Feeble 9600XT even stomped by old 9500Pro!

[Blastman]

You should have kept the 9600 Pro over the 9500 ?

No question about it. From a performance point of view I see no reason to get a 9500 unless you?re thinking of a softmod. A 9500 can almost match a 9600pro in 3Dmark2001 and a 256bit 9500 is probably ~ 10% faster, so it would probably outscore a 9600pro in that benchmark. But when you look at actual gaming scores the 9500 = 9600. ? ? link A 9600pro is 35 - 40% faster than the 9600 so it?s still far ahead of any 9500. The 9500pro was a temporary solution for the midrange market as it was too expensive to produce.

Power consumption is still an issue for a lot of people. Most people buy off the shelf computers and I would think a lot of them only have 200 - 250w psu?s which may rule out a top end graphic cards. A 9600XT could be an ideal low power consumption solution in these cases.

 

[Corporate Thug]

Originally posted by: i82lazyboy

Originally posted by: shady06all the hard article proves is that the 9600 XT is overall a better card than the 9500 pro

yes, well... how old is a 9500 Pro and how young is a 9600XT? Not to mention the 9600XT is just a overclocked 9600 Pro. I might add the 9600 Pro is definatly not faster than a 9500 Pro in any way. If there was a overclocked 9500 Pro called XT, it'll be better than a 9600XT wouldn't it now?

The good for ATI and bad for you comment is very true indeed. Sure it's still a good mid range card, but come on... usually you would expect a replacement product to be better than the fade out model it's replacing. It's fair enough to be critical of the 9600 purely on those grounds.

Nvidia's done even worse in the last 6 months. Their replacment for the GF4 Ti series, FX line (spare the 5900 range) has been bitterly dissapointing as well.

i dont see your point when you say the 9600 XT is just an OCed 9600 pro. isnt a 5950 just an oced 5900 ultra? isnt a 9800 XT just an overclocked 9800 pro? isnt a 9800 pro just an overclock 9800????

is the 9600 pro faster than the 9500 pro? no its not but that wasnt my argument. my argument was showing that Rollo is once again coming to wild conclusion by saying Feeble 9600XT even stomped by old 9500 pro when in fact it is a faster card.

how old is a 9500 Pro and how young is a 9600XT?

once again, lets look at the counterpart. how old was the 9700 pro and how new was the 5800 ultra? the 5800 came out almost a year later and still was not able to match 9700 pro in DX9

also, the 9800 non pro was released what, 9 months after the 9700 pro? so is the 9800 non pro a bad card too because i gets beaten the the 9700 pro that was released a lot earlier?

 

[dguy6789]

Rollo Rollo Rollo, have you learned yet? Just because your beloved 9800 pro is faster then the 9600 Series doest not mean it is feeble. You have the right to your own opinion, but your point is useless. The 9600 Pro is not much slower then the 9500 pro. but the 9600 pro is alot cheaper.

9600 XT stomps 9500 Pro by the way, and most 9600 pros can be clocked to 500Mhz(XT Speeds) so that puts your feeble theory out the door. Add to that, the fact that the 9600 series does DX9 far better then anything nvidia has, just makes it more appealing.

How can a card be feeble, if it is faster then 95% of all cards out, and can run any game at 1280X1024 with AA/AF at over 60 FPS.

And for new games, just turn down the AA/AF and you will have 60+ FPS in future games as well in high settings. The 9600 series also has better compression and a more efficient design then even the 9800XT, not to mention more powerful shaders as specified above. Yes the 9500 Pro is faster most of the time then a 9600 Pro, but the 9500 Pro is alot more expensive as well. And i can guarentee future games will take advantage of the 9600's efficient design and the card will woop the 9500s.

And, to ad to that, on average overclocks for both cards, an overclocked 9600 pro will beat an overclocked 9500 pro Rollo.

And one more thing, the 9600 Pro plays flight simulater 2004 better then any card out right now, including 9800XT and GFFX 5950 Ultra.

 

[Corporate Thug]

And, to ad to that, on average overclocks for both cards, an overclocked 9600 pro will beat an overclocked 9500 pro Rollo

i thing firing squad tested an oced 9600 pro and a 9500 pro and the benchmarks showed they were nearly identical with the 9600 pro eeking out a few wins and the 9500 pro eeking out a few small wins

edit: conclusion from firingsquad

"As you saw in our test results, the overclocked RADEON 9600 PRO is quite a performer. In some cases, not only is it capable of outperforming the overclocked, 8-pixel pipeline RADEON 9500 PRO, it can even give the RADEON 9700 PRO a run for its money! "

 

[Pete]

Originally posted by: TerryMathews
Every time this type of post is brought up, the fact that the 9600 series has effectively double the number of shaders the 9500/9700/9800 does is overlooked/glossed over.

The 9500/9700/9800 has 1:2 Shaders to Pipelines. Meaning, in the case of the 9500 Pro, 9700 series, and 9800 series excluding SE, there are four shaders.

The 9600 series is 1:1 on shaders to pipelines. Combine that with the higher clock rate of the core and you have the ability to do many more shaders on a given frame without bogging down.

In layman's terms, you'll have the ability to cram more special effects into a given frame although you may not have as many frames.

Someone should write a benchmark that runs better on the 9600, just to show it can be done.

Erm, no?

All the 9500P+ cards have one pixel shader per pipeline, just like all of nV's cards. (I believe the pixel shader has to be integrated into the pixel pipeline by definition, though I may be wrong--if not WRT current cards, then possibly WRT future cards that support DX Next's unified shader models.) The 9500P (a R3xx series chip) has two more vertex shaders and also some HyperZ features that the 9600 (a RV3xx series chip) lacks, so it may be a tad faster clock for clock in certain games.

So the 9500P, 9700, and 9800 all have eight pipelines with a pixel shader per pipe, as well as four vertex shader engines. The 9600+ has four pipes with a PS per pipe, as well as two VS engines. I believe the VS was tweaked in the 9600 and possibly 9800 series, so their VS engines may be faster clock for clock than the 9500's.

I appreciate how Rollo overlooked the fact that the 9500P was a temporary card, a wolf in sheep's clothing--really, a gift to mainstream consumers. The 9600XT competes with the 5700U, not the 9500P. (Check some of Digit-Life's ethically-questionable benches of the HL2 leak--the 5700U gets stomped, even with the coveted all-improving "Unified Compiler Tech" Det 52's.). He also overlooked the fact that the 9600XT outpaced the 9500P in a few later games. I was surprised that the 9600XT was slower than the 9500P in the DX9 games, but the XT is still 50MHz shy of theoretical shader parity with the 9500P, and it's also missing some HyperZ elements which should help a lot in memory-intensive shaders.

Rollo, no need to get troll-y with your threads, buddy. We all know the mainstream cards pale to the high end WRT to full effects. Why give the more excitable folks here an excuse? Do you miss DefRef's posts that much? bunnyfubbles was right. Take a #3, and post in the morning.

 

[Corporate Thug]

^^^ LOL, he says "Yep these will be great for HL2!" and doesnt realize the 5700 U gets pwned by the 9600 XT in HL benchies. shame shame shame.

 

[TerryMathews]

Originally posted by: Pete
So the 9500P, 9700, and 9800 all have eight pipelines with a pixel shader per pipe, as well as four vertex shader engines. The 9600+ has four pipes with a PS per pipe, as well as two VS engines. I believe the VS was tweaked in the 9600 and possibly 9800 series, so their VS engines may be faster clock for clock than the 9500's.

It may be wrong in this case, but AIDA32 says that the 9600 Pro has four vertex shaders, not two. That was the point of the 1:1 vs 1:2 post I made earlier.

 

[Tab]

Originally posted by: gotey
I totally agree with rollo. I have a 9500np and used to own a 9600pro. I sold the 9600p because the 9500np was faster in 3dm2k1 by 600 points and slower in 3dm2k3 by 240 points. In real games I could not tell a difference so it had to be pretty close in FPS
.

Okay.......

We know good 3dmark is!

 

[nRollo]

OK first of all the feeble 9600s are 4X1, and the good 9500s are 8X1, and they both have 128bit memory:
9600 castration revealed

the Radeon 9500 Pro was quite a fluke in that it was basically a slightly detuned Radeon 9700 Pro whereas the Radeon 9600 Pro is much more castrated

Anand seems to agree with me:

If we step back and bring the Radeon 9500 Pro into view it's clear that regardless of which of the aforementioned two is faster, neither one of them can hold a candle to the old Radeon 9500 Pro.

Once again we see the GeForce FX 5600 Ultra and the Radeon 9600 Pro offering very similar levels of performance, but the Radeon 9500 Pro puts them both to shame.

 

[Corporate Thug]

funny, i dont see anand say feeble anywhere...

 

[ronnn]

Originally posted by: Rollo
Wow Cindy has agreed with me twice now, and offered opinions that didn't sound like ATI ads.
<retracts assertion Cindy is Rogo and offers apology>

Have things been too quiet lately??
:moon:

 

[Rogodin2]

"Wow Cindy has agreed with me twice now, and offered opinions that didn't sound like ATI ads. retracts assertion Cindy is Rogo and offers apology."

Not to me.


Implying that I spout ATI ads.

Just because I won't sell Nvidia cards (predicated on kant's categorical imperitive, and my own moral standard) doesn't mean I am biased and blind.

I have 5 9500pros in my stockroom and 4 are going into new rigs this week-I've purchased them over 9600XTs because of their intrinsic superiority.

These personal attacks are insipid.

rogo

 

[Pete]

Terry, AIDA is wrong. It sounds like a synthetic overall system benchmark, and I didn't realize it could determine how many vertex shaders a card has on the fly. I don't see how, as different VSs may have different performance characteristics. Also, no one really focuses on VSs when they talk about DX9 performance. We're all mainly fixated on PS2.0 perf, I thought.

Rollo, I don't get it--have things been too quiet around here? This vitriol at the 9600 series is puzzling. You know the 9500P was a temporary deal to serve the mainstream market with a single-chip release, and you know they had to stop sticking expensive VPUs in less-profitable packages, pronto.

 

[dguy6789]

That is true indeed. The 9600 Pro, has 4 Pipelines. The 9500 Pro has 8 Pipelines. This gives the 9500 Pro a fillrate advantage in heavily multi textured games where they come to use. But what about not multi textured games, games that use 2 or 4 pipes? The 9600 Pro would own the 9500 because it has alot higher gpu clock, and if only 2 pipes of each card are being used, the 9600 would have alot more fillrate in the circumstance.

The 4 Pipes is the ONLY weakness of the 9600 series. It is suped up everywhere else, it has compression better then anycard out there right now, a memory controller built specifically for transfering alot of data and saving bandwith on a 128 Bit memory controller.

 

[nRollo]

funny, i dont see anand say feeble anywhere...

Sorry, you're right. He only says "castrated" and "put to shame" and "considering that the Radeon 9500 Pro has twice as many pipelines as the Radeon 9600 Pro, it is unlikely that we will see that sort of a showing from the 9600 core."
More truth

I guess to you Shady06, that stuff means "not feeble".

 

[nRollo]

dguy:

But what about not multi textured games, games that use 2 or 4 pipes? The 9600 Pro would own the 9500

Could you do me a favor and list some of those single textured games the 9600 would be better at dguy? You see, I can't remember any since "Forsaken" about 5 years ago. I'm going away for the weekend, but I'll be interested to see the single textured revolution I've apparently missed when I return tomorrow night.

It is suped up everywhere else, it has compression better then anycard out there right now, a memory controller built specifically for transfering alot of data and saving bandwith on a 128 Bit memory controller.

LOL, that's why it performs about as well as the ancient Ti4600 w/o AA/AF on, and is too slow to be really usable with AA/AF on.

 

[Regs]

Originally posted by: Rollo

Anand seems to agree with me:

If we step back and bring the Radeon 9500 Pro into view it's clear that regardless of which of the aforementioned two is faster, neither one of them can hold a candle to the old Radeon 9500 Pro.

Well of course. Pete all ready stated that the 9500p was a more expensive design for ATi to produce, so consequently it was a more robust architecture. More than needed for what is now considered a mid-range card. The 9600p which is cheaper to produce, respectfully took on the role as the mid range card.

So, undoubtedly, Anand will state that the 9600 is "castrated", and that's why it's 160 dollars, not 300 dollars.

And as always, this is just another "feeble" attempt to down play the 160 dollar 9600 to the 200 dollar 5900u. In other words, I guess, this is another thread to fish for more proof in why customers should just buy a 50 dollar more expensive card. Logic states, we would have to agree with you. So I have to give you some credit for getting the moxy to post a thread that would be inflammatory.

 

[kylebisme]

Originally posted by: dguy6789
That is true indeed. The 9600 Pro, has 4 Pipelines. The 9500 Pro has 8 Pipelines...

you got that much right, but you have pipelines confused with texture units so the rest of your argument is total fud.


ohh and lol at Rollo!

 

[Corporate Thug]

Originally posted by: Rollo

funny, i dont see anand say feeble anywhere...

Sorry, you're right. He only says "castrated" and "put to shame" and "considering that the Radeon 9500 Pro has twice as many pipelines as the Radeon 9600 Pro, it is unlikely that we will see that sort of a showing from the 9600 core."
More truth

I guess to you Shady06, that stuff means "not feeble".

words dont mean much. benchmarks do. the link you provided to the hard article shows that on a whole, the 9600 XT is faster than the 9500 pro

 

[VIAN]

This just in - Benchmarks are useless. Look at my posts in this thread. Get some of Cindy's links too.

 

[Blastman]

In some of the latest rounds of benches the 9600pro is doing pretty good against the 9500pro even though it is considered a slower solution/replacement for the 9500pro. In THG?s 9800XT review the 9600 pro was faster than the 9500 pro in 5/9 (1024x768) game benches and in the 5950/5700 Firingsquad review the 9600pro was equaled the 9500pro in performance in the 6 benchmarks. The 9600pro still isn?t quite up to the 9500pro?s performance, but it?s pretty close and the 9600pro is less expensive than the 9500pro ever was.

THG?s 9800XT review 9600 pro vs 9500 pro ?. ?

UT3 ? ? ? ?
Wolfenstein ET ... ... ? 9600 ? faster
Tomb Raider AOD ?
X2 Rolling Demo ... ... ? 9600
C&C Generals Demo ?
Warcraft III ... ... ... ? ? 9600
Freelancer Demo ... ... ? 9600 ?(95 wins at higher resolutions)
MS Flight Sim 2004 ... ... 9600
BF 1942 - SW WWII ?

In the 5950/5700 Firingsquad review ?

Nascar ? ? ? ? tie
IL-2 ? ? ? ? 9600
Q3 ? ? ? ? 9500
UT3 ? ? ? ? tie
SplinterCell ? 9600
Tomb Raider ? 9500

 

[Tab]

Originally posted by: VIAN
This just in - Benchmarks are useless. Look at my posts in this thread. Get some of Cindy's links too.

Right...

 

[VIAN]

Didn't you read or do you want me to go and get the info for you.

 

[Corporate Thug]

vian. whats the best indicator of performace? benchmarks

whether they are done properly or not is another issue

 

[dguy6789]

Benchmarks are good indicaters of performance, as long as the benchmark is running the full spec of whatever it is running, on all cards. Games are also good testers.

There are plenty of old and new games that do that. heck, on ut2003 map asbestos, the 9600 pro is faster then the 9500.

Yes the 9500 is faster most of the time, what i am saying is, if games were designed to take full advantage of the 9600, then it would run better then on a 9500. Have you seen the 9700 and 9800 demos? My 9600 pro runs them in real time at great fps at 1280X1024 with 2XAA and 16X AF quality, so how can something that renders something so awesome looking be feeble? I can run any demo ati has at those settings in real time, and i do, and it is quite entertaining to watch.