Let's talk about a G80 AMD vs. C2D benching contest [Post Benchmarks here :)]

[jivetrky]

A little background:
This subject was brought about by this thread


So here's the deal. Let's figure out the rules and guidelines of a purely friendly benchmark comparison of G80 based GPU's in both AMD A64 Dual Core and Intel Core 2 Duo based systems.

with this, we can find out how AMD systems run with 8800GTS/8800GTX cards(or SLI combo of either, which would probably be better for our tests).

This will be based on the fun of benchmarking with the side effect of giving us an answer to the question of, "Is a current AMD based system a serious bottleneck of a G80 card's performance?".

So let's figure out the rules and guidelines and then we'll start the real thread. I practically eat and sleep here at OCF, so I'd have no problem being the official records keeper in said thread.

 

[Rattle]

well I think we should try to establish at what resolution the bottleneck dissapears, then we run tests at that resolution, cpu's should be same clocks, same ram timings also.

 

[jivetrky]

well I think we should try to establish at what resolution the bottleneck dissapears, then we run tests at that resolution, cpu's should be same clocks, same ram timings also.

Well I'd think running at a resolution that removes the bottleneck situation would kind of negate the project wouldn't it?

I think the best bet would be to see at what CPU speeds the bottlenecks go away. Because according to Dom's findings (at least with FEAR) the bottlenecks went away on the C2D at around 3.4Ghz.

 

[Rattle]

I dunno I guess that makes sense lol, after 3.3 my stock 8800gtx doesnt do much different with increased cpu speed.

 

[jivetrky]

I dunno I guess that makes sense lol, after 3.3 my stock 8800gtx doesnt do much different with increased cpu speed.

I would think that would change depending on the game though, wouldn't it? I mean if a game was requiring more CPU power, the CPU would have to run faster to still remain a non-bottleneck.

 

[Rattle]

I would think that would change depending on the game though, wouldn't it? I mean if a game was requiring more CPU power, the CPU would have to run faster to still remain a non-bottleneck.

yes you are probably correct, I guess the best thing to do is run whatever games are decided at 4 diff clock speeds, 8800s stock and diff resolutions perhaps?

 

[jivetrky]

yes you are probably correct, I guess the best thing to do is run whatever games are decided at 4 diff clock speeds, 8800s stock and diff resolutions perhaps?

Yep, once we get a few more opinions in here we can figure out which speeds and resolutions would be good.

 

[RollingThunder]

well I think we should try to establish at what resolution the bottleneck dissapears, then we run tests at that resolution, cpu's should be same clocks, same ram timings also.

Rattle,

I'm not very good at this because I don't game or run heavy graphics.

I was thinking more of a "Bring what you got" for either CPUs. I think it's asking too much to have everyone change their timings and overclocks. I realize this somewhat skirts the direct question but I think it's still relative to the Intel and AMD guys to crank it up to whatever you have. Both CPUs are certainly cabable of doing so and if the Core2 Duo really is that much faster it will show overall.

No one here that I see runs their 8800 cards and CPUs stock and it's more "real world" than a regimented Anand type hardware test. I don't know what software tests would evaluate similarly with the games.

I think in the end a pattern will still be obvious. I really don't know how to conduct something like this. Let them have at it and post their "crowing rights." Am I way off base here?

 

[brakezone]

yeah, instead of making conclusions lets get some data in here!

 

[ChinStrap]

this is the kinda stuff OCF should be game for

BTW

If this turns into a flame war, I'll start cracking heads.
PLEASE keep this civil.

-- David

this was great

i'll post what i can. i have a 8800GTS, but shes paired with a D920 @ 3.56Ghz.... like i said tho.... ill do what i can

 

[jivetrky]

Well, just a "Bring what you got" kind of thing would still be fun, but it doesn't really answer the underlying question/debate.

I guess the only thing it will/could show is that an AMD processor at X speeds can run Y game at Z speed. Which would tell ppl that yes, indeed, this AMD system will run the G80 and get more than adequate frames.


But I still think it isn't the original vision of the "contest"

 

[RollingThunder]

this is the kinda stuff OCF should be game for

BTW


this was great

i'll post what i can. i have a 8800GTS, but shes paired with a D920 @ 3.56Ghz.... like i said tho.... ill do what i can

Chin,

I really don't think would ever turn into a preliminary flame war for two reasons:

1. The mods wouldn't let it.
2. The data will speak for itself.

I'm really curious myself if nothing else. I don't take it too seriously.

 

[jivetrky]

I think it'll just be a fun way to try to prove the point yay or nay.

I think most ppl around here are good enough to keep it friendly. But if not their flaming will be reported and hopefully removed to keep the thread on track.

 

[RollingThunder]

Well, just a "Bring what you got" kind of thing would still be fun, but it doesn't really answer the underlying question/debate.

I guess the only thing it will/could show is that an AMD processor at X speeds can run Y game at Z speed. Which would tell ppl that yes, indeed, this AMD system will run the G80 and get more than adequate frames.


But I still think it isn't the original vision of the "contest"

Jivetrky,

You're correct of course but there is no way you could establish logistics, rules, and software requirements everyone would agree too. It's ain't gonna happen because there are too many valid arguments. I think with enough participation and varying hardware (to a point) you'll see enough to draw some valid conclusions.

To answer the original thread that began this thread, perhaps a stricter sets of baseline rules could be established for a test that strictly addresses that original post after this competitive thread runs a while.

These guys here are as good as any of the commercial web sites that do similar testing, I'm sure of that!

 

[ochungry]

As RT says, lets not limit it to 8800 cards. Not too many have that card. A lot of us are budget overclockers and this thread should invite all of oc'ers here, top end, mid, and low ends.

 

[Rattle]

lol

well this was sposed to be about 8800;s i thought

whatever, I got a 7900gto sitting here if thats budget enuff

 

[deathman20]

Clock wise should get benchs for the CPU's from...

Say 2.2-2.8Ghz (100Mhz increase) and go from there at least people with the big cards for either AMD or Intel should have no problems hitting the majority of the speeds.

 

[RollingThunder]

As RT says, lets not limit it to 8800 cards. Not too many have that card. A lot of us are budget overclockers and this thread should invite all of oc'ers here, top end, mid, and low ends.

OC, Rattle, Jivetrky,

Since I am very ignorant about graphics cards, it's logical to get the other high ends cards in on this. I agree not everyone has a $500 8800. Maybe the 7900 guys can also show what they have. It may be a good comparison for performance and what extra you may get for $500.

We need to title a thread that will catch the attention that there is some real (friendly) competition between the Intel / Core Duo / AMD high end graphics cards use.

By the way, it's for sure I won't participate!

 

[ziggo0]

I like how all these threads that pop up regarding AMD vs. C2D wind up with every reply (almost every) flaming on AMD. Makes me laugh even more that there are some in the AMD Section as well..

...however, in simple terms, higher the resolution the more the game will become dependant on the GPU than the CPU. Lower the resolution the more the game will become dependant on the CPU, than the GPU.

 

[brakezone]

well if you run 2X8800's in sli with the SLI performance indicator on, do the bars indicate that the GPU's are not being used 100%??

Sometimes in games like GTR2 while SLI is on, I can see the performance bar indicating the GPU is hardly working, which probably means the bottleneck is on something other than the GPU.