My Gaming System is "Smoother" Than Yours: Weigh In Please

[Zucker2k]

As some of you already know, there is heated debate (call it a fracas) in the AMD section regarding an apparent phenomena plaguing Intel systems so that they are not as smooth as AMD systems in general. However, with the hostility and mob mentality in the AMD section towards people who are perceived to be anti-AMD, and to prevent some clueless person being dragged into one of these frequent fracases and and ending up warned or banned, I am creating this thread to allow a more objective debate; one in which everyone's opinion is tolerated and valued.

In the next few days, I plan to ran a series of tests showing what is going on under the hood (so to speak) when the graphics processing unit (gpu) starts interacting with the central processing unit (cpu) in various scenarios. I will add more as I receive requests. My hardware is in my sig.

So far, the most popular methodology proposed to solving this disease of sorts is to capture frames using fraps and then converting these frames linearly over a timeline in a graph. I question the viability of this method and propose that a frame capture alone is not the answer, we must also be able to show what kind of loads the gpu is putting on the cpu and the subsequent effect of this load on individual frames in realtime. See screenshot below of the above mentioned interaction in 2 synthetic benchmarks. I think this is interesting and should be factored into any broad conclusion regarding this matter. This of course complicates the process, but it is indispensable information to realizing objective results. The truth is, there are many factors one must take into account in testing for this particular phenomena. What is your opinion? Suggestions welcome.

[Movieman]

nah, mine is just faster! 8 cores at 3600/12mb/1600..
OOPS, I don't game!...:rofl:

[Zucker2k]

nah, mine is just faster! 8 cores at 3600/12mb/1600..
OOPS, I don't game!...:rofl:Dave, it's not about fast, or max frames :rolleyes: It's about SMOOOOOOOOTH :up:

[Asgard_thor]

nah, mine is just faster! 8 cores at 3600/12mb/1600..
OOPS, I don't game!...:rofl:

no match for my core 2!! :ROTF::ROTF:

[stevecs]

Not been part of the debate at all and not seeing the issue (or not recognizing it) can you briefly describe the "lack of smoothness" ie, what effects are exhibited? As for your system modeling that is a good start but when dealing with large data moves tracking memory I/O utilization would also be important and a possible congestion point in addition to your discrete components of cpu & gpu.

[Gilhooley]

Hmm, I don't follow your logic in measuring load levels. Smoothness in games = stable framerates. IMO the system with the highest % of grouped fps'es over time is smoother.

For example: System A have max 100fps and min 30fps, but only have a max 20% cluster around 60 fps. System B have max 70fps and min 25fps, but have a max 40% cluster around 45fps.

If you played the game with V-sync OFF, I would say that we would percive System B as Smoother. But with V-Sync on, it's different and also need to be tested.

[CmB]

I'm interested to see how this thread goes...crossing my fingers it doesn't turn ugly!

[Zucker2k]

Hmm, I don't follow your logic in measuring load levels. Smoothness in games = stable framerates. IMO the system with the highest % of grouped fps'es over time is smoother.

For example: System A have max 100fps and min 30fps, but only have a max 20% cluster around 60 fps. System B have max 70fps and min 25fps, but have a max 40% cluster around 45fps.

If you played the game with V-sync OFF, I would say that we would percive System B as Smoother. But with V-Sync on, it's different and also need to be tested.The logic is that an over-burdened component should exhibit or impact the framerate consistency; therefore frame-capture and hardware load levels if matched in real-time should be able to help quickly pinpoint if any of the components have an impact on an irregular frame rendering based on a particular position in the framecapture timeline where said irregular frame-rendering occured. If no such over-burdening occured, then we can look to other sources for a culprit. At this point in time, testing by elimination is the only way we can solve what is behind this phenomena. For now, we have to assume that io operations are not having any negative impact on the tests; of course care must be taken to rule out this occurence as much as possible by working with excellent hardware all around.

Race Driver: Grid. Vista x64 SP1.

Edit: Yes, I experienced 2 pauses, call it stuttering, I'm on an x2. Everest was also in the background, writing to a second hard disk. Game is installed on a third disk, OS on the first.

I run the game to skip the video intro, alt-tabbed into windows, started monitoring app, went back into the game and loaded a 3 lap race and completed it.

[B.E.E.F.]

There's way too much info on that graph.

You don't need to show the hard drive temperature.

Why do you have GPU1 and GPU2? They should be combined. Same with the CPU cores.

[Zucker2k]

There's way too much info on that graph.

You don't need to show the hard drive temperature.

Why do you have GPU1 and GPU2? They should be combined. Same with the CPU cores.The app I use does not combine x2 gpu cores.

[Xello]

If it's the graphics card that is drawing the frames, and not the cpu, and you're playing at a resolution where the cpu does not come into affect, what does the cpu have to do with 'smoothness' ?

[makaka]

If it's the graphics card that is drawing the frames, and not the cpu, and you're playing at a resolution where the cpu does not come into affect, what does the cpu have to do with 'smoothness' ?

totoally agree , when we talk game@ high resolution than the GPU is number 1 and 2 , the come the memory and the processor

[Zucker2k]

If it's the graphics card that is drawing the frames, and not the cpu, and you're playing at a resolution where the cpu does not come into affect, what does the cpu have to do with 'smoothness' ?The argument from the "other" side is that AMD has imc; but apparently Ci7, which also has imc, exhibits this behavior. :shrug: With regard to the other part of the question, since the gpu has to work in tandem with the cpu, a slight hiccup with either will result in this anomaly. Tests showing load levels, could help pinpoint where a potential hiccup may occur based on load data and then matching them to a fps capture showing sudden drops of frame (jerkiness). Of course the "cpu" in this instance would have to stand for not just the cpu, but fsb and nb as well; in effect, it's a platform argument.

[adamsleath]

i thought bottlenecks caused un-smoothness...??????

[XS Janus]

This is just... stu... silly I mean!
Threads on this subject should be under parental control. Cause they will cause kids participating in them to become quite dumber then they were before entering them...

[Zucker2k]

This is just... stu... silly I mean!
Threads on this subject should be under parental control. Cause they will cause kids participating in them to become quite dumber then they were before entering them...Choose your words carefully, some people build their machines based on these observed "smoothness" - it's a feeling thing and you certainly don't want to hurt other people's feelings before we arrive at a verifiable conclusion.

[STEvil]

This is just... stu... silly I mean!
Threads on this subject should be under parental control. Cause they will cause kids participating in them to become quite dumber then they were before entering them...

Yes, please do choose your words more carefully.

I would like this to be a rational and reasoned debate. There are infraction cookies and ban muffins behind the door that hits you in the ass when you leave.

[CmB]

Yes, please do choose your words more carefully.

I would like this to be a rational and reasoned debate. There are infraction cookies and ban muffins behind the door that hits you in the ass when you leave.

:rofl:That is funny! I too would would like a rareasonable debate. I don't think we should put someone down who's trying to get tot he bottom of a heated topic, even if the methodology is flawed (I don't think it's flawed and I honestly haven't looked at said methodology so i'm not saying anything either way).

Numbers and quanitative conclusions sure as heck beat the it's smoother because I say so circular argument...maybe we can say it's smoother because of "X" number of reasons based on repeatable data points or no it's not really smoother at all conclusively and then shut up about it. / rant thanks for listening:)

[Chad Boga]

As some of you already know, there is heated debate (call it a fracas) in the AMD section regarding an apparent phenomena plaguing Intel systems so that they are not as smooth as AMD systems in general. However, with the hostility and mob mentality in the AMD section towards people who are perceived to be anti-AMD, and to prevent some clueless person being dragged into one of these frequent fracases and and ending up warned or banned, I am creating this thread to allow a more objective debate; one in which everyone's opinion is tolerated and valued.
Very sensible course of action trying to have this conversation outside of "The Sanctuary" :clap:

[EliTE22]

"smoother":rofl:, higher FPS= smoother:up:

[EliTE22]

Very sensible course of action trying to have this conversation outside of "The Sanctuary" :clap:
:rofl: they only heard what they want and attack when someone diagrees:shakes:, they just want to be able to say amd is faster, not saying some intel fanboy go over aboard disagreein hint hint:D or think intel is better

[gOtVoltage]

Back on topic,,


Does Higher FSB = Higher smoother ness?

Or

Does Higher HTT = Higher smoother ness?


What ive seen from both sides ,,

FSB/HTT and a good CPU multiplyer play a critical role in Smooth factor,,

Despite how slow or fast a CPU is ,,it comes down to how fast or Smooth the FSB/HTT can talk to devices at the same time .


In all ohnesty,,
I dont care how fast a Pc is "if it Skips ,pauses or lags", in any way, its not smooth.

Ive seen Fast unsmooth pc's and they do give a jerkyness feeling when gaming.

There are so many variables to take in its not even funny.

I give Props to the Man above for (Opening a Big ol Can O worms) charting Smoothness of a PC. I know many people can tell if a Pc is un optimized or (unsmooth just by how it starts or loads apps(Even if it has a Low GHZ CPU.

Smoothness actually applies to any Pc. Its either gonna be smooth when optimized ( no matter the Speed )

or

its gonna be jerky/Laggy/SlowBuggy (no matter the Speed) if its unoptimized.

Being able to chart loss/gain of smoothness is actually helpfull.

Pc's are like performance Cars now a days...Even the Budget Pc's can be tweaked...Smoothness can come from tightening just the ram and or Cleaning you HD . But most of All Bumping up those GHz's/FSB's and HTT's:D

I may be a fan of something , but the Smoothness thing goes for any PC :yepp:

[loften]

the old shuttle in my sig plays all games smoother then the other computer in my sig. it might not be as fast or as full of eye candy but it plays them all much smoother.. i can't even tollerate the newer one. even after using it for a year.

[Soulburner]

the old shuttle in my sig plays all games smoother then the other computer in my sig. it might not be as fast or as full of eye candy but it plays them all much smoother.. i can't even tollerate the newer one. even after using it for a year.
Run them both on the same settings. The faster machine will be much smoother...otherwise any other comparison is invalid.

[B.E.E.F.]

The data from the charts. Can it be imported into excel for analysis?

The best way to quantify smoothness is through statistical analysis. Excel has such formulas.

[loften]

i have tried to run them at as close to the same settings as they will allow due to being different hardware years apart. it just feels like there is something not right with the way it runs at times in games. it screems through tests. in game it feels like it is super fast then random slow, glitchy then fast again. i am not a expert either so i really have no clue what causes the glitchy feeling i experience on it in game.. i know the old shuttle will run pretty much every game out pretty smooth.. i gave it to my son he installs all of them on it and plays... he took it to the intel fest and actually competed on it.. people were like what,, how is that old thing running all these games.. he did pretty well too. when he plays on mine he says this thing sux for how fast it is.. even after playing on it and getting used to it for a while he says it is laggy.. i think it is the video cards or monitor lag but i am not sure as it could be the way things are inside..

[Zucker2k]

Here is an update; the major limitation here is that the apps I'm using start logging at separate times and there's no timeline. But this should give us an idea about what is happening under the hood. It seems the gpu is taking a bashing and there were about 3 very brief pauses. Guys, what do you make of these:

5 laps at Instanbul Park.

[gOtVoltage]

Ok nice work... Based on the Graph @4.3ghz the cpu seems to have the 4870x2 waiting.


What would be the out come on this same chart if you Lower the FSB/ raise CPU one Multi .

Also can you raise the Ram Higher ,, freqency say 1066 / 1150//1333 ,, use what ever timing 5-/ -6/ or -7 latencies.

The reason Behind this i,,because two systems (being compared or even a single PC) may very well perform night and Day using differnt HT/FSB/NB although its at the Same CPU freqency.

This is where the Fun begins.


Try to use the same 4.3ghz cpu freqency but with lower higher FSB/NB/ Ram freqency .

Changing things around should or will show on CPU side of the Ghraph more than the GPU side. your GPU's look like thier waiting for info compard to the what the CPU is giving them.

So a little tweak her and a little tweak there may make things better /worse.

Since you have a base line it would be interesting to see how it performs with Lower and Higher Ram/ Cpu multi's compared to your 4.3ghz baseline.

This will show you what the System wants to perform ,,

Its not allways about" how high" the CPU goes , Its really "how good", you can tweak the Mobo to go with the CPU .

Interesting to see what you come up with:up:
Note: keep the 4870x2 @ stock clocks while testing.

gOtVolTage

[STEvil]

i have tried to run them at as close to the same settings as they will allow due to being different hardware years apart. it just feels like there is something not right with the way it runs at times in games. it screems through tests. in game it feels like it is super fast then random slow, glitchy then fast again. i am not a expert either so i really have no clue what causes the glitchy feeling i experience on it in game.. i know the old shuttle will run pretty much every game out pretty smooth.. i gave it to my son he installs all of them on it and plays... he took it to the intel fest and actually competed on it.. people were like what,, how is that old thing running all these games.. he did pretty well too. when he plays on mine he says this thing sux for how fast it is.. even after playing on it and getting used to it for a while he says it is laggy.. i think it is the video cards or monitor lag but i am not sure as it could be the way things are inside..

Microstutter. Common issue with 3870's and crossfire.

[XS Janus]

...my apologies.

[swale]

Will throw a few up here for intel/nv shortly LOL :)

[largon]

Frames per second has little to do with "smoothness".
Crysis at 25-30FPS is smooth.
Mirror's Edge at 25-30FPS is unplayably choppy.

Frames per 100 milliseconds might show any possible difference, though.

[Bradan]

Frames per second has little to do with "smoothness".
Crysis at 25-30FPS is smooth.
Mirror's Edge at 25-30FPS is unplayably choppy.

Frames per 100 milliseconds might show any possible difference, though.

I was actually saying this to my friend the other day.

1/10 or 1/100 avg/min would be good in this era

[demonkevy666]

Fanboyism isn't a club exclusive to AMD.


Anyway, I think this picture I made should explain things better:
http://apokalipse.googlepages.com/fps-time.PNG
the fps on the left is probably higher on average. but it's very rough, and you're probably not going to have fun playing through it.
The one on the right, however, is quite smooth, and highly playable.

kinda like wash boarding on the road like on myth busters, that was load of data on that show.

btw this off the AMD section.

[Zucker2k]

kinda like wash boarding on the road like on myth busters, that was load of data on that show.

btw this off the AMD section.Those are some wide fluctuations right there. Would be interesting to see some more data; like gpu and cpu load, frequencies, etc. See above.

[B.E.E.F.]

Until the data can be analyzed, you cannot compare systems. This isn't something you can eyeball.

[fiskov]

Been playing CS (NOT Source) now for 7-8 years.. been through quite a lot of PC's in my time.
Been mostly an AMD person myself, but from the Intel systems i've tried at both events/home it has always been a much smoother experience.
I can't really describe it other than Intel systems seem so much more natural.. The flow, its so perfect and consistant.

When your spraying the M4A1 on an AMD you've got this power but it feels rough, unreliable and choppy. But at the same time it's got so much power behind it, that it works well. Doing the same on an Intel system, you've got this consistant, smooth, reliable power allowing you to almost guide each bullet individually.. The main difference though is the instantness of a kill with an Intel system, people seem to die before you've even finished pushing the mouse button.

I've never understood how people have never noticed the differences in CS depending on the system, people say im crazy or spout, "As long as you've got 100fps its no different" Every PC i've ever owned can play it at 100fps, so i guess im crazy...

[WillC]

The GRID pictures posted above almost sounds like the problem I am having over at another thread: http://www.xtremesystems.org/forums/showthread.php?t=217018

Using one card feels choppy even though I average high framerates...

[Eastcoasthandle]

The GRID pictures posted above almost sounds like the problem I am having over at another thread: http://www.xtremesystems.org/forums/showthread.php?t=217018

Using one card feels choppy even though I average high framerates...

You may not have this problem but have you checked your refresh rates (http://www.xtremesystems.org/forums/showthread.php?t=217180) to make sure they are at the correct Hz?
I too noticed stutter at high frame rates (using 1 card though). And it got worst when I used fraps to record. Now I don't notice any difference between in game and frap recording, they are both very smooth.

[WillC]

You may not have this problem but have you checked your refresh rates (http://www.xtremesystems.org/forums/showthread.php?t=217180) to make sure they are at the correct Hz?
I too noticed stutter at high frame rates (using 1 card though). And it got worst when I used fraps to record. Now I don't notice any difference between in game and frap recording, they are both very smooth.

Gameplay feels pretty much the same either at 60 or 75 Hz. Only time I get completely smooth framerates is turning on vsync. Thanks for trying though :)

[Katanai]

:rofl:

I dunno man, from my experience Intel systems run smoother than AMD ones, not the other way around. Especially in other apps outside of gaming.

[FUGGER]

I have identified the source of the problem and it is an easy fix, running it by Intel first for clarification.

This was an exploited weak point on certain setups, not all.

[StAndrew]

Can the architecture of the busses on the moboard affect this? For example, your typical Nvidia moboard has a HyperTransport connection between your NB and SB. When info flows from your CPU (lets stick with FSB architecture for now) it goes through the FSB to the NB, which then splits to the primary/secondary PCI express slot and the SB. When it reaches the SB, it splits to the secondary/tertiary PCI Ex slots (of course the i780 boards and some X58 boards have the nforce 200 chip to also take into account, sending the 3D info through up to 3 diff chips and buss architectures). Im not too familiar with how the AMD boards are designed, but I dont think they incorporate PCI slots on the SB (I can be wrong) eliminating this medium the info has to pass, and I know for dam sure they don’t have the nforce 200. The info should flow a bit better between the proc and chipsets (at least on Nvidia boards) as they use almost the exact same HyperTransport direct connect interface.

The i7 is a diff story, however it still utilizes (some do at least) the dreaded nforce 200 chip. Its so hard to say AMD processors are better than Intels (or smoother, whatever the hell that means) when both run off of radically different platforms. The variables are just too erratic to come up with any conclusive evidence on the levels we are looking. Just keep it simple: What is the output? In this, I think its safe to assume that the Intel procs are superior in every way (except in smooth FPS - ??? - but thats a very minute argument for buying AMD over Intel IMHO). Lets go back to cutting logs and not worry about splitting hairs.

For those interested in getting to the bottom of this, try using a single GPU only to eliminate micro-stuttering, especially with Race Driver: Grid.

[Zucker2k]

I have identified the source of the problem and it is an easy fix, running it by Intel first for clarification.

This was an exploited weak point on certain setups, not all.Hmmm, interesting; I have some questions, but I'm sure you're not allowed to say anything yet?

[Musho]

I have identified the source of the problem and it is an easy fix, running it by Intel first for clarification.

This was an exploited weak point on certain setups, not all.

Any news on this issue yet? I'm eagerly awaiting this fix :up:

[Musho]

Why is this thread dieing? Am I the only one who finds a fix for this problem a very interesting thing? :shrug:

[B.E.E.F.]

So in conclusion which system is smoother and by how much?

[XS Janus]

some might even argue that smoothness is in he eye of the beholder...
hmmm....

[iandh]

I have identified the source of the problem and it is an easy fix, running it by Intel first for clarification.

This was an exploited weak point on certain setups, not all.

I currently own/have owned a LOT of both intel and AMD hardware and have seen intermittent performance lapses on only some of the intel systems in gaming. Some did run butter smooth. If you have any idea what could be causing this that would be awesome.




I love how Zucker2k starts his first post with an insult to the entire AMD section, and then proceeds to talk about how he is surprised that they haven't taken well to him. :rolleyes:

My company supplies thin film deposition systems to both intel and AMD, my brother-in-law and sister both used to work for intel, and a good friend is a process engineer for AMD. I don't have the slightest bit of "bias" or whatever the hell else you may call it. Zucker2k, implying someone is a moron, a nut job, or a fanboy for sharing repeated personal observations they've made is a surefire way to piss people off.

Every discussion thread we've attempted to have in the AMD section starts with users sharing their experiences in a friendly and constructive way, and ends with a swarm of rude intel trolls trying to spin the issue every way they can in repeated attempts to derail the discussion.

If you search through the threads, you'll see that there are a good deal of intel users being treated perfectly well as long as they decided to actually join the discussion instead of mocking it with feigned participation.

[shogo_ca]

1 second = 1000ms
30fps = 33.33ms/frame
60fps = 16.67ms/frame

To know the exact difference, you would need something that calculates the exact time in millisecond for EACH frame. The problem may be within 1fps.

heres an example: you have 30fps. the average time per frame is 33ms. Maybe on AMD it has 33ms every frame and on Intel, you have frame 1 at 10ms and frame 30 at 80ms.

[XS Janus]

Man I was asleep when they were handling those T1000's brains to process that fast.
But I see what you mean now. Equal frame rate and different time of the displayed frame = diferent perception of smoothness... makes sense in theory at least.

[little_scrapper]

My computer is suffering from "unsmoothlessnessism". When I start up Fallout 3, my computer just sits there... starring at me blankly. Like a retarded red-headed step-sister. For about a minute and a half before the game starts. And, after I have been playing for a while and everything is all hot and juicy, and dripping with excitement... it starts to stutter badly in the outdoor scenes.. especially when I turn fast.

So the question that begs to be asked, assuming it hasnt been asked already since I only read the 1st page of this thread, is: How do I resmoothinate my chalifrajilism?

PS What is the HTT thing people are mentioning? Can someone PM me with an excruciatingly painfully detailed explaination as to what it is and does and how it relates to my particular settup. And please use between 1000-1250 words? :)