AMD Phenom X4 9950 BE Black Edition processor review [Guru3D]

Guru3D News

A review on the Phenom X4 9950 BE processor. The BE editions in essence are the 'regular' processor yet they passed quality and analysis for better overclocking performance and have their multiplier...

More...

Stukov · 10-03-2008, 01:33 | posts: 2,247 | Location: South Dakota | User is Offline

I am sometimes considered pro-AMD (actually would have bought a 45nm X4 if they would have ever come out in decent time) but I would think it would have been slightly more fair if you tested the Q6600 overclocked as well. At least overclocked to 3.1ghz like the 9950 got.

Oh, and otherwise another great article.

Hilbert Hagedoorn · 10-03-2008, 01:52 | posts: 11,705 | Location: Guru3D Trenches | User is Offline

Hey Stukov; there's only so many things we can do time-wise really.
Articles like these cost 2-3 days to produce. I try to cover the most important angles, yet each reader has it's own preference when it comes to what they want to see.

If I insert a Q6600 results, then i receive mails .. why are there no dual-core results ? If I use a dual-core to compoare, then I receuive mails .. why is there no Penryn in there ?. If I overclock a Q6600, then I get mails, why wheren't the other X4's overclocked ? And so on and on and on. I get roughly 30 emails per day with the questions "Why ?" or "I would have liked ..."

Fact is, that if we cover all these request angels .. you'd see 1 article per two weeks on Guru3D.com, as all angles cost a LOT of time.

We really can't do it all bro

But stricktly seen, the dynamics between a Q6600 / 9950 both overclocked would show the same performance scaling, just a little faster.

Stukov · 10-03-2008, 02:22 | posts: 2,247 | Location: South Dakota | User is Offline

Quote:

Originally Posted by Hilbert Hagedoorn

Hey Stukov; there's only so many things we can do time-wise really.
Articles like these cost 2-3 days to produce. I try to cover the most important angles, yet each reader has it's own preference when it comes to what they want to see.

If I insert a Q6600 results, then i receive mails .. why are there no dual-core results ? If I use a dual-core to compoare, then I receuive mails .. why is there no Penryn in there ?. If I overclock a Q6600, then I get mails, why wheren't the other X4's overclocked ? And so on and on and on. I get roughly 30 emails per day with the questions "Why ?" or "I would have liked ..."

Fact is, that if we cover all these request angels .. you'd see 1 article per two weeks on Guru3D.com, as all angles cost a LOT of time.

We really can't do it all bro

But stricktly seen, the dynamics between a Q6600 / 9950 both overclocked would show the same performance scaling, just a little faster.

Oh, no doubt, that was my only critisim of the article, I thought you did alot right as you put quad cores in the same price range against each other. Like I said, good arcticle, that was just one point I thought would be good constructive critisim (am I spelling that right? it looks wrong for some reason). The general quality of your articles is what made me want to start my own review site (as well as my need to find some place my long winded posts/ideas

, which I am still in the process of designing with a friend of mine.

aircool

That is pretty damned good value, definatley caught me thinks - where it counts.

Although Q6600's and P35 boards are of similar price, competition

Devolution · 10-03-2008, 03:10 | posts: 620 | Location: Sunny state | User is Offline

Seems like a great processor and all, but I'd like to see the AMD Fusion in action.

Stukov · 10-03-2008, 03:44 | posts: 2,247 | Location: South Dakota | User is Offline

Quote:

Originally Posted by Devolution

Seems like a great processor and all, but I'd like to see the AMD Fusion in action.

Swift will be out for mobile/laptops I think in Feb/March of 09, unless it slips.

jack0077 · 10-03-2008, 04:06 | posts: 3 | User is Offline

Quote:

ironically there's no quad-core platform more power efficient that the combo of a Phenom X4 with a 790GX based mainboard

yeah, right

http://www.anandtech.com/casecooling...oc.aspx?i=3413

Stukov · 10-03-2008, 04:09 | posts: 2,247 | Location: South Dakota | User is Offline

Quote:

Originally Posted by jack0077

yeah, right

http://www.anandtech.com/casecooling...oc.aspx?i=3413

That's funny, the 790GX isn't even in there. Neither is the 9950BE.

jack0077 · 10-03-2008, 04:48 | posts: 3 | User is Offline

Quote:

Originally Posted by Stukov

That's funny, the 790GX isn't even in there. Neither is the 9950BE.

No, but the 9850BE is, and since the 9950BE is the same architecture and made in the same process, and has a declared TDP that is 15W higher, I'm sure you can extrapolate.
Then you can look at the 790FX and 790X (42W and 43W, respectively). Do you think the 790GX will be a lot different?

Then again, these X4s are still 65nm, I'm sure the new 45nm Shanghais will be a lot cooler, and hopefully clock a lot higher also.

Hilbert Hagedoorn · 10-03-2008, 04:57 | posts: 11,705 | Location: Guru3D Trenches | User is Offline

Quote:

Originally Posted by jack0077

No, but the 9850BE is, and since the 9950BE is the same architecture and made in the same process, and has a declared TDP that is 15W higher, I'm sure you can extrapolate.
Then you can look at the 790FX and 790X (42W and 43W, respectively). Do you think the 790GX will be a lot different?

Then again, these X4s are still 65nm, I'm sure the new 45nm Shanghais will be a lot cooler, and hopefully clock a lot higher also.

You are correct in assuming that the peak wattage (TDP) on phenom X4 is much higher (as shown in our article) yet on average, say desktop usage, the ball turns arround. The 790GX chipset allows a lot P-states and clock gating.

So with an X4 / 790GX we used up something like 110-140 Watts for the entire PC while working on that PC. Fact is, there are not a lot of situations where you constantly use up all 4 CPU cores at 100%

That's why we love the power consumption that much. At daily normal usage this combo rocks -- if you think green that is.

But sure, at the other side of the scope, especially overclocked, the CPU is a power drain though, but we mention that in the article as well, in fact show it with some charts AND mention it in the conclusion.

Stukov · 10-03-2008, 05:06 | posts: 2,247 | Location: South Dakota | User is Offline

Quote:

Originally Posted by jack0077

No, but the 9850BE is, and since the 9950BE is the same architecture and made in the same process, and has a declared TDP that is 15W higher, I'm sure you can extrapolate.
Then you can look at the 790FX and 790X (42W and 43W, respectively). Do you think the 790GX will be a lot different?

Then again, these X4s are still 65nm, I'm sure the new 45nm Shanghais will be a lot cooler, and hopefully clock a lot higher also.

Yes the 790GX is sufficiently different. I also don't trust Anand's methodology in that article as I believe wall socket draw (with the right equipment) is more accurate. It won't give you individual component power ratings, but I don't think you can test those accuratly with consumer products. If AMD or Intel were to state some numbers on it, I wouldn't hesitate to beleive it, but this is a review site, not a manufuacterer of the product.

jack0077 · 10-03-2008, 05:27 | posts: 3 | User is Offline

Well, I was looking at that table and comparing the Q9550 with, say, the 9850BE.
Q9550 at any state (idle with power management, idle and load) is consuming half or less.
But I never tested it myself, and you did.
So I'll take your word for it.

theubersmurf · 10-03-2008, 06:04 | posts: 5 | User is Offline

I wish you had reviewed the 125 watt 9950...Er, well, I have no real confirmation of such from any reliable resource, however newegg is advertising one here, also of note was an x2 6000+ 65nm part, at 89 watts. I'm sort of curious if this is a 45 nm part that can't be announced because of the impending deneb release...They had to try the manufacturing process out on something established right? Meh, it's just a suspicion, but I'm curious about them, you should revisit this article with the newer part...be they real.

Iarwain · 10-03-2008, 06:38 | posts: 3,048 | User is Offline

Holy crap my CPU is sucking down 308 watts under full load?

No wonder crossfire stopped working, huh.

mandex · 10-03-2008, 07:35 | posts: 567 | Location: Brazil | User is Offline

Quote:

Originally Posted by Hilbert Hagedoorn

If I insert a Q6600 results, then i receive mails .. why are there no dual-core results ? If I use a dual-core to compoare, then I receuive mails .. why is there no Penryn in there ?. If I overclock a Q6600, then I get mails, why wheren't the other X4's overclocked ? And so on and on and on. I get roughly 30 emails per day with the questions "Why ?" or "I would have liked ..."

A cpu chart \o\

So amd vs intel is like ati vs nvidia, amd being fast/cheap and intel the faster but expensive?

Iarwain · 10-03-2008, 07:53 | posts: 3,048 | User is Offline

Quote:

Originally Posted by mandex

A cpu chart \o\

So amd vs intel is like ati vs nvidia, amd being fast/cheap and intel the faster but expensive?

Those comparisons aren't really fair or accurate. ATI vs Nvidia isn't AMD being cheaper and slower, and Nvidia being faster. ATI has the fastest, most expensive card, and solid performing low-mid range cards.

Nevermind the fact that your analysis changes with every release, and is irrelevant.

mandex · 10-03-2008, 09:08 | posts: 567 | Location: Brazil | User is Offline

yeah, the cpu chart was a joke in fact ;p
is dificult to make the benchs on the same conditions/variable, and new mobos could make one go faster or not.

but are you saying that 4870x2 be the fastest and most expensive?
i was thinking of single chips.(4870 vs 280)

Iarwain · 10-03-2008, 10:25 | posts: 3,048 | User is Offline

Quote:

Originally Posted by mandex

yeah, the cpu chart was a joke in fact ;p
is dificult to make the benchs on the same conditions/variable, and new mobos could make one go faster or not.

but are you saying that 4870x2 be the fastest and most expensive?
i was thinking of single chips.(4870 vs 280)

I'm not gonna have this fight with another person...single GPU doesn't mean anything. The requirements for the 280 and 4870x2 are roughly the same(one PCI-E slot, two power connectors, same installation procedure), and crossfire is supported by almost all games.

Colt M4 · 10-03-2008, 11:23 | posts: 669 | Location: Eden, NY | User is Offline

To use the overdrive overclocking feature do you have to have a AMD video card? Or just the motherboard and obvisouly the CPU. Do you even have to have a AMD chipset could you use a Nvidia Chipset to get the overdrive feature?

mandex · 10-03-2008, 11:39 | posts: 567 | Location: Brazil | User is Offline

Quote:

Originally Posted by Iarwain

I'm not gonna have this fight with another person...single GPU doesn't mean anything. The requirements for the 280 and 4870x2 are roughly the same(one PCI-E slot, two power connectors, same installation procedure), and crossfire is supported by almost all games.

sorry, my bad.I'm thinking of production of the chip.
I agree that for gaming it doesn't change anything.
But my question is who have better production method, something like that.

Norvekh · 10-03-2008, 13:35 | posts: 1,862 | Location: Ohio | User is Offline

When I had to replace some parts recently I thought long and hard whether to waffle back to an AMD system in hopes for the 45nm chips, for one of the few that is DDR2 compatible as I had seen a number of reviews where the 9950 really held well against the Q6600. In the end I decided to stay with Intel for a bit longer. But that darn upgrade bug just wants to have at an AMD system again even now.

Paladinchan · 10-03-2008, 14:03 | posts: 1,934 | Location: FL USA | User is Offline

i just realized that by using the 3870 is better than using a faster one.. because it shows the CPU speed .. right?

ive always kinda had the thought in the back of my mind to get a Phenom.. but their past performance has been kinda lacking behind the Core 2 Duo/Quad .. but how many FPS would you actually loose using this 9950 vs lets say a Q6600 like most ppl said was a better performer..

BlueSkyNIS

Hi guys! This article got my attention and actually made me to join this forum

Can someone, please, tell me what was the motherboard used for testing the 9950? I am hunting a good board for my new Phenom rig so I want to know what board was used for this testing...

Cheers

Iarwain · *Last edited by Iarwain; 10-03-2008 at 18:17.*

Quote:

Originally Posted by Paladinchan

i just realized that by using the 3870 is better than using a faster one.. because it shows the CPU speed .. right?

ive always kinda had the thought in the back of my mind to get a Phenom.. but their past performance has been kinda lacking behind the Core 2 Duo/Quad .. but how many FPS would you actually loose using this 9950 vs lets say a Q6600 like most ppl said was a better performer..

You can see from the review that the Q6600 stock, isn't a better performer than the 9950 stock, in many cases.