are saying that it will be a monster compared to all previous x86 CPUs and that the 45nm with high-K will be very fast and energy efficient process.

[Rickster_64]

I still do not see 65nm getting beyond 3.5. And the reason is they did nothing drastically different in the chemical mix on the shrinks, since 130nm SOI. Clocks then were getting upto 2.4-2.5ghz, and 90nm only picked up what, 400-500mhz? So, I really don't see 65nm doing much more than that unless you make changes elsewhere to inrease frequency.

[Fred_Pohl]

If they used CoreTemp as Nedjo to read temps. then Intel results are in line witch those from Nedjo.

Did you miss Nedjo's measured idle CPU temps? You do realize that it isn't possible for an air-cooled CPU to idle at 8C unless the ambient air is <8C (<46F) don't you?

Assuming a more realistic ambient air temp of 21C (70F) and knowing that the CPU idle temp has to be higher than that by at least a few degrees, we can easily interpret Nedjo's erroneous Core Temp readings. Idle should be in the 25-30C range and load should be 47-52C since CT showed a 22C rise from idle to load. This would correspond much better with Anand's results and reality.

In short, your Brisbane @4.26GHz with 1.4V story is pure FUD.

Nedjo gets 3.1GHz @1.5V using high-end air-cooling, Anand gets 2.9GHz @1.475V using stock cooling and we're supposed to believe that Intel is getting 4.26GHz @1.4V and the same temps using a weak Sempron cooler???

[chrisf6969]

Don't know if any of you noticed but Anandtech reviewed one: http://www.anandtech.com/cpuchipsets...spx?i=2889&p=8

thats the last page about overclocking.

It only got 2.95Ghz

(my) Summary: About the same as 90nm, but a lot cooler & less watts at stock.

[Lightman]

Did you miss Nedjo's measured idle CPU temps? You do realize that it isn't possible for an air-cooled CPU to idle at 8C unless the ambient air is <8C (<46F) don't you?

Assuming a more realistic ambient air temp of 21C (70F) and knowing that the CPU idle temp has to be higher than that by at least a few degrees, we can easily interpret Nedjo's erroneous Core Temp readings. Idle should be in the 25-30C range and load should be 47-52C since CT showed a 22C rise from idle to load. This would correspond much better with Anand's results and reality.

In short, your Brisbane @4.26GHz with 1.4V story is pure FUD.

Nedjo gets 3.1GHz @1.5V using high-end air-cooling, Anand gets 2.9GHz @1.475V using stock cooling and we're supposed to believe that Intel is getting 4.26GHz @1.4V and the same temps using a weak Sempron cooler???

You're blind?? If Nedjo had wrong measurements form Core Temp then Intel guys had same WRONG measurements (in line). That was my point! Not that someone can do over 4GHz on air with quoted temps!
READ -> THINK -> READ AGAIN -> THINK 3x!

[Fred_Pohl]

LOL! This from the guy who posted FUD about 4.26GHz Brisbanes at 1.4V running 35C...

What is your new point? That this nonexistent 4.26GHz Brisbane that Intel is supposedly testing was actually running >50C load like others do at 3GHz because Core Temp doesn't read Brisbane's sensors correctly?

[mofoman]

Let's just get serious here for a minute!! K8 is never gonna do 4GHz, the architecture is not going to suddenly jump through hoops because it got a die shrink and that is not what the shrink is about.

This is about Advanced Micro Devices changing their manafacturing process to 65nm, and that now starts with k8, but that doesn't suddenly make the chip faster or better, sure we may see a couple of hundred mhz added, but nowhere near 4ghz.

Let's just hope the chips can hit 3.5, cause 3.5ghz k8chips in my Quad Fx??? yeah, I'll take that!

[Lightman]

LOL! This from the guy who posted FUD about 4.26GHz Brisbanes at 1.4V running 35C...

What is your new point? That this nonexistent 4.26GHz Brisbane that Intel is supposedly testing was actually running >50C load like others do at 3GHz because Core Temp doesn't read Brisbane's sensors correctly?

I quoted some other guy putting clear that this is UNCONFIRMED, FUD, ETC.
The reason I did that was to bring discussion back on topic (I mention this few posts back)! Read previous 20 or so posts! It's not that I believe in 4GHz K8, but some parts of that story make sense. I don't know how to put this more clear!

[BigJoe]

GREATE WORK Nedjo.... You are pride of Belgrade