View Full Version : ATI X1900XTX vs ATI X1800XT
Hey OWA what was your stock clocks when you got the card ?
the OD utility & ATitool are reporting 500/600 to me as stock , didnt pay attention to which cat version i have though.
i got ~4700 on FX-55 , stock clocks , 512MB DDR400.
The clock speeds threw me a bit when I first got the X1800XT since it showed a lot lower than I expected. Apparently, the X1800/X1900s cards are now using 2D/3D clock speeds as well so when looking at the Overdrive panel, for example, you'll see the current clock speed that is low and you should see a requested clock speed value that is higher. In 3D, it'll use the requested value and that value should be pretty close to the specs they advertise. I say close since mine shows like 1 or 2 under what the spec shows. When using their auto config utility, the overdrive (requested) setting essentially maxes out giving me 687/797.
Edit: Your score is close to what I got at stock. With the CD drivers, I got approx. 4600 at stock settings (that was with the AMD 4000+ that I was using at the time). With the 6.2betas/officials, 3DMark06 went up around 200-300 points. I'll have to look through the threads some more to see what my stock score with the 4800+ was.
500/600 = 2D clockspeeds.
I use the Ati Tool 0.25 beta13 now to adjust fanspeeds becuz the X1900XTX can get pretty hot. Atm i'm using the beta beta 6.2 cats instead of the WHQL's.
Official 6.2 cat : 5309 3d06 points at ~89C
Beta 6.2 cat : 6001 3d06 points at ~75C
fatal1ty
02-15-06, 03:19 PM
Yea - oh well everyone sees something different I guess. To me Nvidias AA was never all that impressive. Maybe thats because I was always using ATIs 6x and never had SLI to run anything higher than 4x on the Nvidia without a huge performance drop - OWA you were using 8xSLI - I never thought having to buy 2 video cards to get decent AA was worth it...
Thats not even what Im talking about...
I'm talking about the Nvidia drivers not applying AA to everything in a scene. I have seen it first hand many times but someone here had posted a shot of Guild Wars ( I believe) and in it there was Lava coming down. Everything else was aliased except the edge of the lava. Others posted their screenshots showing it was working. I think the guy reinstalled the drivers to correct it. <I looked, I can't find it now>
Riptide
02-15-06, 06:12 PM
Apparently, the X1800/X1900s cards are now using 2D/3D clock speeds as well so when looking at the Overdrive panel, for example, you'll see the current clock speed that is low and you should see a requested clock speed value that is higher.FYI ATITool disables this switch and keeps the 3D clock speed for 2D also. I wouldn't recommend that program personally as I've had problems with it hard locking my machine when running 3dmark. At least the last beta did anyway.
Honestly I just use CCC and that's it. The ATI overclocker tool doesn't save settings and apply on startup and w/ATI Tool messed up there isn't much else to use.
Yea - oh well everyone sees something different I guess. To me Nvidias AA was never all that impressive. Maybe thats because I was always using ATIs 6x and never had SLI to run anything higher than 4x on the Nvidia without a huge performance drop - OWA you were using 8xSLI - I never thought having to buy 2 video cards to get decent AA was worth it...
Thats not even what Im talking about...
I'm talking about the Nvidia drivers not applying AA to everything in a scene. I have seen it first hand many times but someone here had posted a shot of Guild Wars ( I believe) and in it there was Lava coming down. Everything else was aliased except the edge of the lava. Others posted their screenshots showing it was working. I think the guy reinstalled the drivers to correct it. <I looked, I can't find it now>
Nah, SLI isn't needed and I rarely use SLI AA. I use 8xS in some games though. The only game where I use SLI AA is Unreal II and that's because the performance SLI mode does nothing and even without the performance mode, it's still way up there in the framerate. So, why not put the 2nd card to work doing AA.
Like I mentioned earlier, you can find games or circumstances where one or the other doesn't look as good, sometimes because of bugs, somethings b/c of the situation.
And to quote saturnotaku who recently switched to the X1800XT...
When you're talking about raw performance, yes. However, on the 7800 cards 4x AA with transparency and gamma correct looks every bit as good, if not better than ATI's 6x and the performance hit is far less than 8xS. The point I'm really trying to make is that NV's AA offers a lot more flexbility, while still looking very good in the process. ATI no longer has the clear advantage it once did.
FYI ATITool disables this switch and keeps the 3D clock speed for 2D also. I wouldn't recommend that program personally as I've had problems with it hard locking my machine when running 3dmark. At least the last beta did anyway.
Honestly I just use CCC and that's it. The ATI overclocker tool doesn't save settings and apply on startup and w/ATI Tool messed up there isn't much else to use.
Thanks for the info. Yeah, I just use the CCC as well. Just feels safer since it is the drivers doing it and it works.
Hey OWA what was your stock clocks when you got the card ?
the OD utility & ATitool are reporting 500/600 to me as stock , didnt pay attention to which cat version i have though.
i got ~4700 on FX-55 , stock clocks , 512MB DDR400.
With the 4800+ at stock and the XTX at stock, I get 5762. So, the combination of upgrading to the 4800+ and the Cat 6.2s (and possibly another gig since I added that when I went to the 4800) bumped my stock score up from 4600 or so to 5762.
With the CPU still at stock and the XTX overclocked to 687/797, I get 5949.
Thanks for the explanation owa. I have always heard that ati had slightly better and sharper image quality but it looks like that isnt the case.
ATI use to be known for better AA and Nvidia use to be known for having better AF. Now it seems things are reversed to a certain extent but in the end, I think they're pretty close (especially true when gaming and not studying screenshots).
I've always thought Nvidia had the "sharper" image though since ATI's AA method seemed to create a "softer" image. Lately though, that doesn't always hold up. When I tried comparing max IQ with Boiling Point (I think the thread is in this part of the forum also) just after I got the X1800XT, ATI did have the sharper image in that case and Nvidia had the softer image. Looked like the Super Sampling might have changed things a little but it usually seems to make the image sharper. I never did track down the issue to find out what caused that.
If there are certain games you're interested in, if I have them, I can try taking a few screenshots so you can see somewhat how it'll look. Sometimes, you really need to see things in motion though to make the best judgement but screenshots at least give you a rough estimate.
In any case, I wouldn't just take my word for it. I'd check various reviews, IQ comparisons, etc. that you can find around the web/forums. IQ is so subjective you should probably get lots of different opinions especially if you're interested in a particular game.
I agree with Riptide , the ATitool is messing up my clocks.
the CCC OD was reporting the correct clocks when i unlocked teh OD feature (650/775) , but then trying the automated setting it gave me different numbers.
i'm quite confused with the clocks right now!
think i'll grab the 6.2 for now & play stock until i fix my memory issue. most likely i'm going to replace the mobo.
FYI ATITool disables this switch and keeps the 3D clock speed for 2D also. I wouldn't recommend that program personally as I've had problems with it hard locking my machine when running 3dmark. At least the last beta did anyway.
Honestly I just use CCC and that's it. The ATI overclocker tool doesn't save settings and apply on startup and w/ATI Tool messed up there isn't much else to use.
There's an option to maintain the CCC 2D and 3D clockspeed settings. Upon first install of Ati Tool there's a popup window which asks you to choose.
I'd advise to use Ati Tool for artifact scanning & fanspeed profiles. I recommend it over Ati Overclocker becuz that program actually sets the memory at relaxed latencies ; which is not a very good solution imho.
Ati Tool should work with all unmodified catalyst drivers, including the 6.2 Beta ; I haven't had any lockups with 6.2 beta + AtiTool 0.25 B13.
From what I've seen from recent X1900xtx reviews is that ~ 685 gpu oc's are pretty good for that card ; which in turn reassures me a bit more about my xtx.
Ati Tool should work with all unmodified catalyst drivers, including the 6.2 Beta ; I haven't had any lockups with 6.2 beta + AtiTool 0.25 B13.
What about AtiTool with the official drivers? I started having issues with Overdrive in the offiicials last night and that made me a little nervous about using it. I'm setting them manually, at the moment, in overdrive. Does ATI Tool automatically detect clocks or is it strictly manual overclocking?
One thing I did find is that gaming at 1920x1200 with a single card is more plausible than I thought.
I think I'm going to have to reconsider my comment. Now that I've got the 37" LCD connected to the XTX, I was going through various games last night to see how things played and games aren't really tracking as well with the benchmarks as I thought they would. The first three I tried, I was in the 20s right off. Doom 3 opening area, in Quake 4 opening area after just going inside the building and in Far Cry on the steam level (although there I had AA+HDR on). So, even though the resolution is a little less than what I was running the benchmarks at (1920x1200 for the benchmarks vs 1920x1080 for some gaming last night), it doesn't look like I'd be able to use 4xAA/16xAF at 1920x1080 as often as the benches would seem to indicate.
Ati Tool's 'find max core' & 'find max mem' are basically the same as the Overdrive functions in CCC.
Only when Overdrive Oc's are locking up becuz Ati Tool is installed ; yes
then it would be better to ditch it.
Ati Tool 'find max core' + official 6.2 = a black screen for me.
Ati Tool (using it for my custom fanspeeds profile) + OC by means of CCC Overdrive, beta 6.2 = :thumbsup:
Riptide
02-16-06, 12:00 PM
AFAIK the artifact scanner in the latest beta does not work properly. At least not w/x1800 or x1900.
What are your temps OWA ?? i reach 90C in FEAR & 80s in COD2/3DMark06 .. thats on stock clocks (650/1550)
the card is surely hot!!
EDIT : i see Knot having nearly same temps as mine .. maybe its a cat 6.2 issue ?
Yeah, my temps with all the drivers (CD drivers, beta 6.2s and now the official 6.2s) have been in the mid-to-upper 50s at idle and in the mid-to-upper 80s under load. This was just while running 3DMark06. I'll try FEAR/CoD2 to see how that is. After the auto-detect process, I've seen 94C. That's the highest I've seen so far.
This is the combo I use with great success:
ATI Tool Beta 13
ATI Tray Tools
CAT 6.2 Beta (haven't tried with the officials yet)
I just bypass CCC completely by using ATI Tray Tools (be sure to disable the overclocking function when it asks you about it during install). I used to use ATI Tools + the Overclocking Tool to oc, but now I just use ATI Tool as it seems give better performance. Just disable the ATI Hotkey and ATI Smart like you would normally. Than set your voltages to whatever you want (1.4V on the core seems to be stock for 3D) and type in the clock speeds you want.
The ATI artifact scanner works very well with the new beta. No more artifacts at stock speeds.
My quick impression on the card so far that it seems as fast as my 7800GTX SLI 256MB system.
no i'm not exaggerating , but from a quick test on FEAR & COD2 i can see that both PCs are running the 2 games as fast as each other.
of course framerate wise the GTX SLI is faster but as i've already said in other threads that SLI needs higher framerate to get smooth experiance (in short 35fps+ for SLI , 25-30fps+ in single card .. for me anyway).
so experiance wise they are both the same (or maybe the XTX is faster , but i need to investgate this further more before giving my final impressions!)
heres a comparsion bench in FEAR :
1600x1200,2xAA,8xAF,noSS
7800GTX SLI (256MB) [HQ]: min=38(0%) , av=61(2%) , max=188(98%)
X1900XTX [HQ , Cat AI enabled]: min=32(0%) , av=64(23%), max=161(77%)
its not all what it seems though as with the 7800GTX SLI the filtering was set to HQ (no shimmering) , on the ATi it was set to HQ , but A.I. = std . so i was seeing some shimmering on the XTX with these settings.
on COD2 i ran the game @1280x1024,4xAA,8xAF .. this one aswell was running fine & in some spots my framerate was higher than the 7800GTX SLI PC!!
(i'm waiting to reach my benchmark point in the game to make a valid comparsion)
next i'll benchmark X3 .. stay tuned! :D
Thanks for sharing your experience SH64.
SH64,
what are your XTX clocks & temps now ?
My X2 4800 = at stock and cat 6.2 Beta for the XTX
XTX 2D mode = 500/600 ~45C idle, running 27% fanspeed by means of Ati Tool 0.25 Beta13.
XTX 3D mode load = 676/786 (by means of CCC OD) ~ 65C to 75C, running 82% fanspeeds using Ati Tool again.
The Omega's 6.1 hard crashed my first attempt to use Ati Tray Tools, so I'm a bit hesitant to use Tray Tools now....
SH64,
what are your XTX clocks & temps now ?
As i said above .. stock clocks (650/1550). at idle its on 50s , with load its on 80s & up to 90c.
thats using the official 6.2 drivers.
Thanks for sharing your experience SH64.
You're welcome! :)
i've just benchmarked X3:rolling demo & heres a very brief comparsion of the scores:
@1600x1200,4xAA,8xAF,HQ,AQ,EAX this is what i get :
4800+/7800GTX SLI = 55.8fps
FX-55/X1900XTX = 55.7fps
amazing , isnt it ?
if anyone's interested on the full list of the scores including the min/av/max for each scene just let me know & i'll paste them here.
btw my temp after the bench was 88C.
My quick impression on the card so far that it seems as fast as my 7800GTX SLI 256MB system.
no i'm not exaggerating , but from a quick test on FEAR & COD2 i can see that both PCs are running the 2 games as fast as each other.
of course framerate wise the GTX SLI is faster but as i've already said in other threads that SLI needs higher framerate to get smooth experiance (in short 35fps+ for SLI , 25-30fps+ in single card .. for me anyway).
Can you run anything higher than 1600x1200 like maybe 1920x1200 or preferably 1920x1080 or is your monitor limited to 1600x1200? I'd like to see how the SLI GTX 256 does at 1920x1080 so I can compare it to the XTX. Right now, it's not handling 1920x1080 very well and I thought maybe the SLI GTX 256s would be a better fit. If the SLI setup isn't any faster, it won't do me any good to try that route.
My monitor can run up to 2048x1536 (not sure if 1920x1080/1200 is supported natively or not though but i'll look into it) , what game(s) do you like to check ?
The ones I compared in the last chart would be ideal (the ones you have anyway). Those were relatively easy to benchmark. I think I did some 1920x1200/1920x1080 with my CRT but it didn't look very good. It also had a max of 2048x1536.
Also, spot checks with CoD2 and NFSW would be good also if possible. I didn't try to benchmark those b/c they were more difficult. But, just a spot check at the start of some track or level would probably be enough to give me a rough idea of how it'll be.
Edit: Thanks in advance but don't spend a huge amount of time on it unless you enjoy it. I'm currently trying to find some of the reviews that included widescreen results also just to see if they had any SLI results listed.
Edit 2: Hopefully you'll have a few of them and be able to run a few. I'm having a hard time finding widescreen results in the reviews. A couple had some higher than widescreen results that I guess give some indication. Disregarding wanting WS results and just looking at other results, it doesn't look like it'll do me any good to switch. It looks like they're pretty close when the framerates are lower and when SLI is really way ahead, the framerate for both solutions is usually acceptable. Couple that with some games doing better with 512M, I don't think my idea was very good.
Edit 3: While I'd still be interested in your widescreen results, I think I've found a solution to my problem. With a little rearranging, a longer DVI cable and if the VGA port on the 37" LCD supports 1920x1080 as well (checking the very poor manual, I think it should), I should be able to have both systems connected to the 37" LCD at 1920x1080 and both connected to the Dell 24" LCD at 1920x1200. That's almost a best case scenario with the only downside (that I can think of at the moment) being that the designated secondary systems will be using analog connections. Analog looks pretty good on the Dell LCD. Haven't tried that on the Westy yet. Actually, I'll probably need a longer vga cable as well. The ports on the Westy are so high, a lot of the cable length is wasted running along the back of the LCD.
vBulletin® v3.7.1, Copyright ©2000-2010, Jelsoft Enterprises Ltd.