The Spin off Smoothness Thread

[keithlm]

Ripped out of the Review Thread and put into it's own.

I dont see how this will prove anything, if you run max threads of prime on a quad core and then start a game every system will output less fps, it doesn't matter if its intel or amd or any other cpu.

I guess you missed the part where I mention adding different things that would tie up various resources until you could determine where the CPU's actually had a difference? Your limited response only addresses one of those resources.

It is true that the i7 would perform better than the older Intel chips and be smoother. BUT it is not true that it would automagically be better because of hyperthreading. The jury is still out on whether hyperthreading is actually useful or detrimental in heavily stressed systems.

A closed mind, or thinking you already know everything there is to know, is never a good place to be. There is not a single person on this site that knows everything there is to know about computer platforms.

That is why I find it amusing when someone jumps in and screams "NO... this can't be real. You can't test for this. Closed case. Everybody ignore this person. <fingers in ears> na-na-na-na-na"

Perhaps it might be difficult to find a test that will allow for this. But if someone just closes their mind and won't even concede that there might be something that could be tested they are not showing a modicum of knowledge on the subject.

i said about a year ago about this whole phenom smoother thing that it had to do with the fast hyper transport speeds and the fact that the cpu has such low latency and it can just whip things out lightning speed. when you have things like super pi or something that sits in the cpu cache and doesn't really exit the cpu that often intel will get better performance. but with the way amd has designed their platform and made the connection speeds so fast it just seems obvious that it would be smoother.

I think you hit he nail on the head. I think eventually this topic will find its way into Intel forums as a discussion about i7 versus C2Q. However the C2Q fans might just accept it when it pertains to the i7 because they are part of the same "fraternity".

I'm switching from 940 to am2 to 775 from time to time, and the most "smooth" setup of those 3 is the am2 setup. Not too much difference, though, and I have a hard time proving anything, but that's just my .02.

Actually that is what initially made me a believer. Initially I didn't consider it possible. Then I went from my 939 to an AM2+ board preparing to get a Phenom. For testing and bios updates I ended up getting an X2 CPU that was the same speed as my old socket 939 CPU. What a difference. I would guess that you couldn't measure that difference with any available benchmark. (I often run Prime95, everest stress, and the NVidia Cascades demo to heat up and stress any new system. On the old socket 939 it worked. On the AM2+ system with the "same" CPU it was dramatically... wait for it... SMOOTHER. (And now I can't use the NVidia demo. Dang it. Although Froblins might someday work for that purpose if they get it working with Crossfire.)

My decades of experience working on various projects such as doing benchmarks on a Cray (SGI) computer or Sun Starfire system made me stop and think: "Hey... there must be a way we could create a repeatable test."

But to do a comparative benchmark and find the difference you must have both types of systems. I have not really put a lot of additional effort into this idea because I don't have an Intel system at home AND the fact that even if/when I came up with a repeatable method of measuring this aspect... you would STILL have closed minded people on forums claiming it doesn't mean anything.

[_Lone_Wolf_]

Taking smoother to mean more stable and consistant frame rates, if fraps or an equvlilent application can catalogue a history of frame render times to a resolution of one thousandths of a second I don't understand how this phenomena (no pun intended) can't be quantified as it's proponents tend to claim it can't. As soon as it becomes a question of faith I'm sceptical. I've tried to question people about this on several forums and haven't be given a satisfactory answer or demonstration posted to youtube.

[gallag]

What I find funny is that for some time we have had to listen about how cpu performance was not all that important and so what if you had to wait a few minutes extra on an encoding job or a few extra raring a file and a few extra fps is also irrelevant BUT apparently a non quantifiable difference off a few milliseconds opining a program is of supreme importance lol

+1 for placebo affect and a p4 1.8ghz with a new hdd and fresh windows install will feel faster than a 3ghzcore2 with a 2year old install

[TL1000S]

What I find funny is that for some time we have had to listen about how cpu performance was not all that important and so what if you had to wait a few minutes extra on an encoding job or a few extra raring a file and a few extra fps is also irrelevant BUT apparently a non quantifiable difference off a few milliseconds opining a program is of supreme importance lol

+1 for placebo affect and a p4 1.8ghz with a new hdd and fresh windows install will feel faster than a 3ghzcore2 with a 2year old install

Agree.

There might be a "techie" explanation for "smoothness" other than fresh install/enough RAM (also GPUMem!).

One should think AMD with the IMC (prior to ci7/Nehalem) should have an advantage when the OS/Software "calls" intstructions for memory-IO from/to RAM/CPU.
Another important aspect (apart from *drivers*) is chipset/PCI-E bus and any differences in hw&sw implementation for accessing/passing data from/to pheripals (like graphics cards/HD's etc).

Personally I have not seen any differences between my AMD/Intel rigs (about 50/50 during the years).
What really make a difference for 3D is the amount of GPUMem where large textures have to be "swapped" in/out.

[roofsniper]

there are other factors that can affect smoothness as well but you can't blame them every single time for why something is not smooth. most of the time when people are doing this testing they are using the same ram and the same hdds. it would be nice if someone could find some method of testing this if anyone has any ideas.

[gosh]

What I find funny is that for some time we have had to listen about how cpu performance was not all that important and so what if you had to wait a few minutes extra on an encoding job or a few extra raring a file and a few extra fps is also irrelevant BUT apparently a non quantifiable difference off a few milliseconds opining a program is of supreme importance lol

I am a very heavy multitasker and i can tell you that it is annoying when the computer isn't smooth. I don't have much user experiance of using a Intel Quad so can't say anything there but C2D I have a lot of experience of. I bought a pretty fast laptop and was planning to work on that, because of the uneven behaiviour it isn't used much more than to surf on.
Starting applications isn't a big problem though. The problems is that the mouse can stop, if you are a used to work with the computer you hardly need to look at the mouse moving to other locations, but if the mouse suddenly make a tiny stop (it is very small but enough) you miss the target. Opening popup menus and you expect the menu to show immediately. But it there are some other work in progress on the C2D it doesn't. Typing in some advanced editor and sometimes things go very fast and suddenly you need to wait, not much but enough that the workflow breaks.

[rk7p5]

That's why i prefer amd. All the intel rigs i used before i built my first amd system had that annoying lag when i would move to anything different. Intel may have all out speed but amd is smooth and steady imo.

[Glow9]

The irony being that when phenoms were hitting the shelves they couldn't OC as high and be stable in Vista so at a slower speed they were some how smoother. I think not. I switch back from AMD to Intel and can tell you a p133 can install and run XP so if that tells you anything I don't thing smoothness is much to brag about in XP which for a long time and now continues to be widely used. Smoothness is just a justification for going with AMD, more of a placebo than anything. It's like being proud you bought a Mustang cause you always wanted one but couldn't afford a V8. So when someone asks you why you went with the V6 you tell e'm well I wanted to save on mileage.
I have a lot of hope for Phenom IIs I think the 2mb of cache was really holding them back I don't care how you sell it 2mb for a Quad is pretty pathetic when you want something highend from AMD. Soon as I get the cash and MCP82 boards come out I'm pickin one up I dunno about you guys but 8mb of cache and a quad from AMD I'm down for gaming, and it's not going to based on smoothness.

[gosh]

Smoothness is just a justification for going with AMD, more of a placebo than anything.

And the "placebo" is the defense for Intel users justifying the higher price and that theirs CPU therefore needs to be better
As I said before, I don't have that much experiance using a C2Q but you need to be blind if you don't notice the differences comparing a C2D (no OC) and A Phenom when some multitasking is in the works. Don't know how the C2D feels if you overclock it.
if I run performance tests on the C2D and the Phenom the C2D wins most of the time.

Before I bought the phenom for the work computer I had Opteron 165. It IS slower than the C2D, you feel that it is much slower but it is also smooth. And that makes it feel pleasant to work with even if tasks takes a bit longer to complete.

I think the main difference is that Intel has gone totally for speed. They have worked on their already fast areas and made those faster. The weak parts is still weak though.
Amd has done the opposite. They have worked on their weak parts and made those faster. The processor doesn't have any weak areas. But it doesn't have any superfast areas either. The CPU is evenly built.

[Glow9]

And the "placebo" is the defense for Intel users justifying the higher price and that theirs CPU therefore needs to be better
As I said before, I don't have that much experiance using a C2Q but you need to be blind if you don't notice the differences comparing a C2D (no OC) and A Phenom when some multitasking is in the works. Don't know how the C2D feels if you overclock it.
if I run performance tests on the C2D and the Phenom the C2D wins most of the time.

Before I bought the phenom for the work computer I had Opteron 165. It IS slower than the C2D, you feel that it is much slower but it is also smooth. And that makes it feel pleasant to work with even if tasks takes a bit longer to complete.

I think the main difference is that Intel has gone totally for speed. They have worked on their already fast areas and made those faster. The weak parts is still weak though.
Amd has done the opposite. They have worked on their weak parts and made those faster. The processor doesn't have any weak areas. But it doesn't have any superfast areas either. The CPU is evenly built.

So your saying in performance tests you'll see a C2D win but in a 939 you'll feel better smoothness. Hrmm yeah I can't say I've seen too many people go from a 939 to a Phenom or C2D back to a 939 again. How are AMDs ratios compared to Intel, cause if AMD runs something inparticularly tighter that could have an effect.
So basing your performance on desktop smoothness I don't think is gonna do it for most people. Again another car analogy, it's like a car salesman being like well ya know that car over there has 300HP and grumbles and shakes a bit but this one hear well it's only 100hp but its smooth as silk when it idles. You want the best performance out of your processor generally meaning the fastest so its prolly going to be obvious the one you'd want to buy even if it is a bit rough around the edges

[gosh]

So your saying in performance tests you'll see a C2D win but in a 939 you'll feel better smoothness.

No! I said that even if the Opeteron 165 is slow it isn't that problematic because it is smooth. I wouldn't switch back from C2D but the smoothness removes some of the slowness.
Advantages for the C2D is enough comparing Opteron 165 even if the CPU is schizophrenic.

The C2D wins most performance tests (they are mostly singlethreaded) against my phenom.

[rk7p5]

So your saying in performance tests you'll see a C2D win but in a 939 you'll feel better smoothness. Hrmm yeah I can't say I've seen too many people go from a 939 to a Phenom or C2D back to a 939 again. How are AMDs ratios compared to Intel, cause if AMD runs something inparticularly tighter that could have an effect.
So basing your performance on desktop smoothness I don't think is gonna do it for most people. Again another car analogy, it's like a car salesman being like well ya know that car over there has 300HP and grumbles and shakes a bit but this one hear well it's only 100hp but its smooth as silk when it idles. You want the best performance out of your processor generally meaning the fastest so its prolly going to be obvious the one you'd want to buy even if it is a bit rough around the edges

Then the amd car has 275hp but it's transmission is smoother and quicker shifting

[roofsniper]

Smoothness is just a justification for going with AMD, more of a placebo than anything.

you rarely see many people pull the smoothness card so i don't see how that is a justification. plus what do we have to prove? we buy our cpus because we do and even if some people buy amd because they can't afford intel it doesn't mean they are going to try to hide it by saying o it was the better choice. plus as i have said before the only people that can say it was smooth were the ones that have had both. so obviously they bought amd over intel for a reason or they were completely neutral and they have no side to stick to and just think the amd is smoother. the problem is that many people just refuse to believe this is true and they have no proof to back it up. they will say something like o well the fps is higher so its smoother. hell what if an intel cpu can pump out 20 frames in half a second and then there is a hitch and it pauses up for a little while and then another 15 frames come out in the last half of the second. so that would be 35 fps. then lets say that the amd has 30 frames throughout the entire second without a hitch or anything. sure the intel system has higher fps but does that mean it is smoother? if you like car analogies lets say you are driving along at 60 miles an hour and you slam on the brakes every 10 seconds and then speed back up again. then lets say another car is going 40 miles an hour without slowing down or speeding up. the car going 40 miles an hour will have a smoother ride even tho the car going 60 will most likely make it farther. the same can be said for fps. intel might get more fps but it won't do it at a smooth rate so it doesn't matter how many fps you can get if its not smooth.

[demonkevy666]

So your saying in performance tests you'll see a C2D win but in a 939 you'll feel better smoothness. Hrmm yeah I can't say I've seen too many people go from a 939 to a Phenom or C2D back to a 939 again. How are AMDs ratios compared to Intel, cause if AMD runs something inparticularly tighter that could have an effect.
So basing your performance on desktop smoothness I don't think is gonna do it for most people. Again another car analogy, it's like a car salesman being like well ya know that car over there has 300HP and grumbles and shakes a bit but this one hear well it's only 100hp but its smooth as silk when it idles. You want the best performance out of your processor generally meaning the fastest so its prolly going to be obvious the one you'd want to buy even if it is a bit rough around the edges

a car analogy could like this

V8 low RPM Horse power, 4 cylinder high RPM same power
the torque lines are much shorted on the 4 cylinder the v8 start very low
V8 can curse at 1,5000 rpm. 4 cylinder hit 60mph at 3,000 rpm (sometimes lower with overdrive on both).

back on topic

phenom got a bad wrap thanks to tom's and anand for fail to to see. phenom clocking isn't good with HTT. cooling will need to sufficient I don't think they've ever used water on one yet they will on core 2's.

the only positive things I've ever read from them about is it's 15% faster then K8.
phenom is only lacking SSE pre-core compered to core2.

[rk7p5]

I picked amd because at the time they just came out with the brisbane and it was like 80 bucks, paired that with a 75 dollar dfi mobo and clocked the x2 to 3ghz and i was very happy. My buddy built a e6400 and bx2 rig around the same time, he spent quite a bit more but he could afford to. Sure his rig is faster but he kept asking me how can he make his rig as smooth as mine.

[Glow9]

plus as i have said before the only people that can say it was smooth were the ones that have had both. so obviously they bought amd over intel for a reason or they were completely neutral and they have no side to stick to and just think the amd is smoother. the problem is that many people just refuse to believe this is true and they have no proof to back it up. they will say something like o well the fps is higher so its smoother.

See it's impossible to know really, look how bogged down this side of the forum is with AMD fanboys. You can scream at me till your blue in the face that AMD is smoother but look at video card reviews, people want to see cards running at their max well obviously they aren't picking AMD as of late.

lets say that the amd has 30 frames throughout the entire second without a hitch or anything. sure the intel system has higher fps but does that mean it is smoother? if you like car analogies lets say you are driving along at 60 miles an hour and you slam on the brakes every 10 seconds and then speed back up again. then lets say another car is going 40 miles an hour without slowing down or speeding up. the car going 40 miles an hour will have a smoother ride even tho the car going 60 will most likely make it farther. the same can be said for fps. intel might get more fps but it won't do it at a smooth rate so it doesn't matter how many fps you can get if its not smooth.

Well I'd say it wouldn't be too hard to find out really. You see a lot of video card reviews where AMD vs. Nvidia and you have the low average and max fps per test. The thing is if you were to do these tests with AMD and then with a Intel chip you'd see which ran the most consistant. The thing is it's really not fair running a 9950 against Intels high end cause I'm sorry theres no way in hell it can compete with a Q9950 or higher, it would be fair running high end vs. high end. But even so its hard all you could do really is clock a older Intel Quad against a Phenom at the same clock speed. It would be interesting to look for but this review of 100 cpus does a good job http://www.matbe.com/articles/lire/1...urs/page29.php
I guess a Q8200 is pretty much on par with a 9950 in games so cant beat low end Intel in gaming.
Microsoft Exel AMD loses again so if you want to talk about desktop smoothness wouldn't you see it in this?
http://www.matbe.com/articles/lire/1...urs/page13.php
And again in photoshop another proggie AMDs highend loses out to a Q8200
http://www.matbe.com/articles/lire/1...urs/page14.php
Multithreaded WAV>MP3 again same thing http://www.matbe.com/articles/lire/1...urs/page23.php

So I'm not saying AMD is bad, or can't have potential for running smoother based on whatever architecture they use but don't you think even in software that isn't gaming you would see them come out ahead? Being more efficient would also equal being faster in terms of processor calculations would it not.

At heart I'm a AMD guy but don't shoot the messenger when people are still in some sort of AM2+ denial. I have hated AM2 ever since they jumped from 939 so I can't say I'm really sad to see it come up with nothing. I'm jumping right back to AMD as soon as I can but I think it would just be rediculous to not acknowledge Intel winning this round, not to mention with obvious wins in desktop based programs and very commonly used programs as well AMD just does not take the lead. Personally I think more people should be worried about Price:Performance rather than what color the box is.

[keithlm]

It is interesting how some people just jump in and immediately dismiss the concept. They haven't seen it so it does not exist. End of story. (Or they dismiss it because they believe it is somehow a tactic used to justify something and not something that many people have experienced.)

Yet the topic is raised by a large number of different people on a lot of different forums. That fact alone means that it would be foolish to dismiss the concept without further investigation.

The people that do not question this concept are the ones that have experienced it. They do not question that such a thing exists. They only question HOW MUCH of it exists.

This topic came up about half a year ago. The same thing happened. People split into a group of people that just want to ridicule it and pretend it doesn't exist post and a group of a few people that post about various experiences they have had.

The discussion was derailed by the detractors. Apparently they have some reason that makes them really not want people to discuss this issue; they often use ridiculous arguments to get people to stop talking about it. I'm getting tired of that tactic. If you don't believe it exists... that is your "right". But if that is the case you have no reason to post anything in this thread. Period. QED.

I think it would be nice to actually discuss methods that might be used to measure such a system attribute and stop all the juvenile bickering.

[Glow9]

It is interesting how some people just jump in and immediately dismiss the concept. They haven't seen it so it does not exist. End of story. (Or they dismiss it because they believe it is somehow a tactic used to justify something and not something that many people have experienced.)

Yet the topic is raised by a large number of different people on a lot of different forums. That fact alone means that it would be foolish to dismiss the concept without further investigation.

The people that do not question this concept are the ones that have experienced it. They do not question that such a thing exists. They only question HOW MUCH of it exists.

This topic came up about half a year ago. The same thing happened. People split into a group of people that just want to ridicule it and pretend it doesn't exist post and a group of a few people that post about various experiences they have had.

The discussion was derailed by the detractors. Apparently they have some reason that makes them really not want people to discuss this issue; they often use ridiculous arguments to get people to stop talking about it. I'm getting tired of that tactic. If you don't believe it exists... that is your "right". But if that is the case you have no reason to post anything in this thread. Period. QED.

I think it would be nice to actually discuss methods that might be used to measure such a system attribute and stop all the juvenile bickering.

Alright I'll admit I like how my 939 handled better than my friends P4 but at the same time show me adequte proof. I'm an AMD guy but I'm gonna force a debate out of you guys with logical reasoning. Cause if its all just a visual thing than theres other things to take into account to. But seems very odd that the scores show 1 winner but another group seems to thing otherwise. Real tests are not showing in AMDs favor here you see what I'm saying. The only thing juvenile is people getting up in arms because people are asking for proof and the best people can offer is well its there cause I've seen it. uh huh... This usually comes from fanboys who don't have both rigs. I'd say invite some prominent extreme members with both systems and ask them to compare. The thing is if you have ever been in university is that you learn time and time again all tests and testers are always biased or have one bias towards something. So thats why these tests are good but the thing is the results don't really seem to backup the argument so far.

Get 2 systems running next to eachother and video tape them and maybe we'll have something.

[roofsniper]

See it's impossible to know really, look how bogged down this side of the forum is with AMD fanboys. You can scream at me till your blue in the face that AMD is smoother but look at video card reviews, people want to see cards running at their max well obviously they aren't picking AMD as of late.

there are many intel fanboys as well but that has nothing to do with the topic. and yes people use intel in video card reviews. why? because in video card reviews you make graphs to show fps. if intel gives more fps then the graph will show that the card has a higher fps. video card reviews have nothing to do with the smoothness of a cpu. most people are trying to show that the latest card has the highest fps as possible so people will buy them.

Well I'd say it wouldn't be too hard to find out really. You see a lot of video card reviews where AMD vs. Nvidia and you have the low average and max fps per test. The thing is if you were to do these tests with AMD and then with a Intel chip you'd see which ran the most consistant. The thing is it's really not fair running a 9950 against Intels high end cause I'm sorry theres no way in hell it can compete with a Q9950 or higher, it would be fair running high end vs. high end. But even so its hard all you could do really is clock a older Intel Quad against a Phenom at the same clock speed. It would be interesting to look for but this review of 100 cpus does a good job http://www.matbe.com/articles/lire/1...urs/page29.php
I guess a Q8200 is pretty much on par with a 9950 in games so cant beat low end Intel in gaming.
Microsoft Exel AMD loses again so if you want to talk about desktop smoothness wouldn't you see it in this?
http://www.matbe.com/articles/lire/1...urs/page13.php
And again in photoshop another proggie AMDs highend loses out to a Q8200
http://www.matbe.com/articles/lire/1...urs/page14.php
Multithreaded WAV>MP3 again same thing http://www.matbe.com/articles/lire/1...urs/page23.php

So I'm not saying AMD is bad, or can't have potential for running smoother based on whatever architecture they use but don't you think even in software that isn't gaming you would see them come out ahead? Being more efficient would also equal being faster in terms of processor calculations would it not.

At heart I'm a AMD guy but don't shoot the messenger when people are still in some sort of AM2+ denial. I have hated AM2 ever since they jumped from 939 so I can't say I'm really sad to see it come up with nothing. I'm jumping right back to AMD as soon as I can but I think it would just be rediculous to not acknowledge Intel winning this round, not to mention with obvious wins in desktop based programs and very commonly used programs as well AMD just does not take the lead. Personally I think more people should be worried about Price:Performance rather than what color the box is.

i think you are forgetting the point. i am talking about smoothness here. not how fast a task can finish that has absolutely nothing to do with how smooth it is. and i am not talking about min and max fps either. what does that have to do with it? lets say you have an amd getting a max of 50 fps and a min of 40 fps and an average of 45 fps. then you have an intel getting a max of 70 fps and a min of 45 fps and an average of 50 fps. how big of a change is the drop from 70 fps to 45 fps going to be? you will be able to tell thats for sure. even if the amd has a lower minimum fps and a lower maximum if it can keep the frames coming out at a more consistent pace and within a reasonable amount of the intel cpu it should be smoother. its not like its that easy of a test to find out which is smoother. the only info we have right now are neutral people saying that amd is smoother when they have tested them side by side and some people switching to amd because they think its smoother. unless you can have some info to disprove this i don't know what you are trying to say. just because one gets a higher ammount of fps doesn't mean it will be smoother. especially if the fps changes around by a dramatic amount. only way i can think of testing this is by taking fps measurements at many different intervals and comparing them. because by comparing max fps average fps and min fps there is no way to tell which one is smoother.

[demonkevy666]

only thing I want to know is why phenom gpu scores are higher then Intel cpu's in 3dmark06 ?
(in tony's thread using an 8800 utlra and clockign phenom to 3.2ghz and Q9650 to 3.2ghz)
cpu scores are lower for phenom sure but if intel cpu scores are higher should not the gpu scores be higher too?

[roofsniper]

only thing I want to know is why phenom gpu scores are higher then Intel cpu's in 3dmark06 ?
(in tony's thread using an 8800 utlra and clockign phenom to 3.2ghz and Q9650 to 3.2ghz)
cpu scores are lower for phenom sure but if intel cpu scores are higher should not the gpu scores be higher too?

i have wondered the same myself but im not exactly sure how it calculates the different scores. i know there are calculators for the overall score but what makes up the individual scores?

[rk7p5]

I'm not sure exactly why amd is smoother, maybe it's the imc or the ht system. I think Gosh is on to something when he says that the fsb is somehow causing the lag on intel systems. intel created i7 for a reason

If i could could afford it i would love to build an i7 rig, best of both worlds, speed and smoothness

[roofsniper]

I'm not sure exactly why amd is smoother, maybe it's the imc or the ht system. I think Gosh is on to something when he says that the fsb is somehow causing the lag on intel systems. intel created i7 for a reason

If i could could afford it i would love to build an i7 rig, best of both worlds, speed and smoothness

i have said before that i believe it is because of the fast I/O speeds of of the cpu. theres a quote of me in the first post. i just don't know how you can prove it but ill try to find a way.

[roofsniper]

http://www.xtremesystems.org/forums/...1&postcount=24
toppc covered up the fps of each individual test but it would be very interesting to see if i7 got higher fps on every gpu test and if amd got lower fps on every gpu test because then we know that 3d mark doesn't base the scores on fps alone and it might base them on something like smoothness.

edit: something i remember from awhile back: http://www.xtremesystems.org/forums/...+phenom&page=4
http://www.xtremesystems.org/forums/...7&postcount=98

[accord99]

i think you are forgetting the point. i am talking about smoothness here. not how fast a task can finish that has absolutely nothing to do with how smooth it is. and i am not talking about min and max fps either. what does that have to do with it? lets say you have an amd getting a max of 50 fps and a min of 40 fps and an average of 45 fps. then you have an intel getting a max of 70 fps and a min of 45 fps and an average of 50 fps. how big of a change is the drop from 70 fps to 45 fps going to be? you will be able to tell thats for sure. even if the amd has a lower minimum fps and a lower maximum if it can keep the frames coming out at a more consistent pace and within a reasonable amount of the intel cpu it should be smoother.

But this is not a case of AMD being smoother or more consistent, it is a case of AMD being slower and therefore unable to push to the same frame rates when the situations switches to be more CPU limiting. Using your criteria, a Phenom 9400 would be even smoother and consistent than a Phenom 9950 as the Phenom 9400, while reaching roughly the same minimum frame rate when GPU limited, would be unable to hit the same maximum frame rate when CPU limited.

only way i can think of testing this is by taking fps measurements at many different intervals and comparing them. because by comparing max fps average fps and min fps there is no way to tell which one is smoother.

JumpingJack has done some investigation into this matter using his own QX9650, Phenom 9850 and 4870X2 and posted results in this thread:

http://www.xtremesystems.org/Forums/...d.php?t=197423

Here are the direct links to some of the results that provide a frame rate plot over time:

http://www.xtremesystems.org/Forums/...&postcount=235
http://www.xtremesystems.org/Forums/...&postcount=285
http://www.xtremesystems.org/Forums/...&postcount=364
http://www.xtremesystems.org/Forums/...&postcount=391