Advice on what graphics card wont be bottle necked by my CPU

[jamesro]

Tired of waiting for an article to come out on this website

wondering if any of you guys can help me

I have a Pentium 4 @ 3.2GHz on an Asus P4S800MX-SE with 2GB of DDR400 RAM and atm
its got a XFX 7600 GT that i have had for a couple of years now.

im wondering, to kick on this computer for a little longer is there a higher graphic card option
such as the X1950 Pro, XT or XTX, OR 3650 i know there is the 3850 as well now,
which card would not be bottle-necked by the CPU and AGP Bus.

I read the early '07 article that the X1950 wasnt bottlenecked @ 4x AGP when compared to AGP @ 8x FPS in F.E.A.R

and this was with Athlon 64 3400+ as the CPU

many thanks in advance

 

[-Fran-]

I'd say that none of those cards will be bottlenecked by your CPU. A 4850/8800GT would make your CPU bottleneck them a little, but i say you wouldn't even notice.

Just get the best one u can get (3850?) and have a blast gaming.

Esop!

 

[L1qu1d]

actually they can bottleneck the cards, if the game was optimized for multi-core. Running programs in the background might also diminish the experience.

 

[jamesro]

do that with my current 7600gt or the next card i get?

 

[-Fran-]

do that with my current 7600gt or the next card i get?

With your next card.

But if you plan on getting a new CPU, don't buy the video card now, since AGP is no longer used at all.

Save up a lil' money and you could buy an AMD system that will perform very well with *really* tight budgets. And i'd say any AMD system is an upgrade from were u stand.

If you don't want to update your whole rig now (or even in a year), get the 3850 IMO. But keep in mind that you'll waste that money since you won't keep the video card for your next build.

Esop!

 

[mactronix]

There is a big differance between the AGP bus being a bottleneck and the CPU being the botttleneck,also the Athlon 64 3400+ is a way better CPU than a Pentium 4 @ 3.2GHz.
The whole bottlenecking thing is a bit misleading anyway, all it basically means is that the CPU wont let the card perform as well as it could with a faster chip. It would mean you can use more features AA/AF etc before it slowed you down.
As Yuka says if you can afford it get the 3850 that way you wont be wondering "what if" Make sure your power supply can handle it though.
Mactronix

 

[jamesro]

MMM

I ran the stress test on CS: Source at different resolutions
and paid close attention to the framerate.
and the frames seemed to stay the same at the same points.
and the overall averages were within 2fps of each other

im guessing this is showing CPU bottleneck?

 

[JAYDEEJOHN]

Back in the day when these cards were new , the 1900 being top dog, they had an article saying a AMD cpu clocked at 2.2 would max the cards out, meaning the cpu would get the best out of the card at that cpu speed. Id guess your cpu close to a 2.0Ghz AMD cpu of that era, so it may not get everything from your card, but itll be close

 

[mactronix]

Can i ask what resolutions you used ? I think JDJ is about right i had a system that was similar to yours and it seemed like mine was balanced on the brink just like yours seems to be.

Mactronix

 

[jamesro]

i typed in fps_max and in console it automatically came up with fps_max 100 which i had never typed before
but changed it to 300 for these tests here.

i used 1920 by 1080 -- everything on HIGH and 4xAA i got 51.24 ave fps --- and 53.22 ave fps with 4xAA disabled
all the way down to like 848 480 ----- 52.96av fps with 4xAA not that it mattters. ----4xAA disabled 51.42av FPS
and in between i used 1360 x 768 ---- 52.91av fps


note that during the test the fps hit the high mark *110-115fps* at the same point in the test every time.

 

[jamesro]

is JDJ talking about this article?

http://www.tomshardware.com/reviews/agp-platform-analysis,1533.html

 

[JAYDEEJOHN]

Not that exact one, but its close. It gives you your ballpark performance. I cant find the one Im refering to, but in that one, they used a FX60 and downclocked it all the way down to like 1.6Ghz, ran all the cards in the test at each setting, 1.6/1.8/2.0/2.2 etc until the bottleneck was removed by the cpu. At 2.2 Ghz, it was found that was the cutoff speed from the cpu as being a bottleneck

 

[jamesro]

ohk

so am i seeing cpu bottleneck 100% in the Source Stress Test?

 

[JAYDEEJOHN]

Compared to? You say 100%, compared to what?

 

[jamesro]

compared to a chance of it being graphic card limitation

 

[JAYDEEJOHN]

Yes. The cpu would be the only bottleneck. You have plenty of ram, and the cards youve listed, only the 1900 or the 3850 would be seen in a cpu bottleneck

 

[jamesro]

im referring to the fps on different resolutions i submitted a few posts ago
in regards to my 7600gt and cpu

 

[mactronix]

im referring to the fps on different resolutions i submitted a few posts ago
in regards to my 7600gt and cpu

Am i understanding correctly that you are using a stress test or even a specific game test at different resolutions to try and see if you have a CPU bottleneck ?
If so i would strongly recommend that you don't use a stress test or a game test as its designed to stress things and find limits, which while that is what we are doing i think it best to test for these things in actual game play as the tests will be putting artificially high demands on the hardware.

I would make a save in a game and then run through for a bit using fraps to note both your high and low FPS. Do this at different resolutions and you will get a real world idea of where your hardware is at. The AA etc doesn't really matter, if you do include it it may give you more detail or it may just cloud the results.


Mactronix

 

[JAYDEEJOHN]

OK, Macs idea is perfect. Didnt know it was a stress test either. And no, your cpu shouldnt bottleneck your 7600GT. Your GT was a fine Midrange card in its day, but still a mid reange, you had to use a top of the line or near it, to see bottlenecks from a cpu. But Macs idea is correct, do that, and youll see, if any at different resolutions, a true cpu bottleneck, if 1 exists, but shouldnt IMHO

 

[jamesro]

ok in STALKER
how do i bring up the framerate info
like in Half-Life etc it is net_graph 3
not sure what it is in stalker

 

[JAYDEEJOHN]

Go here http://www.tweakguides.com/ It gives great advice in many games, and even for xp and vista too

 

[mactronix]

Download FRAPS its a free download that will give you a FPS counter which you can use across differant games, just note the high and low FPS. We dont have to be 100% accurate here, if its that close that you need to be within a frame or two to see the differance then you are pretty much restricted anyway.

Mactronix

 

[jamesro]

thanks for the advice guys

ill let you know how it all goes

 

[jamesro]

tested it out playing stalker
once unplayable to me... tweakguides showed me the way. thanks heaps jdj.

at the lowest resolutions around 820 by 480 the fps hover around the 50-60mark
and then at 1920 by 1080 the frames hover around the 40-50mark rarely touching the low 40s.

both resolutions can hit the 80s in the same easy scenarios of within a room.

i think stalker would be one of those CPU intensive games and i think its showing the CPU bottleneck before the card might be restricted

what is your analysis.

 

[jamesro]

am i seeing CPU limitations in the lower resolutions
and graphic card limitation in the higher resolution
not sure how to interpret that