X800XtPE>X850XtPE???

[charlie]

So I'm thinking about an x850xtPE and have been looking around the ORB and at threads here...
But I don't see a performance advantage to the new x850! In fact, it appears the "old" x800xtPE PCI-E beats the X850xtPE....??????? :confused: :shrug:

Opp's #1 WR is with x800xt.... and what's with ALL THIS issue about the x850 and teh really BAD Nature scores???? Peeps have been changing mem timings on the x850 to make up for the bad core clocks...

So what's the deal??

C

[k|ngp|n]

IMO....the x800xt pci is prolly the card to have for 01' atm. The x850xtpe's are putting out lower drags then the x800's. You need to tref at 5 on these x850's with monster clocks to match x800xtpe drag and nature. I could be missing something, but this has been my results so far.

x850xtpe 540/587 = 265 nature with all my tweaks and stock mem/ fx clocks
x800xtpe 520/560 = around 265 nature with exact same settings / fx / mem clocks

[charlie]

yeah, interesting.
I wonder if the X800XT PCI-E clocks any LESS than the X800XTPE PCI-E?? The latter are getting HARD to find!

[k|ngp|n]

isnt the asus x800pci the only one that is actually an xtpe??

As some have pointed out to me :) , drivers plays a role in the performance gap too charlie.
Although most of my testing was all done using the exact same driver/ cp combo. The newer cards might be optimized diff. and require drivers more suited to those optimizations???

[charlie]

isnt the asus x800pci the only one that is actually an xtpe??

As some have pointed out to me :) , drivers plays a role in the perormance gap too charlie.
Although most of my testing was all done using the exact same driver/ cp combo. The newer cards might be optimized differnetly and require drivers more suited to those optimizations???

yeah, that's right.... OPP used MSI RX800XT, default 500/500.
Wonder if the Asus is any better?
C

[OPPAINTER]

I'm not sure whats up with the 850 and 2001 scores. Now I'm wondering how it compares with the 3D03 and 3D05 benches. Has anyone tested those benches with these 2 cards?

OPP

[G H Z]

We were waiting for you to do that;)

[OPPAINTER]

We were waiting for you to do that;)
You know what, I have the 2 cards right here:D
I'll try them in my regular winchester rig:D

OPP

[G H Z]

Great:D

I see that there are X850 PE's with AGP floating around now.

[charlie]

You know what, I have the 2 cards right here:D
I'll try them in my regular winchester rig:D

OPP

thx, OPP!
Will be VERY interested to see the results!!!


C

[Geforce4ti4200]

the x850 series does overclock better but its got relaxed ram timing which cancels out the benefits anyway. Mike does have a 42.3k score with x850xt I think and 325 nature but you need monster clocks. Ah well for those concerned, hopefully the upcomming r520 is more than enough to satisfy ;)

[OPPAINTER]

The X850 works good with the 3D05 and 03 benches. Does as it's supose to being that it's clocked higher.

http://www.xtremesystems.org/forums/attachment.php?attachmentid=26219&stc=1

I guess the only reason you would want the X800 would be for 3D2001 benches:D Other then that I would take the X850, the card will put out higher overclocks also.

OPP

[k|ngp|n]

Nice compare OPP.....

[Dumo]

bba 850xtpe with stock...

[charlie]

thx, OPP...
Can you clock x850 and x800 the same and see if there's a difference clock for clock???

C

[OPPAINTER]

thx, OPP...
Can you clock x850 and x800 the same and see if there's a difference clock for clock???

C
I'll have to underclock the X850.
I'll see what happens. I hate underclocking:D

OPP

[charlie]

I'll have to underclock the X850.
I'll see what happens. I hate underclocking:D

OPP

lol.....

[charlie]

I would wait with comparing untill the X850AGP is out...

No, it's a good compare, OPP's X800XT is PCI-E....

[perkam]

Dont the X850s go higher on air than the X800s ?? IF they do, even if the X800 is better in 01 clock for clock, you'll still get better results with the x850, right?

Though I imagine the 850s need more modding to go 700+ clocks than the x800s on phase or something...Macci's x850 at txgf looked like he was doing soldering practice on his card :eek: :rotf:

Perkam

[OPPAINTER]

Good call Charlie. The X850 underclocked to 520/560 clocks show a definite lack of power compared to the X800 card.

http://www.xtremesystems.org/forums/attachment.php?attachmentid=26227&stc=1

OPP

[OPPAINTER]

I guess what this means is if you can find a real high overclocking X800 card it would beat out the X850 in all benches then, umm:D

OPP

[Kunaak]

so why is the X850 slower clock for clock??

ram timings?

[charlie]

Hmmmmmm...
Good compare, OPP!

So it SEEMS.... that ATI may have done SOMETHING to allow greater clocks on the X850, without increasing overall performance. Maybe the memory timings?
SO it really seems the x850 really IS a PCI-E X800xt with higher default clocks. We all knew THAT but it looks like to make the higher clocks were achieved at some PRICE. Maybe the x850 uses higher default Vgpu (hence the giant cooling solution) and looser timings on the memory???
People should grab the ASUS X800XTPE PCI-E while they're still AVAIL.

C

[Kunaak]

Opp, while your at it. can you try running at 490/612

thats charlies best with the XL.

I wanna see how the XL compares clock for clock to the XT's.

please :D

[OPPAINTER]

I think since we've done all this testing, I'll put my X800 on my waterblock and try get the clcoks to 540/587 and see how much it beats the x850 by.

OPP

[OPPAINTER]

Opp, while your at it. can you try running at 490/612

thats charlies best with the XL.

I wanna see how the XL compares clock for clock to the XT's.

please :D

What bench?
Also, our systems are different. I'm running a 3000+ at 2600 Mhz 290x9 with ram at 180 divider using 2.5-3-3-5.
It would be hard to compare unless the systems are running the same.

OPP

[Kunaak]

if I remeber right, charlie was running 2.7 ghz with a winchester 3000 at 9x300 with gskill ram cas 2.5-3-3-7....

either way, yours should be close enough for a GPU test only, and since 2005 is mainly GPU anyways, I would like to see a 2005 test, that would do fine for me.
just clock for clock, to see how the XL hangs with it's bigger brothers.

[mikead_99]

Hmmmmmm...
So it SEEMS.... that ATI may have done SOMETHING to allow greater clocks on the X850, without increasing overall performance. Maybe the memory timings?
SO it really seems the x850 really IS a PCI-E X800xt with higher default clocks. We all knew THAT but it looks like to make the higher clocks were achieved at some PRICE. Maybe the x850 uses higher default Vgpu (hence the giant cooling solution) and looser timings on the memory???
People should grab the ASUS X800XTPE PCI-E while they're still AVAIL.

C

Yeah, it ain't higher default Vgpu, 1.4V on the x850's too. It could be default memory timings, a peek with AtiTool might shed light on that. Since the big "enhancement" seen on the X850XT-PE's is memory clocks, IMO, I think that is where the difference will be found.

[Naughtyboy]

We where discussing the differenses between X800 and X850 over at Techsweden.org. First off we checked the differens in timings between both PE cards...and there was a little differens in the settings.
The X800 card was an agp but I can´t see how the memtimigns could differ between agp vs. pci-e....any how, here it is.

X800XT PE

http://upl.silentwhisper.net/uplfolders/upload0/x800_timing.jpg


X850XT PE

http://upl.silentwhisper.net/uplfolders/upload0/x850_timing.jpg


As you can se the only thing that differs is MEM_REFRESH_RATE






Now I tried to change the setting on my X800XT PE...I tried to set it at 20 / 22 / 24 ....and run 3Dmark01. I tried several runs to make sure it was not just the app fluctating...but the performance increase/decrease was constant all the time.
Here ya go.

http://upl.silentwhisper.net/uplfolders/upload5/rate20.JPG


http://upl.silentwhisper.net/uplfolders/upload5/rate22.JPG


http://upl.silentwhisper.net/uplfolders/upload5/rate24.JPG


Now I don´t know how much this particular thing affects the differens in nature test it self.....but it clearly does make a differens...and it could be one of several things that makes the differens.

It would be nice if someone with an X850XT PE would care to try theese changes to and run 01...to see if it affects the performance int the same way as it did with my card above.

[OPPAINTER]

if I remeber right, charlie was running 2.7 ghz with a winchester 3000 at 9x300 with gskill ram cas 2.5-3-3-7....

either way, yours should be close enough for a GPU test only, and since 2005 is mainly GPU anyways, I would like to see a 2005 test, that would do fine for me.
just clock for clock, to see how the XL hangs with it's bigger brothers.


Well here is a score, I got my rig to do 300x9 and mem doing 300 at 2.5-4-4-5 timings.

http://www.xtremesystems.org/forums/attachment.php?attachmentid=26230&stc=1

OPP

[charlie]

if I remeber right, charlie was running 2.7 ghz with a winchester 3000 at 9x300 with gskill ram cas 2.5-3-3-7....

either way, yours should be close enough for a GPU test only, and since 2005 is mainly GPU anyways, I would like to see a 2005 test, that would do fine for me.
just clock for clock, to see how the XL hangs with it's bigger brothers.

yeah, my X800XL benchies are on my 24/7 rig, DFI NF4/3000+/TCCD

9 x 300ish, 2.5-3-3, 1:1 WinXP, nearly NO optimizations other than setting sliders, setting to real-time and "performance" in the device mgr. OH and "Ricky Tweak, too"
Would love to see how XL stacks up against it's big bro.

C

[charlie]

Well here is a score, I got my rig to do 300x9 and mem doing 300 at 2.5-4-4-5 timings.

http://www.xtremesystems.org/forums/attachment.php?attachmentid=26230&stc=1

OPP

cool, looks to be the same here...
http://www.xtremesystems.org/forums/attachment.php?attachmentid=26231

[OPPAINTER]

We where discussing the differenses between X800 and X850 over at Techsweden.org. First off we checked the differens in timings between both PE cards...and there was a little differens in the settings.
The X800 card was an agp but I can´t see how the memtimigns could differ between agp vs. pci-e....any how, here it is.

X800XT PE

http://upl.silentwhisper.net/uplfolders/upload0/x800_timing.jpg


X850XT PE

http://upl.silentwhisper.net/uplfolders/upload0/x850_timing.jpg


As you can se the only thing that differs is MEM_REFRESH_RATE






Now I tried to change the setting on my X800XT PE...I tried to set it at 20 / 22 / 24 ....and run 3Dmark01. I tried several runs to make sure it was not just the app fluctating...but the performance increase/decrease was constant all the time.
Here ya go.

http://upl.silentwhisper.net/uplfolders/upload5/rate20.JPG


http://upl.silentwhisper.net/uplfolders/upload5/rate22.JPG


http://upl.silentwhisper.net/uplfolders/upload5/rate24.JPG


Now I don´t know how much this particular thing affects the differens in nature test it self.....but it clearly does make a differens...and it could be one of several things that makes the differens.

It would be nice if someone with an X850XT PE would care to try theese changes to and run 01...to see if it affects the performance int the same way as it did with my card above.
The refreash rate on this x850 card here is the same as on your x800 it looks like.

http://www.xtremesystems.org/forums/attachment.php?attachmentid=26232&stc=1

OPP

[GazC]

I'm betting that it's a driver bug. Remember that when the X800s first came out that they were not making full use of their 256Mb of memory. Maybe the X850s device ID needs adding to the relevant driver switches to make it work at full capacity?

[Pedro Rocha]

There is definitively a issue with the 850 and Nature.

Here is one quick bench wth my 24/7 system (Silence Vapo with fans at very low speed almost noiseless), TCCD mem and Asus X850XT PE 618-654 stock cooler (fans@75%), case closed.
Win2k/DX8.1 - 4.12 drivers

38758 - http://service.futuremark.com/compare?2k1=8444330

Cars, and specially 176 in Car High are not bad for 100% stock graphic card

Lobbys not bad 2.5-3-3-6 mem

Dragos are nothing special .. and Nature are quite low, just 295 :(

If you compare, for example, against one X800XT PE @ 604-620 - http://service.futuremark.com/compare?2k1=8370561 - you can see better Nature score (314) even with less graphics and CPU

Any ideas, sugestions ?

With the same stock cooling card speed , already try 5.1 and 5.2 with worst Drago and Nature scores :(

[eshbach]

0x45 is the best refresh rate i've found. it is about 1.2 fps faster than 0x22 on my system.

edit: actually with a little more testing, i've found 0x75 is slightly faster yet! 0x80 or more causes artifacts.

[GazC]

There is definitively a issue with the 850 and Nature.

Here is one quick bench wth my 24/7 system (Silence Vapo with fans at very low speed almost noiseless), TCCD mem and Asus X850XT PE 618-654 stock cooler (fans@75%), case closed.
Win2k/DX8.1 - 4.12 drivers

38758 - http://service.futuremark.com/compare?2k1=8444330

Cars, and specially 176 in Car High are not bad for 100% stock graphic card

Lobbys not bad 2.5-3-3-6 mem

Dragos are nothing special .. and Nature are quite low, just 295 :(

If you compare, for example, against one X800XT PE @ 604-620 - http://service.futuremark.com/compare?2k1=8370561 - you can see better Nature score (314) even with less graphics and CPU

Any ideas, sugestions ?

With the same stock cooling card speed , already try 5.1 and 5.2 with worst Drago and Nature scores :(

I've never seen such a high nature score for 604/620, you are only two frames behind me with my card at 687/645. You haven't got ATi/Nvidia speed up enabled in the bios have you? Either way, that is an amazing score for that clock speed. The low latency ram helps of course!

[Saliman]

Hi opp, can you please try to set the cards to more aggressive timings and see how they stack up against each other?

thanks in advance ;)

[Naughtyboy]

The refreash rate on this x850 card here is the same as on your x800 it looks like.

http://www.xtremesystems.org/forums/attachment.php?attachmentid=26232&stc=1

OPP

Hmm..odd..I´ve checked with several X850XT PE users and they all had 20 on the refresh_rate.
Would you care to try chaning the setting as I did and make some runs on your sytem?
Would be interesting to see how/if it affects your performance.

[Rabbi]

0x45 is the best refresh rate i've found. it is about 1.2 fps faster than 0x22 on my system.

edit: actually with a little more testing, i've found 0x75 is slightly faster yet! 0x80 or more causes artifacts.

Great, with 0x75 I gained 1fps in nature and about 800 points in aquamark with a x800xt pe :toast:

[Pedro Rocha]

I've never seen such a high nature score for 604/620, you are only two frames behind me with my card at 687/645. You haven't got ATi/Nvidia speed up enabled in the bios have you? Either way, that is an amazing score for that clock speed. The low latency ram helps of course!

Seems there is a misunderstanding - my score is URL: http://service.futuremark.com/compare?2k1=8444330

Let's compare my stock cooler score with yours, I'm using a litle more CPU, but same mem latency (booth TCCDs) you have much better natures and Drago low ..
Booth using same OS, DX and drivers..

http://pedrorocha.planetaclix.pt/X850vesusX800.jpg

You : X800pro vivo @ XTPE 687/645 - 316 - Nature
I : X850XT PE stock 618/650 - 295 Nature

Can the 70Mhz core diference explain the 21 fps diference on Nature?

I have a DFI nF4, so no ATi/Nvidia speed up to enabled

[Kunaak]

Opp thanks for the benchmarks.

thats very interesting.
seems those cards are nearly identical in speed for speed settings.
so assuming a XL can handle 1.8 volts or more, seems no reason why a XL couldn't do nearly the same thing a XT can do...
thats a hell of a bargain for a $300 card ;)

[Kunaak]

Someone Try a X800 XT PE Driver manually installed, on the X850 XT PE chipset.
the issue may be driver related.
ATI might be optimizing the 850 drivers for DX9 benchmarks or games, and it's hurting the 3dmark 2001se scores...

someone take a look at that.

if anyone needs help manually installing a driver, I'll write up a little guide.

[Pedro Rocha]

Someone Try a X800 XT PE Driver manually installed, on the X850 XT PE chipset.
the issue may be driver related.
ATI might be optimizing the 850 drivers for DX9 benchmarks or games, and it's hurting the 3dmark 2001se scores...

someone take a look at that.

if anyone needs help manually installing a driver, I'll write up a little guide.

What you mean by XT800XT drivers?
Booth cards use Catalysts...
Can you explain please?

Do a quick run with 5.2 cats, DX9.1, Winxp, same speeds - even less nature fps - 287 :(

[misteroadster]

same problem here , solution: my X850 is just for 2003 or 2005 benchies now :D
I've tried to flash to X800XT Pci-e but i had a lot of artefact at stock XT frequency.
Weird.

[GazC]

Seems there is a misunderstanding - my score is URL: http://service.futuremark.com/compare?2k1=8444330



Yeah, sorry. I thought both runs belonged to you. Looking at the second run you posted though ( the one that I got confused over) I don't believe the declared card clocks in his description. Nature is way way too high.

[Mr. Tinker]

maybe rivatune and do a manual device id?

[eshbach]

Great, with 0x75 I gained 1fps in nature and about 800 points in aquamark with a x800xt pe :toast:

glad to help...

someone should try 0x75 on an X850.

[Kunaak]

What you mean by XT800XT drivers?
Booth cards use Catalysts...
Can you explain please?

Download a ATI driver. extract it to your desktop or somewhere easy to find.

double click, so it extracts the file, but cancel it before actual install.

from there go to Start> My computer> right click> properties> Hardware> Device manager> Display adaptor... select your videocard, then select "update driver"

when there... select "install from specific location"

"dont search, I will choose driver to install"

"Have Disk"

"browse"

then direct it to the extracted ATI driver file that I had you do in step 1.

find the C2 or CX .INF File.
it will be really easy to find.

from there, go through the long list of videocards, and instead of selecting the usual X850 XT PE driver, select the X800 XT PE driver.

it will then say "this isn't compatible" or something similar.
just hit OK and ignore it, it's almost never worth listening too.

similar cards have extremely similar drivers.
like a 5900 Ultra can use a 5950 Ultra driver with no problems.
but a Ti 4200 couldn't use say a 5700 ultra driver.

the X850 and X800 are so close, there should be no issues with manually installing different drivers.

sometimes different videocards have better optimizations.
like for my 6800 NU scores in my sig, I used the 6800 Ultra drivers (renamed to 6800 NU) which was about 200 points better then a standard 6800 driver was at 3dmark2001se.

driver modding is one of the best simlpe ways to get some extra performance from a videocard, cause say you had a X800... theres no real reason ATI is gonna spend much time optimizing thier mid line cards drivers, if it works, thats good enough.
but having the fastest videocard in the top market is a feather in the cap of the company that can pull it off... so theres more reason to optimize the high end videocards drivers.
this is why I usually try driver mods, cause theres sometimes a nice little gain to be had from it.

this may be true in the case of the X800XT PE, and X850XT PE too...
cause the X800XT PE has had alot of time to be optimized, however the X850XT PE has not, thats one reason the X850 is alittle slower the older X800XT PE...

it's worth trying. :)

[GazC]

Seems there is a misunderstanding - my score is URL: http://service.futuremark.com/compare?2k1=8444330

Let's compare my stock cooler score with yours, I'm using a litle more CPU, but same mem latency (booth TCCDs) you have much better natures and Drago low ..
Booth using same OS, DX and drivers..

http://pedrorocha.planetaclix.pt/X850vesusX800.jpg

You : X800pro vivo @ XTPE 687/645 - 316 - Nature
I : X850XT PE stock 618/650 - 295 Nature

Can the 70Mhz core diference explain the 21 fps diference on Nature?

I have a DFI nF4, so no ATi/Nvidia speed up to enabled

Just looking at the scores again Pedro, I noticed that your Car Low score is huge. Do you run it earlier in the test or do you have a special tweak for it? ;)

You don't have to tell me what it is if there is a tweak, I just wanted to know.

[gocchin]

I was so excited when I first picked up my X850 XT and a GPU cooler from k|ngp|n. I was ready to crush my x800xt pe score that was done on water and way lower clocks. So far with the same CPU same memory and timings I just barely eeked by even with way bigger clocks on the video card.... I did however use cat 5.1 for the x850xt as I kept locking up using 4.12.

X800XT PE @ 675/635 on water with MSI Neo2 Platinum FX55 12x258 BH-5 1.5-2-2-5

VS

X850 XT @ 749/688 R507 Prommy with DFI Ultra-D FX55 12x258 BH-5 1.5-2-2-5

http://www.jpop.com/x800vsx850.gif

[$0m#0n#]

I was so excited when I first picked up my X850 XT and a GPU cooler from k|ngp|n. I was ready to crush my x800xt pe score that was done on water and way lower clocks. So far with the same CPU same memory and timings I just barely eeked by even with way bigger clocks on the video card.... I did however use cat 5.1 for the x850xt as I kept locking up using 4.12.

X800XT PE @ 675/635 on water with MSI Neo2 Platinum FX55 12x258 BH-5 1.5-2-2-5

VS

X850 XT @ 749/688 R507 Prommy with DFI Ultra-D FX55 12x258 BH-5 1.5-2-2-5

http://www.jpop.com/x800vsx850.gifreally weird

[gocchin]

Oh just a quick note, while I'm struggling with my x850 on 2001, 03 and 05 benches seem where they should be :

17531
http://service.futuremark.com/compare?2k3=3767070

8084
http://service.futuremark.com/compare?3dm05=643788

I'm going to have a go with the x850XT tonight again with the 4.12's and 2001

[Franky 4 Finger]

dissapointing to say the least

[Pedro Rocha]

Just looking at the scores again Pedro, I noticed that your Car Low score is huge. Do you run it earlier in the test or do you have a special tweak for it? ;)

You don't have to tell me what it is if there is a tweak, I just wanted to know.

Have a "special twek" - Ricky & Pedro Rocha Tweak;

Testing Version 2 now, gave me good results ..specially with 3DMark2005 - I will tell you all how to do it AFTER the bench in the cascade in the next days :)

[G H Z]

gocchin:)

Really nice compare & edits you made there. Part of it looks like you didn't get off a good CL test, but without a doubt the X850 is dissapointing to say the least.

I think someone needs to do what Kunaak mentioned and try running the exact same timings from an X800 on a X850.

[gocchin]

gocchin:)

Really nice compare & edits you made there. Part of it looks like you didn't get off a good CL test, but without a doubt the X850 is dissapointing to say the least.

I think someone needs to do what Kunaak mentioned and try running the exact same timings from an X800 on a X850.

My car low was terrible... I ran it many times and couldn't do much better than that which surprised me... CPU speed is pretty good and I was running very tight timings across the board (including secondary timings)....

[HiJon89]

Has anyone tried flashing X850 to an X800? This seems like it might be a good solution, it combines whatever is making the X800 faster with the higher overclockability of the X850.

[charlie]

I was so excited when I first picked up my X850 XT and a GPU cooler from k|ngp|n. I was ready to crush my x800xt pe score that was done on water and way lower clocks. So far with the same CPU same memory and timings I just barely eeked by even with way bigger clocks on the video card.... I did however use cat 5.1 for the x850xt as I kept locking up using 4.12.

X800XT PE @ 675/635 on water with MSI Neo2 Platinum FX55 12x258 BH-5 1.5-2-2-5

VS

X850 XT @ 749/688 R507 Prommy with DFI Ultra-D FX55 12x258 BH-5 1.5-2-2-5

http://www.jpop.com/x800vsx850.gif

And to make it even worse, your gain in '01 there was PROBABLY due to the performance boost from NF4.......

C

[IvanAndreevich]

Holy.. that's a LARGE difference in core clocks for the X850. Can ya post the memory timings on each of the cards?