Another: x1900xt 256MB or 7900gto

MSI 7900gto
Sapphire x1900xt 256MB
I'm probably going to go ahead and answer my own question, because I love NFS Carbon, and it's one of the main reasons I want a new videocard. Simply because GeForce 7*** cards SUCK at NFS Carbon. However, I still would appreciate any input, I'm buying tomorrow. Specs are in sig. I will also be getting a new PSU, I was thinking the Antec SmartPower 2.0 500watt, but any input on any other PSU would be appreciated as well. I have 400$ to spend CAD, so it doesn't have to be these cards, but they look to be the best deals. For anyone who wants, use www.pricecanada.com for finding cheap parts from Canadian online retailers.

The games I play most are:
FEAR
Quake 4
Prey
NFS Most Wanted
NFS Carbon
Source games (HL2:E1, etc..)
Age of Empires 3

I Personally would go with the 7900GTO ( no offence to ATI) Because the GTO version of 7900 is exactly the same as the GTX
just clocked down. All you Have to do is clock up the memory speeds to GTX or just flash the GTX Bios and it should run like a GTX.

better performance and is cheaper

You Dont need a power supply over 500 Watt to run a single X1950.
In this article this guy used 2 8800GTX and over clocked it and it only reached about mid 400W. http://www.theinquirer.net/default.aspx?article=36066

Um, the GTO is not exactly like the 7900GTX. It's like the original first run of 7900GTX's that had issues with the memory clocked so high. You are NOT guaranteed GTX speeds on a GTO, if you were they would have sold them as GTX's. We have discussed this here many time before.

Amongst the games you mentioned MW, Carbon, FEAR and maybe even AOE3 and EP1 should receive little to no benefit from the extra 256MB at not so high res unless you plan to use Tripple Buffering. So for these games 512MB VS 256MB becomes secondary, but for Quake 4 the difference will be substancial and you will need to reduce the eyecandy in Quake 4.


I don't know if the performance is improved or not, 1900xt's play Carbon very well and atleast the initial performance for Geforce 7 was horrible, may have(probably) improved by now.

AOE3 is a game which will run much much better on ATI cards. A 1950XTX is probably faster than a 8800GTS and close to a 8800GTX in this game, so a 1900xt becomes a pretty obvious choice here as long as we are not considering the VRAM.

EP1, in case lack of VRAM is of no significance, the 1900xt should be better. As a simple rule, if a game doesn't benefit measureably from the additional VRAM, the 1900XT will be faster. If you had to choose between a 1900xt 512 and 7900gtx 512 both at the same price, a 1900xt is usually the prefered choice.
Now, Quake 4, ATI's drivers opengl are supposed to be improved by now, but still 512MB 7900GTO should better.

1900XT
Better IQ
HDR + AA
Faster card

7900GTO
512MB(a huge, night and day benefit for many current and near future games)
Better experience with nvidia(atleast mine)

Hey thanks alot for your input akshayt and Pauldh. You guys definitely confirmed what I've been thinking. ATI's current x19** cards can pretty much stomp Nvidia's GeForce 7 series now in most games, especially the more popular ones (FEAR, AoE3, Carbon, Oblivion, etc..). Except for OpenGL games. However, while the 7900gto would outperform an x1900xt 256MB in Quake 4 at high quality, in Prey, the x1900xt would do better. And I only play games at 1280x1024 max, so an x1900xt will have no problems with Quake 4. Frames are alot more important to me then IQ. I run FEAR right now at 1024x768, med quality, 8xaf, with shadows and volumetric lighting off. I absolutely must have high frames especially for FPS games.

An example of how worthless Nvidia's GeForce 7 cards are in Carbon, I ran painful race at 1280x1024 everything max (0xaa 0xaf), and by observing my frames, my min was probably 8fps, average was about 19fps. I rarely went over 30fps. While my buddy can run the same game everything cranked and he has AMD Opteron 165, 1gb DDR400, x1950pro, and he would almost never drop under 30fps. Even though Nvidia's counterpart is quite a bit faster then my 7800gs, it still goes to show it's weakness.

While I never directly went out and asked the question that has been asked a million times, I'm glad you guys took the initiative. Does the extra 256MB really help? I don't think it does, although I'll probably still run games that require huge amounts of memory at an average texture quality. I have the G15 keyboards, and I can monitor how much memory is being used while I play. And so far the only games I can't run cranked due to lack of memory is Doom3, Quak4 (ultra quality), and FEAR. When my system memory gets used up I notice when I turn a corner the games pauses for a split second to load the textures. Again, to remedy this I just turn down texture quality, that doesn't bother me. I rather have a SHARP image then one with alot of eye candy (FEAR @ 1280x1024 shadows off vs 1024x768 shadows on, I rather have 1280, but my card can't even do that )

And thanks for the link to the 7900gto review that has both cards in question Pauldh. While in almost all benchmarks the cards stay within 10% of eachother, but the x1900xt does dominate NFS Carbon, Oblivion. I don't see the 7900gto dominating in anything. In fact in Quake 4 it only beats the x1900xt by 4.1fps at my res of 12x10.

Thanks again for your helps guys.

While we discussed the x1900xt to be the better deal, what PSU should I go with then? I might just go with the OCZ GameXStrem 600watt, but I don't want to spend the extra money if I don't have to. These are the PSUs I'm looking at:

Silverstone SST-ST50EF-Plus 500Watts
Thermaltake TR2 500Watts
Enermax NoiseTaker II EG-495P-VE 485Watts (No active PFC )
Antec Smartpower 2.0 500watts

Also, one more question, x1950xt vs x1900xt, is it worth 40$ more to get the x1950xt?

And even if you do get the GTO to GTX speeds it's still a lesser card in the most edemanding games, especially the ones he seems to be interested in.

Xbit once again looks at the XT versus GTX-512 in their 7950GT review;
http://www.xbitlabs.com/articles/v [...] 950gt.html

and even thought it's 512MB on the XT that isn't as much of an impact as the GTX-512, which is way faster than the GTX-256

In this case even with an insance OC the improvement isn't enough to tip the balances, but the savings may be something to consider, but if it's about the money then I'd say wait a few weeks for the new products to launch an redefine the market again.

I would stay away from all ATI cards. I have always used nvidia and never had a problem, but, like you, heard that the x1950xt out-performed the 7 series nvidia cards, so I bought one...

It doesnt work, so i returned it for another....

this one doesnt work...

I have been looking through several different forums for a solution and found many people with the same problem.

Anyway, I will never buy another ATI card no matter how much better they are said to be over nvidia. Their drivers sucks, their compatibility sucks and thier customer service sucks.

ATI's 30 Amp 12v rail requirement is assuming a loaded system and is also talking about rated spec. I assume they know that PSU's aren't operating at 100% efficiancy and may not sustain those ratings. That Antec 500W SP 2.0 has a 17 amp and 19 amp 12v rail (36 Amp total), which should have no problem powering a X1900.

I have an X1800XT that is running in an A64 system with the smart power 2.0 450W power supply without issues. I have also run a 7800GT in an A64 for a while on the SP 2.0 400Watt PS. I actually like what those power supplies offer for the price. Plus they are quiet. That's one reason I buy Antec TX640B cases and SOnata 2's as they come with SP 2.0 400W and 450W PSU's. I did upgrade my TX640B to a True Control (Truepower) 550W though.

What kind of issues? Can't imagine a 550W not powering that rig uless somethings wrong with it or a compatibility issue. :?

To be honest I probably have already built 20+ systems around one of the Smartpower 2.0 power supplies without stability issues or one failing yet. I have 6 more new ones in stock right now. Is that pushing my luck? 8O

But a buddy of mine who has also used a bunch of them had a 400W go bad, running an A64, 7800GT, 2 HDD's, 2 Optical's. But it was stable even in that loaded system until it blew.

I'd have to agree that most people completely blow the power requirements, for most machines, completely out of proportion. I am using a 380watt cheap-a** power supply to power what I have now, so telling me an x1900xt will require an additional 170watts minimum is ridiculous. However, having extra power is alot better then having just enough. But what will happen, is I will probably build my box to something actually a little more reputable, and build new once DX10 becomes solid. My build will probably look a little something like this:

Case: Unsure of at the moment, maybe Praetorian 731.
*PSU: Whatever PSU I decide to buy today, probably the Antec Smartpower 2.0 500watt.
Motherboard: 650i chipset, unsure of exact make/model
*CPU: My C2D 6300
Memory: 2GB of something.
*Videocard: X1900xt
*HDD: WD 2500KS X2
*DVD+/-RW: LG 16x

*Already own (or will own hopefully beginning of next week)

Once that machine is built, I wont be really adding to it. Like I said, I'll just start over.

Thanks again for everyones help! I never got such overwhelming and speedy responses from you guys!

yes, now do you mind giving me $250 usd

I really could use the money right about now

ha, right now the best psu's on the market are pcp&c, enermax, silverstone, ocz, and I guess kinda but not really, rosewill (offer very nice prices for their performance)

Yeah, a few games like Carbon, Oblivion, or Double Agent, the X1900XT would creme the GTO and beat a GTX too. I tell ya, I went through this with a good friend of mine (favors NV) when he spent quite a bit more on a GTO than he could have bought the 256MB X1900XT for. Oh well, I tried, but he fell victom to grasping at the then only bright spot in the GF7900 line(bang for buck wise), and the talks of OC'in to GTX speeds. Nothing wrong with the GTO, just rarely (if ever) has it offered the best performance for the buck.

Good link there BTW. I wish they would do one of those for about 10-15 mid to top AGP cards. Sweet to include so many games. Intersting they do state X1900XT >= 7900GTX and they didn't even test Carbon or Double Agent. Also weird when some sites show X1900XT destroying the 7900GTX in Oblivion and others show them closer. Makes me feel the ones where they are close aren't the most stressful outdoor areas.

Yeah, currently I say X1900XT over GTO(stock speeds or both OC'ed), come the new product launches, price and availability changes, etc. Who knows if we will still be talking about either of these cards.

With motion blur, I'd definately agree. And he should use swingman view for the bet as that gives a little fps hit too.

I'm just amazed how well the X19xx's do in that game and how poorly the X1800XT did.

My own non overclocked results of a 2 lap circuit in swingman view: (each tested 3 times with same results)

1280x1024 4x/16x max quality but no motion blur.
X1800XT ave 28 fps
X1950XT ave 43 fps

1280x1024 4x/16x max including motion blur.
X1800XT ave 22 fps
X1950XT ave 34 fps


Using the in car view resulted in 3-5 fps more fps over those results.

Well I wouldn't agree based on what you wrote above, but the spirit of your reply I do agree with. Let me explain, you said the X1900XTX, and the post is about the X1900XT as I was talking about. For the X1900XTX it actually uses the same Samsung BJ11 memory chips as the X1950XT so the memory overclocking potential is the same (spec'd at 1.1ns/900Mhz), however the X1900XT (not XTX) uses BJ12 memory which is spec'ed at 1.2ns nearer 800Mhz. So the X1900XT has a hard limit on potential so would be better than the original recommendation, but the X1900XTX should have the same potential OC-ability as the X1950XT. The X1950XTX has the new GDDR4 BC09 chips rated at 0.9ns or 1100Mhz, so it has alot more headroom still, and achieves it at less power draw.

But I don't know it any X1900XTXs are close to the same price as an X1950XT, but your argument is good for the advantage of the X1950Xt over the X1900XT though.

I have not even tried to OC the X1950XT. I don't even have the HIS software installled for running the card at it's advertised Turbo speeds. But the X1800XT I OC'ed. It added a good 6-7 fps to those Carbon results, whick made a big differnece for that game in playing maxed without motion blur.

Yeah I agree, even Digit-Life hasn't revamped their AGP list since last March, and it has nowhere near the number of tests that Digit-Life does (imagine the hours for 10-15 AGP cards times that many test runs).

You'd pretty much have to have parrallel setups, because one rig doing all that would take weeks, even without breaks for driver cleaning and re-installs, etc.

Exactly! The thing about Oblivion is that people monkey around with things like grass settings and then test at lower levels to achieve 30fps avg @ 16x12 as if that's more valid than 40fps with things enabled at 1280x1024. Why are you buying Oblivion if you aren't going for the immersive environement? If it's just for story line and combat then people should play with OldBlivion mods.

The most important thing about Xbit's reviews is they show the min FPS, and that's what really shows in Oblivion, and also what show the X1600 and X1900's strengths, because while they may offer 30fps, it's a consistent 30fps with dips to like 24 fps, meawhile the GF7s tend to jump around more, especially outside.

Look at this [H] review;
http://enthusiast.hardocp.com/arti [...] VzaWFzdA==

I mean who plays Oblivion with NO GRASS! Sure make the grass bigger if you want, but removing it, you might as well play wireframe!
At least [H] points out that's an issue with the screenies at the bottom of the page, but some reviewers just turn stuff off/down to eliminate the low end drops on the GF7 series, and that's ridiculous. I just wish they played oblivion apples to apples to show the stress drops between systems.
Anyone who's played Oblivion knows it's about the environment as well as the rest of the stuff, and good solid FPS not 100fps followed by 5 fps, is what matters. Don't get me started.

Yep, exactly. I like the chances of the GF8600 series, and the X2600 series. I of course would also like to know what their laptop prospects are. BTW, I finally decided to try and give Vista a pass for the first little while. I thought some of the DRM stuff would dissapear, but nope, and I like the media side of things too much to allow my system to get that locked up, and I speak as someone with over 300 store bought DVD and more than that CDs (and DVD-Audio, AND btw, I suggest you get Love by the Beatles if you have DVD-Audio, AWESOME, and cheap [you can get it bundled in a 2 disc set, CD&DVD-Audio, for $2 more than just the CD!]).

Speaking of Cheap the PSP is selling at WalMart up here for $148 CAD, it's such a good price I had to buy a second one.

Oh no problem, it obviously happens to the best once in a while.

I tell ya, you had me worried and I was about to yank the iceQ to see for myself. My satisfaction in the card would go way down if I didn't have 1.0ns as I do plan to OC it someday. Seriously, i am lovin' it, but if it was 1.2ns I'd feel jipped.

Anyway, I see Tekbunker shows 1.0ns for the Sapphire, but the His reviewer doesn't seem to yank the cooler. *sigh of relief*

ooh...i've never done any benches :S. and i hate the motion blur, so i guess thats maybe a reason why ^^. what program did you use to bench ?

awesome.. thx!

Here are some NFS Carbon benchmarks I did. While my 7800gs isn't no top-performer, it is an Nvidia GeForce 7 card. Here are the settings I ran it at, unless stated otherwise:




First run: Everything maxed, World detail set at high, not maximum, and 8x af was enabled through Nvidia's control panel.
Avg: 26.456 - Min: 21 - Max: 35

Second run: Same as above, motion blur off.
Avg: 36.959 - Min: 30 - Max: 48

Third run: Same as above, motion blur off, shader detail at medium
Avg: 46.221 - Min: 37 - Max: 60
*This is playable to me. Nothing else is smooth enough for game play

Fourth run: Same as first run, but camera view is from inside car
Avg: 27.303 - Min: 22 - Max: 39

I raced on the:

using the Nissan 240SX from the bonus cars section. (Wow that thing has uber handling, so does the other unlockable cars. The Eclipse and the MR2)

so I had no opponents, and traffic was turned off, so keep in mind that these benchmarks have pretty high framerates for those reasons.

EDIT: And just for the hell of it, here is the game MAXED (1280x1024, 4xaa, 8xaf, everything set as high as possible)
Avg: 15.894 - Min: 12 - Max: 20

haaah, you are right!.. i put on blur along with the fraps, managed to sustain 15-20 fps . oh well, good thing i hate motion blur!

are you going pci-e or agp 8x? the banwidth on that asrock board is not pci-e 16x it is 4x which will considerably diminish the performance of any pci-e card that you install.

cool, congradulations on the new home.