9600XT vs siftmod 9500?

[blownsix]

Ok whitch one is better i bought a 9600xt and my buddy said he would trade me the 9600xt for a 9500s that can be softmoded for the 8 pipelines whitch one sould i go with would i notice a diffrence?

 

[harryinny3]

IF, and i say IF

That 9500 can be modded it will mod to 9700 performance. The XT is suppose to be close to 9700Performance anyway. But if i had the choice and that 9500 CAN be modded id take the 9500.


Harry

 

[El<(')>Maxi]

I think the key here is weather the 9500 he has is a 256bit bus version. If it is and it softmod's perfectly, and will run better than average overclocks, then it is a better card than the 9600XT.

Another good way to determine this would be to compare 3DMark '03 scores.

 

[ManOfKnight]

if you can get a 9500 that is guaranteed a softmod I would get it...as the 9600xts and the 9500's are now running about the same price. If you cannot get a guarantee...i would go 9600xt. Although you may not get the performance of the 9500 modded...you have a stable card that comes with HL2

 

[harryinny3]

Now usually

The soft mod will work. The soft mod is to Actually Enable the extra 4 pipes. The easiest way to do this is to use the Omega Drivers. They include the soft mod as an install option. After the mod is applied your card should read as a 9700 In Display properties.

Now, Even with the 8 pipes active, Your still going to be neck and neck with the 9600XT.

IF the 9500 has the L shaped memory pattern on the board, The cards Bios should be able to be flashed to 9700pro Speeds. That would make the Card Much more worth the Trade.

This trade is up to you, I cant make a choice for you about a newer card for an older card. Its your call. If the 9500 can be modded Your in the same Catagory with both. IF it can be flashed to 9700 specs, Your a big winner and You got the better of the deal.

Your call Man. Good Luck.

Harry

 

[james.miller]

a 9600xt isnt going to be as fast. Most places are showing the xt to be as much as 15% slower or more than a 9700pro when you turn the res up or use aa/af.

 

[harryinny3]

I was Saying

That after soft mod the 9600XT and 9500 modded will be neck and neck. Once flashed if it can be, It will destroy the 9600XT.


Harry

 

[james.miller]

not by the benchmarks. as ive said, everything ive seen has shown the xt to be slower than the 9700pro by upto 15% or more, and the difference increase with res or aa/af. It'll be very close without any aa/af at 1024x768, but i bet it wont be as fast with the eye candy on, and the 9700np isnt that much slower.


Everybody is welcome to prove me wrong though. Infact i'm willing to do some side-by-side benchmarks to see just what the difference really is.

 

[Matthias99]

The 9600XT is, indeed, significanly slower than a true 9700Pro.

However, it's about the same speed as a 9700 non-pro -- or, equivalently, a 9500 that's been softmodded. If you can softmod AND overclock a 9500 (thus turning it, effectively, into a 9700PRO), it will beat the 9600XT. But not all of them can be pushed that far.

 

[james.miller]

you just dont understand. the 9700's are more efficiant than the 9600's. They may well be similar in speed at stock 3dmark settings (1024x768 no aa or af) but with any eye candy the 9600xt's don't keep up. Every review i've read has shown this, and lets face it - who uses no aa or af thesedays?

if you want proof - look at amd me's 9600xt benches further down the page.

 

[Matthias99]

http://www.digit-life.com/articles2/digest3d/index.html

They don't have the 9700-series cards in there anymore, but the 9700Pro and 9800NP are very close in performance. The 9600XT doesn't seem to lose much ground to the 9800NP with AA/AF enabled, so I don't buy your argument.

 

[james.miller]

you want proof? ok then.

amd me's 9600xt.

http://www.ocforums.com/showthread.php?s=&threadid=253909

3dmark2001se best score:
http://service.futuremark.com/compare?2k1=7369693

UNMODDED Amd 1800+ [email protected]=2500mhz on water, UNMODDED Retail ATI 9600XT 558/364

3dmark:15100

The closet score i could find of mine own was this:

http://service.futuremark.com/compare?2k1=6829760

Softmodded r9500 @ 340/290. faster than a 9700pro core by 10mhz, slower ram by 40mhz.
Amd xp1800+ @ 214 x 10 = 2157mhz.

3dmark:15422

Now look at the difference in cpu speed and look at that overclock he's got on his 9600xt. Not convinced yet?

http://service.futuremark.com/compare?2k1=7070902

335/295 core/ram, 225x10.5 cpu

mine again. slower core than last time, slightly faster ram, but still slower than a 9700pro (295 vs 310 for the pro)

3dmark:16782

....And thats still below 2.4ghz on the cpu. now i am running 25mhz faster on the fsb, but that isnt going to bring me nearly 2k over him, especially when i'm losing over 100mhz on the cpu. Also note he was using a 16bit z-buffer.


STILL don't believe me?

 

[harryinny3]

LOL My whole point WAS(its lost now)

That The 9600XT Is ABOUT the SAME as a SOFT modded(NOT FLASHED) 9500NP. I Know full well A Modded and FLASHED 9500Np will Whoop up on a 9600XT. Because effectivly A Modded and flashed 9500NP Is a 9700PRO. Thats all i was trying to say.

And i will state that this 9600Pro im useing is Just as Even in benches as my buddys 9500Pro in Benches. He has Maybe 50 More 3Dmarks than i do in both benches. Same systems ext. I built both so they are exact.

Harry

 

[james.miller]

That 9500 i used for comparison is NOT FLASHED. I should know - its my card.

And i will state that this 9600Pro im useing is Just as Even in benches as my buddys 9500Pro in Benches. He has Maybe 50 More 3Dmarks than i do in both benches. Same systems ext. I built both so they are exact.

Harry

a softmodded 256bit r9500 is much faster than either the 9500pro or 9600pro.

 

[harryinny3]

His is a BBA 9500PRO

We did the benches when i built both systems at the same time. So you tell me Why They are almost even then?


Harry

 

[james.miller]

because a softmodded 256bit r9500 is NOT the same as a 9500pro, or 9600pro.

My softmodded r9500np has a 256bit memory interface, and 8 open pipelines. A 9500pro and 9600pro both only have 128bit ram and 8 pipeline. That's a big difference in memory bandwidth and if you you'd just look at those pics i posted, you'd see it......

the real question, as asked before, is if its a 256bit card or not. if it isn't, it'll be identcle to a 9500pro. If it is, it'll destroy the 9500pro's and all 9600's.

 

[Matthias99]

....And thats still below 2.4ghz on the cpu. now i am running 25mhz faster on the fsb, but that isnt going to bring me nearly 3k over him, especially when i'm losing over 100mhz on the cpu.


STILL don't believe me?

Uh, yes, I still don't believe you. You showed that a flashed-and-overclocked 9500 is 10-15% faster than a 9600XT, which is what I was saying!!!

*you* said (emphasis added):

you just dont understand. the 9700's are more efficiant than the 9600's. They may well be similar in speed at stock 3dmark settings (1024x768 no aa or af) but with any eye candy the 9600xt's don't keep up. Every review i've read has shown this, and lets face it - who uses no aa or af thesedays?

And I posted a benchmark showing that the 9600XT keeps up just fine with AA/AF on. What exactly is that post full of 3DMark01 (non-AA non-AF) benches supposed to prove?

 

[ColtIce]

the nature test tells me the whole story.it seems(to me)that the 256 bits really makes a difference in that test.

 

[james.miller]

you really dont listen do you?

that card IS NOT FLASHED. it's been running the same bios that it always has since the day i bought it. NOT FLASHED.

Your right - i did say "They may well be similar in speed at stock 3dmark settings (1024x768 no aa or af)", but looking at my comparison, i'll have to admit - i was wrong. The softmodded r9500's are still faster.

You see to be forgetting that his card is also overclocked, and overclocked a lot more than mine for that matter.

And I posted a benchmark showing that the 9600XT keeps up just fine with AA/AF on. What exactly is that post full of 3DMark01 (non-AA non-AF) benches supposed to prove?

that if im THAT MUCH faster without aa or af, how much faster would my card be WITH it. every review is pointing to around 20% at least - you do the math.


Another comparison.

http://service.futuremark.com/compare?2k1=7083613
q149's 9600 (another member here).

he actually beats me by 200 points, but look at his specs - 3.6ghz p4 @ 300mhz fsb, while im still running slower than a 9700pro with my little amd.

 

[Matthias99]

When did you even say you card wasn't flashed? You clearly had it overclocked to near-9700Pro speeds, so I assumed you had flashed it. Sorry. I should have said "softmodded-and-overclocked 9500" above.

I *really* don't like trying to extrapolate from 3DMark01 results, especially when the base systems are so different (I'd rather see someone with a similar box and overclock running a 9600XT, rather than someone with a vastly different box whose score is close to yours). But we can try.

In the first test, things are even more muddled because his card is overclocked and his CPU is higher than yours. However, your FSB is higher (which makes a big difference in 3DMark01). The only test you really beat him badly in is Nature, which is likely due to the higher memory bandwidth of your video card and system. I'd wager that if his video card was clocked closer to yours, it'd be 10-15% slower. Of course, there's no way to *tell*, which is the problem with this -- that's why I posted benchmarks from a site that tests the cards on the same machines.

In the second test, I have no problem attributing your lead to your 10% higher FSB -- and you're less than 2K ahead of him (1700 points, actually, just over 10% faster), not "3K higher". And again, most of your lead is in the Nature test, where clearly a high memory bandwidth is making a big difference. But I don't know how well that will translate into other games.

Do you have any 3DMark03 results? Those are much more video-card dependent and would thus provide a less biased view of performance.