Printable View
Just ran a comparo in 3Dmark01 nature test with a 9800GX2 - On an Intel X48 I get about 950fps and on Nvidia 790I board about 1250fps. Of course I had the same clocks on both boards.
Is this normal - if so WHY? :confused:
Or does the P5E3 Premium have a problem with the 9800GX2 - I am confused why the FPS drop is so significant.
Did a quick search but haven't come across anything useful yet.
have you searched HWBOT
there are no big differences in FPS between intel and nvidia chipsets
could be just your board or something screwy with your install
do u have the same memory settings for both, the 790i uses some tighter sub timings to make up for the slight added latency so it gets better bandwidth with defaults
also u reinstalled windows right
zanzabar
it won't matter at all
nature '01 is purely a GFX card bench
he could have had a Single core Celeron in there and nature would still not change much
there is a definitely problem with the setup and it's not Intel vs nvidia chipset related......it is either the board or his setup has screwed up >> like OS/drivers/card or something of that sort
thanks for the headup Dinos. It's actually my second install with the same problem. But one thing I never did was to update XP - maybe one of those service packs will do the trick.Quote:
Originally Posted by dinos22
At least now I know it's not a Intel in general problem ( maybe with the p5e3 - but doubt it )
There seems to be a bandwidth problem with the PCI-E slot on the P5E3 Prem. - As soon as I am raising the PCI-E freq. on the X48 board the FPS goes up. I wonder if the card is running PCI-E X8 or something along the lines. Can't tell from everest - GPU-Z says PCI-E 16 2.0.
Looks like there is a compatibilty problem here.
is your 790i raising the pci-e bus since u have an NV card
PCI Express running at 8x compared to 16x would maybe give you a 5% hit in nature
you are talking about 5x more performance loss here
maybe your GFX is somehow throttling the asus board. i remember reading that some asus motherboards overheat with GX2 installed. to be honest i am not really sure what the deal is but have a look at top5 3DMARK01 scores on HWBOT
first > kingpin 790i
second > pro Gigabyte X48
etc
http://img187.imageshack.us/img187/2...board01rf7.jpg
no it doesn'tQuote:
Originally Posted by zanzabar
i fact with GX2 you'd be hard pressed to raise PCIe much at all and you will see no gain at all either
I had played with the PCI-E clocks on 790i - Like Dinos said makes no difference ( no gain ) on the 790i and is actually making things a lot more unstable the higher you.Quote:
Originally Posted by zanzabar
Thanks Dino for the screeny - I didn't take the time to test the car on a Asus X38 - but something is certainly way off with the P5E3 Prem.
Argh - at least the Asus board was giving me decent FSB on wolfdale (585 ) - now looks like I have to find another 790i ( my current one is stuck @ 540)
Did you upgrade bios to 401?
I tried both 0204 and 0401 - same results with both. Man I love the board - but I also love the GX2 - :mad:Quote:
Originally Posted by metro.cl
I experienced the same difference between a P5E3 Deluxe and the 790i. The 790i would score 1000+ in 3dmark06 over the P5E3. Noticeable runs for me were grand canyon where you would see almost 40+fps on the nvidia board.Quote:
have you searched HWBOT
there are no big differences in FPS between intel and nvidia chipsets
could be just your board or something screwy with your install
Don't know why.
RLM
the ONLY other thing i can think of is LOD
you may have been comparing LOD vs non-LOD runs which is exactly where it's at
i did some benchig on my card just now and with 783/1944/1188 on GX2 and 4GHz E7300 i get around 950fps from memory on non-LOD run and 1300 with LOD
No I'm pretty sure it's not LOD. In addition with LOD runs myself I've found only minor differences in 3dmark 06. I know in others like 3dmark03 it makes a big difference.
It's got me wondering a bit. Is the 790i somehow tweaked to get more out of the nvidia cards?
Does this somehow go towards explaining vid corruption issues at high fsb. Could be well off, but has entered my mind.
Anyway no worries:D
RLM
I have the same problem on P5E64 WS Evolution x48.
Its not bad install or something like that. i re-installed 3 times, and tryed all posible drivers,bios etc.
I am around 950fps at 700/1100 on this evolution board, and on my P5E WS PRO x38 i am at 1200+ same clocks.
Also...i ofen have problems just getting the x48 boards started with gx2 in...but if i let it stand for a wile, it will boot.
Something is deff strange........
Hope we can work this out.......
:)
If you give me full info i can email Asus directly i have a pretty good contact there.
Quote:
Originally Posted by metro.cl
thx:) But i am currently talking to Marcus (Kinc) about the problem. he works at ASUS aswell :)
good news that someone at asus is aware of it - emailed support - but as usual no response...
That is great, i didnt know he worked with them. I guess there isnt anyone better than him to fix this :)Quote:
Originally Posted by SIOUX
Marcus have not worked there for a long time, nearly started as technical PR...Quote:
Originally Posted by metro.cl
You guys with the contacts please do update here, when you hear anything. Have you tried the latest drivers at all?
RLM
ANy updates on this? Finally can reach 600FSB with Wolfsdale on X48 - but can't bench my GX2...Quote:
Originally Posted by SIOUX
stable 600MHz FSB?Quote:
Originally Posted by mikeguava
what cooling?
I have also an issue with ASUS P5E3 premium and 9800GX2.
Framerates in Crysis or CoH benchmarks are way below what they should be (up to twice less performing compared to what can be found with those games in 9800GX2 reviews).
Also I have poor 3dmark06 score:
http://img263.imageshack.us/img263/7...0gx2bd1.th.jpg
Especially the Canyon Flight test is very low. I have not seen any 9800 GX2 in futuremark ORB that performs so BAD.
mikeguava, can you provide your 3dmark06 detailed score?
Anyone who have tested with PATA HDD and clocked up the PCIEx bus? Just to see if the performence loss evens out.
I have tested with evolution but since I dont have any other boards than the ones mentioned in this thread I cant see any perf difference. I tested with frist booting the VGA in the second blue slot, then soft power of and switch to first. It gives no post at first since it is relocating the VGA to a new irq, it it do not post after a min reset power and it will post. It did show a slight difference in 3dmark 03.
Anyone else who have tested both of the blue PCI-Ex 16X slots?
Just run Nature in 01, and you willbe around 900-1000fps if the bug is pressent. and around 1200fps you dont have the bug :)Quote:
Originally Posted by Kinc
The thing is ....its not all the same boards that does it. my own EVO i have no problem. but the one i did review on had the problem. No matter what PCI-E slot i used.
I tryed with ATA as i told you on msn, and had PCI-E at 125mhz wich gives an unatrual boost of about 80fps (nature 01) wich is not nearly enough to even out the normal scores (wich is 200fps behind)
Err now after playing with my bios to get a proper OC, I ran 3dmark06 again and the bug is back ;-(
I have lost the +25fps I had in Canyon Flight with 3dmark06 and am back to square 1...
There must be something that triggers the bug but do not count on Asus to solve the issue.
My board's a Maximus rather than a P5E3, but I saw a similar problem. The first time I installed the 9800GX2 in my Maximus Formula (with the Rampage's BIOS) I got pretty bad Nature FPS as well (~980). Just the other day I tested it again using an XP installation that was imaged from the 790i board's installation and then driver-cleaner'ed. This time I got 1349 FPS in Nature. I don't know what was causing that problem before. For what it's worth, my card is installed in the lower PCIe x16 slot (farther from the CPU) because that's the only way to make the DVI cable clear my radiator since nV found it necessary to put the two ports in the center of the card stacked vertically instead of side-by-side. Maybe try re-installing XP (the only updates I've applied to XP are the ones included in SP2) and install the just VGA driver, rivatuner, and 3d01 and nothing else and see what happens.
What's interesting is that I've also seen a similar problem on 790i and re-installing the OS (from the image) fixed it so I chalked it up to the 790i data corruption problem. The Nature FPS was far worse when the 790i fouled up too - it was actually worse than what I get normally with SLI forced off (and would get even lower if I forced SLI off when the problem was occurring) so I don't think it was the same root cause.
The CPU and GPU clocks were about the same on these two runs. The second one is a little bit better tweaked but not nearly enough so to account for the difference in Nature.
3d01 run first installation on X38 board: 981 FPS Nature
http://service.futuremark.com/compare?2k1=9355626 (sorry no screenshot)
3d01 run new installation on X38 board: 1349 FPS Nature
http://service.futuremark.com/compare?2k1=9382068
http://i164.photobucket.com/albums/u...r_bob/94k7.jpg
I tryed re-installing many times...problem is still there i´m affraid....
Errr... I get only 416fps in nature test of 3dmark01se :confused:
I have windows vista x64 sp1 btw
Is there some kind of frame limiter or what?!
I will be very thankful that reinstalling fixed it for me then. Wish I knew what the source of the bug was.Quote:
Originally Posted by SIOUX
Vista and 3d01 don't get along very well at all. I don't have any runs in Vista with the 9800GX2 but I get 409 in Vista with Crossfired HD3870s. I'll run it in Vista now and see what happens with the 9800GX2.Quote:
Originally Posted by MateoTTR
Edit: I get 666 FPS for Nature under Vista Ultimate x64 with the 9800GX2.
http://i164.photobucket.com/albums/u...ture-vista.jpg
I guess this has to be the normal score using Vista x64?
May I ask what is your mobo? 790i or X38-based?
That score is on X38 (with DDR2). I have a 790i as well, but I don't have Vista installed on it. Within XP my scores are withing 3% going from one board to the other with the same CPU and GPU settings though.Quote:
Originally Posted by MateoTTR
ok so it seems you are not affected by this bug...
I found some other evidence of this problem:
http://www.xtremesystems.org/forums/...d.php?t=182344
As far as I can see, the problem would occur only on Asus brand X38 & X48 DDR3 boards... But not all of them.
I'm also having the same problem with my P5E WS Pro and GX2 setup. I thought it was a PSU issue since i used the 6 to 8 pin adapter but today I changed the PSU to a Corsair TX750W and the problem remains.
I'm really getting worried about this, i already PMed SIOUX since he had the same board but he told me his was working fine so it seems that it only happens to some boards :(
I have read somewhere on a french forum that some people got the problem sorted out by using a different bios version for the P5E3 WS pro.Quote:
Originally Posted by ]siNNer[
Basically 0302 & 0306 would work well, while 0502 would not.
Can I ask what bios you use? Can you confirm the above?
it might happend with all the V. Cards i had some issues with my 8800GT when i left settings at auto, then set primary settings on bios manually and all was good again. I think in only shows a lot on 9800GX2 couse it really needs the extra BW.
What 'primary' settings did you change exactly such that performance of graphics card went right?!
Not sure we talk about the same issues.
Again it's the P5E WS PRO, X38 DDR2 version, not the P5E3! I've already tried bios 0404, 0507 & 0603 with no luck.Quote:
Originally Posted by MateoTTR
i have a slowdown problem with my P5E3 DLX board 0703 bios used currently
dinos can you specify what kind of slowdown you experience?
Is it the same as reported earlier in this thread (nature in 3dmark01 or canyon flight in 3dmark06).
The problem seems to happen only on ASUS X38+X48 boards... How come they did not find and solve this issue yet.
nature '01
i also tested it on Gigabyte P35-DS3P and it was even worse
i used a driver cleaner and installed 173.67 again and the frames jumped up with GB board but i am still not happy
have a look at nature
it's still at least 100FPS lower than it should be
http://www.hwbot.org/result.do?resultId=741043
Diabolical:shocked:(worst 1 liner ever)Quote:
Originally Posted by Me1tDown
1333 for nature looks good, but I trust you it should be higher for the oc you have.Quote:
Originally Posted by dinos22
Maybe still not as severe as what the thread starter experienced (900 instead of 1300).
Would you have by chance a 3dmark06 bench?
In canyon flight people experiencing the bug get -40fps on average for this specific test.
This is literally insane !
I unmounted my E8500 to replace it with my QX9650.
I also plugged out my 9800GX2 and plugged it back in the same slot again (closest to CPU).
I rebooted, launched 3dmark06 canyon flight: score went up from 79fps to 93fps !
I smell a crazy bug with the pe53 premium bios... I dumno if it has something to do with the CPU being used or the fact to re-detect the VGA card on a new slot when booting.
Or maybe the fact to switch CPUs forces the bios to reconsider some internal settings / timings, and based on that the bug shows off or not.
I had that motherboard briefly and from what I could tell graphics performance wasn't as good as I thought it should be no matter what cards I used. I didn't have a GX2 to try.
I have been investigating a little bit further:
Everytime I boot my computer, the situation can be different.
After each boot, I run systematically 3dmark06 canyon test and I can see immediately if the performance is normal or bugged.
Then if I restart, without even entering bios, the bug can show off again, or maybe not, it's just pure random.
Once I have booted up bug free with the right graphics performance, I can run the test ok mutliple times, everything is consistent until the next restart.
I am struggling to understand if it's a board / bios issue, graphics card issue or something related to my install (the latter I doubt).
Me1tDown, mentions the 790i Image fixing the problem on his board. I wonder if it has something to do with intel chipset drivers or similar. Or whether there's a driver that xp installs on the intel board that it doesn't on the nvidia board.
Would it be worth a fresh os install of XP, install graphics drivers and then futuremark stuff and give it a whirl? Prior to updates and further driver installs.
RLM
I really don't see how re-installing windows would help in my case, but who knows...
I have made the following test 10 times in a row:
- Power on the computer and boot windows
- Run canyon flight test 3dmark06 right after getting to desktop
- Shutdown the computer completely, wait 10s, then start again
4 times I booted without the graphics performance bug, 6 times I had it.
I am not touching any bios setting, nor changing anything to my OS install / drivers.
This board behaves completely erratic. You boot once, it's not right, switch off, boot again, then it's ok... Makes no sense at all.
Once I have booted up in the 'right' configuration, everything is ok until the next restart/shutdown.
far out man
i am so relieved that corruption issues on 790i are over
last night i fired up my replacement board and clocks my GPU to lesser clocks it was on previous setup and my nature was 1410 ...... with higher clocks it would obviously increase
there is definitely an issue of some sort even on that P35 board even if performance hit is smaller
I had the same problem believe it or not a P5E64 WS Pro (X38) with 2900's in CrossFire. The nature in 01 was abysmal, though still not as bad as what you guys are seeing with the GX2's relatively. It was about 120 FPS too low. I never solved it, chalking it up to a faulty board. Single card performance was fine though.
The AM3 also seemed to suffer considerably.
P5E64- http://www.hwbot.org/result.do?resultId=682892 309k
BlackOps- http://www.hwbot.org/result.do?resultId=707977 317k
I think I have made considerable progress in understanding when the graphics performance issue happens with the P5E3 premium + 9800 GX2 combo.
In fact I have no problem when I power on the computer and go straight into windows.
I can run 3dmark06 and will get the correct fps. Even after multiple hours of up time I can launch the benchmark again and get the expected performance again.
The bug shows off as soon as I restart the computer, without powering off.
I will also get the bug if I power on, enter the bios, save changes and reboot without powering off.
The bug has definitely something to do with the board adopting a different behavior under reset.
I am convinced about this, as I spent the last few days running Canyon Flight in all kind of situations to come to this conclusion.
What I do now as a workaround is power off everytime I want to restart, or power off after each modification of the bios, wait 10s and then start the computer.
I would really like to see if others could confirm the above, and possibly draw the attention of asus engineers on this issue...
hmmm very interesting
have you got another PSU around by any chance???
I do have another PSU on my older rig, but I fear it doesn't provide enough wattage (enermax 550) for 9800 GX2 + QX9650 but I might be wrong.
It might not have the 6 + 8 pin plugs, but it can worked around.
Do you think it is worth testing my other PSU?
The Zalman 1000-HP I got could be the culprit?
sure it does
in fact very easy even with max OC on air for both
The beta bios 0503 seems to have solved my performance issues with 9800 GX2.
I will do more testing to confirm...
Yes, this one fix bad performance :)
pmp I know you have close contacts with Asus.
Do you care to elaborate what was wrong with the previous bios and what was then fixed in 0503?
Just wanna understand.
If i understood you correctly, you would like to know from technical side what were the reasons of slow performance of 9800GX2 ?
absolutely.
Sorry my post was not so clear.
Sorry as well if you have no clue what the technical problem was and how it was solved.
In my P5E WS Pro the problem seems to continue with the new 0704 & 0801 bios. I just gave up and bought a P5Q Deluxe, lets see if the P45 is any good and if I don't have the same performance issue. Anyway, I'm still keeping the P5E WS Pro for the time being and waiting for Asus to release a bios that fixes it. :rolleyes:
As I thought, with the P5Q Deluxe there's no bug. Nice to finally see the GX2's full power :D
AFAIK there's a beta BIOS for the P5E WS Pro that fixes the 9800GX2 issue.
There are beta BIOSes coming out for the 9800GX2 affected boards coming out soon ( we already have BIOS fixes for the P5E3 series )
Which bios is that ?Quote:
Originally Posted by BenchZowner
P5E3 Premium v0503
P5E64 WS Evolution v0501
For the "non-deluxe" and "pro" and boards I don't have the BIOSes, and don't know if they're available yet, because I haven't requested them.
:worship: pmp :worship:Quote:
Originally Posted by pmp
really thankful for that, gonna change the P5Q for the P5E right away :up:
hi
i have this problem with my ASUS P5E64 WS PROFESSIONAL
i was thinking to sell the board but sinner as told me that is possible to correct de bios.
i love my 9800gx2 can someone help me?
You just need to flash your board with the new BIOS M4TR4 ;)
Use the one pmp linked.
my board is a p5e64 ws pro and sinner's board is p5e ws pro, mine is ddr3 his ddr2, do they booth use the same bios?Quote:
Originally Posted by BenchZowner
You need the new BIOS for YOUR motherboard.
You can't flash a BIOS for a different motherboard.
Go Go Go... ;-)Quote:
Originally Posted by M4TR4
ASUS P5E64 WS Professional BIOS build 0802
http://i169.photobucket.com/albums/u...07/802test.jpg
well iam a bit desapointed de bios did not work as planed...
i evan got worse score....
Problem solved, thanks pmp!
Now benching @ 4500, gx2 @ 700/1100 and 21k on 3dmark 2k6 :D
Is there a fix for the P5E3 Deluxe, does anyone know?
RLM
*brb*
true man it's retarded how bad it is lolQuote:
Originally Posted by RLM
It looks like there's not a 9800GX2 fix BIOS for the P5E3 Deluxe yet.
This may be a long shot..I'm having similar issues with my GX2 and my P5B Deluxe. Are there any planned fixes for that board?
I only get 13k in 3D Mark 06 with the spec in my signature :shocked::down:
PMP, do you happen to have any newer version of P5E WS PRO BIOS than 0802?
I'm having strange problems with a GTX280 & P5E WS PRO, using 0802.
When using 0801, I get PCIe 1.1 & 8X speed.
With 0802, I get PCIe 2.0 & 8X speed, better but still not 16X speed.
And on top of this, with 0802 I get ~30-60 seconds delay before cold boot-up (the 1st beep).
I replaced the motherboard, still no go :(
A 9800GX2 worked OK though, full PCIe 2.0 16X speed - and no delay at boot up.
If you have access to a newer BIOS than 0802 for P5E WS PRO, pls let me try it... this is driving me nuts.
ASUS P5E3 Deluxe BIOS build 1401Quote:
Originally Posted by RLM
Check this out and lmk :)
ASUS P5E WS Professional BIOS build 0803Quote:
Originally Posted by nusihb
Please check and PM me about the results. I don't have P5E atm to check it myself.
Hi guys
Sorry to intrude but noticed you guys are talking about a similar issue I've had with my P5E3 based board.....however I have the P5E3 standard, non deluxe which has a 1MB as oppose to 2MB BIOS on all the other P5E3 boards (So I can't flash to anything else).
Do you know of any BIOS newer than the 1002 (which is the latest BIOS on their site and ftp)?
I do not have a GX2, but I have a G92 GTS and have noticed huge variations of 1200 points with my 3dmark06 score, Game Test 3 with the zeplin has a horrible dip to sub 25FPS. (lower than my old P4 with 7950GT!)
I have noticed that the PCI-E Bandwidth changes ocassionaly from X16 2.0 to X16 or x4. I found that raising the NB voltage did alieviate that issue to some extent but I still get poor fluctuations in 3dmark06.
Thanks guys
John
Hey cheers PMP;). Might not be a while though before I get the P5E3 back in.Quote:
ASUS P5E3 Deluxe BIOS build 1401
Check this out and lmk
Anyone else tried this?
RLM