Deneb does 4.3GHz on air
Forum CPU & Components : CPUs - Deneb does 4.3GHz on air
Another thread edited and locked due to the fanboy voting game.
Message edited by turpit on 10-08-2008 at 08:48:24 PM
I find it extremely difficult to believe. Given the process technology AMD uses, its simply no match to Intel's more advanced HK/MG process.
But of course, if there's actually proof for it (CPU-Z shot, machine shot), then it should be believed.

Reply to Yomamafor1
| Yomamafor1 wrote : I find it extremely difficult to believe. Given the process technology AMD uses, its simply no match to Intel's more advanced HK/MG process.
|
Lol, same here. Without high k material, AMD chips face a serious physical wall on oc that high. It's not just "computer stuff," there are laws of physics that need to be followed.
Reply to dagger
AMD's Deneb's do use HK/MG. Time for someone to do their homework. THat's why AMD's Denebs are doing well.
yep AMD and IBM have been using it for some time. IBM i guess developed their own version of high k and gave it to AMD.
i'll try to dig up the article i found it at.
edit: here the link http://blogs.computerworld.com/node/4462
someone teach me how to link?
Message edited by teh_boxzor on 10-06-2008 at 02:39:34 AM
Reply to teh_boxzor
Actually no. Deneb is still built on 45nm SOI technology from IBM. They will make the migrate probably next year.
| Quote : Time for someone to do their homework. |
Right back at ya

Reply to Yomamafor1
Sounds dubious, but good news for everyone other than Intel if it does turn out to be true...
if it is true, great. i couldn't help notice that site pretty much circles the wagons around AMD products so i would take that article with a grain of salt.
| teh_boxzor wrote : yep AMD and IBM have been using it for some time. IBM i guess developed their own version of high k and gave it to AMD.
|
I'm sorry, but your article mentions nothing about AMD moving to HKMG with Deneb.
Here is a report this year:
http://www.xbitlabs.com/news/cpu/d [...] ssors.html
| Quote : Key innovations within AMD’s 45nm process are scheduled to include Immersion Lithography, Fourth-generation Strained Silicon, Ultra-low-k Dielectrics, and High-k/metal Gates. However, the initial 45 nm CPU generation will not use high-k dielectric metal gate technology. AMD believes that it can be “competitive” without high-k at this time, but high-k is an option to be introduced at a later stage of 45 nm production. |

Reply to Yomamafor1
| roofus wrote : if it is true, great. i couldn't help notice that site pretty much circles the wagons around AMD products so i would take that article with a grain of salt. |
Well, they're not doing AMD any favors by trying to build up hype to an untenable level. It'll be the same thing that happened to Phenom. Overinflated expectations leads to crushing disappointment. ATI's 4000s graphics card is the opposite. No one expected much, but received high performers.
Reply to dagger
i 200% agree dagger. that's pretty much my point. dropping links and posting from obscure, subjective sites could lead to a major let down. AMD kept it pretty quiet on their 4000 series cards and had pure gold. the "leaks" and hype for the Deneb are reminiscent of the Phenom.
Go here to find out.High-K is later on.
http://www.amd.com/us-en/0,,3715_1 [...] dir=45nm01
See the Press Presentation .pdf and the Q&A on 45nm.
When did clock speeds matter in comparision to other very different chips?
| roofus wrote : i 200% agree dagger. that's pretty much my point. dropping links and posting from obscure, subjective sites could lead to a major let down. AMD kept it pretty quiet on their 4000 series cards and had pure gold. the "leaks" and hype for the Deneb are reminiscent of the Phenom. |
The same could be said of the sites reporting that Nehalem will OC over 4 Ghz on air. I'm skeptical that this article is true just as I am skeptical of the articles about Nehalem. Until the production processors are released and we see test results from non-ES chips it's all just conjecture.
Regarding AMD's use of the HK/MG process, I suspect that this will be implemented with the AM3 Denebs slotted to come out in Q1-09 and the AM2+ Denebs that will be released this year will be the last SOI parts.
| roofus wrote : the "leaks" and hype for the Deneb are reminiscent of the Phenom. |
But Deneb doesn't have a high performance bar to jump over, unlike Nehalem. People are also more skeptical of AMDs claims this time round (remember the Barcy 40% over Clovertown?), so I don't think there will be as much hype. More hope, for sure, but the hype has too much salt this time.
Message edited by randomizer on 10-06-2008 at 03:38:21 AM
And the hype machine goes into overdrive!![]()
I like the months before something new gets released, its exciting with so many rumours, leaks and claims of grandeur... of course with those highs often come the inevitable lows when release time comes and you realise all the hype was blown way out of proportion.
I'll be honest, without a CPU-Z screeny I am quite sceptical of these claims, especially the part about power consumption being better than Intel 45nm HKMG, but it would be amazing if true.
However, I'm sure most of us here remember the hype surrounding Phenom before release... and we all know how that went.
Message edited by epsilon84 on 10-06-2008 at 03:48:17 AM
| jj463rd wrote : Go here to find out.High-K is later on.
|
The link you provided and both the Press Presentation and Q&A simply mention HiKMGs as future 45nm options, not something being used. In fact, in the Press Presentation PDF, it simply states in slide 3 "First to volume production with low-K interconnect material", and in slide 4 "High-k/metal gate / future 45nm".
But back OT -
4.3Ghz is impressive if true. But what if Deneb cannot come close to this supposed mark? I just hope this isn't an attempt to put AMD's upcoming product in a spotlight that it might not be able to handle.
2:

Reply to NMDante
AMD is still SOI. They wont go HKMG until next year at best. Oh and by the way the Nehalem at 4 GHZ have pics.
And Intel's CURRENT Quad Core Processors are going 4ghz and Beyond on Air. So to believe their next gen can meet current levels is not surprising.
its not good theory.
Reply to zenmaster
| kassler wrote : http://www.nordichardware.com/news,8195.html
|
Considering Nehalem clock for clock gains a performance gap as wide as the gap between Core 2 and Phenom, and the fact that a phenom cant even compete with an ancient Core 2 Q6600 (since we ARE talking overclocked figures - 3ghz on stock, 3.6 on an aftermarket easy), and we are talking about Intel here, the memory performance+latency king (Intel 440BX anyone?) - i say good luck AMD.
Nehalem has real performance figures from 10+ sources, your AMD has no REAL figures - post when you have real figures and not BS.
8gb Kingston 800mhz
Gigabyte EP35-DS3P
XFX 8800GT/512
Reply to apache_lives
So NOW its hype. OK, sure it is. So is coolalers findings. You think Andreas would print this if he didnt think it had merit? I know one things for sure. Using HKNG doesnt let you over volt that high, as itll burn up when you try, so having this cpu volted high wont have the same effect as it would on Intels chips. And this isnt hype any more than what weve seen about i7
Reply to jaydeejohn
| randomizer wrote : (remember the Barcy 40% over Clovertown?) |
The alleged 40% performance over is based on 4U server performance. They never say that Agena will out-perform Penryn. Stars core design is for servers, and AMD just didn't bother to design and optimize it for desktop. Ci7 is out to battle AMD on 2U server, while we are just about to see 45nm 24-core vs 16-core battle raging very soon.
| jaydeejohn wrote : So NOW its hype. OK, sure it is. So is coolalers findings. You think Andreas would print this if he didnt think it had merit? I know one things for sure. Using HKNG doesnt let you over volt that high, as itll burn up when you try, so having this cpu volted high wont have the same effect as it would on Intels chips. And this isnt hype any more than what weve seen about i7 |
The 4.3ghz claimed for Deneb on air is already achieved by current Intel offerings, so Nehalem will need to take a big step backward.
The same 4.3ghz on air, however, isn't anywhere near the max oc of current Phenoms, of which Deneb shares the same architecture and is a die shrink of.
So "this isnt hype any more than what weve seen about i7" isn't right. It ranks much higher on the hype scale.
Reply to dagger
| jaydeejohn wrote : And this isnt hype any more than what weve seen about i7 |
Except that in one corner we have actual independent benchmarks of actual hardware and in the other corner we have the word of some guy at AMD and an anonymous 'source'...
I'd love to see AMD come out with a CPU that's competitive with Nehalem, but I don't see any reason to see this as more than hype right now.
Message edited by MarkG on 10-06-2008 at 04:47:37 AM
If this is true, would it kill some people around here? I think all of this is good news. If its hype about i7, or hype about Deneb, if both are true, its all good.
Message edited by jaydeejohn on 10-06-2008 at 05:00:15 AM
Reply to jaydeejohn
Except you need to say this to Andreas, and email him about it. So coolaler is better than Andreas? Or is this just fanboyism at its best? No one considered the 4xxx series to be what it is either, so if true it wouldnt surprise me. This is good news , I dont see why everyone has an excuse. Sure it may be lies, just like getting i7 to anything higher than 3.6 unless its a XE could be lies as well. Dont give me the "because I like or prefer this one over that one as an excuse. These are exciting times
Reply to jaydeejohn
Coolaler is very respectable. So to answer your question, Yes Coolaler is better than Andreas
And they had the 4xxx series as 800 shaders and were crazy then too. Maybe people remember for the wrong reasons?
Reply to jaydeejohn
Its so funny as I type nothing but truth I get downgraded, at the same time Im trying to be fair and am truly happy if both Intel and AMD from what weve heard and seen comes true. Some people are not real enthusiasts I guess, just bandwagoners. I was just as excited when I, yes me, posted the Annands article on then Nehalem. I guess people do like to remember for the wrong reasons
Reply to jaydeejohn
Bah... 4.3ghz is so old school. Where the 10ghz CPU's?
LOL, I think Intels got some special P4s waiting in the wings
Reply to jaydeejohn
Wow, lots of down votes on JDJ. I voted them back up, he hasn't said anything that should be down-voted.
On subjekt, the possibility of Deneb hitting 4.3Ghz is fantastic. Competition good, overclocking potential, which we haven't seen out of AMD since the early X2 days is awesome. Have any AMD chips ever gone over 3.5Ghz on air reliably, if at all? This is hype, but it's great news if it's true.
Its hype for sure, but if true, all those people waiting on this cpu will be very happy. And maybe well see better competition from now on, which translates into better deals/performance for us
Reply to jaydeejohn
There is an Intel fanboy on the loose in the voting system. Stand your ground, vote back! Viva la AMD!
*Gets hit with negative vote*
VIVA LA CYRIX!!!
Viva La Vacuum tube!!!!
Reply to jaydeejohn
what would happen if everyone stopped acting like 2 year olds...
the world would end
| dagger wrote : The 4.3ghz claimed for Deneb on air is already achieved by current Intel offerings, so Nehalem will need to take a big step backward.
|
Ummmm Phenom's perform clock for clock WORSE then an Intel (65nm Core 2's even), and do you forget what happend for AMD when they went from 90nm to 65nm - ZERO, high end models remained based on the old 90nm etc as the jump didnt help much, apart from yields/costs etc.
Mind you in the change AMD might do a few tweaks to bring it to a more competitive level but that might only bring them to stock Intel peryn speeds etc
Nehalem isnt hype we all have the figures - google it to see the performance, vs this paper tiger named deneb.
The voting system - wtf are we voting for? More like a Fanboy attack system or somethin
8gb Kingston 800mhz
Gigabyte EP35-DS3P
XFX 8800GT/512
Reply to apache_lives
Looks like the early precurser of the P4 heheh, just not as powerful or fast.... but probably hotter tho
Reply to jaydeejohn
i was mistaken i guess AMD will be using hi-k next year with the AM3 Denebs.
don't hate me. just trying to inform. =\
Reply to teh_boxzor
| gators1223 wrote : what would happen if everyone stopped acting like 2 year olds... |
AMD would reach 6GHz on air with SOI.
^+10
I think that AMD just didn't have very good 65nm tech. And somewhere it was stated that this was "more than just a process shrink". No reason to think there haven't been minor adjustments, and some of those could lead to 1. Higher clockspeeds and 2. Higher IMC/L3 speeds. Both should yield significant benefits. Phenom's arch scales well up in speed, it just isn't very fast yet.
As of 2004, a chip of silicon measuring 0.02 inches (0.5 mm) square holds the same capacity as the ENIAC, which occupied a large room.
As old school as that is, its still very impressive and interesting thats where it all started. HAIL the vacuum tubes!!
I wonder what a cpu cooler would look like by todays standards if we were still using vacuum tube technology?...
...the size of a volkswagen perhaps?
Message edited by englandr753 on 10-06-2008 at 07:41:07 AM
| jaydeejohn wrote : So NOW its hype. OK, sure it is. So is coolalers findings. You think Andreas would print this if he didnt think it had merit? I know one things for sure. Using HKNG doesnt let you over volt that high, as itll burn up when you try, so having this cpu volted high wont have the same effect as it would on Intels chips. And this isnt hype any more than what weve seen about i7 |
I personally don't think this is hype, as it has not been confirmed by any believable sources. As for Coolaler's finding, I still don't believe its hype, because 1. Intel didn't authorize it, 2. Coolaler doesn't get paid to do so, and 3. Coolaler doesn't benefit financially from it. Its similar to someone posting a Phenom @ 3.6Ghz with cascade cooling. Its more like a statement of self-achievement rather than hyping the capability of the processor itself.
But of course, since you believe that it only matters how you feel, so I guess its useless for me to convince you if you refuse to be convinced.

Reply to Yomamafor1
| the last resort wrote : AMD's Deneb's do use HK/MG. Time for someone to do their homework. THat's why AMD's Denebs are doing well. |
No they don't Maybe AM3 will but more than likely the 32nm AMD CPUs will.
| Just_An_Engineer wrote : The same could be said of the sites reporting that Nehalem will OC over 4 Ghz on air. I'm skeptical that this article is true just as I am skeptical of the articles about Nehalem. Until the production processors are released and we see test results from non-ES chips it's all just conjecture.
|
The difference is the fact that the Core i7 OCs have CPU-Z shots from more than one place.
This is reminisant of that 4GHz stock Deneb Phenom FX80 screenshot that everyone thought was real when it was actually fake.
| jaydeejohn wrote : So NOW its hype. OK, sure it is. So is coolalers findings. You think Andreas would print this if he didnt think it had merit? I know one things for sure. Using HKNG doesnt let you over volt that high, as itll burn up when you try, so having this cpu volted high wont have the same effect as it would on Intels chips. And this isnt hype any more than what weve seen about i7 |
Without any real screenshots or CPU-Z verifications, yes it is hard to believe and feels like someone is hyping it. With Core i7 we have screen shots showing the 4GHz OC so that is not so much hype as it is fact.
| jaydeejohn wrote : And they had the 4xxx series as 800 shaders and were crazy then too. Maybe people remember for the wrong reasons? |
Actually you would be suprised how many people believed that ATI could do 800 shaders. Remember though the way its set up its 160 shaders x 5 giving the full 800 shaders. I for one know ATI can do it. The difference between ATI and nVidia and Intel and AMD is that ATI is tipically very good at coming back really big, normally because the gap between the technologies is not that big. The HD2900 series was not that far off of a 8800 series to start and driver uptates have put them on par.
The last few OC screens I could find from Deneb were at 3.4GHz on air but needed 1.5v+ for stability. Unless this is the 3GHz stock version. But until I see a validation or at least a CPU-Z screen I can't believe it.

Reply to jimmysmitty
Actually, they took a big hit to their credibility for saying it was 800, maybe sorta like here with Deneb? I wonder how much Andreas got paid for printing this? I wonder where and AMD official has allowed this? I wonder why both are hiding their sources? If AMD was closer in performance than it is now, you can bet, these "leaks" weve seen wouldnt have happened, or for sure thered be greater action taken. The 2900 was all software driven with their AA resolve, which is what put them so far behind, so of course it took awhile for them to catch up. Sure, it looked halfway decent with no AA on, but once the AA was turned on, if the drivers werent specifically tuned for that game, it flopped. Same with the 3xxx series. But the 4xxx series, they changed their AA resolve, went to a HW solution, and added 150% shaders, got rid of their ringbus etc. It was far more than a tweak, and why poeple still look forwards to driver enhancements for these cards, as newer drivers will add such improvements
Reply to jaydeejohn
| apache_lives wrote : Considering Nehalem clock for clock gains a performance gap as wide as the gap between Core 2 and Phenom, and the fact that a phenom cant even compete with an ancient Core 2 Q6600 (since we ARE talking overclocked figures - 3ghz on stock, 3.6 on an aftermarket easy), and we are talking about Intel here, the memory performance+latency king (Intel 440BX anyone?) - i say good luck AMD.
|
ancient core 2 q6600 : http://www.guru3d.com/article/amd- [...] r-tested/1
AMD keep schtum about what theyre up to these days until its release time. 4000 series radeons example of this.
There are 6 identified and unidentified users. To see the list of identified users, Click here.
Please mind
You are about to answer a thread that has been inactive for more than 6 months.
If you still wish to proceed, please ensure that your posting is original and does not duplicate or overlap any prior responses to this thread.
