AMD 9950

[BTRY B 529th FA BN]

9950 I'd like to see some reviews over this chip.

 

[Idontcare]

Originally posted by: BTRY B 529th FA BN
9950 I'd like to see some reviews over this chip.

Anandtech has one: AMD's Phenom X4 9950, 9350e and 9150e: Lower Prices, Voltage Tricks and Strange Behavior

 

[TC91]

nothing too exciting

 

[Christobevii3]

It is at least a decent upgrade from people on x2's now. If they can get to 45nm and under 100 watts i'll definitely be buying.

 

[Sylvanas]

At XS it seems this chip with a bit of tweaking hits 3.4ghz on most 790FX boards.

 

[Markfw]

Originally posted by: Sylvanas
At XS it seems this chip with a bit of tweaking hits 3.4ghz on most 790FX boards.

Which still can't touch the $199 Q6600@3.4, which is almost guaranteed.....On a $100 motherboard

 

[BTRY B 529th FA BN]

Originally posted by: Idontcare

Originally posted by: BTRY B 529th FA BN
9950 I'd like to see some reviews over this chip.

Anandtech has one: AMD's Phenom X4 9950, 9350e and 9150e: Lower Prices, Voltage Tricks and Strange Behavior

thanks, didn't see the article

 

[BTRY B 529th FA BN]

Originally posted by: Markfw900

Originally posted by: Sylvanas
At XS it seems this chip with a bit of tweaking hits 3.4ghz on most 790FX boards.

Which still can't touch the $199 Q6600@3.4, which is almost guaranteed.....On a $100 motherboard

So, we can quote you on this 3.4Ghz comparison?

EDIT: Opps, i didn't see the money comparison, which btw, i haven't compared.

 

[myocardia]

Originally posted by: BTRY B 529th FA BN

Originally posted by: Markfw900

Originally posted by: Sylvanas
At XS it seems this chip with a bit of tweaking hits 3.4ghz on most 790FX boards.

Which still can't touch the $199 Q6600@3.4, which is almost guaranteed.....On a $100 motherboard

So, we can quote you on this 3.4Ghz comparison?

You can quote me, if you'd like. A clock-for-clock advantage doesn't care what speed the processors are running, as long as they're running @ the same speed, and Kentsfield is still faster than Phenom/Barcelona clock-for-clock, unless you're running memory bandwidth benchmarks.

 

[Sylvanas]

Originally posted by: myocardia

Originally posted by: BTRY B 529th FA BN

Originally posted by: Markfw900

Originally posted by: Sylvanas
At XS it seems this chip with a bit of tweaking hits 3.4ghz on most 790FX boards.

Which still can't touch the $199 Q6600@3.4, which is almost guaranteed.....On a $100 motherboard

So, we can quote you on this 3.4Ghz comparison?

You can quote me, if you'd like. A clock-for-clock advantage doesn't care what speed the processors are running, as long as they're running @ the same speed, and Kentsfield is still faster than Phenom/Barcelona clock-for-clock, unless you're running memory bandwidth benchmarks.

Can you please provide a link to 3.4ghz Phenom and 3.4ghz Q66 comparison? There are alot more things that go into gauging Phenom performance opposed to Intel- there is NB speed, HT frequency/multiplier and memory speed/timings (more sensitive to memory timings on AMD's IMC). I will see if I can dig through Google later on when I have the time as I am interested in the results.

 

[myocardia]

Originally posted by: Sylvanas
Can you please provide a link to 3.4ghz Phenom and 3.4ghz Q66 comparison? There are alot more things that go into gauging Phenom performance opposed to Intel- there is NB speed, HT frequency/multiplier and memory speed/timings (more sensitive to memory timings on AMD's IMC). I will see if I can dig through Google later on when I have the time as I am interested in the results.

Why would you need a 3.4 Ghz comparison? If either of the CPU's has an advantage when both are running the same speed, that same chip will still have the advantage at any speed, as long as they're both running at that speed. And since we all know that a 2.4 Ghz Q6600 is faster than a 2.4 Ghz Phenom, that's all you need to know, to know which will be faster at any speed.

 

[Triskaine]

You now, there is that little thing called ''clock scaling'' which makes your point moot as you can't just assume that performance scales the same way all the way to 3.4 GHz and as Sylvanas said, you have to factor in the other components of the processor then you look at a Phenom. It wouldn't surprise me if the Phenom is actually faster than a Kentsfield at that speed, because it can potentially scale better than Core 2 because of it's IMC and Hypertransport. The only way to know for sure is to test it. So please no fast assumptions.

 

[BTRY B 529th FA BN]

Ahhh, you know what fooled me without realizing they are at different speeds, is that they are marketed similar. First of all, quads, one true quad and one dual core quad, both similar prices. Then the benchmarks comparing them together, ...forgot it was a 2.6 vs a 2.4 ...first distraction though was that they are both able to hit 3.4Ghz...

 

[jones377]

Originally posted by: Triskaine
You now, there is that little thing called ''clock scaling'' which makes your point moot as you can't just assume that performance scales the same way all the way to 3.4 GHz and as Sylvanas said, you have to factor in the other components of the processor then you look at a Phenom. It wouldn't surprise me if the Phenom is actually faster than a Kentsfield at that speed, because it can potentially scale better than Core 2 because of it's IMC and Hypertransport. The only way to know for sure is to test it. So please no fast assumptions.

How do you think the Phenom scales from 2.4GHz to 3.4GHz with the L3/IMC running at 2GHz (or is it 1.8GHz?) in both cases? This is assuming the 9950 is overclocked by changing the unlocked multiplier.

Contrast this with the Q6600 which is overclocked using a higher FSB speed.

 

[SlowSpyder]

I know that the Intel stuff is very popular right now, so AMD's processors really don't get as much exposure. So a lot of people I don't think realize how decent the Phenom can perform when the IMC/L3 is overclocked. From what I saw when the IMC/L3 reaches the 2.4GHz range the Phenom seems to be quite close clock for clock to the Q6600.

But, because the Phenom BE's have an unlocked multiplier most reviewers simply use that to get their overclock and call it a day.

I think AMD's biggest problem isn't the performance of the chip, more is certaily better, the Phenom is still plenty fast for the vast majority of users. It's biggest problem is power draw for the performance you get. Hopefully 45nm will be a big help with that.

At the end of the day though, even though I had to buy a $150 motherboard and I get less overall performance then a Q6600, I'd have to say I'd buy this Phenom again today after having it for a while now. It easily handles anything I throw at it. My power hungry 2900Pro @ 800MHz and my power hungry Phenom @ 2.8GHz seem to easily run on my little (by a lot of people's standards here) 550 watt power supply.

Anyway, I'd say the 9950 will probably be a very nice upgrade for people who have an existing AM2 board that can handle it's power demands. Obviously for most people that are building a system from the ground up, Intel has some very compelling options.

 

[jones377]

Originally posted by: SlowSpyder
I know that the Intel stuff is very popular right now, so AMD's processors really don't get as much exposure. So a lot of people I don't think realize how decent the Phenom can perform when the IMC/L3 is overclocked. From what I saw when the IMC/L3 reaches the 2.4GHz range the Phenom seems to be quite close clock for clock to the Q6600.

But, because the Phenom BE's have an unlocked multiplier most reviewers simply use that to get their overclock and call it a day.

I think AMD's biggest problem isn't the performance of the chip, more is certaily better, the Phenom is still plenty fast for the vast majority of users. It's biggest problem is power draw for the performance you get. Hopefully 45nm will be a big help with that.

At the end of the day though, even though I had to buy a $150 motherboard and I get less overall performance then a Q6600, I'd have to say I'd buy this Phenom again today after having it for a while now. It easily handles anything I throw at it. My power hungry 2900Pro @ 800MHz and my power hungry Phenom @ 2.8GHz seem to easily run on my little (by a lot of people's standards here) 550 watt power supply.

Anyway, I'd say the 9950 will probably be a very nice upgrade for people who have an existing AM2 board that can handle it's power demands. Obviously for most people that are building a system from the ground up, Intel has some very compelling options.

Some people said the exact same thing about Prescott back in the days....

 

[SlowSpyder]

Originally posted by: jones377

Originally posted by: SlowSpyder
I know that the Intel stuff is very popular right now, so AMD's processors really don't get as much exposure. So a lot of people I don't think realize how decent the Phenom can perform when the IMC/L3 is overclocked. From what I saw when the IMC/L3 reaches the 2.4GHz range the Phenom seems to be quite close clock for clock to the Q6600.

But, because the Phenom BE's have an unlocked multiplier most reviewers simply use that to get their overclock and call it a day.

I think AMD's biggest problem isn't the performance of the chip, more is certaily better, the Phenom is still plenty fast for the vast majority of users. It's biggest problem is power draw for the performance you get. Hopefully 45nm will be a big help with that.

At the end of the day though, even though I had to buy a $150 motherboard and I get less overall performance then a Q6600, I'd have to say I'd buy this Phenom again today after having it for a while now. It easily handles anything I throw at it. My power hungry 2900Pro @ 800MHz and my power hungry Phenom @ 2.8GHz seem to easily run on my little (by a lot of people's standards here) 550 watt power supply.

Anyway, I'd say the 9950 will probably be a very nice upgrade for people who have an existing AM2 board that can handle it's power demands. Obviously for most people that are building a system from the ground up, Intel has some very compelling options.

Some people said the exact same thing about Prescott back in the days....

Hehe... well, I keep hearing that Phenom is the new Netburst. Seriously though, I would like to see a review where the reviewer pushes the IMC/L3 up... not sure if this has been done or not, but I haven't really seen one on the BE chips.

 

[Triskaine]

Originally posted by: jones377

Originally posted by: Triskaine
You now, there is that little thing called ''clock scaling'' which makes your point moot as you can't just assume that performance scales the same way all the way to 3.4 GHz and as Sylvanas said, you have to factor in the other components of the processor then you look at a Phenom. It wouldn't surprise me if the Phenom is actually faster than a Kentsfield at that speed, because it can potentially scale better than Core 2 because of it's IMC and Hypertransport. The only way to know for sure is to test it. So please no fast assumptions.

How do you think the Phenom scales from 2.4GHz to 3.4GHz with the L3/IMC running at 2GHz (or is it 1.8GHz?) in both cases? This is assuming the 9950 is overclocked by changing the unlocked multiplier.

Contrast this with the Q6600 which is overclocked using a higher FSB speed.

It's hard to draw any clear conclusions, because there aren't any reviews or tests which really occupie themselves with the Phenom. Most of them just test everything at stock without any tweaks and then conclude: Phenom=Crap.I know, at total stock settings the Phenom consistently lags behind Intel's Quad's, but I'm not getting tired of repeating that the Phenom can be a diamond in the rough, if you tweak it the right way, and that is without even overclocking the cores themself.
Back to the scaling issue. The performance gains through a higher FSB aren't much to write home about, they are more like nonexistant. You can get a greater benefit through stuff like the TRD for example. But I'm still sure that the Phenom has the potential to outscale a Core 2 Quad, especially through increasing the NB/L3/IMC clock domain. What we really need is a major hardware site which occupies itself with that issue and brings out a good Phenom tweak article. It's just a shame that the Phenom is ignored and looked at like the greatest processor turd since Prescott, which it can be sometimes but it doesn't have to. Really guys, give him a chance at least those of you that aren't out for maximum performance.

 

[Idontcare]

Originally posted by: SlowSpyder
Hehe... well, I keep hearing that Phenom is the new Netburst. Seriously though, I would like to see a review where the reviewer pushes the IMC/L3 up... not sure if this has been done or not, but I haven't really seen one on the BE chips.

It may be a fatally flawed analogy but I always figured the Phenom/Conroe -> Deneb/Yorkfield -> Deneb/Nehalem situation would be similiar to the Willamete/Northwood/Prescott situation where Willamete sucked balls to AMD at the time (lacked clockspeed), then Northwood hit a pretty good balance between clockspeed and IPC and Intel did alright, then Prescott fell on its face while the X2 took off.

I'm thinking Deneb will actually do well (45nm and revamped K10 design) against Yorkfield, but Nehalem will be a Deneb ASP destroyer.

Regarding everyone's laments about Phenom not getting a decent working over by the enthusiast sites...these are not non-profit sites people. Someone pays the bills. How many AMD commercials you see on TV these days? Want to guess what kind of an advertising budget Intel has versus AMD for "endearing" the enthusiast sites to spend extra time with their Phenom reviews?

On the flipside the fault first and foremost lies squarely on AMD. Just because no credible enthusiast review sites want to delve into a decent review of Phenom does not mean AMD can't do it for them. The fact that even AMD cares so little about ensuring Phenom users have a good tweak guide in their hands that they won't bother having even one PR person pull such a tweak guide together is the problem.

AMD's silence is deafening.

Don't blame the for-profit reviewers for doing what is in their best interests.

Blame the for-profit CPU manufacturer AMD for not doing what is in their best interests.

 

[zsdersw]

Originally posted by: BTRY B 529th FA BN
First of all, quads, one true quad and one dual core quad,

They're both "true" quad-core CPUs.. in that they both have 4 cores in one package. If you're going to talk correctly about the differences in how they get to 4 cores in one package, then you use terms like "native" and/or "MCM".

 

[BTRY B 529th FA BN]

Originally posted by: zsdersw

Originally posted by: BTRY B 529th FA BN
First of all, quads, one true quad and one dual core quad,

They're both "true" quad-core CPUs.. in that they both have 4 cores in one package. If you're going to talk correctly about the differences in how they get to 4 cores in one package, then you use terms like "native" and/or "MCM".

MCM?

then again you are breaking it down as "in that they both have 4 cores" if a car has dual exhaust coming out the back, the pipe from front to back may be one pipe, if it has two pipes all the way back, its called True two. AMD has a monolithic true 4-in-one-core design where Intel has 2 dual cores side by side from 2 cores and if we are using the word QUAD CORE we are only using a singular word for core. No one suspects the CORE from being apart *not A Part* of a totally different idea that the word QUAD CORE implies.

 

[zsdersw]

Multi-chip module.

 

[Lonyo]

Originally posted by: BTRY B 529th FA BN

Originally posted by: zsdersw

Originally posted by: BTRY B 529th FA BN
First of all, quads, one true quad and one dual core quad,

They're both "true" quad-core CPUs.. in that they both have 4 cores in one package. If you're going to talk correctly about the differences in how they get to 4 cores in one package, then you use terms like "native" and/or "MCM".

MCM?

then again you are breaking it down as "in that they both have 4 cores" if a car has dual exhaust coming out the back, the pipe from front to back may be one pipe, if it has two pipes all the way back, its called True two. AMD has a monolithic true 4-in-one-core design where Intel has 2 dual cores side by side from 2 cores and if we are using the word QUAD CORE we are only using a singular word for core. No one suspects the CORE from being apart *not A Part* of a totally different idea that the word QUAD CORE implies.

Does it even matter?
If it's so bad to use two smaller dies on a single package to make a quad core, why (IIRC) are AMD planning to use two quad cores on a single package to make their octo core chips?
All this crap about "true" quad core vs using dual dual cores is marketing crap from AMD which has no real relevance other than whether/how it impacts performance.

The empirical evidence suggests that while AMD's solution might be monolithic, it's also pretty moronlithic because it's not as good as Intel's "hack job".

Posted from a system running an AMD Athlon X2 4200+ with Radeon HD4850 graphics card.

 

[BTRY B 529th FA BN]

Originally posted by: Lonyo

Originally posted by: BTRY B 529th FA BN

Originally posted by: zsdersw

Originally posted by: BTRY B 529th FA BN
First of all, quads, one true quad and one dual core quad,

They're both "true" quad-core CPUs.. in that they both have 4 cores in one package. If you're going to talk correctly about the differences in how they get to 4 cores in one package, then you use terms like "native" and/or "MCM".

MCM?

then again you are breaking it down as "in that they both have 4 cores" if a car has dual exhaust coming out the back, the pipe from front to back may be one pipe, if it has two pipes all the way back, its called True two. AMD has a monolithic true 4-in-one-core design where Intel has 2 dual cores side by side from 2 cores and if we are using the word QUAD CORE we are only using a singular word for core. No one suspects the CORE from being apart *not A Part* of a totally different idea that the word QUAD CORE implies.

Does it even matter?

Opps.

To me? In the end, no.

It began from a quote from a sequence string over how I ended up getting confused by the review. It was no, 'this is better than that..."

 

[zsdersw]

Originally posted by: BTRY B 529th FA BN
then again you are breaking it down as "in that they both have 4 cores" if a car has dual exhaust coming out the back, the pipe from front to back may be one pipe, if it has two pipes all the way back, its called True two. AMD has a monolithic true 4-in-one-core design where Intel has 2 dual cores side by side from 2 cores and if we are using the word QUAD CORE we are only using a singular word for core. No one suspects the CORE from being apart *not A Part* of a totally different idea that the word QUAD CORE implies.

CPUs are not car exhaust pipes. There is no such thing as a "false quad-core", so there also cannot be a "true quad-core". A CPU is either quad-core or it isn't. If it has 4 cores in one package, it's a quad-core CPU. If not, it's not. If you're going to talk about the differences between two quad-core CPUs then you use a term like "native" or "monolithic".. not "true".