Problems with my x1800XT (Prev. "Are these decent 3DMark scores?")

[jimbob200521]

Originally posted by: Avalon
Why is everyone here telling him it is his CPU? I am so sick and tired of you people auto-responding to every low 3dmark05 score that it is the person's CPU. The benchmark is a GPU benchmark, not a system benchmark like 01 was. His CPU is not going to cause any noticeable change in score. He is NOT supposed to be getting that low of a score, and it is NOT because of his P-D.

It took you guys 30 posts to finally recommend he try reinstalling his drivers. Jesus.
Well, since a format and reinstall did not work, and reinstalling drivers didn't work. This leads me to believe it is a hardware issue.

I did not see it in the thread, but did you post your GPU temps?
How are the voltage rails on your powersupply? I know Enermax makes good stuff, but it's always good to check to make sure your unit isn't flaking out. Do you have any other PCI-e card laying around, or one you can borrow from a friend?

Something just to throw into your system and run 3dmark05 on, to see if everything works correctly and you are getting the score you are supposed to for that card.

Do you have a spare PC with PCI-e or a friend that has such a PC? I'd recommend trying the card in a somewhat similar system.

agreed.

procs do not usually limit in 3dmark 05, i have a 3.6ghz prescott (oc'd from 3.0), and a 7800gt oc'd to 486/1150. my scores are below. im not sure, but they seem pretty average, so theres something to go by

 

[Addle]

Here is what Rivatuner is telling me:

----------------------------------------------------------------
$ffffffffff ATI specific display adapter information
$ffffffffff ----------------------------------------------------------------
$0900000000 Graphics core : R520 (16x1)
$0900000002 Memory bus : 256-bit
$0900000001 Memory type : DDR3 SGRAM
$0900000003 Memory amount : 512MB
$0900000004 Core clock : 594.000MHz
$0900000005 Memory clock : 693.000MHz (1386.000MHz effective)
$0900000006 Reference clock : 27.000MHz

Does this all look right? I thought my card should be running 625/1500?

Also, this is the powersupply I'm using.

 

[Elfear]

That looks like a decent PSU. I loved the Enermax I used to have.

When you open Rivatuner there is a little tab you can open (found a picture here) at the top right of the box. The second icon from the right, which should be a magnifying glass and a piece of filmstrip, will be the hardware monitor you need (here is a picture of what it should look like). Just open that up and than run 3DMark05 again to verify that the 3D clocks are working okay. You should see the clocks jump up to 625/1500 when a 3D app starts running.

 

[Addle]

Okie, ya, the clocks got up to 621/1494 while running 3DMark. However, I noticed the core temp was as high as 82C at one point. Is that normal?? To me that seems like one tick away from meltdown, but I'm still a novice compared to some of you guys.

Lemme know what you think.

 

[Powermoloch]

LOL bunch of noobs in this forum, the CPU has no impact in his score. It's the card.

 

[mwmorph]

Originally posted by: Powermoloch
LOL bunch of noobs in this forum, the CPU has no impact in his score. It's the card.

lol you must be referring ot youraself. why would the goddamn orb and all reviews tell a different story? i guess everyone else is a noob. hmm?

no 3dmark here but you can see how 3d performance is much lower

3dmark 01,03,05 and Aquamark here show big drops in scores with a top of the line card, in this case a 7800gtx, which is comparable to the x1800xt. Similar score as the op once you figure in the 200mhz clock boost huh?

hey! whatda ya know? 3dmark05 results are in once again and guess who's lagging behind? this time they're using "only" x850xts and the pentium d is still cpu limited!

Now with p-d840EE. see how it's still slower than other cpus and the variation in scores? even with a much slower x800xt?

now op i hadnt realized you were having problems with BF2. that might be a problem. have you tried to use fraps to see exactly how low your fps go? did you know the latency at the moment of the slowdown? Do you know your temps for cpu and gpu at moment of slowdown? How long does the slowdown last? just for a split second during firefights or for a long time, maybe half a munute or more?

 

[eva2000]

Originally posted by: Addle
These are the scores I got for my new gaming system, wasn't really sure if these were average given my system specs. This was my first time using 3DMark.

Scores:
'03 - 15817
'05 - 6907

System:
Pentium D 830
Ati x1800 XT
d945 Intel Mobo
2GB PQI DDR2533 CL3

You're cpu limited you need a much beefier cpu to take full advantage of the card i.e. 9-10k 3dmark05 is pretty easy with AMD64 Opteron or Sandiego 1MB cache cpu at 2.8+ ghz.

My 512MB HIS X1800XT results at http://i4memory.com/showthread.php?t=1842 even managed 11,007 3dmark05 when oc'd

 

[Crescent13]

Your scores suck because of the pentium D. It's absolute crap and it's slowing down the rest of your system. Seriously, if you want to get all the performance you paid for with that new card, get an AMD.

I get 4055 with my 6600GT and AMD 3200+. Get a clue.

 

[Elfear]

Originally posted by: Crescent13
Your scores suck because of the pentium D. It's absolute crap and it's slowing down the rest of your system. Seriously, if you want to get all the performance you paid for with that new card, get an AMD.

I get 4055 with my 6600GT and AMD 3200+. Get a clue.

Boy, who peed in your Wheaties this morning?

Originally posted by: Addle
Okie, ya, the clocks got up to 621/1494 while running 3DMark. However, I noticed the core temp was as high as 82C at one point. Is that normal?? To me that seems like one tick away from meltdown, but I'm still a novice compared to some of you guys.

Lemme know what you think.

Weird. Well at least you verified that the card is clocking properly. Have you tried ocing your chip just to see if that would add anything? I wish I had more experience with Intel chips. I used to have a 3.2GHz Northwood and after switching to a 3200+ A64 my 3DMark05 score only went up 100-200 points when paired with a 6800GT. Doesn't seem like your cpu should affect you that much.

If it was just 3DMark stuff than it wouldn't be that big of deal but it sounds like your games are suffering too.

EDIT: Have you verified that the Catalyst AI is turned on or off? I'm pretty sure it's supposed to be off but I can't remember off the top of my head.

 

[Avalon]

Originally posted by: mwmorph

Originally posted by: Powermoloch
LOL bunch of noobs in this forum, the CPU has no impact in his score. It's the card.

lol you must be referring ot youraself. why would the goddamn orb and all reviews tell a different story? i guess everyone else is a noob. hmm?

no 3dmark here but you can see how 3d performance is much lower

3dmark 01,03,05 and Aquamark here show big drops in scores with a top of the line card, in this case a 7800gtx, which is comparable to the x1800xt. Similar score as the op once you figure in the 200mhz clock boost huh?

hey! whatda ya know? 3dmark05 results are in once again and guess who's lagging behind? this time they're using "only" x850xts and the pentium d is still cpu limited!

Now with p-d840EE. see how it's still slower than other cpus and the variation in scores? even with a much slower x800xt?

now op i hadnt realized you were having problems with BF2. that might be a problem. have you tried to use fraps to see exactly how low your fps go? did you know the latency at the moment of the slowdown? Do you know your temps for cpu and gpu at moment of slowdown? How long does the slowdown last? just for a split second during firefights or for a long time, maybe half a munute or more?

Your first link is a big duh. Of course gaming performance will be impacted by a slow CPU, but 3dmark05 scores will NOT be. The OP has a genuine issue, from the looks of things.

Here, have a little proof.
Text
Text

Hmm...a whopping 75 points decrease by dropping my CPU down nearly a full ghz to a 1.5ish chip that certainly will not be stellar in games. This is out of almost 3000 points. I don't care which GPU you have. Your X800XT is going to get an even smaller point ratio.
The OP is missing a few THOUSAND points.

Your second link I can truly trust.

For your third link, that's just the CPU score. Not the 3dmark05 score.

Fourth link, I'm not going to believe some random site claiming a 700 point deficit 3dmark05 overall score when my testing shows only 75 points. Sorry, but I have not seen a single person on these boards bring this up to note.

 

[eva2000]

Proof that cpu makes a difference straight from http://i4memory.com/showthread.php?t=1842

512MB X1800XT stock speed 621/747

3dmark05

Opteron 146
@stock 2000mhz = 8,636
@2800mhz = 9,235
@3000mhz = 9,287
@3100mhz = 9,321


now 512MB X1800XT @ 708/801

3dmark05

Opteron 146
@stock 2000mhz = 9,106
@2800mhz = 10,163
@3000mhz = 10,221

 

[Elfear]

CPU speed does make a difference like the other posters have stated. I don't know if that's the OP's problem but in general the cpu does affect your 3DMark05 score. At stock speeds (2.0GHz) I score ~9600, with the cpu@2.9GHz I get ~12,600 with no change to the gpu clocks.

 

[Avalon]

Originally posted by: eva2000
Proof that cpu makes a difference straight from http://i4memory.com/showthread.php?t=1842

512MB X1800XT stock speed 621/747

3dmark05

Opteron 146
@stock 2000mhz = 8,636
@2800mhz = 9,235
@3000mhz = 9,287
@3100mhz = 9,321


now 512MB X1800XT @ 708/801

3dmark05

Opteron 146
@stock 2000mhz = 9,106
@2800mhz = 10,163
@3000mhz = 10,221

So you're saying if he replaced his 2.8 P-D with a stock Opteron 146, he's going to gain 2000 points? I don't think so.
Granted, I'm willing to trust your scores for that high end bracket, but I have a hard time believing this for the OP's case having a 2000 point deficit compared to yours.

Reason being, is I have not kept up with 3dmark05 scores since the trickle of X800XTs and GTX512 cards have kicked in. The reason why I was getting so irritated earlier is that typically we'd find a thread where someone had a 6600GT/X800pro or the likes, have a score 1000 points lower than normal, have an Intel CPU, and everyone would scream "It's your CPU!" when it was always either a driver problem, their AGP GART was deleted, etc.

 

[eva2000]

well opteron 146 @2000mhz is about equivalent to P4 3.4ghz cpu if i compare my P4 3.4E which clocks to 4.2ghz - my opteron 146ghz @2800mhz is faster than my P4 3.4E @4.2GHZ

i.e.

1x 256MB Powercolor X800XT @623/599 with AMD64 3500+
@2805mhz = 7,135 3dmark05 http://service.futuremark.com/compare?3dm05=677823

versus

1x 256MB Powercolor X800XT @615/602 with P4 3.4E
@4199mhz = 6,938 3dmark05 http://service.futuremark.com/compare?3dm05=441584

yeah GPU clock a tad higher but it illustrates the performance different AMD64 have over P4s

my results with

1x 256MB BFG 7800GTX @564/1410 with Opteron 146
@3130mhz = 9,644 3dmark05 http://service.futuremark.com/compare?3dm05=1452613
@3001mhz = 9,403 3dmark05 http://service.futuremark.com/compare?3dm05=1289781
@2800mhz = 9,211 3dmark05 http://service.futuremark.com/compare?3dm05=1271876



newer 7800GTX and X1800XT's performance is far ahead that from reviews even a FX-57 @3GHZ (P4 4.4-4.8ghz) would possibly be a bottleneck for these - especially in SLI /AMR config

 

[RussianSensation]

You guys are right that cpu does affect 3dMark scores. But besides 3dmark01, it shouldn't affect it by a large amount like Avalon states. Unless 3dmark05 takes into consideration actual CPU benchmark score when it calculates the overall benchmark score (which I am not aware of).

CPU Scaling Article <-- you can see there that under 3dmark05, A64 1.4ghz gives almost the same score as A64 2.2ghz (yeah they are using older cards but even 3dmark03 doesnt show cpu dependence that severe and it scores 15000+ points. So I just cant seem to believe that 3dmark 05 which is even more dependent on gpu will show you 30% difference due to cpu speed, and that is at still limiting 9000+ score (with top end cards).

Even with what Eva2000 has shown, I find it very difficult to believe that P4 3.0ghz (or D840) will lose 29% in 3dmark05 benchmark (which is far more gpu limited than any game on the market at 1024x768 noAA/noAF. Simple logic suggests that GPU should be drained the most. If the previously linked database scores of 6200+ and 6400+ performed on D830 show that X1800XL was able to reach these scores, then surely X1800XT should score much higher than 6900 points.

We can also analyze the scores obtained with D series Pentium and 7800GTX card. Here is what I found:

8434 marks - Pentium D 840 stock (3.2ghz), 7800GTX @468/1197
another user got:
6721marks - Pentium D 840 stock (3.2ghz), 7800GTX @ 468/1197
Notice, both used identical Nvidia drivers and their 3dmark cpu score in itself is almost the same. So how can we explain 25% difference between what seem to be identical setups?
Then I got really stumped when i saw this:
8652marks - Pentium 4 D 4.2ghz, 7800GTX @540/1316
Shouldn't this score be way faster than the top score of 8434?

This leads to a conclusion that other factors can influence the score besides specs alone.

1) was the computer restarted with all applications turned off (ie. anti-virus, background tasks) when the benchmark was ran?
2) was the motherboard bios properly optimized? (ie. proper videocard aperture size, ram timings, turbo modes, etc.)
3) did each system have proper intel chipset drivers installed?
4) is the operating system itself properly and freshly installed?
5) is the hard drive fully defragmented?

Finally, even when I overclock my slow 8500 in 3dmark01, I get a score faster at 301/301 (from 275/275) than 306/306 since the latter actually overheats my card and it runs slower but I dont necessarily see glitches or artifacts.

 

[Addle]

I would still like to know if my card running at 82 degrees celcius is as much of a problem as I believe it to be (that's almost 180 fahrenheit!). Even if Rivatuner didnt show a dropoff in core clock speed it is still a possibility that my GPU could be throttling, no? At the very least, it shows that my card is not working 100% correctly.

 

[otispunkmeyer]

Originally posted by: grimlykindo

Originally posted by: Corporate Thug

Originally posted by: grimlykindo

Originally posted by: Addle
These are the scores I got for my new gaming system, wasn't really sure if these were average given my system specs. This was my first time using 3DMark.

Scores:
'03 - 15817
'05 - 6907

System:
Pentium D 830
Ati x1800 XT
d945 Intel Mobo
2GB PQI DDR2533 CL3

That 05' score is WAAAAY LOW

I get 7569 with my 7800GT & Venice 3200+

:Q

venice v pent D

gameover

I didn't think the CPU had a big effect on 3dmark 05??

I was just making the point that he has a WAY better video card than me and still has a lower score...

seems to

i score more than 1000 extra just by overclocking my 3200 venice to 2.5Ghz

 

[eva2000]

Originally posted by: RussianSensation

We can also analyze the scores obtained with D series Pentium and 7800GTX card. Here is what I found:

8434 marks - Pentium D 840 stock (3.2ghz), 7800GTX @468/1197
another user got:
6721marks - Pentium D 840 stock (3.2ghz), 7800GTX @ 468/1197
Notice, both used identical Nvidia drivers and their 3dmark cpu score in itself is almost the same. So how can we explain 25% difference between what seem to be identical setups?
Then I got really stumped when i saw this:
8652marks - Pentium 4 D 4.2ghz, 7800GTX @540/1316
Shouldn't this score be way faster than the top score of 8434?

answer is right in those compare urls, P5ND2-SLI is well known to severely suck for performance compared to P5WD2-Premium motherboard

but yes it can come down to optimisations... at OS level by the end user as well

 

[mwmorph]

Originally posted by: Avalon

Originally posted by: mwmorph

Originally posted by: Powermoloch
LOL bunch of noobs in this forum, the CPU has no impact in his score. It's the card.

lol you must be referring ot youraself. why would the goddamn orb and all reviews tell a different story? i guess everyone else is a noob. hmm?

no 3dmark here but you can see how 3d performance is much lower

3dmark 01,03,05 and Aquamark here show big drops in scores with a top of the line card, in this case a 7800gtx, which is comparable to the x1800xt. Similar score as the op once you figure in the 200mhz clock boost huh?

hey! whatda ya know? 3dmark05 results are in once again and guess who's lagging behind? this time they're using "only" x850xts and the pentium d is still cpu limited!

Now with p-d840EE. see how it's still slower than other cpus and the variation in scores? even with a much slower x800xt?

now op i hadnt realized you were having problems with BF2. that might be a problem. have you tried to use fraps to see exactly how low your fps go? did you know the latency at the moment of the slowdown? Do you know your temps for cpu and gpu at moment of slowdown? How long does the slowdown last? just for a split second during firefights or for a long time, maybe half a munute or more?

Your first link is a big duh. Of course gaming performance will be impacted by a slow CPU, but 3dmark05 scores will NOT be. The OP has a genuine issue, from the looks of things.

Here, have a little proof.
Text
Text

Hmm...a whopping 75 points decrease by dropping my CPU down nearly a full ghz to a 1.5ish chip that certainly will not be stellar in games. This is out of almost 3000 points. I don't care which GPU you have. Your X800XT is going to get an even smaller point ratio.
The OP is missing a few THOUSAND points.

Your second link I can truly trust.

For your third link, that's just the CPU score. Not the 3dmark05 score.

Fourth link, I'm not going to believe some random site claiming a 700 point deficit 3dmark05 overall score when my testing shows only 75 points. Sorry, but I have not seen a single person on these boards bring this up to note.

What i dont think you understand is that there is a proverbial wall that your cpu will be limited at. at 1.5ghz, your wall will still be higher than 2.7k. What i'm saying is that a2.8, the pentium d in his system hits the wall at high 6, low 7 k scores as verified by the second review and all the ORB sources. that means his top of the line card can create a 8k,9k or 10k score but his cpu is now holding back the card at the wall of 6k.

If you get, say a 7800gtx, and clock it at 3,0ghz, yes you will get 8-9k, but if you declock it to say 1.9 or even 2.0 ghz, i bet you you will lose a big part fo that score, dropping probably to mid to low 7k.

Your system itself is not a good comparison because the 2.7k score is too low to cause a cpu bottleneck unless you have a horrible maybe lress than 1 ghz cpu.

sorry if i seemed like an asshole in my last post, but i was just so damn frustrated by russiansensation.

btw, the 4th link was a 700pt difference between cpus and like i said the 840 can do better than 5.9k or so, so that's why there seems to be no change. if they used 7800gtx sli or x1800xt or x850xt xfire, you would see a big difference i gurantee you.

Originally posted by: Addle
I would still like to know if my card running at 82 degrees celcius is as much of a problem as I believe it to be (that's almost 180 fahrenheit!). Even if Rivatuner didnt show a dropoff in core clock speed it is still a possibility that my GPU could be throttling, no? At the very least, it shows that my card is not working 100% correctly.

not really. i think most cards have a cutoff in the 100-120* c range. 82 is hot and i would recomend a new cooler like a zalmanvf700, but it wouldnt throttle at 82.

 

[Avalon]

Originally posted by: eva2000
well opteron 146 @2000mhz is about equivalent to P4 3.4ghz cpu if i compare my P4 3.4E which clocks to 4.2ghz - my opteron 146ghz @2800mhz is faster than my P4 3.4E @4.2GHZ

i.e.

1x 256MB Powercolor X800XT @623/599 with AMD64 3500+
@2805mhz = 7,135 3dmark05 http://service.futuremark.com/compare?3dm05=677823

versus

1x 256MB Powercolor X800XT @615/602 with P4 3.4E
@4199mhz = 6,938 3dmark05 http://service.futuremark.com/compare?3dm05=441584

yeah GPU clock a tad higher but it illustrates the performance different AMD64 have over P4s

my results with

1x 256MB BFG 7800GTX @564/1410 with Opteron 146
@3130mhz = 9,644 3dmark05 http://service.futuremark.com/compare?3dm05=1452613
@3001mhz = 9,403 3dmark05 http://service.futuremark.com/compare?3dm05=1289781
@2800mhz = 9,211 3dmark05 http://service.futuremark.com/compare?3dm05=1271876



newer 7800GTX and X1800XT's performance is far ahead that from reviews even a FX-57 @3GHZ (P4 4.4-4.8ghz) would possibly be a bottleneck for these - especially in SLI /AMR config

So if the OP was willing to overclock his chip from 2.8 to something like 3.4 for testing purposes, he would see a sizable increase of 500-750 points or so?

Mwmorph, I apologize for sounding rude. This was news to me. I didn't know there was a wall at 7k.

 

[Elfear]

OP- Did you check to see if Catalyst AI was turned off yet? Seems like another X1800XT user over at Xtreme forums was having issues and traced the problem down to the AI.

I wouldn't be too worried about the temps. 7800GTX's regularly run in the mid to high 70's. Besides that I believe there is a known issue with the temps being reported wrong on the X1800 series.

 

[Addle]

I turned off the Catalyst AI and it got me an additional 50 points, so now I'm up to 7004. Should I turn off the AI just for benching or all the time? Also, are there any more theories as to what might be going on? I think I may try to exchange the card for another x1800 XT and see if that changes anything.

 

[Elfear]

Originally posted by: Addle
I turned off the Catalyst AI and it got me an additional 50 points, so now I'm up to 7004. Should I turn off the AI just for benching or all the time? Also, are there any more theories as to what might be going on? I think I may try to exchange the card for another x1800 XT and see if that changes anything.

I don't know Addle, seems like you've tried about everything and even verified that the card is clocking properly. I did a quick Google search for other Pentium D users and what kind of scores they're getting. Here are a few of the results I found:

If you look through a few of the posts here, it looks like guys with SLI'd 7800GTX's are scoring in the 8000 range where generally guys with 2.6GHz+ AMD procs are scoring around 11,000. Some of the decrease is obviously due to SLI overhead making things even more cpu limited, but even my 7800GTs at stock speeds scored ~9700 with an Opteron@2.0GHz.

Here is a compare link to a guy running a 3.0GHz Pentium D and an X1800XL/XT. Even if the card is an X1800XL, low 7000's is average for these cards with an AMD proc.

Here is another person running almost the same setup as the guy above.

Look at post #28 and #29 in this thread. The guy has almost the exact same setup as you except his proc is running at 3.66GHz and the memory on his card is only running at 1350MHz. Even with the decrease in memory bandwith, the score still seems low compared to the results obtained with AMD procs.

Not trying to bag on Intel chips at all, I'm just thinking we may have exhausted all the other possibilities and it may just be your proc. Maybe try overclocking it and see if your score goes up substantially. Might point towards a cpu bottleneck.

 

[Cooler]

My rig gets about 7K and gets about 6200 none oc it was not a very big differnce at all.
You must have a driver problem.

 

[Munky]

I tried 3dmark05 at various cpu speeds on my rig, and the difference between a 1.8ghz A64 and a 2.6ghz A64 was somewhere around 200 points. Of course, that was on my old x800 vanilla, and I'm guessing a fast card like the x1800xt can be more sensitive to cpu bottlenecks, but still I doubt a P4 would account for a 3000 point difference. Either the card is defective, or there's a software/bios glitch. Have you tried any other game benchmarks to see if your scores are close to what the tested x1800xt gets?


Edit: crap, did you see this?
http://www.theinq.net/?article=28135

Compare the A64 score to the P4 score. I know these results are just averages, and could have a bunch of statistical oversights, but looks like the average score for a x1800xt with a p4 @ 3.4 is around 7200. Almost 2000 points lower than on an A64