Where are benchmarks for single core cpus + 8800 GTS/GTX ?

[ShreddedWheat]

I know that a single core cpu is going to be a bottle neck for the video card but for those of us that don't have limitless amounts of money for upgrading to a C2D, etc. what kind of performance can we expect to get with our current systems with a 8800 GTS for example until we upgrade to dual core cpu?

I currently have a Opteron 146 oced to 2.75 and 19" LCD so figuring 8800 GTS 320meg card would be sufficient. Currently have AGP x800xt @ 518/558

Links with Reviews and Benchmarks along with opinions are welcome.

 

[lavaheadache]

I was playing around 2 weeks ago with my rig. I swapped my back up 3400+ @ stock in my rig(in sig) and scored 5600 in 3dmark 06. I get 9600 with the FX-60. As you can see the 3400 was a huge bottleneck. BTW, this was in Vista HP, My XP scores were around 200 or so points higher

 

[ShreddedWheat]

That is a big difference in 3dmark....any thoughts/experiences on actual game play, etc....

 

[SPARTAN VI]

Originally posted by: ShreddedWheat
That is a big difference in 3dmark....any thoughts/experiences on actual game play, etc....

Right, you'll only perceive this bottleneck in 3dmark and if you benchmark at low settings in-game. I went from a X1900XT to a 8800GTS on my lowly 3700+ San Diego. I compared my X1900XT results with someone on a E6600 and a X1900XT. In Medieval II: TW, we set up the exact same custome battle. Same settings, same amount of units, same benchmark and almost identical performance (give or take 1fps). When I jumped up to a 8800GTS, it flat out stomped my old X1900XT score and was a stone's throw away from the 8800GTX guys.

I'll see what I can do about pulling up that thread.

 

[SPARTAN VI]

E6600 and X1900XT

1280x1024. Unit Scale Huge.
6xAA
16xAF
Shadows on
Bloom on
Reflections on
All settings maximum.

4 Units Landshnekt, 4 Units arqubusiers, 4 units Basilisk heavy cannon, 2 units halberd militia, 2 units Dismounted Gothic Knights, 3 Units Teutonic Knights, 1 Unit Generals bodygaurd

Versus,

random Scottish Full Stack. Around 4080 men in total.


Grassy Plain, Foggy, couldnt fix that unfortunatly.

Min 20
Max 42
Avg 27.9

This is my 3700+ and X1900XT at the same settings:


Frames: 8167 - Time: 300000ms -

Min: 11
Max: 43
Avg: 27.223

I'm guess the guy didn't zoom in on the action, caused my PC to "hitch" when zooming in explaining the 11FPS.

Here it was a 1680x1050 2xAA 8xAF (ie my normal settings)


Frames: 10834 - Time: 300000ms -
Min: 14
Max: 52
Avg: 36.113

 

[JPB]

I'd like to see some benches also with these GPU's

 

[SPARTAN VI]

Duh, I forgot the most important part of my test... the 8800GTS results.

8800GTS & 3700+: 1280x1024 8xAA 16xAF HQ, Huge unit scale, Max everything
2006-12-18 13:39:35 - medieval2
Frames: 11796 - Time: 300000ms - Avg: 39.320 - Min: 14 - Max: 60

X1900XT & 3700+: 1280x1024 6xAA 16xAF HQ, Huge unite scale, Max everything
2006-12-09 14:07:24 - medieval2
Frames: 8167 - Time: 300000ms - Avg: 27.223 - Min: 11 - Max: 43

X1900XT & C2D E6600: 1280x1024 6xAA 16xAF, Huge unit scale, Max everything
Avg: 27.9 - Min: 20 - Max: 42

And the units:

4 Units Landshnekt, 4 Units arqubusiers, 4 units Basilisk heavy cannon, 2 units halberd militia, 2 units Dismounted Gothic Knights, 3 Units Teutonic Knights, 1 Unit Generals bodygaurd

Versus,

random Scottish Full Stack. Around 4080 men in total.

 

[thegimp03]

Interesting...

I'd like to see what the difference would be on a game like Supreme Commander. I've heard that game absolutely kills cpus, but no experience myself.

 

[ShreddedWheat]

Spartan thanks for the benchmarks. Even with the 3700+ and 8800 that was a big jump.

I can understand that when a new video card comes out that review sites such as Anandtech, firingsquad, tomshardware, etc....want to use the highest end cpu to push the vid card but include some benchmarks with a middle of the road processor so that the majority of users can see what kind of gains they can get with a newer vid card until they can upgrade their processor.

My 2 cents

 

[TanisHalfElven]

Originally posted by: ShreddedWheat
Spartan thanks for the benchmarks. Even with the 3700+ and 8800 that was a big jump.

I can understand that when a new video card comes out that review sites such as Anandtech, firingsquad, tomshardware, etc....want to use the highest end cpu to push the vid card but include some benchmarks with a middle of the road processor so that the majority of users can see what kind of gains they can get with a newer vid card until they can upgrade their processor.

My 2 cents

not really. look at the min frame rates instead of the avg ones.
it shows how inmportant the cpu is.

 

[deadseasquirrel]

Originally posted by: tanishalfelven
not really. look at the min frame rates instead of the avg ones.
it shows how inmportant the cpu is.

In *that* game. It doesn't always work out that way. Bottlenecking isn't just about CPU power vs. GPU power. Resolution/settings and the particular game go into the equation as well. Had he run, for example, FEAR at 1600x1200, the results would be almost identical with the E6600 or the 3700+.

When determining if your CPU is going to be holding back your GPU, you need to consider:

1) your resolution and settings you play at.
If you have an LCD that limits you to 1280x1024, you'll find that a powerful card like the 8800GTX will be held back in many games by a slower CPU. LCDs pushing high resolutions such as 19x12 with AA/AF, will notice pretty much no benefit from a faster CPU in most games. Which brings us to this point...

2) what games you like to play?
RTS games such as Supreme Commander, the Total War series, etc will often see a very good increase in playability with a faster CPU, even at higher resolutions. FPS games usually don't, except in lower resolutions.

3) checking benchmarks for performance indicators.
Let's use Spartan's benches from above for Medieval2 as an example. Yes, the C2D with the XT has a better min framerate (20) than the 3700+ with the GTS (14), but the average framerate shows a 40% difference (the comparison is a tad off since 8xAA is enabled in one and not the other), in favor of the GTS/3700+ combo. That's quite a bit, especially for an RTS game at 12x10 res. IMO, if you have that 3700+ and an XT, you're better off going with the GTS as an upgrade than keeping the XT and getting a C2D, in this example.

For me, it all comes down to bang-for-buck. The cost of upgrading to a C2D, plus new ram and mobo far exceeds just buying the GTS and gaining 40% (again, using Spartan's example).

For ShreddedWheat, you'll see a huge gain going from an x800 to a GTS, even at your LCD's resolution. It will give you the best bang-for-buck upgrade right now. However, your single-core Opty, even at 2.75, will begin to give you problems as more games begin taking full advantage of multiple cores, if Supreme Commander is a sign of things to come.

 

[SPARTAN VI]

Originally posted by: deadseasquirrel

Originally posted by: tanishalfelven
not really. look at the min frame rates instead of the avg ones.
it shows how inmportant the cpu is.

In *that* game. It doesn't always work out that way. Bottlenecking isn't just about CPU power vs. GPU power. Resolution/settings and the particular game go into the equation as well. Had he run, for example, FEAR at 1600x1200, the results would be almost identical with the E6600 or the 3700+.

When determining if your CPU is going to be holding back your GPU, you need to consider:

1) your resolution and settings you play at.
If you have an LCD that limits you to 1280x1024, you'll find that a powerful card like the 8800GTX will be held back in many games by a slower CPU. LCDs pushing high resolutions such as 19x12 with AA/AF, will notice pretty much no benefit from a faster CPU in most games. Which brings us to this point...

2) what games you like to play?
RTS games such as Supreme Commander, the Total War series, etc will often see a very good increase in playability with a faster CPU, even at higher resolutions. FPS games usually don't, except in lower resolutions.

3) checking benchmarks for performance indicators.
Let's use Spartan's benches from above for Medieval2 as an example. Yes, the C2D with the XT has a better min framerate (20) than the 3700+ with the GTS (14), but the average framerate shows a 40% difference (the comparison is a tad off since 8xAA is enabled in one and not the other), in favor of the GTS/3700+ combo. That's quite a bit, especially for an RTS game at 12x10 res. IMO, if you have that 3700+ and an XT, you're better off going with the GTS as an upgrade than keeping the XT and getting a C2D, in this example.

For me, it all comes down to bang-for-buck. The cost of upgrading to a C2D, plus new ram and mobo far exceeds just buying the GTS and gaining 40% (again, using Spartan's example).

For ShreddedWheat, you'll see a huge gain going from an x800 to a GTS, even at your LCD's resolution. It will give you the best bang-for-buck upgrade right now. However, your single-core Opty, even at 2.75, will begin to give you problems as more games begin taking full advantage of multiple cores, if Supreme Commander is a sign of things to come.

Definitely. I'm almost certain if we benchmarked Rome: Total War with 10,000 units on screen, his E6600 anx X1900XT combo would wipe the floor with my 3700+ and 8800GTS combo simply because we're looking at the CPU topping out. In M2TW however, I was surprised. Identical results.

Now the minimum FPS, I can almost explain. There was a noticable split-second hitch during gameplay when I *zoomed* in on the action, then it returned to fluidity. I don't even have to tell you it was a split-second hitch, this is apparent in the data.. if it were longer than a second, it would've impacted my average much more.

Now I don't know if the guy with the E6600 zoomed.. for all I know, he was hovering far away from the battlefield just looking at sprites. I was going through the demo as if I were actually playing the game.