Pre-release Tests Conducted on AMD Denreb

[vojc]

My E3110 Xeon (E8400) does 4Ghz at 1.4125V (24/7 stable)

You got an awesome chip.

u all forget something, we talk abaut 4 cores here, not 2 cores, tell me how many q6600 or q9*** can work on 4ghz at 1.25V? my q6600 need 1.475 for 3.2-3.4GHz

 

[DaedalusHelios]

u all forget something, we talk abaut 4 cores here, not 2 cores, tell me how many q6600 or q9*** can work on 4ghz at 1.25V? my q6600 need 1.475 for 3.2-3.4GHz

Q6600 is 65nm. Not 45nm.

 

[Kei]

The AMD dual core processors are still pretty awesome in their Black editions. When will Intel get those awesome Multi's.

A Phenom at the same clocks as the dual core AMD chips is significantly faster. I run my Phenom in X2 mode majority of the time as I don't need the power all day and I am VERY much faster than any of the Athlon 64 X2 chips at the same speed. As long as we can get the same clocks the Phenom is much faster...in order to equal the Phenom in speed the Athlons need to be clocked significantly higher, so in a round about way you just complimented the Phenom and we thank you haha.

K

 

[DaedalusHelios]

A fanboy is someone who feverently believes that they chosen company is better than the others even if the other company has been proven to be better in certain things, or overall.

There are still Intel fanboys who believe that you can't get better than Intel, and in raw performance they aren't half right, but having used Intel for over 18 months now as impressed with the raw performance and incredibly high benches, I'm looking for something more stable that will last me longer (IE no socket changes or being forced to change motherboard to use the latest tech).

They are behind in game benches..... thats not synthetic... thats real performance.


Pentium D was crap. Early Phenoms were crap. You win some, you lose some. I just find it funny when people dodge the facts.

 

[WarEagleAU]

That was the biggest thing I noticed, the oc potential. Its a hell of a lot more than the current phenoms, even their black editions. I also like the 6mb of l3 cache, much needed for them. @45nm this engineering sample is a tad bit higher on volts than Id like, but most AMD chips have higher than normal voltages compared with Intel. This is the first step, I foresee alot more chips with better potential from AMD.

Also, SuperPI was never AMDs strong point. the p4 vs a64 shows that.

 

[DaedalusHelios]

A Phenom at the same clocks as the dual core AMD chips is significantly faster. I run my Phenom in X2 mode majority of the time as I don't need the power all day and I am VERY much faster than any of the Athlon 64 X2 chips at the same speed. As long as we can get the same clocks the Phenom is much faster...in order to equal the Phenom in speed the Athlons need to be clocked significantly higher, so in a round about way you just complimented the Phenom and we thank you haha.

K

So you would pay the price for a quad, and have to compare it to the same companies old tech, to make it look good, rather than compare it to the opposing companies product line?

~Largest run-on sentence in the world^^^^

 

[rhythmeister]

Good god that thing is making me want to wait for a release date instead of going Phenom 8750 and ocing in a 780 based mATX board!

Good old AMD, how much would the intel fanbois be paying for their "quad" cores without AMD nipping at their heals and giving us all superb value for money?

I've got an old 2500 XP-M that takes more than 1.65V to touch 2.3GHz and these new beasts are hitting 3.4GHz with four cores? It's insane I tells ya!

 

[cdawall]

They are behind in game benches..... thats not synthetic... thats real performance.


Pentium D was crap. Early Phenoms were crap. You win some, you lose some. I just find it funny when people dodge the facts.

i have an early phenom i wouldn't call 2.8ghz crap

 

[DaedalusHelios]

i have an early phenom i wouldn't call 2.8ghz crap

My main reasoning was the errata.

PS You could make whatever processor you picked up clock well. You aren't scared of frying anything. You bought a Celeron just to murder it. That phenom better do what you want it to, or you will decide to murder it and grab another. Its a massacre!

 

[Paintface]

Ill be finally upgrading to quadcore when these 45nm phenoms come out.

The performance is more than good

Most important though remember the prices are of the CPUs you compare it to, along with the cost of the mainboard, both which AMD is usually cheaper then intel comparing the same performance.

I think its wrong to compare the best intel has to the best AMD has but neglecting the price difference, but sadly many look at who has the best performing CPU and than go down the list of what CPU they can actually afford.

Im not a die hard fanboy but i do prefer AMD, i had their CPUs since 1994, they made it even possible for me to afford a PC back then, they always had good price/performance even when they wrecked the competition with their athlon 64s, which cant be said of Intel.
Also even if the fastest AMD is half the speed of the best intel, games dont require quadcores or that kind of performance, so its not an issue for me at all either.

Glad to see they are back on track, ill get it along with a new mobo and a 4870 end of the year.

 

[johnnyfiive]

Phenoms are solid processors. They are not huge overclockers and we all know that. When they were first released they were the much cheaper quad core option. If your not an overclocker or your a content builder its a good choice. I'm pretty satisfied with mine. Crysis performs much better on it then people give it credit for. With my CrossFired 3870 setup I get 35fps in HIGH at 1680x1050, this is at 2.3GHz. Thats not bad at all. I've seen people with the same GPU setup with a Q6600 at 3.4-3.6GHz get the same fps. So that shows you that the Phenom is not bad in ALL AREA's.

 

[Darkrealms]

I really hope AMD has a lot more work planned for these : ( They don't look BAD but they need something impressive at this point. Decent and price point will only take them so far . . .

Common AMD give us something to look forward to!


Fanboy note AMD/Nvidia, I want something good to come from AMD

 

[kid41212003]

It's not too bad, it's a reasonable release. It may not out-perform the Intel current 45nm Quad-core, but It'll probabaly have a really attractive prices. And all the current users who have AM2 motherboard still can use this CPU.

 

[cdawall]

My main reasoning was the errata.

PS You could make whatever processor you picked up clock well. You aren't scared of frying anything. You bought a Celeron just to murder it. That phenom better do what you want it to, or you will decide to murder it and grab another. Its a massacre!

meh TLB fix is off and i have yet to have an issue however no idea what you mean i have yet to kill a chip

i just have no problem cranking volts through things everyone i see is a pansy and gets scared with the smallest increase lol no problem with 2v through my 5000BE if it lets me hit the WR on it

 

[mamisano]

This is quite good news, considering how much better the CPUs are supposed to overclock with AMD's upcoming SB750 southbridge.

Deneb + 790GX + SB750 + HD4700x2 would be a sweet system!

 

[newtekie1]

But who cares about pure number crunching if the processor costs a arm and a leg. I would rather sacrifice a encoding a video 5 seconds slower if it means I pay £50 less. Honestly are you going to care if your application launches a nanosecond faster or if you get 150 FPS in a game opposed to 149.9 seconds if you can't afford the extra £ required or for a better Super PI result which didn't have any baring in the applications you really use everyday.

A processor doesn't have to cost an arm and a leg to crunch number or to outperform this processor. A Q6600 is sub-$200 and crunches numbers like crazy.

Well according to the article phanbuey sent earlier, despite most of the benchmarks being synthetical the 9850 BE was still performing equivalently, in some benchmarks better in some worst, in some just trailing behind. If you were to measure the two CPU's using SuperPI it would show a huge gap between the two CPU's in favour of the Q6600 maybe 10 seconds between? in most tests according to phanbuey's article the performance difference wasn't abnormally different, which completely contradicts the readings Super PI would of gave.

Where you looking at the same benchmarkes I was? Most of them were not synthetic, or at least as non-synthetic as a benchark can get. They were testing real work applications. Just because they weren't games, doesn't mean they aren't real world. The world consists of a lot more than games.

ProShow is a real world application, and the Q6600 destroys the 9850. Sony Vega is a real world application and the Q6600 beats the 9850 again. Excel is a real world application, and again the Q6600 hands the 9850 it's ass. Cinebench is a benchmark program based on a real world 3D Modeling app, essentially what a time demo would be to a game, and this is the only real world benchmark the 9850 managed to equal the Q6600 in one of the two tests. POV is another real world app, and the Q6600 beats even the 9950 in one of the tests, and just barely loses to the 9850 in the other.

I agree, that measuring the two with SuperPi doesn't give the whole picture, but it does give part of the picture. SuperPi cores are no where near the end all and be all of tests, but to a lot of poeple they are important. Different processors are always going to be better at different things.

Right now, all we have to go on is SuperPi times. As more details emerge, we will get a better idea of the performance of these processor. But for right now, all we have to go on are the SuperPi times, and they suck. Yes, they are better than the previous AMD processors, but saying they are better than the worst* isn't really saying much is it?

*I'm not saying AMD processors are the worst, I'm just saying they have the worst SuperPi times.

 

[aGeoM]

Nice to see some earlier tests on Deneb, it will be my next CPU to replace the B2.

Nice work AMD, now put them out.

 

[holy_]

I'm happy that AMD will release 45 nm Quad
Hope can get this when I have money

 

[robodude666]

Those super-Pi times are pretty bad, I would be embaressed.

Remember a couple of things. SuperPi is hungry for Cache which AMD processors lack in comparison to Intel. The Intel CPUs will always get better SuperPi results regardless of speed.

To get these types of SuperPi results with an Athlon X2 or other processor you'd need to OC to some CRAZY high speeds with LN2. My X2 3800+ @ 3GHz w/ 1000MHz memory was only getting 30s from superpi. So 23s at 3GHz is a free 7s drop which will mean nicer gaming performance.

The 3rd thing to remember is that SuperPi is not the only benchmark. There are other bunchmarks which are important to take into account. Maybe the new Phenoms are bad at SuperPi but they a ton faster at other important real world tests?

So far I like what I am seeing. Might have to dust off my 790FX-DQ6 later and get one for a test bed.

-robodude666

 

[Thefumigator]

It is a guage on how a single one of those cores performs, which we can then use to get an idea on how all 4 will perform. If 1 core performs like crap, adding 3 more crappy performing cores just gives you a quad-core processor that performs like crap compared to other quad-core processors.

Yes, still they could run 4 instances of superpi at the same time, each in their respective core.

 

[Tatty_Two]

Good thing there are a lot of people that follow AMD in this thread!

If you've been paying attention the last two forevers AMD's do not put up the same type of numbers as Intel processors in SuperPi so stop expecting it to put up a 10 second time because they NEVER have!

Now if you use 1% logic and 2% effort you'll go over the old SuperPi results from previous AMD processors and notice that this is a very nice improvement for these processors! My best SuperPi time to date is 24.679s in the 1M test with everything turned up as high as I can get it (3.1Ghz)...the new processor even with nothing turned up will nearly match that at only 2.8Ghz!!! That's a fantastic increase overall and I'm damn happy with it. For those wondering (and waiting to throw more mud) my setup for that test was... (all air cooling Xigmatek HDT-S1283 on low)

Phenom 9850 @ 3.1Ghz 1.37v
Patriot Extreme 1150Mhz 4-5-5-15 2T
Northbridge 2.354Ghz
HT Link 2.14Ghz

Anyway I'd love to see some other tests done with this but right now I'm sold already ESPECIALLY with the L3 6Mb cache as that'll help a lot in other apps we really needed the speed in.

On another note and without sounding terribly rude but...why do people with dual cores keep posting what clocks you can achieve on "x" voltage...nobody cares what your DUAL core can do. You quad guys on the other hand are a different story though because you actually have apples to apples going.

K

I agree completely but if you look at it from a slightly different perspective then....well, I will leave you to decide, I had one peach of a single core Athlon 4000+ San Diego chip about 20 months ago and got it to about 3.2gig which at the time I think got me the top AMD superPI score in these forums, thing is, even today, a 2008 AMD chip running superPI at 3.2gig will not post a time greatly faster, why, because the basic architecture, instruction set and to some degree ...efficiency in AMD's chips has not moved forward at any real pace, now if you go back those same months, see what time in superPI a 3gig P4 chip was posting, compare that with a wolfdale at 3gig and their is a VAST difference because Intel has made HUGE leaps in technology, architecture and efficiency.

Dont get me wrong, I am not a fan of artificial benchmarks and I am a fan of AMD and always will be, I was brought up on them, where superPI does work though.....is in identifying those performance hike's by making realistic comparisons between architectural changes.

 

[vojc]

hmmmm....well intel did two steps backwards on c2d CPUs (FSB )
so that why pentium 3 was faster than P4 on same clock C2D is mor like P3 in general, ok they have sse3,4......

 

[yogurt_21]

wtf did I just get lumped in as an intel fanboy? wow, that's a stretch. just because my system has an intel cpu doens't mean i wouldn't jump back over to amd at the first sign of good performance. the problem is that no phenom out now or about ot be relerased can best my current cpu. so I keep the intel.

it's funny that the term fanboy has been thrown about so easily because my last intel before this one (in my primary rig) was a pentium 2 since then it's been nothing but amd's till the q6700. I went from pentium 2 to a duron 600, to an athlon t-bird 950MHZ, to a 1.4GHZ athlon t-bird, to an athlon xp 2000, to an athlon xp 2500 barton, to an athlon xp 2600m, to an athlon 64 3500 newcastle, to a 4000 sandiego, to an fx-62. then finally the q6700.


I've followed amd in both good times and bad and this has to be the longest period I've seen them not be competitive. the k-6's flopped but they resurrected themselves one year later with the Athlon, the most successful period in amd's history. (yes revenue and stocks were higher than the athlon 64 empire) the athlon 64x2's became dated and amd decided to launch the phenom, now more than 2 years later and still no competitve cpu.

fanboyism is one thing, but seriously I need some proof that amd is doing more than just play around and highlight ati. I need the k7 day amd back. ability to offer a product cheap that beats intel in gaming but loses in multimedia. (as with modern gpu's, lack of multimedia performance can be offloaded by the gpu)

I'm not saying I'm taking this as the definitive performance os the new phenoms, I'm just saying that I think amd needs to move on past the phenoms and make something better, the k6's were set for a 2 year run, amd cut that off and intro'd the k7 early to stay competitive. they need to do the same here.

 

[Thefumigator]

A Phenom at the same clocks as the dual core AMD chips is significantly faster. I run my Phenom in X2 mode majority of the time as I don't need the power all day and I am VERY much faster than any of the Athlon 64 X2 chips at the same speed. As long as we can get the same clocks the Phenom is much faster...in order to equal the Phenom in speed the Athlons need to be clocked significantly higher, so in a round about way you just complimented the Phenom and we thank you haha.

K

I agree. I'm an owner of both an Athlon 64 X2 and Phenom X4, both systems are 2.2Ghz.
I work in video encoding and even in 2 cores, the Phenom is faster, so worth the upgrade on the cheap.

TMPGEnc 5000 frames Xvid file (virtual dub as frameserver)
Athlon X2 4200 2.2Ghz:

1 threads...133s
2 threads...71s

Phenom X4 9550 2.2Ghz:

1 threads...108s
2 threads...55s
3 threads...50s
4 threads...50s

Phenom X4 9550 2.2Ghz
2 instances of TMPGEnc (+2 instances of Virtual dub server):

1 threads...207s
2 threads...106s
3 threads...79s
4 threads...59s

 

[candle_86]

Agreed the problem is very simple though if you compare overall designs you end up with something scary and it makes sense.

The AThlon 64 is an Athlon K7 class chip at its heart, it has 64bit interngers and an IMC but at its roots its a K7, but thats not to bad honestly.

Core2 Duo at its heart is actully a pentium Pro, its P6 based design tweaked.

Now we all say the last major P6 design lost to AMD i correct you, AMD K7 lost to Coppermine in every gaming test or multimedia test, why did they well so well, well because the coppermine launched behind sechudle and topped out at 1ghz as the 1.13 was unstable and by the time the P3 was fixed the P4 was out and the AThlon XP was months away. But P6 was always stronger than K7 thats where Intel got them, AMD has to go back to thinking like they did then, how to undercut intel how to sell say a 3ghz chip for 50% less than intel ect