Dual vs Quad - Lab Benchmarks Vs Reality?

[Emo]

Can you selectively disable the cores on a quad? Say 1 and 4 vs 1 and 2, then benchmark it and see if there is a difference. I thought the Q9550 is just two C2D slapped together with shared 6MB L2 between each pair.

 

[Modelworks]

Comparing multiple cores using examples like antivirus and other programs installed on the pc as a excuse to get multiple cores really doesn't hold up unless those programs are something that is running constantly. Things like AV usually doesn't consume much at all of the processor time unless it is explicitly being used to do something. Many of the programs loaded on a pc just sit in the background using memory, but very little cpu time. Load up all your background apps and then pull up task manager while the programs are loaded but not being used directly by the user to see if multiple cores really will make a difference in things like that for you.

I use quad cores for rendering 3d scenes . It does not max out all cores at 100%. While one core is rendering another is getting data to render, so the usage is all over the place. It is faster with quad cores, but usage is not anywhere near what benchmarks show because they are usually not real world scenarios.

One benchmark that is close to real world is cinebench.
http://www.maxon.net/pages/download/cinebench_e.html

When I run the test as multiprocessor using 4 cores is 3.43 times as fast, not 4 times as fast as some people would think because of the overhead of shifting around data.

 

[taltamir]

Originally posted by: Emo
Can you selectively disable the cores on a quad? Say 1 and 4 vs 1 and 2, then benchmark it and see if there is a difference. I thought the Q9550 is just two C2D slapped together with shared 6MB L2 between each pair.

go to task manager (ctrl+alt+del). Right click on a task (say, game.exe) and select "thread affinity"
You can choose which cores it is allowed to run on. By limiting it to only two cores... its not 100% accurate because the other two cores can still run all the background tasks and there is the issue of shared cache

 

[cusideabelincoln]

Originally posted by: taltamir
i did publicize it, you simply lack the technical understanding to comprehend them. The result is that there is OBSERVABLE MICROSTUTTER thatis PROVEABLE based on frame dumps in fraps, that microstutter is observed at 1920x1200 resolution AND at 720x480 resolution and anything in between, that microstutter is observable using an E8400, OC to 3.6ghz, and at quads with two cores disabled. it is observable on a 4850, a gtx260 and other video cards.

And that simply going to a quad core fixes the microstutter.

I don't know what more do you want from me, unless you want me to post the entire framedumps...
The frame dumps are large and unwieldly and it is impractical to post them.
IF you are saying I am lying, no point, they could easily be faked.

So really, Nothing further to discuss here.

I love how you call me out when you are the one not understanding what I am saying. What a fucking hypocrite. But I'll repeat myself just one more time:

1. "Publicized" does not mean a forum post spread out over multiple replies. Like I said I am not wading through 70 posts of bullshit and bickering. Post your findings continuously - make a new thread or hell make a Myspace page for all I care (I don't, though).

2. We are arguing for the same thing. Whether or not one specific game - Mass Effect - is a part of the data is irrelevant because the list is not all-inclusive nor comprehensive, and it was never meant to be. It's just a demonstration.

3. And quite frankly I only wanted to use trustworthy sources that showed significant gains. SIGNIFICANT! The fact that PCGH's results show little difference in a quad over a dual means I do not trust you.

 

[Modelworks]

Originally posted by: cusideabelincoln

I only wanted to use trustworthy sources that showed significant gains.

I can show you one title that has added multicore support and it does make a difference. Team Fortress 2 .

This is a test I did the other day when they enabled multi-core support. http://i44.tinypic.com/ftk93b.jpg

Nothing else was loaded, no AV, firewall software or anything running but the game.

Usually a game that is not smp will show 25% across all 4 cores.
TF2 showed nearly 100% on core 0, and about 40-50% on the remaining cores. None of the cores ever dropped below 20% so even with just the remaining 3 cores usage would have been at least 60% on the second core of a dual core cpu. There are points when the other 3 cores total 150% so that would have been beyond what a dual core could handle.

 

[apoppin]

Originally posted by: Modelworks

Originally posted by: cusideabelincoln

I only wanted to use trustworthy sources that showed significant gains.

I can show you one title that has added multicore support and it does make a difference. Team Fortress 2 .

This is a test I did the other day when they enabled multi-core support. http://i44.tinypic.com/ftk93b.jpg

Nothing else was loaded, no AV, firewall software or anything running but the game.

Usually a game that is not smp will show 25% across all 4 cores.
TF2 showed nearly 100% on core 0, and about 40-50% on the remaining cores. None of the cores ever dropped below 20% so even with just the remaining 3 cores usage would have been at least 60% on the second core of a dual core cpu. There are points when the other 3 cores total 150% so that would have been beyond what a dual core could handle.

now let's see your graph with a Dual core running it - for comparison purposes

and it makes ZERO *practical* difference that TF2 has multi-core support as anyone with a reasonably fast GPU and a dual core will have GREAT frame rates - maxed out .. what does Quad over dual give you at normal resolutions - from 10x12 to 19x12?
- no one cares about 10x7[period]

"more maxed out" for TF2?

Use World in Conflict as a much better example .. there a Quad is playable at the same resolution a Dual chokes
- 19x12 and 16x10; with maxed out details even plus 4xAA/16xAF
-- but you better have really fast graphics
[want proof? = click my sig's link; it is well detailed and charted there, i think]

 

[Modelworks]

Originally posted by: apoppin


now let's see your graph with a Dual core running it - for comparison purposes

and it makes ZERO *practical* difference that TF2 has multi-core support as anyone with a reasonably fast GPU and a dual core will have GREAT frame rates - maxed out .. what does Quad over dual give you at normal resolutions - from 10x12 to 19x12?
- no one cares about 10x7[period]

"more maxed out" for TF2?

Actually that isn't what people are reporting. Before multicore support people were getting hitching and stuttering on servers that have high tick rates and 32 players. Now it is always smooth , even in the most extreme of fights.

This was at 1680x1050, 8xAA, all options at max.

I'll do a dual core comparison later and graph it for a comparison. I'll set TF2 to run on cores 2 & 3.

You should consider graphing cpu usage as well as frame rate. It is the only way to see how the cpu cores are being used.

I was running windows 7 but vista has almost the same graphing functions built in so no software has to be installed.

 

[taltamir]

Originally posted by: cusideabelincoln

Originally posted by: taltamir
i did publicize it, you simply lack the technical understanding to comprehend them. The result is that there is OBSERVABLE MICROSTUTTER thatis PROVEABLE based on frame dumps in fraps, that microstutter is observed at 1920x1200 resolution AND at 720x480 resolution and anything in between, that microstutter is observable using an E8400, OC to 3.6ghz, and at quads with two cores disabled. it is observable on a 4850, a gtx260 and other video cards.

And that simply going to a quad core fixes the microstutter.

I don't know what more do you want from me, unless you want me to post the entire framedumps...
The frame dumps are large and unwieldly and it is impractical to post them.
IF you are saying I am lying, no point, they could easily be faked.

So really, Nothing further to discuss here.

I love how you call me out when you are the one not understanding what I am saying. What a fucking hypocrite. But I'll repeat myself just one more time:

1. "Publicized" does not mean a forum post spread out over multiple replies. Like I said I am not wading through 70 posts of bullshit and bickering. Post your findings continuously - make a new thread or hell make a Myspace page for all I care (I don't, though).

2. We are arguing for the same thing. Whether or not one specific game - Mass Effect - is a part of the data is irrelevant because the list is not all-inclusive nor comprehensive, and it was never meant to be. It's just a demonstration.

3. And quite frankly I only wanted to use trustworthy sources that showed significant gains. SIGNIFICANT! The fact that PCGH's results show little difference in a quad over a dual means I do not trust you.

1. You don't NEED to read 70 posts, just the following paragraph:

The result is that there is OBSERVABLE MICROSTUTTER thatis PROVEABLE based on frame dumps in fraps, that microstutter is observed at 1920x1200 resolution AND at 720x480 resolution and anything in between, that microstutter is observable using an E8400, OC to 3.6ghz, and at quads with two cores disabled. it is observable on a 4850, a gtx260 and other video cards.

And that simply going to a quad core fixes the microstutter.

Thats it, thats the result, that is all that is needed. I also posted a few other snippets at the same posts with more specific data... What ELSE could you POSSIBLE want that would ANYTHING to the discussion? You want me to use word to make some charts? maybe in color? tell me what data is MISSING specifically?

2. I don't consider arguments and discussions as a war where I must win and form alliances to do so. I try to learn and teach. That means that when "my side" has a bad argument, I point it out, when I am shown my argument is wrong I "switch sides", and when I feel that there is nothing for me to learn to teach I just stop doing it. Victory or Defeat are not the purpose.

3. Extremely annoying microstutter, that spurred me to upgrade my hardware, is significant. Reliable? well, that is a judgment call each person makes based on what they discuss, anyone can make a site or write a book, what matters is what that person says and how they demonstrate their knowledge and integrity. I think I am both knowledgeable and honest, as well as the people whom I discussed it with in the thread I posted. (it isn't 70 posts of bickering, its mostly analysis, there are just two stupid trolls who were pestering us there) if you disagree then feel free to ignore my findings.

 

[apoppin]

Originally posted by: Modelworks

Originally posted by: apoppin


now let's see your graph with a Dual core running it - for comparison purposes

and it makes ZERO *practical* difference that TF2 has multi-core support as anyone with a reasonably fast GPU and a dual core will have GREAT frame rates - maxed out .. what does Quad over dual give you at normal resolutions - from 10x12 to 19x12?
- no one cares about 10x7[period]

"more maxed out" for TF2?

Actually that isn't what people are reporting. Before multicore support people were getting hitching and stuttering on servers that have high tick rates and 32 players. Now it is always smooth , even in the most extreme of fights.

This was at 1680x1050, 8xAA, all options at max.

I'll do a dual core comparison later and graph it for a comparison. I'll set TF2 to run on cores 2 & 3.

You should consider graphing cpu usage as well as frame rate. It is the only way to see how the cpu cores are being used.

I was running windows 7 but vista has almost the same graphing functions built in so no software has to be installed.

Whenever i hear of improvements in MP-only games i am suspicious of the many other factors that are involved playing on the Internet.

i am looking for the possible practical improvements of running Quad over dual in a game; ultimately for me it is not the CPU-usage chart that says anything, but the frame rates at the resolution i play that most interests me. One of our other editors did a CPU usage comparison of TF2 and also concluded it was using all 4 cores.

EDIT: One other thing, is i really a fair comparison to disable 2 cores of a Quad? it is not a comparison to e8600 as the cache is now different and other programs may be running on the quad's "extra" cores

 

[Tempered81]

its pretty sad that people still have to argue the viability of using a quadcore for games. Its been 3 years now...

This is a never ending argument until all new games perform like fc2 and gta4.

 

[Modelworks]

Originally posted by: apoppin
[

EDIT: One other thing, is i really a fair comparison to disable 2 cores of a Quad? it is not a comparison to e8600 as the cache is now different and other programs may be running on the quad's "extra" cores

I don't have any other way to do it since all I have on my pc are quad cores.

 

[exar333]

Originally posted by: Scholzpdx

Originally posted by: taltamir

You don't encode videos, run your spyware and defrag, stress test your cpu, install your games at the same time you game. When you game, all you do is game. If not, you're a fool.

Torrents are the most convinient example of other things...
Spyware / other shit is the most COMMON exaple of background tasks that most people don't have the knowledge or skill to get rid off...
Anti virus - something that is NEVER used in benchmarks, but is always there for most people.
Auto windows defrag - always on unless specifically disabled
windows indexing - always on unless specifically disabled.

Real life is very different from benchmarks... I wanna see a game benchmarked with all this crap in the background.. every single one of the things I listed... and maybe a few i haven't.

Good point. I ALWAYS have Avira, AIM, Steam, VENT, Fraps, GPUz and RealTemp while i am gaming.

Why not? If that is what you want to run while you game, and your system is beefy enough to handle it, awesome. When I had a single-core machine years ago, doing a virus scan, encoding, or buring a CD/DVD tied up the whole machine. With dual core, I can do other tasks while these are running. On my computer now, I do MORE of those things and still play 90% of games with no issues. If you really need the max fps in Crysis or something, then you can shut down the background processes.

Your arguement is the same as those that said "dual core is a waste, I just leave my computer sit when I do anything processor intensive."