[Testing] Sapphire X2900XT

[Syn.]

i also had to convert the .BMP image from FRAPS to .JPEG

I know you were not trying to start anything , thats why i also said ""dont get me wrong""


regards

You dont have convert them, you can just choose in fraps to save SS in BMP, PNG or JPEG.

[DilTech]

should i sell my 8800GTX that cant overclock real good, and get this card???

To be blunt about it? No, you shouldn't even consider it.

[bluep3ace]

To be blunt about it? No, you shouldn't even consider it.

it seems like GAR is considering to overclock his GPU. If you see Kinc's and Denny's results on their R600 overclocks, they can easily beat stock 8800gtx's.

It's perfectly reasonable for GAR to consider selling his 8800gtx.

[DilTech]

it seems like GAR is considering to overclock his GPU. If you see Kinc's and Denny's results on their R600 overclocks, they can easily beat stock 8800gtx's.

It's perfectly reasonable for GAR to consider selling his 8800gtx.

Maybe in 3dmark, but not in games.

[mascaras]

You dont have convert them, you can just choose in fraps to save SS in BMP, PNG or JPEG.

i know , but only if you have a full version of program

[Xion X2]

Maybe in 3dmark, but not in games.

In some games, an overclocked XT can catch or pass a stock GTX. I've been benching my card against several guys over at the Nvidia forums. My OC'd XT is ahead in R6 Vegas and is neck and neck in Oblivion. DIRT performance is on par w/ a stock GTX:

http://www.nvnews.net/vbulletin/show...&postcount=413

Benches on the XT were done on the Alphas (8.37.4.2) and 8.38s. That's a rather long thread, but there are tons of benchmarks between the cards across different games if you want to have a look. We've tested Lost Planet, Oblivion, R6 Vegas, DIRT, Riddick, and several other games. The 2900XT is more than holding its own.

Don't believe everything that you read in reviews; a lot of them are probably on Nvidia's payroll. Hardocp and a lot of those other review sites weren't using these performance drivers; they were using very unoptimized sets. It's crazy that ATi let them get away with it, but they did.

[DilTech]

In some games, an overclocked XT can catch or pass a stock GTX. I've been benching my card against several guys over at the Nvidia forums. My OC'd XT is ahead in R6 Vegas and is neck and neck in Oblivion. DIRT performance is on par w/ a stock GTX:

http://www.nvnews.net/vbulletin/show...&postcount=413

Benches on the XT were done on the Alphas (8.37.4.2) and 8.38s. That's a rather long thread, but there are tons of benchmarks between the cards across different games if you want to have a look. We've tested Lost Planet, Oblivion, R6 Vegas, DIRT, Riddick, and several other games. The 2900XT is more than holding its own.

Don't believe everything that you read in reviews; a lot of them are probably on Nvidia's payroll. Hardocp and a lot of those other review sites weren't using these performance drivers; they were using very unoptimized sets. It's crazy that ATi let them get away with it, but they did.

The guy in question that we're speaking about already owns a 8800GTX, he asked if he should sell his GTX and buy a HD 2900xt, which honestly he shouldn't and I think we can both agree on that.

As for forum numbers, I now have a hard time believing anyone making claims, even with screenshots showing said claims...

http://www.xtremesystems.org/forums/...=147978&page=2

I'm not saying you can't trust forum members numbers, I'm just saying that after that I'm having a rather difficult time believing claimed numbers from people regardless of which card they have.

[Xion X2]

The guy in question that we're speaking about already owns a 8800GTX, he asked if he should sell his GTX and buy a HD 2900xt, which honestly he shouldn't and I think we can both agree on that.

Why even bring this up? I never implied the guy should trade his card in.

As for forum numbers, I now have a hard time believing anyone making claims, even with screenshots showing said claims...

http://www.xtremesystems.org/forums/...=147978&page=2

I'm not saying you can't trust forum members numbers, I'm just saying that after that I'm having a rather difficult time believing claimed numbers from people regardless of which card they have.

Well I don't know what to tell you. In my personal opinion, Hardocp doesn't have a clue what they're doing. My #s were not forged, and I can tell you for a fact that HDR or AA was not "broken" on Oblivion like some people have claimed. Both were working just fine. I even benched the game again w/o AA to prove it. The framerate shot up (almost doubled) and jaggies were much more apparent than on the screens w/ AA enabled. And there was absolutely no comparison between an OC'D GTS and OC'd XT as far as performance goes. The XT beat it handily.. and outperformed a stock GTX on a few occasions.

I've actually owned an 8800GTX before this card (long story), and the IQ between the two looks identical to me on Oblivion.. a game I played all the time on my 8800GTX. And performance between the two is very, very similar.. almost identical. I would say the GTX is probably a little quicker but not by much.

I realize why you would be skeptical; I guess all I can do is state my case and results. It's up to you whether or not to believe them over some review site that could be receiving benefits under the table from a major card distributor.

But Hardocp's results are very fishy, to say the least. The 8.37.4.2s brought me 400% increase in Oblivion with the exact same I.Q. I benched both back to back with screenshots to prove it. I uninstalled-->reinstalled the driver multiple times for a control measure. There's a large gap in performance between the two. Settings were as follows:

8xAF, 4xAA w/ Adaptive (both enabled in driver)
HDR (in-game)
Maximum Settings, draw distances maxed (including grass)
Default INI (no tweaks)
Catalyst A.I. = Low, Off (no performance difference w/ either)


So it's my opinion that Hardocp is full of ****. After viewing their behavior, I will never, EVER read another one of their reviews or link to their lousy website. I think those who forge benchmarks are pathetic, myself. I'd never do something like that.. why would I? What good is it to do something like that when you know yourself that it's not true?

[DilTech]

Why even bring this up? I never implied the guy should trade his card in.



Well I don't know what to tell you. In my personal opinion, Hardocp doesn't have a clue what they're doing. My #s were not forged, and I can tell you for a fact that HDR or AA was not "broken" on Oblivion like some people have claimed. Both were working just fine. I even benched the game again w/o AA to prove it. The framerate shot up (almost doubled) and jaggies were much more apparent than on the screens w/ AA enabled. And there was absolutely no comparison between an OC'D GTS and OC'd XT as far as performance goes. The XT beat it handily.. and outperformed a stock GTX on a few occasions.

I've actually owned an 8800GTX before this card (long story), and the IQ between the two looks identical to me on Oblivion.. a game I played all the time on my 8800GTX. And performance between the two is very, very similar.. almost identical. I would say the GTX is probably a little quicker but not by much.

I realize why you would be skeptical; I guess all I can do is state my case and results. It's up to you whether or not to believe them over some review site that could be receiving benefits under the table from a major card distributor.

But Hardocp's results are very fishy, to say the least. The 8.37.4.2s brought me 400% increase in Oblivion with the exact same I.Q. I benched both back to back with screenshots to prove it. I uninstalled-->reinstalled the driver multiple times for a control measure. There's a large gap in performance between the two.

I think those who forge benchmarks are pathetic, myself. I'd never do something like that.. why would I? What good is it to do something like that when you know yourself that it's not true?

I mentioned what I did about 3dmark because someone asked if they should trade their 8800gtx for a HD 2900xt. Follow the quotes starting from my post and you'll see what I mean.

It wasn't hardocp's numbers that I was linking that for. I posted it because there's people giving fake numbers by lowering settings to make it appear as if it's running faster than it is maxed out. Look at v_rr's post in that thread(that's why I linked to page 2, and not page 1), and then read mine afterwards.

That's why I'm having trouble trusting peoples numbers on forums, because I've seen this occur now on multiple sites. In one case, someone was bragging about 193fps in HL2:Lost Coast, and he wasn't even running it maxed out, and it was with no AA.

[Xion X2]

I can see where you're coming from, but I guess my next question would be:

"What do you accept as proof, then?" or "Who do you believe?"

Because some of these review sites could be just as corrupt or misleading. You know that some of them are getting perks behind the scenes to say one thing or another, just as some forum member may be adjusting settings in order to feel better about his purchase.

I honestly didn't do that but also realize that I can't expect you to believe me. Those Oblivion screens were all taken w/ the settings mentioned in my last post. HDR + AA was working just fine.. it looked identical to the screens on the G80. Sunlight was glaring off of surfaces like crazy and shining through tree branches.. the dynamic weather was in full effect just as it was on G80. And AA was clearly working. Here is a shot I took of AA and AF both disabled and then enabled on the second shot.

http://www.nvnews.net/vbulletin/show...&postcount=725

There were still a few jaggies showing, but they were evident in the G80 screens as well. 4xAA doesn't completely get rid of aliasing in Oblivion. Higher levels are needed for that.

But you can clearly see the difference.. especially to the left of the green light.

[cadaveca]

LoL. I think DilTech=DailyTech. They seem to be of the same opinion, always pro-nV.

[Xion X2]

Here was a Lost Planet (DX9, XP) run between an OC'd XT (875|975) and a stock GTX. Both cards were ran w/ the same in-game settings:

Mine (80FPS):
http://www.nvnews.net/vbulletin/show...&postcount=658

Sammy Sung (89FPS):
http://www.nvnews.net/vbulletin/show...&postcount=661


Lost Planet runs the same on the Alphas (8.37.4.2) as it does the 8.38s, from my experience, with no visible IQ difference between the two.

[DilTech]

Here was a Lost Planet (DX9, XP) run between an OC'd XT (875|975) and a stock GTX. Both cards were ran w/ the same in-game settings:

Mine (80FPS):
http://www.nvnews.net/vbulletin/show...&postcount=658

Sammy Sung (89FPS):
http://www.nvnews.net/vbulletin/show...&postcount=661


Lost Planet runs the same on the Alphas (8.37.4.2) as it does the 8.38s, from my experience, with no visible IQ difference between the two.

I'd be interested in seeing you guys do those runs with higher cpu clocks, mainly because the game is as taxing as it is on the cpu.

[Xion X2]

I'd be interested in seeing you guys do those runs with higher cpu clocks, mainly because the game is as taxing as it is on the cpu.

Yeah, both of those were w/ E6600s at stock. The Cave benchmark is much more CPU intensive than the Snow benchmark is, though.

[DilTech]

Yeah, both of those were w/ E6600s at stock. The Cave benchmark is much more CPU intensive than the Snow benchmark is, though.

I'm just curious how well both of them scale. I'm sure I'm not the only one.

[Xion X2]

By far not the only one. I'm anxious to see how much of a gain I'll see w/ my XT on all of these titles w/ an overclock.

I just installed this Bad Axe a few weeks ago. It's a pain to overclock because you have to custom make your own BIOS, which is why I'm still at stock speeds for now. I'm looking into it, though.

[Zytek_Fan]

Performance wise, what should I get, 8800GTS or 2900XT? I would like to hear some comments that don't sound like they are playing favourite

[cadaveca]

Still a bit hard to decide at this point...HD2900XT is faster for me in alot of games that I personally play, so yeah, I cannot help but "play favorites".


But at the same time, I'm hoping for decent driver improvements that will make it outshie GTS in DX10 titles, however, that's what I bought the card for, DX10. In DX9, i think the GTS is a better buy...

[Zytek_Fan]

Still a bit hard to decide at this point...HD2900XT is faster for me in alot of games that I personally play, so yeah, I cannot help but "play favorites".


But at the same time, I'm hoping for decent driver improvements that will make it outshie GTS in DX10 titles, however, that's what I bought the card for, DX10. In DX9, i think the GTS is a better buy...

I am kinda leaning toward the 2900XT, with every driver release, the performance seems to improve

[cadaveca]

It does. In fact, I had some issues with some apps that have righted themselves after installing the whiteout demo. It seems that DX update is what it partially needed, based on what was installed when installing the demo...

However, I'll be plain...I have these cards in Vista, not XP. In Vista, alot of DX9/8 apps, just plain don't work. SO I have a rig for DX9(X1950 Crossfire), and one for DX10.

That being said, I have really only used apps in Vista with these cards...not XP...so my perspective on this issue is slightly skewed towards a specific set of needs form this card, and although I feel personally that that's where this card should be, in Vista, and not XP, alot of others that are ocmplaining of issues and poor performance are simply just not using Vista.

SO, the driver I use is different, the apps I'm using behave a bit different, and alot of other factors come into play when considering this card's performance, or it's lack thereof.

Altogether, I'm very happy with these cards, despite the slight noise(which i never notice when playing a game with sound cranked anyway), the lack of hardware decode for HD content(to me, shader decode is still software decode, I don't care who says otherwise), and poor Crossfire drivers.


Then again, I own just about every product ATI has ever relased, in some form or another, so I'm jsut a wee bit of a fanboy in that regard...I would have bought cards even if they really really sucked...because without hte bad, you cannot know what good really is!

[Eastcoasthandle]

I mentioned what I did about 3dmark because someone asked if they should trade their 8800gtx for a HD 2900xt. Follow the quotes starting from my post and you'll see what I mean.

It wasn't hardocp's numbers that I was linking that for. I posted it because there's people giving fake numbers by lowering settings to make it appear as if it's running faster than it is maxed out. Look at v_rr's post in that thread(that's why I linked to page 2, and not page 1), and then read mine afterwards.

That's why I'm having trouble trusting peoples numbers on forums, because I've seen this occur now on multiple sites. In one case, someone was bragging about 193fps in HL2:Lost Coast, and he wasn't even running it maxed out, and it was with no AA.

What people? What people are faking numbers??
I think you are over exaggerating the situation. From what I've read in that thread there were no "people" making fake numbers. You have been making diggs at the HD 2900XT way before those pics were posted. Therefore, I don't believe your trust issues stem from that thread. For example:
"I am torn. X2900XT because I'm an ATi fanboy, or 8800GTS640 for the price?"
this post (in the same thread)
"should I do it? 2900XT for 8800GTS 640"
"8800, but which one"

As you can see you've been making diggs at the HD 2900XT way before that thread was created.

[ahmad]

2900XT destroys in CoH.. thats why my buddy got it

[Xion X2]

Here are some benches I just did of Test Drive Unlimited. Plays at a consistent 50+FPS on an overclocked XT (875|975). Only dropped below that point on a few occasions for a split second:

http://www.nvnews.net/vbulletin/show...&postcount=806

http://www.nvnews.net/vbulletin/show...&postcount=807

[dnottis]

Im tried of Nvidias drivers. They just keep stalling.... 165.15 pulled.... argh..

While ATI has improved performance, and the new hotfix is supposed to be a beta of 7.7...

[mascaras]

Medal Of honor Airborne

- 1680x1050
- All Game Settings High
- Q6600 G0 @ 3.5Ghz
- HD2900XT Default (CAT 7.8)
- 2Gbs Mushkin XP2-8500
- Windows XP




























High resolution clik images: