ATI X1900XTX vs ATI X1800XT

[OWA]

Just in case anyone has a X1800XT and was wondering how much improvement they'd get by purchasing a X1900XTX, I thought I'd post a few quick comparisons from my system. I recently sold my X1800XT and purchased the X1900XTX to replace it. Nothing else in the system was changed except the Cat drivers since I had to use the ones from the X1900XTX CD.

Stock settings for the system and cards, resolution is 1280x1024. Looks like I should probably try to increase the resolution, maybe settings, to hopefully show more of a difference.

3DMark06
X1800XT = 3689 vs 4644 X1900XTX

3DMark05
X1800XT = 8892 vs 10,220 X1900XTX

FEAR 4xAA/16xAF, no soft-shadows
X1800XT = 27/50/120 (0/21/79) vs 27/61/167 (0/32/68) X1900XTX

Doom 4xAA/8xAF, Ultra Quality
X1800XT = 72.7 vs 76.8 X1900XTX

HL2 Lost Coast 4xAA/16xAF, Max settings
X1800XT = 56.33 vs 62.93 X1900XTX
Edit: Lost Coast results obtained using timedemo lost_coast (not the video stress test).

http://img232.imageshack.us/img232/3880/x1900xtx0012pc.th.jpg (http://img232.imageshack.us/my.php?image=x1900xtx0012pc.jpg)

[Lfctony]

Well, the X1900 has almost no advantage in some games and a lot of difference in others. Like in FEAR, the average FPS are 20% higher, but min FPS are the same. Thus it's not really a significant advantage. Anyway, overall it's a much better card than the X1800XT, but I'm getting the X1900XT and not the XTX, I don't think the little extra performance is worth it.

[Henno]

Nothing else in the system was changed except the Cat drivers since I had to use the ones from the X1900XTX CD.

The driver most reviewers used is 6.2 beta which should give you a boost over the modified 5.13 (which is on the CD).
6.2 WHQL should be here soon.

[OWA]

Thanks I was wondering about that and was getting ready to look for info on the drivers. I had initially just tried installing the Cat 6.1s but they didn't recognize the card. I'm not sure how to read the driver version that the CD installed.

Edit: See the attachment.

[OWA]

Stock settings vs Overdrive settings, resolution is 1280x1024, CPU is still stock. Seems to be stable at the overdrive recommended settings of 687/797. The temp seems similar. In the 50s at idle and around upper 70s low 80s under load. That's about the same as the X1800XT as well.

3DMark06
X1900XTX = 4644 vs 4774 overclocked

3DMark05
X1900XTX = 10,220 vs 10,351 overclocked

FEAR 4xAA/16xAF, no soft-shadows
X1900XTX = 27/61/167 (0/32/68) vs 26/63/175 (0/28/72) overclocked

Doom 4xAA/8xAF, Ultra Quality
X1900XTX = 76.8 vs 78.4 overclocked

HL2 Lost Coast 4xAA/16xAF, Max settings
X1900XTX = 62.93 vs 66.85 overclocked
Edit: Results obtained using timedemo lost_coast

[OWA]

CPU overclocked to 2.58, XTX still oc'ed at 687/797.

Stock CPU, X1900XTX overclocked vs Overclocked CPU (2.58), XTX oc'ed

3DMark06
X1900XTX = 4774 vs 4876 overclocked card and cpu Edit (latest): 4909

3DMark05
X1900XTX = 10,351 vs 10,869 overclocked card and cpu Edit (latest): 10,928

FEAR 4xAA/16xAF, no soft-shadows
X1900XTX = 26/63/175 (0/28/72) vs 30/65/173 (0/28/72) overclocked card and cpu

Doom 4xAA/8xAF, Ultra Quality
X1900XTX = 78.4 vs 80.7 overclocked card and cpu

HL2 Lost Coast 4xAA/16xAF, Max settings
X1900XTX = 66.85 vs 68.15 overclocked card and cpu
Edit: Results obtained using timedemo lost_coast

[Henno]

Xbitlabs shows a much bigger difference between the 1800XT/1900XTX, but that's without FSAA. :o

http://www.xbitlabs.com/images/video/radeon-x1900xtx-gameperformance/hl2lc_pure.gif



Btw, more sites like Driverheaven, firingsquad and Techreport use no AA with TLC. WTH is that i don't know.

[OWA]

I wonder why they didn't do something other than pure speed for that.

Edit: Also, I just realized that HL2 Lost Coast has a video stress test option. My results are *not* from that. I was using timedemo lost_coast and it does something different. To tell you the truth, I thought CS:Source was the only one that had the video stress test. I guess I've just always missed that option in HL2 Lost Coast.

Here are my results from the video stress test in Lost Coast: 89.2 fps.

[OWA]

Looks like I don't have the locking issue with Quake 4 in Ultra Quality mode that HardOCP mentioned. I was worried since I thought that had been fixed already. It works but it's not playable with just 1 gig of RAM due to the hitching. I really need to get another gig for that system.

[jAkUp]

How is IQ? Techreport is noticing some strange things with the IQ particularly in 3DMark06.

[hazindu]

How is IQ? Techreport is noticing some strange things with the IQ particularly in 3DMark06.I haven't noticed any IQ problems with my x1900xt, although I don't play 3dmark. In fact, the Quake 4 problem I had with the x1600pro is gone.

[radekhulan]

There seems to be problem with Crossfire AA as well, NVIDIA SLI has much better picture (from FiringSquad.com):

http://radekhulan.cz/img/1/20060130-kvalita-ati-nvidia.jpg

http://radekhulan.cz/img/1/20060130-ati-nvidia-2.jpg

[eli]

[

FEAR 4xAA/16xAF, no soft-shadows
X1900XTX = 26/63/175 (0/28/72) vs 30/65/173 (0/28/72) overclocked card and c

Hi m8, what resolution is that for fear, 1280x1024?

[Medion]

radekhulan - To be honest, I can't tell much of a difference between either set of screenshots. Granted, it's partly due to them being resized, but regardless.

NV's screentshots look just a hair sharper, but I'd be hard pressed to notice a difference without them side-by-side. Just further proof, IMO, that these IQ arguments are pointless, no matter which camp you're on.

[OWA]

[

FEAR 4xAA/16xAF, no soft-shadows
X1900XTX = 26/63/175 (0/28/72) vs 30/65/173 (0/28/72) overclocked card and c

Hi m8, what resolution is that for fear, 1280x1024?
Yeah, all my results were for 1280x1024 since the main LCD I use for that system is 1280x1024. I'll try to run the benches on the Dell LCD soon so I can use 1920x1200. I think that should show a little more difference between them.

How is IQ? Techreport is noticing some strange things with the IQ particularly in 3DMark06.
I took a quick look at that and don't think I saw any problems. I think I'm using different drivers though. The cards apparently ship with modified 5.13 drivers and the reviewers are using newer drivers than that (what someone here told me anyway). That might explain why I didn't see the 3Dmark06 issue and why I didn't have any problems with Quake 4 in UQ mode.

[radekhulan]

radekhulan - To be honest, I can't tell much of a difference between either set of screenshots. Granted, it's partly due to them being resized, but regardless.

NV's screentshots look just a hair sharper, but I'd be hard pressed to notice a difference without them side-by-side. Just further proof, IMO, that these IQ arguments are pointless, no matter which camp you're on.

There is a huge difference, ATI does not draw several objects, and NVIDIA has much sharper picture. Maybe matter of taste, but important for myself, have been considering X1900XT CF, but .. no way. Will wait for 7900 in SLI.

[fatal1ty]

There is a huge difference, ATI does not draw several objects, and NVIDIA has much sharper picture. Maybe matter of taste, but important for myself, have been considering X1900XT CF, but .. no way. Will wait for 7900 in SLI.

CF is in it's infancy still - SLI didn't work right for along time let alone out of the box. Give ATI a chance to put out a few sets of drivers and Im sure everything will be resolved. Nvidia has had alot more time to play with drivers and dual cards.

[fatal1ty]

Hey OWA -

replaced the stock X1900XTX cooler on my card with the Zalman VF 700 and some arctic silver 5. Load temps tops about 63 deg c, idle is 36 c. I'm running max CCC of 690 / 800 with the XTX.

3dmark 03 = 20,242; 3dmark 05 = 12,484; 3dmark 06 = 5,590.

Just wondering how happy you are with yours!? Hopefully you got a good OC'r this time around!

(nana2)

[OWA]

It seems to overclock better than the X1800XT although I haven't found what the max is yet. It initially chose the max for both (using Overdrive). Aftering running it for a while to heat things up and running the clock-config again, it chose close to the max for the core but the memory was close to the default (I may have that backwards so I'm running it right now to see). It hasn't locked up on me with any of the overclock settings I've tried so far.

So far, the difference between the X1800XT and the X1900XTX hasn't been as big as I expected but I think it's b/c of running it at 1280x1024 with just a XP 4000+. I plan to try it with the 24" LCD soon so I can use higher resolutions. I think that should show more of a difference between them. I hope to get a X2 CPU for that system soon.

Wow, your loads temps are really good. Mine hits the upper 70s and low 80s. BTW, are you using the drivers from the CD or have you been able to grab the ones that the reviewers were using? Also, have you been able to submit any of the X1900 results to Futuremark? It won't let me submit anything.

Edit: Overdrive is consistently choosing max settings now even after running 3DMark06 and going back and running the clock config utility right after it finishes. So, I'm not sure what happened earlier.

[Medion]

There is a huge difference, ATI does not draw several objects, and NVIDIA has much sharper picture. Maybe matter of taste, but important for myself, have been considering X1900XT CF, but .. no way. Will wait for 7900 in SLI.

Not seeing it. Even examining the pictures more closely, this is all I could find.

First set.
+NV - NV image looks a tad sharper, crosshair is semi transparent in NV picture as opposed to being entirely opaque in the ATI shot, clump of grass to the left of crosshair is drawn correctly
+ATI - AF looks smoother (NV shot looks like termites, and whenever I see that, it usually has an annoying crawling effect in motion)

Second set:
+NV - Image looks a tad sharper
+ATI - left railing looks like AA was applied better.

That's me nitpicking, but overall, they look the same.

[fatal1ty]

It seems to overclock better than the X1800XT although I haven't found what the max is yet. It initially chose the max for both (using Overdrive). Aftering running it for a while to heat things up and running the clock-config again, it chose close to the max for the core but the memory was close to the default (I may have that backwards so I'm running it right now to see). It hasn't locked up on me with any of the overclock settings I've tried so far.

So far, the difference between the X1800XT and the X1900XTX hasn't been as big as I expected but I think it's b/c of running it at 1280x1024 with just a XP 4000+. I plan to try it with the 24" LCD soon so I can use higher resolutions. I think that should show more of a difference between them. I hope to get a X2 CPU for that system soon.

Wow, your loads temps are really good. Mine hits the upper 70s and low 80s. BTW, are you using the drivers from the CD or have you been able to grab the ones that the reviewers were using? Also, have you been able to submit any of the X1900 results to Futuremark? It won't let me submit anything.

Edit: Overdrive is consistently choosing max settings now even after running 3DMark06 and going back and running the clock config utility right after it finishes. So, I'm not sure what happened earlier.


I managed to get the reviewers set which is 8.20.3 (I think.) the drivers contain 8.20.4 so I think the ones I am using are actually older but they work. My load with the stock cooler was in the 90s, the Zalman kicks butt but the Case has to have decent airflow for it to work so well.

I can submit for futuremark - but it says the drivers arent approved.

[jAkUp]

CHeck it out guys:

http://www.techreport.com/onearticle.x/9324

It appears to have some IQ problems as of now, likely driver related though.

[Xasz]

There is a huge difference, ATI does not draw several objects, and NVIDIA has much sharper picture. Maybe matter of taste, but important for myself, have been considering X1900XT CF, but .. no way. Will wait for 7900 in SLI.

I'm going to have to agree with Medion - I don't see much of a difference, as in barely anything. Bigger shots are needed.

[OWA]

I'm going to have to agree with Medion - I don't see much of a difference, as in barely anything. Bigger shots are needed.
The first set looks like the same issue that was reported with the X1800XT when it was released (i.e. in HL2, the LOD distance appeared to be different on ATI cards for some people so it didn't end up rendering everything).

[fatal1ty]

CHeck it out guys:

http://www.techreport.com/onearticle.x/9324

It appears to have some IQ problems as of now, likely driver related though.

Thats old info - it's been out since last week. It is driver related.

6.2 should fix it and offer a performance increase for X1900 owners :D