New 45nm Phenom FX Deneb spotted ~ 4.4GHz...?

W

wizzard0003

Limp Gawd
Joined
Dec 6, 2005
Messages
198
Deneb sighting, core blimey
Up to 4.4GHz

By INQUIRER Newsdesk: Wednesday, 27 August 2008, 10:42 AM

A SMALL SITE reports a big sighting of AMD's upcoming 45nm Phenom FX Deneb core.

Reviewage.co.uk says here are two chips in AMD's pipeline, the FX-80 at 4GHz and the FX-82 at 4.4GHz.

And according to the numbers the sits has its mitts on, the FX-80 at stock out performs a 5.0GHz Kentsfield.

The chip will come with 6MB of L3 cache and, being FX flavoured, should be completely unlocked and over-clocking friendly. µ
Click to expand...

Source: http://www.theinquirer.net/gb/inquirer/news/2008/08/26/deneb-core-blimey

Reviewage.co.uk link: http://reviewage.co.uk/content/view/33/1/

WhoooHooo...! :cool:

Hope this is true... :p
 
If this is true, lol, that would be crazy. AMD pulling a graphics card and a processor out of it's ass all within a few short months? w00t? :p I hope it's true, but I've seen bs like this before.. I won't get too excited.
 
A quadcore processor at 4.4 GHz stock speed?
That's almost 2 GHz above the stock speed of the fastest quadcore today.
The words "No way in hell" leap to mind.
3 GHz I would buy, 3.4 GHz is doubtful, but anything above that is just ridiculous.
The most compelling argument I can give for that is that 4+ GHz speeds require long pipelines to be physically possible (think Pentium 4), regardless of process. The pipelines in the Phenom are even shorter than those in Core2 (which is hard to get over 4 GHz on a more mature 45 nm process).
Aside from that, AMD is tripling the cache, which should make it even harder to scale clockspeeds because of the sheer diesize/complexity.
As far as my knowledge of chip design and manufacturing goes, I say that the Phenom simply isn't designed for 4+ GHz speeds.
 
What part of AMD's design team were you on that gives you the knowledge that the new 45nm Deneb isn't capable of this?
 
w1retap said:
What part of AMD's design team were you on that gives you the knowledge that the new 45nm Deneb isn't capable of this?
Click to expand...

I didn't know that you have to be on AMD's design team... I assumed the laws of physics, chip design and manufacturing were universal. In which light, my above analysis holds.
 
lulz.

I'd rather wait and see what is released than base internet theories about physics and design spec limits off of one's assumptions.
 
w1retap said:
lulz.

I'd rather wait and see what is released than base internet theories about physics and design spec limits off of one's assumptions.
Click to expand...

So you'd rather buy into some CPU-Z screenshot that is most likely fake, than listen to technical facts?
I'm not assuming anything, what I say are facts in the real world. There are physical limits to the maximum propagation time of a transistor. And since you cannot make transistors switch faster (aside from supercooling the chip, which is not going to happen in a stock setup for obvious reasons), you need to create shorter dependent chains of transistor logic if you want to increase the clockspeed (obviously the entire logic chain has to propagate within the time of one clock cycle). This translates into implementing the execution pipeline in more stages. Which leads to the well-known rule-of-thumb that higher clockspeeds require longer pipelines.
You might want to grab a book on the introduction to digital technology and chip design. You'll see that what I say is all well-known and understood in this world.

Lulz indeed.
 
hah, I'm well aware. I'm in my senior year of electrical and computer engineering at the university of michigan, and I'm working at a nuclear power facility on the I&C/Electrical Component Systems Engineering department.

I'm saying, you don't know how they designed everything down to the tee. It's better to wait and see what they put out on the shevles than make assumptions that make you look like an asshat. Of course the screenshot could be fake, but we've seen time and time again companies can pull a rabbit out of a hat when everyone thinks something is impossible.
 
We already know that the pipeline is the same length as the current Phenom.
So we have the following information:
1) Deneb has a shorter pipeline than Penryn.
2) AMD's 45 nm process is less mature than Intel's 45 nm process.
3) Penryn is unable to reach clockspeeds of 4.4 GHz, not even close.
4) The 65 nm version with less cache tops out at around 2.6 GHz, which is considerably less than what the 65 nm version of the Core2 tops out at.

Simply put, AMD doesn't have a hat that large.
Intel doesn't even have a hat that large, and their hats are usually larger than AMD's anyway.
 
Scali2 said:
We already know that the pipeline is the same length as the current Phenom.
So we have the following information:
1) Deneb has a shorter pipeline than Penryn.
2) AMD's 45 nm process is less mature than Intel's 45 nm process.
3) Penryn is unable to reach clockspeeds of 4.4 GHz, not even close.
4) The 65 nm version with less cache tops out at around 2.6 GHz, which is considerably less than what the 65 nm version of the Core tops out at.

Simply put, AMD doesn't have a hat that large.
Intel doesn't even have a hat that large, and their hats are usually larger than AMD's anyway.
Click to expand...
I'll choose to wait and see rather than just dismiss it. I'd like hard evidence in the form of a Phenom FX when it's released, not inferences based off of current cores that people are running. You don't have the tech spec documents and design diagrams, you yourself did no testing with this, and you have zero evidence of anything just as I don't. Continue on..
 
I do believe that 45nm Phenom will bring some significan improvements, but I think > 4 GHz on a quad core at stock is nothing more than wishful thinking.

Seriously, I'll be happy with the ~50% power savings mentioned in other threads. My problem with Phenom is that I have no problem justifying buying a CPU that's cooler but slower, and I have no problem justifying the purchase of a CPU that's faster but hotter. I just can't recommend a CPU that's faster *and* hotter. If this 50% power savings is true, they'd have made the thermals much more acceptable, and performance will likely get a boost too :)

If this is an unlocked engineering sample, it is possible the multiplier was simply raised for a suicide screenshot. I really doubt it's anything more than that.
 
w1retap said:
I'll choose to wait and see rather than just dismiss it. I'd like hard evidence in the form of a Phenom FX when it's released, not inferences based off of current cores that people are running. You don't have the tech spec documents and design diagrams, you yourself did no testing with this, and you have zero evidence of anything just as I don't. Continue on..
Click to expand...

Then why do you choose to argue?
I've given plenty of solid arguments why I think it's highly unlikely that there will be a 4+ GHz Deneb anytime soon.
I do this because I don't want people to get their hopes up for nothing (yet again, the original Barcelona/Phenom were hyped as the Second Coming aswell, and fell well short, to the disappointment of many. You might want to read back my posts on that matter. I warned people long before the introduction, and my predictions turned out to be pretty much spot-on when the CPU was finally launched. So either I was really lucky all the time, or I actually do know what I'm talking about).
Choose what you want to believe, but don't keep posting that you don't agree with me, we already know, and you've not added any solid arguments why what I say would be wrong, and a 4+ GHz CPU is possible. So basically none of your posts have added anything to the discussion.
You don't agree, duly noted, leave it at that. There are also people who agree with me apparently, such as LstOfTheBrunnenG and Mav451.
 
That's why I said I'd rather wait and see... not simply dismiss it right off the bat. In addition, nobody really cares about peoples hopes and dreams. This is a computer processor we're talking about here, lol. If it is what the "screenshot" shows, great.. if it isn't, oh well, things will remain the same with slight improvements. No big deal.. don't get so upset over all this.
 
God just when I was thinkin about jumping ship and doing an Intel I7 build, I keep hearing these rumors... I hate waiting both companies just need to get this stuff on the market already...
 
w1retap said:
but we've seen time and time again companies can pull a rabbit out of a hat when everyone thinks something is impossible.
Click to expand...

I've seen plenty of companies pulling rabbits out of there hats, but this isnt a rabbit. This is like pulling out a 1000 pound angry gorilla out of a hat. Not saying its not possible, but If I was a betting man i'd definitely be betting against this one than for it. You wont see me holding my breath for this rumor.
 
DemonDiablo said:
I've seen plenty of companies pulling rabbits out of there hats, but this isnt a rabbit. This is like pulling out a 1000 pound angry gorilla out of a hat. Not saying its not possible, but If I was a betting man i'd definitely be betting against this one than for it. You wont see me holding my breath for this rumor.
Click to expand...
I agree.. that's why I referenced it to bs in post#4. :p
 
Jutsu said:
God just when I was thinkin about jumping shit and doing an Intel I7 build, I keep hearing these rumors... I hate waiting both companies just need to get this stuff on the market already...
Click to expand...

Well, I'm not sure if price means anything to you, but I don't think Deneb and Nehalem will be in the same pricerange.
I think Deneb systems will be cheaper in general.
 
I'll be ready for this processor if its AM2+ backwards compatible though the ~300W TDP I'm not sure a 790FX/GX system could handle.
 
Scali2 said:
4) The 65 nm version with less cache tops out at around 2.6 GHz, which is considerably less than what the 65 nm version of the Core2 tops out at.
Click to expand...
http://forums.guru3d.com/showthread.php?t=267708
This CPU is really kicking butt now! It took a bit for the AS5 to cure, but it's now running at 3.485GHz at 46*C idle and 55*C full load...(full load at 3.51GHz was 63*C - too high to be all the time, but again, stable at that speed)
 
Scali2 said:
Well, I'm not sure if price means anything to you, but I don't think Deneb and Nehalem will be in the same pricerange.
I think Deneb systems will be cheaper in general.
Click to expand...

Price really wont matter with my next build. It is about 3 years over due, and I have had my current system too long. It still runs like a champ but, I don't feel [H] enough knowing that I am running single core...:(

I have just been waiting and waiting, first it was for this years new GPUs, then it was I7, now I keep hearing good things about Deneb. I am going to try to hold out and buy myself a nice x-mas gift this year, hopefully by then I7 and Deneb will be out and both platforms with be stable and mature...
 
Just make it a new year's resolution to build a PC. =D

By then, a lot of new stuff should be rolled out due to the Christmas rush.
 
kassler said:
http://forums.guru3d.com/showthread.php?t=267708
This CPU is really kicking butt now! It took a bit for the AS5 to cure, but it's now running at 3.485GHz at 46*C idle and 55*C full load...(full load at 3.51GHz was 63*C - too high to be all the time, but again, stable at that speed)
Click to expand...

I was comparing stock speeds. If I'm not mistaken, the fastest Phenom (9950) is 2.6 GHz at stock speed.
And the FX-80 and FX-82 would allegedly be 4.0 GHz and 4.4 GHz respectively, at stock speeds.
So that is leap of 1.4 to 1.8 GHz in stock speeds.
 
ill believe it when I see it, or someone trustworthy puts it through its paces.
 
well w1retap and scali2 , u guys are way to [H] for me . Only thing i can say , about 8 or 9years ago when i build my first computer 1ghz was supposed to be the limit , 40gb HDD were huge and some said its impossible to build 100gb HDD... well now we have quadcores that easily clock to 3ghz, 1.5TB HDD ...lets see what comes next. 4.4ghz quadcores ? why not , just a matter of time.
 
game14 said:
well w1retap and scali2 , u guys are way to [H] for me . Only thing i can say , about 8 or 9years ago when i build my first computer 1ghz was supposed to be the limit , 40gb HDD were huge and some said its impossible to build 100gb HDD... well now we have quadcores that easily clock to 3ghz, 1.5TB HDD ...lets see what comes next. 4.4ghz quadcores ? why not , just a matter of time.
Click to expand...

Matter of time, yes... But more in terms of years, while Deneb is going to be released in a few months.
We didn't go from 40 GB HDDs to 1.5 TB HDDs overnight, just like we didn't go from 1 GHz to 3 GHz in a few months.
Likewise AMD isn't going to get a 1.4 to 1.8 GHz boost in a single die-shrink. It takes a lot more than that (all you have to do is look back at all the other die-shrinks that AMD and Intel did on their architectures, and detect the trend of improvement they got out of it. This requires a jump of about 70%).
 
game14 said:
Only thing i can say , about 8 or 9years ago when i build my first computer 1ghz was supposed to be the limit , 40gb HDD were huge and some said its impossible to build 100gb HDD... well now we have quadcores that easily clock to 3ghz, 1.5TB HDD ...lets see what comes next. 4.4ghz quadcores ? why not , just a matter of time.
Click to expand...

Exactly... ;)

And 1TFLOP G'Cards were things of science fiction... :p

Stock 4.4GHz quad's...

Improbable ~ Maybe...

Impossible ~ No...

I'm enjoying thinking that this chip may possibly be real even
if it's still just a fantasy... Please don't burst my bubble just yet... :p
 
game14 said:
...lets see what comes next. 4.4ghz quadcores ? why not , just a matter of time.
Click to expand...

For the most part because everyones to busy it seems trying to build more and more efficient cycle for cycle clock for clock cpus. And if they arent concerned with work efficiency per clock cycle, then they are off trying to make the chips cooler and cooler.

It seems like building a 10ghz chip is a thing of the past, for now, for the industry leaders. I'm sure sooner or later the ghz's war will start up yet again but I dont see it happening until amd's k11 generation.
 
Waffles. The site started less than 2 weeks ago and is just trying to lure in suckers^h^h^h^h^h^h readers.
 
Even if a new AMD quad OC's in the area of 4GHz, that would be great.
lol.gif
 
DemonDiablo said:
For the most part because everyones to busy it seems trying to build more and more efficient cycle for cycle clock for clock cpus. And if they arent concerned with work efficiency per clock cycle, then they are off trying to make the chips cooler and cooler.
Click to expand...

That's true, if you look at the previous architecture and the current, then both Intel and AMD stepped down.
Intel went from 3.8 GHz max (Pentium 4/D) to 3.2 GHz (Core2).
AMD went from 3.2 GHz max (Athlon X2) to 2.6 GHz (Phenom).
In that sense both are actually further away from the 4 GHz mark than they were a few years ago.

Having said that, my brother runs a Pentium D system, aircooled, at about 4.2-4.3 GHz. So it was already possible to break the 4 GHz limit, even on air... and 4.4 might be doable aswell, if he uses a bigger cooler (Zalman Fatal1ty on there now, not the biggest cooler there is). But as far as I know, neither Phenoms nor Core2 Duo/Quads can get over 4 GHz on air (some Core2 Duo's just barely, I believe, but not near 4.4)
 
Scali2 said:
THaving said that, my brother runs a Pentium D system, aircooled, at about 4.2-4.3 GHz. So it was already possible to break the 4 GHz limit, even on air... and 4.4 might be doable aswell, if he uses a bigger cooler (Zalman Fatal1ty on there now, not the biggest cooler there is). But as far as I know, neither Phenoms nor Core2 Duo/Quads can get over 4 GHz on air (some Core2 Duo's just barely, I believe, but not near 4.4)
Click to expand...

Err have you even been following the E0 Wolfdales? People are breaking 4.4-4.5Ghz easily on air. For the most part, C0 wolfies got around 4-4.2Ghz fairly easily, though getting above that started requiring some heavy voltage.

Regardless, Ghz as a benchmark alone is obsolete (nice seeing you Netburst :D) - the A64 saw to that. IPC relative to Ghz, that's what actually matters.
 
Maybe 4.0 and 4.4 GHz OC'd, I'd believe that.

But 4 stock? No way.

My thought on the subject is that if the Core2 can't make it there, how would the Phenom core do it, even at 45nm?
 
Mav451 said:
Err have you even been following the E0 Wolfdales? People are breaking 4.4-4.5Ghz easily on air. For the most part, C0 wolfies got around 4-4.2Ghz fairly easily, though getting above that started requiring some heavy voltage.
Click to expand...

Apparently not. I knew about the early Wolfdales getting just above or below 4 GHz, but I didn't know about the E0 stepping at all, nor that overclocking in general has improved significantly since the first Wolfies. I did hear that the Yorkfields weren't all that successful in overclocking though.
Even so, I doubt you'd get there with a stock cooler and stock voltage, which is what is claimed about the Deneb in the articles linked in this thread.

Mav451 said:
Regardless, Ghz as a benchmark alone is obsolete (nice seeing you Netburst :D) - the A64 saw to that. IPC relative to Ghz, that's what actually matters.
Click to expand...

Yes, as nice as my brothers Pentium D is at 4.2 GHz, it's nowhere near as fast as my E6600 at 3 GHz. In fact, it has trouble keeping up with my E6600 even at the stock speed of 2.4 GHz. So the 4.2 GHz doesn't amount to much in the end. Oh well, at least he can still play Crysis with it :)
 
This sure feels a lot like the rumors that WD is coming out with a 20krpm Raptor "soon." There is a small chance that it's possible, but I wouldn't take the bet if I was gambling with my money.
 
You must log in or register to reply here.
Back
Top