8800GTX in X2 4400?

[TheUnk]

Would I be limiting the 8800GTX by using this processor or would I do just fine? Prefer to game in 1680x1050, mainly FPS types.

 

[Nightmare225]

I think you'll be ok. At that high a res, you won't really be CPU-limited.

 

[BenchZowner]

Corrected...at 1680x1050 and at least 2xAA/8xAF you won't have any 'real' CPU limitation

 

[Fadey]

I got a 4600+ @ 2.7 and love my 8800gtx with it.... i get 175 fps in cs at 1920x1200 16xcs aa and 16xaf under vista my mate with a 6700 at same settings get like 180ish. So unless you really need those few fps you should be fine.

 

[TheUnk]

Well that's good to know.. Now just to decide if I should wait for prices to go down or buy now >< A few games out their I'm holding off on playing cause my current card can't run them so well. A little hard justifying $600 to play games on

 

[LOUISSSSS]

you'll be fine. yea i can't justify $600 video card on top of the price u paid on the rest of ur parts just to play games... its just not worth it...

 

[ArchAngel777]

Originally posted by: BenchZowner
Corrected...at 1680x1050 and at least 2xAA/8xAF you won't have any 'real' CPU limitation

I just have to say I really like your website layout and the reviews you run... Absolutely some of the best reviews I have read... I really like how you test many resolutions (even the low ones) and test all various settings of AA/AF. It really makes for an extremely thorough review. Major thumbs up.

 

[Zenoth]

I got an X2 4400+, at 2.4Ghz, from 2.2Ghz at stock, and the CPU is bottlenecking the card (GTS), so I would imagine that a GTX would also be bottlenecked by it. But one thing I never noticed is if I had to increase my resolution to be "less limited" by the CPU. I usually play at 1280x960, since my monitor cannot go much higher (it can, but then I have to reduce the refresh rate below 85Hz and it gets annoying on my eyes, it's a CRT, so I need to couple with that until I get an LCD). I know that 3DMark05 and 06's default benchmark resolution is 1280x1024, and even at that resolution it is bottlenecking the card. Unfortunately my 4400+ is a lemon, it refuses to over-clock above 2.42Ghz without crashing and such, or else it'd be running at 2.8Ghz, at least.

 

[TheUnk]

How significant is the bottleneck for you, Zenoth? A few fps?

 

[Matt2]

I wouldnt worry about the bottleneck. Zenoth is running a lower resolution and plus 3Dmark scores are waaaay too dependent upon the CPU.

In the real world with some AA/AF, I dont think you'll see much of a difference.

 

[betasub]

C'mon Zenoth, you can't just say something is "bottlenecking" without stating the app/game. How about FEAR, Oblivion or STALKER?

What are you using to judge this bottleneck? Don't say 3DMark, because (as Matt2 has already said) it includes CPU-only results (non-GPU accelerated) in the final score.

 

[MarcVenice]

I've seen reviews where they swapped overclocked c2d's at 3.4ghz out for a x2 4400+ and the results in benchmarks didn't differ much if at all. Most people will tell you that it's all about the gpu, not the cpu. If you're a gamer, skimp on the cpu, spend big bucks on the gpu.

 

[BenchZowner]

Originally posted by: MarcVenice
I've seen reviews where they swapped overclocked c2d's at 3.4ghz out for a x2 4400+ and the results in benchmarks didn't differ much if at all. Most people will tell you that it's all about the gpu, not the cpu. If you're a gamer, skimp on the cpu, spend big bucks on the gpu.

You can easily 'create' a difference performance-wise between a Core 2 Duo E6600 and a Athlon 64 x2 4800+...easily.
Run the game at 1024x768 without Anti-Aliasing & Anisotropic Filtering.
You'll have at least 10 fps difference, and depending on the games' dependency on the CPU you can have a huge gap as in 30fps or so.

Then, run the game at 1280x1024 4xAA / 4xAF and come back to me stating the...almost inexistent performance gap

 

[daveybrat]

You shouldn't waste your money on an 8800GTX anyways IMO. Just buy the 8800GTS 320 or 640MB version. That should be more than adequate for your processor and the games you play and also won't set you back to much financially.

 

[Zenoth]

Ok guys relax, I'll give you numbers ...

Using FRAPS's benchmark system (average / minimum / maximum frames-per-second):

-Oblivion + Better Distant LOD Textures + OOO 1.32 + Qarl's Textures Pack 2
-All in-game settings maxed except for no Shadows Filtering (all the Shadows are on, just not the filtering for it), using 4xMSAA and 16xAF, with HDR
-Location is Imperial Prison Sewer Exit looking towards the first ruin on the other side of the river. The in-game weather is clear, sunny day.
-Same system speeds as sig except for varying GTS speeds (for testing purposes)

-GTS Stock Frequencies (513Mhz / 792Mhz)

1024x768 @ 100Hz = 22 / 16 / 28
1280x960 @ 95Hz = 26 / 19 / 34
1600x1200 @ 75Hz = 21 / 14 / 24

-GTS over-clocked at 620Mhz / 920Mhz

1024x768 @ 100Hz = 24 / 19 / 30
1280x960 @ 95Hz = 27 / 21 / 35
1600x1200 @ 75Hz = 22 / 16 / 25

Method for benchmarking was the following: I load up the game at Sewer Exit, leave character there in first-person-view, start the benchmark, leave it running for more or less two minutes, and by itself the game will get out of first-person view, go into third person (after a while if the player does nothing) and then the camera will automatically rotate around the player, showing a greater area all around in a 360º turn. Then I usually stop the benchmark after one or two such camera turns are done.

Here a screen-shot for this thread, taken while benching: http://img508.imageshack.us/im...761/screenshot1ix6.jpg

It shows that at lower resolutions (1024x768) I am getting a lower frame-rate than at a higher resolution like 1280x960, which means that at lower resolution I am CPU limited. Then from 1280x960 it doesn't go much up in frames even if I over-clock the card significantly, which means that, I will repeat it, after a certain resolution, whatever I do to squeeze more juice out of my card it won't work much since the CPU is bottlenecking the whole thing. I've used the FRAPS benchmark in all of my personal tests and I can say with certainty that my 4400+ isn't enough for my GTS, even OC'ed by an extra 200Mhz.

Whoever asked for a FEAR benchmark I'm sorry I can't give I don't have that game. I got a bunch of others though. If the topic creator wants more benches from me just ask and you'll see what I'm talking about. And I mentioned 3DMark because it shows the same thing, after a certain resolution the score only goes up by 10 Marks or 20 Marks even with significant over-clocks on the card, the 4400+ at 2.4Ghz forces a "wall" upon the card after a certain resolution, whether it is 3DMark or Oblivion or FEAR ... it's going to do the same thing each time.

 

[SolMiester]

Originally posted by: TheUnk
Well that's good to know.. Now just to decide if I should wait for prices to go down or buy now >< A few games out their I'm holding off on playing cause my current card can't run them so well. A little hard justifying $600 to play games on

Try NZD $1100-$1200 mate, you guys have got it good!!

 

[TheUnk]

http://www.overclockers.com/articles1390/index02.asp

 

[ArchAngel777]

Just using mathematic alone tells me your numbers are off. You are doing something wrong, and/or oblivion or the nVidia drivers are doing something funky.

Under NO circumstance will 1024 x 768 offer inferior performance than 1280 x 960. You are pushing fewer pixels. 786,432 versus 1,228,800. Almost double the ammount of pixels it must push through and you are reporting that 1,228,800 is FASTER than 786,432. Sorry, man, something is clearly wrong with your test. That is a good reason why FRAPs is not the most reliable method for benchmarking, especially in a game like oblivion where even a slight difference of a view will affect frame-rates. You need to use something that is consistant all the way around (HL2: Lost Coast for example).

Also it is the shader clock that matters on the GTS, not the core clock. Certain core clocks increase the shader clock. In some cases, the GPU will slow down if you overclock to high. Such is the case with my 7800GTX. It actually performs WORSE when I overclock it, far worse.

A 4400 X-2 is a perfect fit for a GTS-640.

 

[TheUnk]

Seems to me that it all depends on the game as to whether or not the CPU turns into a bottleneck. And in most cases when it does, it's not really a significant difference in FPS.

 

[ArchAngel777]

Originally posted by: TheUnk
Seems to me that it all depends on the game as to whether or not the CPU turns into a bottleneck. And in most cases when it does, it's not really a significant difference in FPS.

Yep, that about sums it up.

On a side note, I am not sure why people are obessed with not being CPU bottlenecked. In most cases a CPU + Ram + Motherboard is FAR more expensive than replacing the graphics card. Look at this way, you are either going to be CPU limited or GPU limited. The end result? The end result is that something is bottlednecked.

The only reason to be concerned about a CPU bottleneck is when you are currently CPU bottlenecked with your current GPU and (qualifier) are planning to upgrade the GPU. In other words, if you truly are CPU bottlenecked with a 7800GTX, then it would be foolish to purchase a 8800GTS, as it would not increase performance at all.

 

[chinkgai]

Originally posted by: TheUnk
Seems to me that it all depends on the game as to whether or not the CPU turns into a bottleneck. And in most cases when it does, it's not really a significant difference in FPS.

this statement, is what i agree with.