Intel Q9450 vs Phenom 9850 - ATI HD3870 X2

Quote:

---

Originally Posted by Boschwanza

@gosh

The Phenom used for the grid scores dont represents a standard Phenom. I remember it was clocked with 3.2 GHz and a insane (at least at this time) NB frequency of 2.7 GHz, DDR2 1100 4-4-4-5.

---

Ok, but Intel should win by a margin. It wins big on resolutions 1024x768 or maybe 1280x1024.
You can't clock the L3 cache also if I am right. If you look at raw processor speed Intel will win at the same clock because of the cache design. And it does win on single threaded applications, it wins big. But as soon as you go outside the processor it is loosing its advantages. And at some point depending on whats going on AMD is passing Intel.

I don't think Grid is using that much memory, PS3 cores have some specific cache that they can use but they can't go outside this. Think this is 256kb. Don't know but have read that one core is doing som filling etc for this cache so the other cores have memory that they are going to use. The key to make the PS3 fast is to use all these cores and make sure that they have fresh data.

If the game is starting to use memory and intel will need to fetch more data in memory then there will be big problems if it also need to send huge amounts of data to the GPU.
What I have seen on game tests comparing 4850 and 4870 in crossfire at very high resolutions is that they almost perform similar if it is a demanding game. That could be that the processor (Intel) isn't able to feed them with the data they can process.

I wonder how much AMD is paying this guy? He has this same stuff posted everywhere trying to get people to buy AMD.... :shakes:

Quote:

---

Originally Posted by road-runner

I wonder how much AMD is paying this guy? He has this same stuff posted everywhere trying to get people to buy AMD.... :shakes:

---

hahaha :D
I think that Intel has much more people on salary ;)
There is a need to explain how these processors work. Intel has done a great jobb marketing the "FSB doesn't mattter, we are faster because we win big on low res" etc.

Quote:

---

Originally Posted by gosh

About this list
LIST - Processors Bottom to Top

They are running Grid at 640x480 and AMD seems to perform better on that very low resolution. The reason for this might be that on that VERY low resolution the cache (L2) for AMD is enough. Maybe you don’t need that very LARGE L2 cache then and that makes AMD even to Intel. AMD synchronize threads faster and that might be the reason why you get a bit better performance.
When you increase the resolution the game is needing more memory and AMD is going to the memory sooner than intel, it is first going to the L3 cache but there is one performance hit for more than 3 times slower. Going to memory the performance hit is even larger. If they had tested on 800x600 intel would probably win, it would win on 1024x768, as the resolution goes up AMD will gain performance again and at some point it will pass Intel.

---

if your so fixated on that list and dont accept other results your hopeless. This list is practically useless. To many config with to many variables. One of the biggest misstake -> noone tested intel rig wiht a 3870 in cf.

Jack already showed you that grid behaves exactly like all other games. The higher the resolution the closer the both cpus come to getter, caused be the gpu bottleneck.

Your running around in circles, you relay on that one one post and try to imply upon that results that phenom is better at highres gaming then core2....

Quote:

---

Originally Posted by gosh

What I have seen on game tests comparing 4850 and 4870 in crossfire at very high resolutions is that they almost perform similar if it is a demanding game. That could be that the processor (Intel) isn't able to feed them with the data they can process.

---

you have seen... -> link please that shows me that there is a cpu limit @ 2560x1600 with aa.

I guess if AMD is doing such a great job with it's processors, then its $5 stock should reflect that. You're the only one arguing against results that everyone else has accepted. If this was September 2007, your arguments might have had more merit, but not today.
I'm not going to argue that Phenom isn't a great chip, but comon, it's essentially a K8 spun into a Quad with L3 cache, the K8 being an architecture that's been around since 2003. You can argue that Core 2 is based off of PIII, but I can guarantee you that the similarity between K8 and K10 is far greater. With Nehalem being shipped in a couple months, AMD will lose it's last foothold in the market, the multi-CPU servers.

The longer this argument drags on, the more painful it is going to be for you when it is over.

Quote:

---

Originally Posted by gosh

I will soon have one my self and will test that on 2560x1600. Also going to get a 790GX card and will compare that with 790FX

What is your explanation of this test?
LIST - Processors Bottom to Top

---

So this is what I mean by a hodge podge of data, nothing is shown for consistency. There is also nothing in the thread that tells me how to be consistent. I don't doubt the numbers themselves, but I doubt that any interpretation can be made.... give me a bit and I will show you what I mean with respect to GRID.

Here is their first problem:

Quote:

---

Edit: As soon as it switches to green pressure on Screenshot button in Fraps.

---

They take an instantaneous frame rate rather than average over a period of time in the game.

The difference is I will report to you frame by frame via the FRAPs output.


jack

Quote:

---

Originally Posted by gosh

hahaha :D
I think that Intel has much more people on salary ;)
There is a need to explain how these processors work. Intel has done a great jobb marketing the "FSB doesn't mattter, we are faster because we win big on low res" etc.

---

It's not that FSB doesn't matter, it does... it just doesn't matter within the context of the interpretation you are giving it... this has been shown in this thread.

I will show you how to force the situation to make FSB matter....


Jack

Quote:

---

Originally Posted by gosh

If the game is starting to use memory and intel will need to fetch more data in memory then there will be big problems if it also need to send huge amounts of data to the GPU.

---

This is where you are going wrong. The explanation is longer than I have time for at the moment. I will be back to explain.

Quote:

---

Originally Posted by JumpingJack

So this is what I mean by a hodge podge of data, nothing is shown for consistency. There is also nothing in the thread that tells me how to be consistent.

---

Or you don't know how to draw conclusions. If you know how the hardware works and have some knowledge about the program you should be able to predict results.

Quote:

---

Originally Posted by JumpingJack

This is where you are going wrong. The explanation is longer than I have time for at the moment. I will be back to explain.

---

If you do, could you use some numbers etc (keep it short), not just words. I can se when you don't know how the hardware works. But I have trouble to explain etc because english is my second language and I am terrible at spelling.

Quote:

---

Originally Posted by Boschwanza

@gosh

The Phenom used for the grid scores dont represents a standard Phenom. I remember it was clocked with 3.2 GHz and a insane (at least at this time) NB frequency of 2.7 GHz, DDR2 1100 4-4-4-5.

If you raise NB frequency you lower latencies, maybe thats why Phenom is ahead there.

@Jack

Can you run some test to confirm or disprove the numbers given for WiC, Crysis and Company of Heroes in http://www.overclockersclub.com/reviews/intel_q9300/

Same settings would bei nice ;)

---

Let me study the link and I will get back to you.... it may take a day or two.

jack

Quote:

---

Originally Posted by gosh

Or you don't know how to draw conclusions. If you know how the hardware works and have some knowledge about the program you should be able to predict results.


If you do, could you use some numbers etc (keep it short), not just words. I can se when you don't know how the hardware works. But I have trouble to explain etc because english is my second language and I am terrible at spelling.

---

Now you are getting condescending again.

Maybe gODJO can fill you in on whether or not I understand how the hardware works :) :)

Now ... here is my point. Read through the thread and look at how they are collecting their data. They set the res at 640x480, startup a game and simply press 'screenshot' when the race count down hits green. This is hardly a representative frame rate you would expect from simply playing the game. heck, even my little 15 second run through is not a good representation overall.

Quote:

---

Originally Posted by JumpingJack

Now ... here is my point. Read through the thread and look at how they are collecting their data. They set the res at 640x480, startup a game and simply press 'screenshot' when the race count down hits green. This is hardly a representative frame rate you would expect from simply playing the game. heck, even my little 15 second run through is not a good representation overall.

---

Ok, so I should listen to the results that you have got instead? What others posts is to complicated? Is that what you are saying

With that kind of ignorance, allow me to have some serious doubts that he'll even read your results and the text...

Quote:

---

Originally Posted by BenchZowner

With that kind of ignorance, allow me to have some serious doubts that he'll even read your results and the text...

---

Read this:
http://www.xtremesystems.org/forums/...0&postcount=31

No you are right if you are writing to me.
Also the test on the german site, testing the FPS when lights go green is probably done in order to do same on all computers. They are trying to do same as all others do. Testing a game at 640x480 isn't something that say much anyway for total game performance.

You know guys, ask your self this question...is AMD bothered about not being as fast as Intel?

I talked to AMD around 1 yr ago, maybe longer, they told me they wanted to build a competitively priced platform, with good dependable technology that improves and runs day in day out. I think they showed us the way they are going with 4XXX GPU, SB750 and the newer chipsets due down the line. The CPU's are getting better also... i know they will not match i7 but really i don't care, as long as they do what i want them to do does it really matter?

overall those looking to surf the net at 4.5GHZ do so on Intel, I surf the net at 1GHZ on a 6000+ with convection watercooling and i bet my pages load faster than the 4.5GHZ systems using the 20mb BB I have here LOL

Each to their own on this, I like Intel and AMD, both are good, I use both every day and have no issues...everything I expect the systems to do they do well and both do it fast.

Quote:

---

Originally Posted by Tony

You know guys, ask your self this question...is AMD bothered about not being as fast as Intel?

---

Yeah, pretty sure they are .... they do not have any pricing leverage to bring ASPs up. AMD has had it rough for the past 2 years or so, but their execution has really stepped up to the plate, the ATI division is hitting on all cylinders which is good to see.

Now, does it matter to a gamer... nope. A Phenom will give you every bit as good a game play as an Intel CPU. I am not trying to argue what is the damnedest best CPU ... this is more like addressing a gap with someone who obviously has some crazy notions :)

Quote:

---

Originally Posted by gosh

Ok, so I should listen to the results that you have got instead? What others posts is to complicated? Is that what you are saying

---

Nope.... just that one frame shot of an instaneous grab by someone on a keyboard is not reproducible and statistically valid.

http://www.techreport.com/r.x/radeon...2/grid-fps.gif
http://www.techreport.com/articles.x/15293/9

This is a time run (via FRAPs) of a race around an entire track.

Take the 4870X2 trace, solid orange for example. Look at the first dip at 79, then eyeball an average (it is actually 94.1) ... a single frame from the anywhere within the run is not representative of what the frame rate actually is for this game.

The data just isn't a valid data set to analyze to any conclusions.

Now notice that the Techreport data shows GRID for a 3.0 GHz 9650 -- again, at high resolution, if this were an issue ... do you think they would getting in the 100's FPS? Doesn't this sorta knock the data you showed out of the water :) ....

Heck, even Techreport's data set can be debated .. why? Because no two traces are the same, the author relies on one lap around the track but it is impossible for a user to reproducibly run the track exactly the same way for ever test case. Examine the lowest trace, at around 75-80 seconds into the race it takes a huge dip and stays flat for 10-15 seconds (he likely crashed and had trouble recovering) and this is factored into his calculation. Thus his framerate for that run is way off.

BTW even these traces are showing normal, random gaming behavior, there is not hint of any bottlnecking... there is good reason for this.

This is kinda fun...

jack

Quote:

---

Originally Posted by gosh

Thats like saying cache doesn't matter. Do you know why cache is so important?
VERY high bandwidth and low latency.
If you are going to multitask (use more than one application etc) then memory is important. But if you time one single but complex operation on desktops those normally depend on processor speed.

---

Cache is what hides the memory latency and decreases the dependency on the FSB, this is why Intel puts large caches into their chips.

This last statement is just bunk.

Quote:

---

Originally Posted by gosh

I will soon have one my self and will test that on 2560x1600. Also going to get a 790GX card and will compare that with 790FX

What is your explanation of this test?
LIST - Processors Bottom to Top

---

FYI -- 790GX and 790FX are chipsets, you probably meant motherboards....

Quote:

---

Originally Posted by JumpingJack

This is a time run (via FRAPs) of a race around an entire track.

---

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.

Is it possible for you to do a CPU intensive task such as CineBench or similiar.

This should help clarify some things

Quote:

---

Originally Posted by gosh

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.

---

who can guarantee you that every person pressed at the exact same time, just look at the data from techreport. You have hughe fps jumps from one seconed to another, also did they all use the same map? same ammount of computer controlled cars?.... there are so many unanswerd questions that makes your "proof" worthless.

Again if your whole hypothesis is based on this dataset its meant to fail...

Quote:

---

Originally Posted by Hornet331

who can guarantee you that every person pressed at the exact same time, just look at the data from techreport. You have hughe fps jumps from one seconed to another, also did they all use the same map? same ammount of computer controlled cars?.... there are so many unanswerd questions that makes your "proof" worthless.

---

eehhh. You can't guarantee anything. The person that reads information need to understand and that means that the person need to understand what’s uncertain. Checking more results will make the more reliable but you will probably always be able to find doubts in all tests.
What I think is a bit funny is that some doesn’t understand that you need to check the bottlenecks when games are tested. I wrote about how threads could switch cores on C2Q and that will invalidate the cache if the thread is moved from one core on C2D to the other C2D. This is a performance hit that you will not see on tests. But if this happens only 10 times each hour the gamer will probably experience problems if that bottleneck is enough to make the game slow.

Is 100 FPS always better than 30?
If 90 are rendered the first half of the second and 10 the other half on the 100 FPS computer.
The 30 FPS computer renders all frames evenly. Witch computer would you choose?

http://www.xtremesystems.org/forums/...2&postcount=95

Quote:

---

Originally Posted by gosh

Witch computer would you choose?

---

http://markteipe.com/wp-content/uploads/witchhunt.jpg

Sorry but I had to do that :D
{ yes, it's just humor, I know that english isn't your native language }

As for the unevenly rendered FPS... do you have any proof at all ?

As for the invalidated cache... well... I have yet to see that happen in any game... maybe you should ask a company to make one so you can "prove" your point and try to invalidate all of us with that one :p:

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.