Intel Q9450 vs Phenom 9850 - ATI HD3870 X2

Printable View

Show 100 post(s) from this thread on one page

08-11-2008, 09:02 AM
Viper666

Its kinda funny on every forum ,Anyone that seems to understand the differences,gets banned or deleted,if they dont fall for the Intel marketing hype

".. an ounce of honest data is worth a pound of marketing hype." - Spec.Org< answers are there!
08-11-2008, 09:20 AM
Viper666

Quote:

Originally Posted by Scubar

That pretty much says it all.

Goshs theory that the C2D and C2Q are only good at single threaded apps is possibly one of the most undeniably stupid things i have ever heard anyone say. Maybe i should buy a Phenom for all my Video encoding as thats all multithreaded.

Actually my Bribane 2.5 45watt dual core and my 4850 ATI using avivio will slaughter your box encoding. Changing a DVD to DIVX about 2 minutes I think

And its a fact that Intel does very well on single threaded apps, but on muti threads not so well(more than 2 threads), the advantage it has on single thread doesnt scale per core the same, on very multi threaded stuff and heavy loads, it cant because all 4 cores talking over FSB and RAM,not processor cache like Phenom.

Anyway why do you thinK the new Intel copied the AMD Idea ? Can you take a guess? lmao

I would find an Open Source Video Encoder and you might become very unhappy ! Also go look at spec.org standardized tests and bench marks with no cheating
08-11-2008, 11:49 AM
Hornet331

Quote:

Originally Posted by Viper666

Actually my Bribane 2.5 45watt dual core and my 4850 ATI using avivio will slaughter your box encoding. Changing a DVD to DIVX about 2 minutes I think

And its a fact that Intel does very well on single threaded apps, but on muti threads not so well(more than 2 threads), the advantage it has on single thread doesnt scale per core the same, on very multi threaded stuff and heavy loads, it cant because all 4 cores talking over FSB and RAM,not processor cache like Phenom.

Anyway why do you thinK the new Intel copied the AMD Idea ? Can you take a guess? lmao

I would find an Open Source Video Encoder and you might become very unhappy ! Also go look at spec.org standardized tests and bench marks with no cheating

:shakes:
i save my breath to correct your mistakes. There are so many its not even funny...
08-11-2008, 02:49 PM
gOJDO

Quote:

Originally Posted by Viper666

Actually my Brisbane 2.5 45watt dual core and my 4850 ATI using avivio will slaughter your box encoding.

Actually same clocked Core2 Duo or Quad will wipe the floor with your Brisbane @2.5GHz for the purpose you've mentioned above.

Quote:

Changing a DVD to DIVX about 2 minutes I think

:D What's the time and the quality you are converting for 2 minutes?

Quote:

And its a fact that Intel does very well on single threaded apps, but on muti threads not so well(more than 2 threads), the advantage it has on single thread doesnt scale per core the same

Core2 Duo have better performance scaling than K8. Phenom has better performance scaling than Core2 Quad, but it performs slower at same clock, including the apps that are utilizing more than 2 threads.

Quote:

, on very multi threaded stuff and heavy loads, it cant because all 4 cores talking over FSB and RAM,not processor cache like Phenom.

What you are saying is not true. Intel always had and still has much more advanced cache architecture than AMD ever.

It's core-to-core communication on their Core2 Duo is much more sophisticated than on both K8 and Phenom.

Core2 Quad is a MCM of 2 dualcore chips. The cores on the same chip are communicating 2x faster than the cores on Phenom. For frequencies over 2GHz the cores of the different dies of the MCM, a C2Q with FSB1333 is communicating slower than the cores of a same clocked Phenom.

Quote:

Anyway why do you thinK the new Intel copied the AMD Idea ? Can you take a guess? lmao

:rofl:AMD idea? :rofl: Intel had a x86 CPU with on-die-memory-controler and on-chip-video 15 years ago, before AMD has designed their first x86 CPU.

Quote:

I would find an Open Source Video Encoder and you might become very unhappy !

LOL, you are funny. Before you can be so sure, why don't you check for some benches online?

Quote:

Also go look at spec.org standardized tests and bench marks with no cheating

SPEC.org are not measuring desktop performance, and if you haven't noticed the title yet, we are talking about desktop CPUs here. So, face it, on the desktop Intel is owning AMD and that will remain to be the case for sure in the following 2 years.
08-11-2008, 04:15 PM
JumpingJack

Quote:

Originally Posted by Viper666

And its a fact that Intel does very well on single threaded apps, but on muti threads not so well(more than 2 threads), the advantage it has on single thread doesnt scale per core the same, on very multi threaded stuff and heavy loads, it cant because all 4 cores talking over FSB and RAM,not processor cache like Phenom

That's odd, it sorta contradicts the data.... could you link this up?

I would like to see that. Every test I have ran is exactly the opposite, or are you arguing scaling factor?

Jack
08-11-2008, 04:27 PM
gosh

Quote:

Originally Posted by JumpingJack

I would like to see that. Every test I have ran is exactly the opposite, or are you arguing scaling factor?

What tests?
08-11-2008, 04:28 PM
JumpingJack

Quote:

Originally Posted by gosh

What tests?

Just link the data..... if this is true, and it is a widely known fact, there must be ample examples.

EDIT: OOOPs, what test have I ran? Hang tight, I will them...

EDIT2: All ran comparing 9600BE@2.3, 9850@2.3, QX9650@2.3, 9850@2.5, and QX9850@2.5 these are all the benches I have ran, some single threaded, some multithreaded.

3DMark01
3DMark03
3DMark05
3DMark06
Aquamark3
Blender
Cinbench2003
CinbenchR9.5
CinbenchR10
Company of heros 1.4
Commanche4
crysis
crystalmark
CPU rightmark
CPUMark2.1
Doom3
DVDShrink3.2
Euler3D
Everest (benchmark suite version 4.5)
FarCry
FEAR
Fritz Chess (1-4 cores)
Geekbench2
H264 Bench Tech ARP version 1 and 2
HexusPiFAST
HL2-LC
LameMP3 Multthreaded both MS and Intel compiled
LostPlanet (which I have linked above)
MainConcept (1 and 2 Pass HD264, MPEG2 render 4 movie files)
MaxiPI (weird behavior on Phenom here, one core is always 10% slower than the other 3)
Multitask suite 1 (4 single threaded apps)_
Multitask suite 2 (4 multithreaded apps)
n-bench3 (this is AMD's benchmark, runs faster on Intel quads)
Ortho CPU benchmark
Paint.NET (benchmark suite)
PCMark04
PCMark05
Povray3.7B23 (Phenom does very well here)
Prey
QW-Enemyterritory
RightMarkMA
SciencMark2.0 (again Phenom does very well here)
SeriousSam2
Unreal Tournament 3
Windows Media Encoder 9
World in Conflict
WPrime
08-11-2008, 04:37 PM
gosh

Quote:

Originally Posted by JumpingJack

Just link the data..... if this is true, and it is a widely known fact, there must be ample examples.

EDIT: OOOPs, what test have I ran? Hang tight, I will them...

Here is one:
http://connexitor.com/blog/pivot/entry.php?id=191

Code:

Clients Opteron 875 Opteron 2347 Xeon 5345 1 3.1s 2.9s 1.6s 2 6.7s 6.5s 12.6s 3 12.1s 10.5s 20.5s 4 18.1s 14.9s 44.6s

There are some links to scaling samples in the thread also
08-11-2008, 04:50 PM
JumpingJack

Quote:

Originally Posted by gosh

Here is one:
http://connexitor.com/blog/pivot/entry.php?id=191

Code:

Clients Opteron 875 Opteron 2347 Xeon 5345 1 3.1s 2.9s 1.6s 2 6.7s 6.5s 12.6s 3 12.1s 10.5s 20.5s 4 18.1s 14.9s 44.6s

There are some links to scaling samples in the thread also

EDIT: Let me add..

Ohhhhh, dual socket server... hmmmmm, sure. Ok.... that's fine, this I agree with... it really has nothing to do with ability to multithread, but really just gross overall bandwidth. There are a few server and many HPC applicaitons where the working sets are huge and the raw throughput is tremendous.... AMD's BW design really shines through here.

However, how does that transcend the bulk of the market? By this, I mean, where desktop and single socket clients... you know, the ones we use.
08-11-2008, 04:52 PM
JumpingJack

Quote:

Originally Posted by gosh

Here is one:

There are some links to scaling samples in the thread also

where? All I saw was your data from some random forum? Also, did you see the lost planet data ... can you explain why doubling the FSB speed doesn't have an effect?
08-11-2008, 05:04 PM
gosh

Quote:

Originally Posted by JumpingJack

Ohhhhh, dual socket server... hmmmmm, sure. Ok.... that's fine, how about single socket client... you know, the ones we use.

I am doing a lot of server programming. Could you please inform me about the differences? How much more communication do you think there is between cores on servers compared to desktops?
If you take one Web application for example, each request is usually handled in one single thread. The communication between threads is not that big, each thread is managing its own state. It is much more difficult to design applications if threads are talking to each other in different ways.

Keep your eyes open for games optimized for Playstation 3. The cell processor is a bandwidth monster. But you need to take advantage the cores (if you could call them that) to get speed. More and more game engines seem to have learned to optimize for this. XBOX works like a PC and if the game is optimized for PC then they will focus on Intel.
On servers there is often applications developed in Java and .NET. These are using a lot of memory; C++ is normally not that big in memory use.
08-11-2008, 05:14 PM
gosh

Quote:

Originally Posted by JumpingJack

EDIT2: All ran comparing 9600BE@2.3, 9850@2.3, QX9650@2.3, 9850@2.5, and QX9850@2.5 these are all the benches I have ran, some single threaded, some multithreaded.

Why not testing applications that use memory? Databases exist on desktops also. running more than one applications will use more memory, running five will use more than two. Web browsers etc it using a lot of memory.
If you only test single threaded applications or applications that are tiny in memory use you don’t need to test, you can predict the result and save some time. High frequencies and one BIG L2 cache will be faster.
08-11-2008, 05:16 PM
JumpingJack

Quote:

Originally Posted by gosh

I am doing a lot of server programming. Could you please inform me about the differences? How much more communication do you think there is between cores on servers compared to desktops?.

This I cannot do, i do not have the data for server.... what i do have is benchmark data from several different applications common to desktop, clock for clock AMD does not win one, single or multithreaded, or even multitasking.

Also, second request can you explain why doubling the FSB BW does not show any appreciable difference in lost planet, if this is not convincing for you then I can do what ever game you want....

EDIT: Hang tight, I think I can link up a publicly accessible paper on this topic.... it should be informative

Jack
08-11-2008, 05:17 PM
JumpingJack

Quote:

Originally Posted by gosh

Why not testing applications that use memory? Databases exist on desktops also. running more than one applications will use more memory, running five will use more than two. Web browsers etc it using a lot of memory.
If you only test single threaded applications or applications that are tiny in memory use you don’t need to test, you can predict the result and save some time. High frequencies and one BIG L2 cache will be faster.

Provide me with what you would like to see and I will do it. I don't run database intensive apps myself, nor do most people. Those are typically done via servers.

jack
08-11-2008, 05:21 PM
gosh

Quote:

Originally Posted by JumpingJack

This I cannot do, i do not have the data for server.... what i do have is benchmark data from several different applications common to desktop, clock for clock AMD does not win one, single or multithreaded, or even multitasking.

Also, second request can you explain why doubling the FSB BW does not show any appreciable difference in lost planet, if this is not convincing for you then I can do what ever game you want....

How do you know that the FSB is a bottleneck on servers but not on desktops?
Show me the code for Lost Planet and then we can see why
08-11-2008, 05:27 PM
gosh

Quote:

Originally Posted by JumpingJack

I don't run database intensive apps myself, nor do most people. Those are typically done via servers.

Yes you do, the thing is that you don't know it. The registry in windows is one type of database. Applications are often using some sort of database to store data between sessions etc. WinFS that didn't make it to Vista is a database
08-11-2008, 05:35 PM
JumpingJack

Quote:

Originally Posted by gosh

Yes you do, the thing is that you don't know it. The registry in windows is one type of database. Applications are often using some sort of database to store data between sessions etc. WinFS that didn't make it to Vista is a database

Well, that is true ... :) part of the OS. I haven't tried to install Oracle and query my house hold expenses.

So what do I need to do to benchmark the registry? :) :) :) :rofl::rofl::rofl:

I kinda think the XP registry (or even Vista registry) is a bit small to overwhelm a processor, considering even a Sempron can run it .
08-11-2008, 05:37 PM
JumpingJack

Quote:

Originally Posted by gosh

How do you know that the FSB is a bottleneck on servers but not on desktops?
Show me the code for Lost Planet and then we can see why

Isn't it clear, you linked the data.

You don't need to see the code for the data, if the FSB is a problem, then why if I double it it has no effect? you keep going back to this code thing, what is it in the code that makes you think you are going to find an answer?

So again, if per your hypothesis, the FSB is saturating at high resolutions ... why at high resolutions can I effectively double (or 1/2) the effective FSB BW and not change the result?
08-11-2008, 05:42 PM
gosh

Quote:

Originally Posted by JumpingJack

I kinda think the XP registry (or even Vista registry) is a bit small to overwhelm a processor, considering even a Sempron can run it .

This is normally what matters for users in normal use. Quick responses and fast load times, switching between applications etc. There are few computer users that run applications with tasks that last for a long time. But if they do that then they don’t sit and wait, they are probably surfing or doing something else while the application that needs time is doing its work.
I don’t know about you but if I am extracting some big file from some compressed format and that takes ten minutes. I don’t sit and wait for that. I do other tasks.
08-11-2008, 05:45 PM
gosh

Quote:

Originally Posted by JumpingJack

you keep going back to this code thing, what is it in the code that makes you think you are going to find an answer?

:eek:
what do you think?
Do you think that the code is unimportant in order to know why some application behaves in a specific way?
08-11-2008, 05:52 PM
JumpingJack

Quote:

Originally Posted by gosh

:eek:
what do you think?
Do you think that the code is unimportant in order to know why some application behaves in a specific way?

No, I think you don't know what you are talking about :) ....

Look, your claim in this thread is that in high res gaming, that the FSB cannot handle all the little threads poking around and that Intel's CPU performs poorly because of the FSB, thus Phenom is best.

I will not argue that high res gaming shows them neck and neck based on the data. But if your FSB argument is true, you do not need to see the code to make an emperical observation. Based on your hypothesis, the FSB is saturated, so if I lower the FSB speed I should take a performance hit....

What you are really observing is the GPU is limiting the FPS, you have hit the GPU limited regime.... the load variation on the GPU is deterined by the resolution you select and the oversampling chosen for the aliasing. You reject this notion, yet I showed you data that proves this notion.... so what I am asking you to do is explain the data.

My suspcion is that any time someone provides you with the data that refutes your claims you will fall back on "I need to see the code", which is in and of it's self code for "I don't know".

jack
08-11-2008, 05:53 PM
JumpingJack

Quote:

Originally Posted by gosh

This is normally what matters for users in normal use. Quick responses and fast load times, switching between applications etc. There are few computer users that run applications with tasks that last for a long time. But if they do that then they don’t sit and wait, they are probably surfing or doing something else while the application that needs time is doing its work.
I don’t know about you but if I am extracting some big file from some compressed format and that takes ten minutes. I don’t sit and wait for that. I do other tasks.

I see, we are going to 'its smoother' arguments now.
08-11-2008, 06:01 PM
gosh

Quote:

Originally Posted by JumpingJack

Look, your claim in this thread is that in high res gaming, that the FSB cannot handle all the little threads poking around and that Intel's CPU performs poorly because of the FSB, thus Phenom is best.

I will not argue that high res gaming shows them neck and neck based on the data. But if your FSB argument is true, you do not need to see the code to make an emperical observation. Based on your hypothesis, the FSB is saturated, so if I lower the FSB speed I should take a performance hit....

If you are talking about games you should test a game that is communicating with the GPU and is using memory at the same time (there will be conflicts in the FSB that increase latency, amd has no conflicts there). Then you will see big improvements for AMD. More and more advanced games will run better and better on AMD. If the game is using more than one thread to render data then AMD will perform better.
I haven’t read what you have written about Lost Planet. I need to check some tests before I can tell you something about that game. How it scales, how it is using threading etc

EDIT: Checked two tests on Lost planet

Here is one: http://techreport.com/articles.x/14424/4
http://techreport.com/r.x/phenom-x4-...lanet-cave.gif
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?)
08-11-2008, 06:05 PM
JumpingJack

Quote:

Originally Posted by gosh

If you are talking about games you should test a game that is communicating with the GPU and is using memory at the same time (there will be conflicts in the FSB that increase latency, amd has no conflicts there). Then you will see big improvements for AMD. More and more advanced games will run better and better on AMD. If the game is using more than one thread to render data then AMD will perform better.
I haven’t read what you have written about Lost Planet. I need to check some tests before I can tell you something about that game. How it scales, how it is using threading etc

Pick the game. I will test it.

Core scaling for Lost Planet:

http://forum.xcpus.com/gallery/d/660...reScaling2.jpg
08-11-2008, 06:06 PM
gosh

Quote:

Originally Posted by JumpingJack

Pick the game. I will test it.

Race Driver Grid

Show 100 post(s) from this thread on one page