Intel Q9450 vs Phenom 9850 - ATI HD3870 X2

[gosh]

It's really obvious and not difficult to see that you're un-negotiable and you're here just to try to convince everybody that your logic and saying is right and everybody else is wrong.

I am VERY negotiable but if someone says something that I have explanation of or if they present facts but don’t know how to explain facts. Or if they see facts, don’t understand how to interpretate it and because of this just say that it doesn’t count. These facts may say something to me.
It could also be that someone shows facts and that person draws different conclusions or draws conclusion like it is 100% GPU bottlenecked. If that person dosen’t understands that everything in a chain of reactions that takes time will add to the total time. How should I replay?
Also I have trouble digging in to complicated discussions because of the language.

[BenchZowner]

All I'm seeing here is you trying to convince the readers that JumpingJack doesn't know how a computer works ( which is obviously not true ).

I've already seen some valuable data, and waiting for more, and might even jump in and do some testing as well.
But I have a feeling that you'll reject everyone's data if it doesn't suit your opinion.

[gosh]

All I'm seeing here is you trying to convince the readers that JumpingJack doesn't know how a computer works ( which is obviously not true ).

Then he needs to show that he understands. It is easy to make someone believe that you understand if that person doesn’t know as much. You only need to be good at writing/talking. What JumpingJack is good at is writing and talking. But when he talks to someone that has some knowledge there need to be more than words.

Reading his first messages in this thread made me wonder if he was one unsecure child.

[highoctane]

Then he needs to show that he understands.

Maybe you need to take your own advice and prove that YOU understand since you are the one trying to make a claim that you have yet to prove yourself. You have lead us ALL think you are the one that doesn't understand anything.

[bowman]

Then he needs to show that he understands. It is easy to make someone believe that you understand if that person doesn’t know as much. You only need to be good at writing/talking. What JumpingJack is good at is writing and talking. But when he talks to someone that has some knowledge there need to be more than words.

Reading his first messages in this thread made me wonder if he was one unsecure child.

So, wait, let me grasp this.. JumpingJack has posted numerous links, done numerous time-taking tests on various systems, and he's the one 'only good at writing/talking'. While you have yet to prove anything yourself let alone refute his data - what does that make you then?

[gosh]

So, wait, let me grasp this.. JumpingJack has posted numerous links, done numerous time-taking tests on various systems, and he's the one 'only good at writing/talking'. While you have yet to prove anything yourself let alone refute his data - what does that make you then?

Does JumpingJack know that it is very important to check bottlenecks (slow parts ) testing games? I haven't seen that from him

[JumpingJack]

This thread is about comparing processors. When you are playing a game you don't want the game to slow down that much so you could notice that the game lags. If you play the game and calculate fps for one whole race that will not say more that testing the fps for one specific event in the game.

The most important thing to test on the game in order to make sure that the game is smooth is the slowest parts of the game. If I remember right the start in Race Driver Grid is using a lot of processor power.
If you have one game that has an average of 1000 FPS but slows down to 10 FPS for 0.5 seconds once every minute the game will be very annoying. When you test games you need to check the slowest parts in order to know if the game will run smooth.

Gosh --
This thread is about debunking your assertion that the FSB is crippling Intel in high resolution gaming. Comparing two processors is a different topic all together, and it is widely shown and been sufficiently demonstrated that Intel currently produces the superior part. This does not make the Phenom (Agena or Barcelona) a bad processor, it is a good processor.

In terms of good game play, this criteria simply resides in the ability of the HW to produce frame rates such that the minimum is above the refresh rate of the monitor.

Your post here http://www.xtremesystems.org/forums/...2&postcount=17 is what is at issue. Nobody, not even me, will argue that the FSB implementation is not old and antiquated, however for client desktop applications it is not the critical factor you are asserting.

Intel and AMD take two approaches to feeding the core(s) with instruction or data from system memory -- a fast memory access will smaller cache or a large cache with slower access to memory. Intel has designed and implemented a caching system that works around the FSB deficiencies compared to AMD's IMC approach.

It works like this ....

When a processor crunches data and instructions it fetches that information into the queue. Those instructions proceed through a general progression fetch, decode, reorder, execute and retire. As it moves through the fetch buffer and requires the next block, it goes to cache first if it is there it generates a cache hit, if it is not there it is a cache miss. The hit rate for Intel cache is higher than it is for AMD cache, due to size and a few other variables.

The miss rate on cache is directly proportional to the size, the large the size the lower the miss rate. Each cache miss generates a penalty for going to main memory in which case the processor needs to wait X cycles for that data. Intel also has more aggressive prefetchers which are working to populate the cache with the next needed instruction to avoid misses that would require the FSB.

AMD generates > misses than Intel does, but each miss has less penalty. So which is better, a fast memory connection with high miss rates or a slower memory connection with low miss rates. Overall each approach accomplishes the same thing, and considering the body of data on the net showing clock for clock Intel significantly outperforming AMD in most any application then it is sufficient to conclude that Intel's archaic FSB technology is not a problem.

Where the FSB does become a problem is when the workload and memory footprint exceeds the capacity to keep the cache miss rate low... this is the case for high throughput applications found in both server and HPC applications. In cases where BW is the limiter AMD will win, in CPU bound workloads Intel will win. DT is all CPU bound workloads, BW is hardly a factor, I have found only a handful of cases where I could point to the BW of the FSB the culprit.

If you wish to see this coming from someone other than me, find a library, read and learn.
http://ieeexplore.ieee.org/xpl/freea...number=4536342
Their conclusions are exactly what I just told you above and have been telling you for the past 5 days.

Now, I have showed you data which indeed demonstrates that changing the FSB BW has no impact on the observed performance in high resolution gaming. There is a reason for this the rate determinant is the ability of the GPU to finish it's work -- the GPU is the bottleneck.

This is due to your lack of understanding how a PC behaving in a graphically intensive gaming environment. Part of this lack of understanding is you do not grasp or comprehend that the video card GPU is high through put and as a result, modern 3D cards put high speed RAM on the card and the data that is needed is loaded to this RAM before the game even starts, the traffic over the FSB is small during actual game play. The only time it does become a factor is when the amount of textures needed exceeds the memory capacity of the video card, in which case you will get a stutter ... this will happen on both AMD and Intel systems, neither NOT even AMD, has enough main memory BW to satiate the throughput needs of a GPU. Here is a less academic site that explains this:
http://www.yougamers.com/articles/13...ly_need-page2/

This is also the reason why nVidia and ATI cards are now coming with more and more ram, as high as 1 GB.

You can test this... download the Doom3 demo for HOCbench.com and run it using their demo file... Doom3 is bad about not precaching their textures. Run this on your uuuubber 4870X2 and Phenom rig... it will stutter and jerk, it will make no difference what speed you run the processor or what graphics card you use. Run it through twice, you will always get higher frame rates the second time because the first round has put all the textures that weren't there into video memory.

Finally, you had better hope you are wrong --- for AMD's sake. Because if the antiquated FSB is indeed holding back hidden potential from the Intel Core microachitecture, then what happens when Intel removes this roadblock?

Nehalem will be out by the end of the year, DT will show impressive gains especially in multithreading to be sure, but server (where the FSB BW issues are truly a problem) it will be:

Nehalem will be nothing short of a miracle – with performance gains of 2X or better.

http://www.realworldtech.com/page.cf...208182719&p=10

Jack

[JumpingJack]

Does JumpingJack know that it is very important to check bottlenecks (slow parts ) testing games? I haven't seen that from him

Unfreaking believable.... I showed you that here:
http://www.xtremesystems.org/forums/...6&postcount=59

it takes a lot for someone to cause me to call them stupid, you are only a few more posts away.

[gOJDO]

[gosh]

Gosh --
This thread is about debunking your assertion that the FSB is crippling Intel in high resolution gaming. Comparing two processors is a different topic all together, and it is widely shown and been sufficiently demonstrated that Intel currently produces the superior part. This does not make the Phenom (Agena or Barcelona) a bad processor, it is a good processor.

In terms of good game play, this criteria simply resides in the ability of the HW to produce frame rates such that the minimum is above the refresh rate of the monitor.

Your post here http://www.xtremesystems.org/forums/...2&postcount=17 is what is at issue. Nobody, not even me, will argue that the FSB implementation is not old and antiquated, however for client desktop applications it is not the critical factor you are asserting.

Intel and AMD take two approaches to feeding the core(s) with instruction or data from system memory -- a fast memory access will smaller cache or a large cache with slower access to memory. Intel has designed and implemented a caching system that works around the FSB deficiencies compared to AMD's IMC approach.

It works like this ....

When a processor crunches data and instructions it fetches that information into the queue. Those instructions proceed through a general progression fetch, decode, reorder, execute and retire. As it moves through the fetch buffer and requires the next block, it goes to cache first if it is there it generates a cache hit, if it is not there it is a cache miss. The hit rate for Intel cache is higher than it is for AMD cache, due to size and a few other variables.

The miss rate on cache is directly proportional to the size, the large the size the lower the miss rate. Each cache miss generates a penalty for going to main memory in which case the processor needs to wait X cycles for that data. Intel also has more aggressive prefetchers which are working to populate the cache with the next needed instruction to avoid misses that would require the FSB.

AMD generates > misses than Intel does, but each miss has less penalty. So which is better, a fast memory connection with high miss rates or a slower memory connection with low miss rates. Overall each approach accomplishes the same thing, and considering the body of data on the net showing clock for clock Intel significantly outperforming AMD in most any application then it is sufficient to conclude that Intel's archaic FSB technology is not a problem.

Where the FSB does become a problem is when the workload and memory footprint exceeds the capacity to keep the cache miss rate low... this is the case for high throughput applications found in both server and HPC applications. In cases where BW is the limiter AMD will win, in CPU bound workloads Intel will win. DT is all CPU bound workloads, BW is hardly a factor, I have found only a handful of cases where I could point to the BW of the FSB the culprit.

Now you are talking!

some tiny notes...
How the system design works depends on the application. You can’t say that one big cache has that hit rate and another cache has another. The hit rate depends on the application. Small caches have rather large hit rates. I think that you will see very high hit rates even for tiny caches in most applications. But the performance penalty going to memory is very big for some system design so they are using VERY large caches
Running more applications will degrade advantages for caches pretty fast. Of course you will always ses much speed improvements having caches even if the number of threads is huge.
The exception from this is applications that isn’t using that much memory and have controlled tasks. Applications heavy on calculations etc

[JumpingJack]

Now you are talking!

some tiny notes...
How the system design works depends on the application. You can’t say that one big cache has that hit rate and another cache has another. The hit rate depends on the application. Small caches have rather large hit rates. I think that you will see very high hit rates even for tiny caches in most applications. But the performance penalty going to memory is very big for some system design so they are using VERY large caches
Running more applications will degrade advantages for caches pretty fast. Of course you will always ses much speed improvements having caches even if the number of threads is huge.
The exception from this is applications that isn’t using that much memory and have controlled tasks. Applications heavy on calculations etc

Depends if the threads are active. Windows at any given time can have up to 30-70 processing running in the background. There are always finite resources available to the processor. Even AMD can saturate their BW (and they do with their current quads and HT3.0) they simply saturate at a higher point to yield higher performance. But yes the miss rate is app dependent, I link a reference below.

However, this is not the case in desktop and certainly not in gaming. The data on that is clear. The BW advantage AMD holds is only good when the application at hand demands it. Otherwise, a fast large cache is sufficient to negate the delta in BW between the two processors.

Again, this is very clear in the data.
laughingly, someone has compared the relative hit rates between an older K8 an C2D and wrote a disseration on it:
http://etd.lsu.edu/docs/available/et...ash_thesis.pdf
This guy got a masters degree for benchmarking a Core 2 Duo


Back to the point at hand, ironically, there has been only one case on today's contemporary gaming scene that demonstrates a condition where the intercommunication limitation is clearly demonstrated: http://www.anandtech.com/video/showdoc.aspx?i=3372&p=9 and it occurs at 2560x1600 max AA for GRID. Unfortunately, this is not the condition that your link provides.


So, gosh, why don't you list out for a high resolution gaming system, where all the potential bottlenecks may occur?

[Hornet331]

Again, this is very clear in the data.
laughingly, someone has compared the relative hit rates between an older K8 an C2D and wrote a disseration on it:
http://etd.lsu.edu/docs/available/et...ash_thesis.pdf
This guy got a masters degree for benchmarking a Core 2 Duo

that got him a master degree?
Damn it i study the wrong subject. My bachelor work was three times as long.

[JumpingJack]

Oh ..... X2's on their way... your prediction on GRID?

[JumpingJack]

that got him a master degree?
Damn it i study the wrong subject. My bachelor work was three times as long.

Don't laugh too hard, it's actually well done... he compares the L1 and L2 hit rates on SPEC for both the Core 2 and K8 cores.... very informative.

EDIT: Actually, when I ran across the thesis and saw the title, I did let out a little chuckle

[Loser777]

I simply have nothing to say because gosh is too bigoted to care.

This thread becomes more fun everyday.

[WeldZilla]

Well Many are looking at this thread all wrong. I have been doing this since 1979 and I thankyou jack for some excellent and informative posts.
I learned from you, that is why I come to forums, to help where I can and learn all I can. Computers have always fascinated me. I work in the
construction industry and always have. Computers are just a passion. I run a weld shop. (big Surprise) So while gosh has contributed little if anything to
his Thread, you have really laid it out there and I appreciate it. I will say though that video editing paid for many of my systems. Although I work to steady now
to do too much of that.

Oh and all 3 of my e8400 systems run daily at 4500mhz.. Why??? Because they can!!!

WZ

[JumpingJack]

EDIT: Checked two tests on Lost planet

Here is one: http://techreport.com/articles.x/14424/4

Comparing QX9650 and QX6850, isn't the main difference cache size? That would mean that Lost Planet likes a big cache (is there other differences between these two processors?)

Just saw your edit...

I don't understand your point here nor what it has to do with your assertion that we are working on ... of course a larger cache is going to yield better overall, because Cave in Lost Planet is CPU bound, thus it responds to CPU performance. IPC goes up with cache size (as explained above), and because the CPU is the throttle in Lost Planet for this particular scene, you would expect (and you in fact point out as an observation), that it should show a response. I have always had a bad taste in my mouth when I see someone post "games love cache" ... anything loves cache, cache will improve just about anything, some more than others, some hardly noticeable... but larger cache usually means good (until it gets so large, latency negates the advantage).

The difference in the QX9650 and a QX6850 is cache size and a few other architectural tweaks, but the influence here is completely based on the difference in cache most likely. I can link several citations for you that show how cache improves performance. As explained above, and as you obviously agreed, the miss rate in cache varies with applications and the functionality of the cache size to miss rate goes as a power of the size. For example, miss rate = k*X^(-power) where k is a proportionality constant, X is the cache size, and power is a number (typically between 1 and 1.8, cannot remember I can dig up the reference) ... EDIT: though the rule of thumb has always been 1/sqrt(size), but this is not really accurate today. It depends on the associtivity, whether it is exclusive or inclusive, yada yada. and does not always track a sqrt(2) proportionality. Just wanted to make that clear. (I think it is this one http://portal.acm.org/citation.cfm?id=327132 but am not logged into my server at work to view it).


Now, using your same link as a reference ...



I have shown (see prior posts) and I have argued vehemently that Snow is GPU bound. This data also shows that same trend. Look at how interesting this is... for changes that address CPU performance there is a FPS response in Cave, but not for Snow .... this is what makes this particular game so good to use to study difference between CPUs and GPUs, I have scripts within two different regions of the game that stresses the two computational sources in the same benchmark run.

Your edit more or less helps make my point.

A side point here is to take a careful take home message. Each script is only about 2-3 minutes of game play in two different levels/locations within the game, in one case I would see CPU's all look the same even at 1152x864 (which is what Techreport used) -- i.e. GPU limited -- yet if I concluded based on Cave then I would be able to make a statement between the relative performance of the two CPUs in question. In other words -- if I only took the Snow data, I would make a false conclusion of the relative performance of various CPUs -- I would think they all are basically the same; however, Cave shows this is not true ... I would have fooled myself.

Games and benchmarks can be selectively created to force a situation if one does not pay attention, questioning what you read is a good habit.

Ok -- so how do I evaulate the game/code/BW/etc concerns of a CPU to make an honest assessment? Well ... take the GPU completely out of the equation. Doing so allows me to see how the two CPUs behave based on the way they run the code necessary to make the game work. Is this realistic ... nope, not really ... but I do want to know this information such that, for example, if I want to upgrade to a faster video card in the future... I don't need to re-address my CPU choice, I will now what performance to expect. Ideally, a good review for a CPU will show both the ultra high res stuff -- like they way we want to play the game as well as the ultra low res stuff so I can 'future' proof my rig.

EDIT2: I am sorry, last edit and one last point. I do not have a test from my experiments that mimics the QX9650 runs that Scott did in his review, so I cannot compare directly. Nor do I mimic his graphics settings ... (if you read the link in pages earlier to my LP article, I document the exact settings). With this in mind, I am matched at stock condition for the Phenom 9850 and I did collect a screen shot many moons ago for that article I linked above, and so here is my 1152x864 run:



As I said, the Graphics settings are not the same -- so Snow will be different - again, because Snow is GPU bound. However, Scott (the TR link you provided) gets around 79.2 FPS, I get 77.7 for the Cave segment ... that is within 1.5 FPS (1.9% difference)... this is what I would consider, within the margin of error ... the same. The point is ... I can reproduce the Tech report values for both the Phenom series and the QX9650 as well as the QX6850 .. though I have not done them all, I did do a few spot checks informally (no screen dumps, just checks to see if I could match that data) which was just a few tests and his numbers are spot on to what I can get with the same HW -- I have 99% confidence that, as he reported his settings, that his numbers are accurate.

What I am demonstrating and the point I want to make with this is that unless you can match configurations, ensure everything is the same, and account for the details ... then a direct compare is not possible .... THIS is why I call your GRID data bunk. I cannot reproduce it. No matter how hard I try there is no information in that thread that allows me to take the same HW, set up the same settings, and reproduce the 'hodge-podge' of results. As such, this invalidates the data set. This is also why my posts get so long, especially the lost planet experiment I produced for you ... I want to ensure that there is more than enough detail such that anyone who wants to and had the capability can reproduce my result... this provides credibility to the data by that fact alone. I was the same level of detail and verbosity in the GRID experiment, and so yes I expect you should believe my numbers over the other guys.

Without reproducibility, there is no truth.


Jack

[gosh]

Depends if the threads are active. Windows at any given time can have up to 30-70 processing running in the background.

Hmmm... When I describe functionality I do believe that the person that reads the text can think. Of course threads needs to be active.

[JumpingJack]

Also I have trouble digging in to complicated discussions because of the language.

Take your time then ... if I post something that the language barrier may prevent you from getting a good grasp, I will rephrase, just ask.

Jack

[gosh]

Take your time then ... if I post something that the language barrier may prevent you from getting a good grasp, I will rephrase, just ask.

I have work to do also, don't have the time. In the end people will think as they want anyway. If you have enough people that say something then it will get more people to say the same thing.
Have you heard about religion? Nobody has seen God but a lot of people know(?) that he/she/it exists.

[thebanik]

I should not post here, coz i have nothing valuable to add here, but still i wanna say, jumpingjack, you got a fan here....quite a few things I have been learning in the last few days from you and this thread......

[JumpingJack]

I have work to do also, don't have the time. In the end people will think as they want anyway. If you have enough people that say something then it will get more people to say the same thing.
Have you heard about religion? Nobody has seen God but a lot of people know(?) that he/she/it exists.

Well, my profession is a bit different and doesn't mesh... there is the spiritual, and then there is the concrete, that which is provable through observation. It is important to keep those separate, otherwise violent eruptions will occur.

[gosh]

Well, my profession is a bit different and doesn't mesh... there is the spiritual, and then there is the concrete, that which is provable through observation. It is important to keep those separate, otherwise violent eruptions will occur.

Just test these new cards that you have ordered. I wouldn’t be surprised if you will see very similar FPS on 1680x1050, AMD would probably be a little higher on 1920x1200.
AMD would also have higher on min FPS. I know that it can differ depending on what race track that is used. If it is a track that isn’t heavy then this will favor Intel.
If the memory on the computer are slow than this will hurt AMD more than Intel

EDIT: You don’t need to test two 4870X2 cards, it’s enough with one. Why did you buy two when you only(?) are playing at 1920x1200 ?

[Hornet331]

Don't laugh too hard, it's actually well done... he compares the L1 and L2 hit rates on SPEC for both the Core 2 and K8 cores.... very informative.

EDIT: Actually, when I ran across the thesis and saw the title, I did let out a little chuckle

sorry for the bad phrasing, i was not laughing about his work. I was laughing about myself for doing more work then necessary.

btw. jack you really have two 4870x2 incoming?

[gosh]

JumpingJack: One tip!
If you do some testing in different games comparing AMD and Intel using 4870X2 and present the results somewhere. I think you will get LOADS of hits. Make sure that the server is able to handle the traffic