Early 45nm Deneb Spotted

[aigomorla]

Extelleron.

As i said my friend has a Neha sitting 3 houses from me.

And, its FAST.

Thats all i need to say.
To me FAST is REALLY FAST.

 

[Gikaseixas]

Originally posted by: aigomorla

Originally posted by: Lonyo

Also, please 'invest' in Firefox. It has a built in spellcheck.
(And try and be less of an obvious Intel fanboy who needs to scream it from the rooftops, and behave in a slightly more respectable way befitting a moderator on the forums).

how in anyway am i showing fanism?

Show me a phenom that truely does perform.

Also show me a AMD that performs remotely close to its competition.

Bah, seems like you people just dont get it do you.

Okey, i wont say anymore, buy your phenoms. Ive had one. 9850BE. Was fairly nice, clocked at 3.015ghz, ran a bit too hot, consumed roughly the same electricty as a kentsfield @ 3.45ghz.

But hey, i dont know my processors right?

Aigo all we're saying is that AMD might turn this thing around, well not turn around but come close to Yorkfield. Nobody knows what the future holds, nobody knew that a 4870 would perform so close to the 280 GTX, i bet all of us where surprised. On the same note, AMD fans didn't expect CD2 to be this good bet, everything's possible.

 

[aigomorla]

Originally posted by: Gikaseixas

Aigo all we're saying is that AMD might turn this thing around, well not turn around but come close to Yorkfield. Nobody knows what the future holds, nobody knew that a 4870 would perform so close to the 280 GTX, i bet all of us where surprised. On the same note, AMD fans didn't expect CD2 to be this good bet, everything's possible.

gpu is another story.

I love ATI.

I have 4 HD3870 cards, 2 of which are gona be replaced to HD4870X2's when they come out.

I have nothing bad to say about ATI. They make awesome cards for the price.

Also i have no problems with AMD chipsets either. I personally think there much better then intel chipsets. I really perfer the AMD boards over Intel Boards. AMD was totally ment for the overclocker, but...

The CPU is what needs work.



Okey Gikas, yes i do hope AMD gets better however, these tests arent showing anything special.

When the C2D was still a baby in mobile platform, YONAH, people were using S479 boards to push them.

We saw the first sub 14 sec super PI off a mobile C2D. And we based it off that thinkn the desktop version would be uber.

There isnt anything AMD has shown us that shows this kind of performance path. And messing up on the southbridge and going OOPS, here people buy another 200 dollar board with the correct south bridge is not a good way to win back your customers.

This last part pissed me off. Its not nice having a 250 dollar ASUS M3A32-MVP-DLX only to tell them OOPS, we gave you a cheaper crapper Southbridge that cant overclock. And NO RMA's allowed.

 

[Extelleron]

Originally posted by: aigomorla
Extelleron.

As i said my friend has a Neha sitting 3 houses from me.

And, its FAST.

Thats all i need to say.
To me FAST is REALLY FAST.

Of course it's fast. Unless Intel messed up very badly, it's faster than Yorkfield and Yorkfield is the fastest thing currently available.

The question is how much faster. In Cinebench....it'll be like night and day comparing it to Yorkfield. But how about in something like Crysis that doesn't scale beyond 2 threads? IMC + QPI + architectural improvements should make it faster than Yorkfield, but by how much? That's what we will have to see come Q4.

Nehalem is a server CPU at heart IMO. Every improvement in Nehalem is more important in the server space than on the desktop. HyperTransport has given AMD an advantage in servers for years now, but it hasn't done much for them on the desktop. I don't think QPI will do much for Intel on the desktop either. SMT makes sense for servers as I said earlier, but not for desktops at this point.

 

[aigomorla]

Originally posted by: Extelleron

Originally posted by: aigomorla
Extelleron.

As i said my friend has a Neha sitting 3 houses from me.

And, its FAST.

Thats all i need to say.
To me FAST is REALLY FAST.

Of course it's fast. Unless Intel messed up very badly, it's faster than Yorkfield and Yorkfield is the fastest thing currently available.

The question is how much faster. In Cinebench....it'll be like night and day comparing it to Yorkfield. But how about in something like Crysis that doesn't scale beyond 2 threads? IMC + QPI + architectural improvements should make it faster than Yorkfield, but by how much? That's what we will have to see come Q4.

i know the answers to this, but i cant tell you guys.

im sorry.

But what i can say is, you'll go WOW.

Once NDA comes out, i'll show you guys.

He's gonna let me drop water on his machine and bench it.

Okey let me appologize, i do kinda sound like a fanboy if you dont know me.

So Let me explain,
I only make statements from what ive seen. Im very honest with my results, and i dont tweek or candy coat anything.

I am hated/blacklisted by a few vendors because of my "bluntness". If i see a problem, i'll yell it out, instead of quietly let them know.

Im quick to point out faults.

Ive seen Neha, ive played, touched, drooled.
After that wonderful experience, i gave away erinyes to setup a new bench. This is how much of a WOW i said after i came back.

 

[SlowSpyder]

aigomorla, just out of curiosity, you keep saying the Phenom is 'crap'... why do you think that? Is it that the C2D's bench much faster or was there something your Phenom couldn't do well enough? Let me be clear on this, I know that the C2D's are faster then the Phenom. Intel hit a home run and then some. They use reasonable power, they have reasonable thermals, they overclock as well as you could want, and they perform stellar. No doubt about it, the C2D is better all around then the Phenom.

But, I must have the fastest Phenom in the world or something, I have yet to run into anything at all that it cannot do. I game on it, nothing seems to slow down a bit. I have been really, really into Titan Quest lately, I think the Diablo 3 trailre got my hack & slash juices running or something. I went from using my PC maybe .5-1 hour a day for general use to gaming a few hours a day (loading the CPU/GPU) and can honestly say that my electricity bill has not budged. As I mentioned above, the C2D IS faster, sometimes by a long shot, yet I feel like we've gotten to a point where CPU's have so much juice that faster is just for bragging rights and synthetic benchmarks at this point. Don't get me wrong, I love to race my car, I understand the desire to have faster just to have faster. But, I just have not found any instance where my Phenom isn't good enough... no where have I yet found a piece of software that when I try to run it I feel that I need more CPU.

I'm not trying to argue, certainly the C2D is a great buy. I am just trying to understand what your Phenom system couldn't do. Or get an idea of what people feel it is bad at so I can understand where this is coming from. I'm going back to Titan Quest now.

 

[ultra laser]

20 seconds for a 3.4GHz CPU is terrible.

 

[harpoon84]

SuperPi is a fun benchmark for enthusiasts, but its quite useless for measuring real world performance, especially Intel vs AMD, but even comparing performance between the same family. Its incredibly sensitive to cache size, so its comes as no surprise that Deneb with 6MB L3 dominates a Phenom with only 2MB L3.

FWIW, Wolfdale is some ~10% faster than Conroe in SuperPi, but real world gains are about 5%. Conroe 4MB is also some ~9% faster than Conroe 2MB, but real world gains are about 3.5%. I'm not sure how this correlates with Phenom, but surely the 3x larger L3 on Deneb has to have a significant impact, probably more so than it might otherwise have in 'real world' apps.

 

[aigomorla]

spyder

i am whats known as a DC junkie.

for me cpu power is my main goal.

I crunch for a good cause and for points.


I had my friend put up his neha rig, and i was quite suprised at how well it did on the DC net. This is why i was sold when i saw the PPD

PPD = points per day.


I cant say anymore then that.


The AMD, it had problems matching PPD with my Kentsfield @ 3.2ghz. And it was very noticible. That is why i pulled it. Not because i hate AMD. It just couldnt provide the PPD i was expecting/need.

You guys know i recomend AMD when it fits. AMD has a niche. I am always a type that matches the person with the machine. I have 2 AMD rigs i use on the side, not PPD machines.

But Erinyes was suposed to be a PPD machine. Unfortunately, she didnt live up to her expectation.

And guys i really did take a lot of flame from friends when i bought the spyder. But as i said, no one can give out advice without owning the hardware. First hand knowledge is always the best. So i wanted to make sure i had the chance to play with one incase i ever spoke.

 

[VirtualLarry]

Originally posted by: aigomorla
The AMD, it had problems matching PPD with my Kentsfield @ 3.2ghz. And it was very noticible. That is why i pulled it. Not because i hate AMD. It just couldnt provide the PPD i was expecting/need.

Yes, but aren't all of the DC apps optimized for Intel, primarily. That seems like an unfair basis on what to compare CPUs by.

I mean, look at the PPD for the 4850 versus the 8800GT? The 4850 is arguably a faster GPU in all senses, but because the GPU2 client is optimized for NV's architecture, and not the AMD's, the AMD gets half of the PPD as the NV.

Would you use that as the basis to recommend GPUs for someone for gaming purposes?

Needless to say, after that admission by you that you consider F@H to be the ultimate benchmark, then I'm going to take your recommendations with a HUUUUUGE grain of salt. You obviously don't see the bigger picture.

 

[SickBeast]

I have a feeling that clock-for-clock these new 45nm AMD chips could be as fast as the 45nm intel chips. The problem is that intel seems to have a 4.4ghz ceiling on their chips, whereas these top out at 3.4ghz. It may be too soon to judge, but it looks as though AMD might continue sliding downward.

 

[bharatwaja]

Originally posted by: SlowSpyder
aigomorla, just out of curiosity, you keep saying the Phenom is 'crap'... why do you think that? Is it that the C2D's bench much faster or was there something your Phenom couldn't do well enough? Let me be clear on this, I know that the C2D's are faster then the Phenom. Intel hit a home run and then some. They use reasonable power, they have reasonable thermals, they overclock as well as you could want, and they perform stellar. No doubt about it, the C2D is better all around then the Phenom.

But, I must have the fastest Phenom in the world or something, I have yet to run into anything at all that it cannot do. I game on it, nothing seems to slow down a bit. I have been really, really into Titan Quest lately, I think the Diablo 3 trailre got my hack & slash juices running or something. I went from using my PC maybe .5-1 hour a day for general use to gaming a few hours a day (loading the CPU/GPU) and can honestly say that my electricity bill has not budged. As I mentioned above, the C2D IS faster, sometimes by a long shot, yet I feel like we've gotten to a point where CPU's have so much juice that faster is just for bragging rights and synthetic benchmarks at this point. Don't get me wrong, I love to race my car, I understand the desire to have faster just to have faster. But, I just have not found any instance where my Phenom isn't good enough... no where have I yet found a piece of software that when I try to run it I feel that I need more CPU.

I'm not trying to argue, certainly the C2D is a great buy. I am just trying to understand what your Phenom system couldn't do. Or get an idea of what people feel it is bad at so I can understand where this is coming from. I'm going back to Titan Quest now.

+1 :thumbsup:

I doubt, beyond a point how we can experience the actual speed of processors in a real world scenario rather than a synthetic benchmark.... video encoding and stuff probably? but what a few seconds worth, like, would it be 10-15% slower at most? we could live with that.

The real competition comes from AMD when ppl realize this and get these quads for their price point, and then intel has nothing but to reduce the price of their line up of procs, which ultimately benefits consumers, it doesnt matter who is on top.... it matters who reaches out to every desk.... and if u still feel u want the "fastestestestest" even if jus for bragging rights, then no one prevents you from going with Intel......

I hope AMD gets more competitive for this reason.... No one expects AMD to turn the tables around.... but a lil' competition could go a long way....

 

[aigomorla]

Originally posted by: VirtualLarry

Originally posted by: aigomorla
The AMD, it had problems matching PPD with my Kentsfield @ 3.2ghz. And it was very noticible. That is why i pulled it. Not because i hate AMD. It just couldnt provide the PPD i was expecting/need.

Yes, but aren't all of the DC apps optimized for Intel, primarily. That seems like an unfair basis on what to compare CPUs by.

nah I use WCG Bionic. And it held my Opteron quite well.

People even ran dual opty's on it, and got very good results.

Also F@H on a Cell processor > Intel. Just incase you didnt know
http://arstechnica.com/news.ar...owns-the-pc-in-fh.html

im seriously not being biased. Just wait and see.

i tasted the colder water. And yes its very yummy.


EDIT: Seeing how im pissing everyone off, i'll keep out of this thread.
However when Neha is full revealed, you'll all understand.

If not i'll preview it when i get my system up. :T

 

[Soulkeeper]

Originally posted by: aigomorla

Originally posted by: VirtualLarry

Originally posted by: aigomorla
The AMD, it had problems matching PPD with my Kentsfield @ 3.2ghz. And it was very noticible. That is why i pulled it. Not because i hate AMD. It just couldnt provide the PPD i was expecting/need.

Yes, but aren't all of the DC apps optimized for Intel, primarily. That seems like an unfair basis on what to compare CPUs by.

nah I use WCG Bionic. And it held my Opteron quite well.

People even ran dual opty's on it, and got very good results.

Also F@H on a Cell processor > Intel. Just incase you didnt know
http://arstechnica.com/news.ar...owns-the-pc-in-fh.html

im seriously not being biased. Just wait and see.

i tasted the colder water. And yes its very yummy.

I too also run WCG, looking forward to the deneb release, i'm getting bored with intel
had two P4's and two C2D's
havn't toyed with amd since the athlon xp (owned atleast 10 of them)

einstein@home is another project that runs much better on an amd cpu than intel

i got the money set aside, if deneb quad can more than double my c2d setup, then it's getting replaced


btw checking newegg, for the price of the cheapest 45nm intel quad cpu you can buy TWO amd duals and TWO full featured amd motherboards

intel is making a killing, i'm ready to give amd the benefit of the doubt for the next release

 

[dmens]

Originally posted by: Extelleron
Nehalem is a server CPU at heart IMO. Every improvement in Nehalem is more important in the server space than on the desktop. HyperTransport has given AMD an advantage in servers for years now, but it hasn't done much for them on the desktop. I don't think QPI will do much for Intel on the desktop either. SMT makes sense for servers as I said earlier, but not for desktops at this point.

heh, if nehalem was yorkfield + CSI, it would've been out a year ago. it's a lot more than that.

 

[Triskaine]

Originally posted by: dmens

Originally posted by: Extelleron
Nehalem is a server CPU at heart IMO. Every improvement in Nehalem is more important in the server space than on the desktop. HyperTransport has given AMD an advantage in servers for years now, but it hasn't done much for them on the desktop. I don't think QPI will do much for Intel on the desktop either. SMT makes sense for servers as I said earlier, but not for desktops at this point.

heh, if nehalem was yorkfield + CSI, it would've been out a year ago. it's a lot more than that.

If it is, than I must have missed something. Nehalem is Yorkfield+IMC+SMT+New Cache Structure+QPI+some minor changes to the frontend. So nothing revolutionary or totally new. It will certainly be a killer chip for servers and for anything that can make use of its eight threads, like encoding, rendering, you name it. But the peformance gains in the average desktop apps and games will be limited.

 

[Somniferum]

Originally posted by: SlowSpyder
But, I must have the fastest Phenom in the world or something, I have yet to run into anything at all that it cannot do. I game on it, nothing seems to slow down a bit. I have been really, really into Titan Quest lately, I think the Diablo 3 trailre got my hack & slash juices running or something. I went from using my PC maybe .5-1 hour a day for general use to gaming a few hours a day (loading the CPU/GPU) and can honestly say that my electricity bill has not budged. As I mentioned above, the C2D IS faster, sometimes by a long shot, yet I feel like we've gotten to a point where CPU's have so much juice that faster is just for bragging rights and synthetic benchmarks at this point. Don't get me wrong, I love to race my car, I understand the desire to have faster just to have faster. But, I just have not found any instance where my Phenom isn't good enough... no where have I yet found a piece of software that when I try to run it I feel that I need more CPU.

The funny thing is, I feel the same way about my 3800+ @2.53GHz. Could it be that CPUs haven't really advanced all that much in the past couple of years, when it comes to real world performance? Sure, SuperPi might be fearing for its life, but last time I checked, pi was a constant. Fact of the matter is, CPUs hit a wall around 3GHz and have not really made much progress since. Some gains have been made around the edges, with cache sizes and such, but ultimately the bitter truth is that Moore's law has been revoked.

 

[Gikaseixas]

Originally posted by: SlowSpyder
aigomorla, just out of curiosity, you keep saying the Phenom is 'crap'... why do you think that? Is it that the C2D's bench much faster or was there something your Phenom couldn't do well enough? Let me be clear on this, I know that the C2D's are faster then the Phenom. Intel hit a home run and then some. They use reasonable power, they have reasonable thermals, they overclock as well as you could want, and they perform stellar. No doubt about it, the C2D is better all around then the Phenom.

But, I must have the fastest Phenom in the world or something, I have yet to run into anything at all that it cannot do. I game on it, nothing seems to slow down a bit. I have been really, really into Titan Quest lately, I think the Diablo 3 trailre got my hack & slash juices running or something. I went from using my PC maybe .5-1 hour a day for general use to gaming a few hours a day (loading the CPU/GPU) and can honestly say that my electricity bill has not budged. As I mentioned above, the C2D IS faster, sometimes by a long shot, yet I feel like we've gotten to a point where CPU's have so much juice that faster is just for bragging rights and synthetic benchmarks at this point. Don't get me wrong, I love to race my car, I understand the desire to have faster just to have faster. But, I just have not found any instance where my Phenom isn't good enough... no where have I yet found a piece of software that when I try to run it I feel that I need more CPU.

I'm not trying to argue, certainly the C2D is a great buy. I am just trying to understand what your Phenom system couldn't do. Or get an idea of what people feel it is bad at so I can understand where this is coming from. I'm going back to Titan Quest now.

Well said, i share the same opinion, i have two very different setups and like i said before my Phenom rocks @ 3.0ghz, it really flies and i have a E8500 @ 3.8ghz so this is saying something.

 

[zsdersw]

Originally posted by: Triskaine
If it is, than I must have missed something. Nehalem is Yorkfield+IMC+SMT+New Cache Structure+QPI+some minor changes to the frontend.

What you listed is a lot more than Yorkfield+CSI, which is exactly what dmens said.

Nehalem being "a lot more than that" doesn't mean it has to contain anything "revolutionary or totally new".

 

[dmens]

Originally posted by: Triskaine
If it is, than I must have missed something. Nehalem is Yorkfield+IMC+SMT+New Cache Structure+QPI+some minor changes to the frontend.

yeah you missed something, unless minor changes to the frontend takes years of work.

 

[Idontcare]

Originally posted by: SlowSpyder
.
.
.
No doubt about it, the C2D is better all around then the Phenom.

But, I must have the fastest Phenom in the world or something, I have yet to run into anything at all that it cannot do. I game on it, nothing seems to slow down a bit.
.
.
.
I'm not trying to argue, certainly the C2D is a great buy. I am just trying to understand what your Phenom system couldn't do. Or get an idea of what people feel it is bad at so I can understand where this is coming from. I'm going back to Titan Quest now.

Hey SlowSpyder I sent you PM as I am really interested in collaborating with someone who is on-the-ball with their Phenom system so as to generate some real-world benchmark results (Kentsfield vs. Phenom) with my particular financial trading package of interest Metatrader 4.

 

[Triskaine]

Originally posted by: dmens

Originally posted by: Triskaine
If it is, than I must have missed something. Nehalem is Yorkfield+IMC+SMT+New Cache Structure+QPI+some minor changes to the frontend.

yeah you missed something, unless minor changes to the frontend takes years of work.

Yes I have forgotten to mention the new TLB and second-level branchprediction the added support for unaligned-cache requests and some other stuff which I would put under 'minor'. It is definitely not trivial to put all that into a scalable Architecture, test and debug it etc. Especially QPI seems to have taken a while to complete.

 

[dmens]

Originally posted by: TriskaineYes I have forgotten to mention the new TLB and second-level branchprediction the added support for unaligned-cache requests and some other stuff which I would put under 'minor'. It is definitely not trivial to put all that into a scalable Architecture, test and debug it etc. Especially QPI seems to have taken a while to complete.

yeah those are pretty minor. too bad that's not the end of it. do you really think intel would tell everything they did to realworldtech?

 

[extra]

I just hope amd gets their act together so we get some good competition going again like there is in the video card market. Hopefully amd can under promise and over deliver with some processors like they did with the 48xx video cards. I doubt very much that they will steal the performance crown back from intel, but if they can get competive and then win based on features and motherboards then that is enough for us all to win (i have no brand preference buy only on value heh, and right now, that's intel imho..but by this winter who knows?)

 

[Foxery]

AMD knows it won't claim any raw performance wins any time soon. Nobody but their marketing department seriousy believes that enthusiasts would suddenly jump ship from Intel just for a die shrink. (Technically, a "Tick" release.)

Releasing a 45nm part is just good business practice for saving money on prouction costs, (thus improving the profit margns on what little they sell,) and getting ready for the next product by transitioning to the new process in advance. Cutting power usage and heat output will also entice some corporate customers to buy a few Server-class (i.e. multisocket) systems.