Jetway X1950Pro, Your thoughts on this card?

[Kasamax]

Hi guys, new to the forum but long time trawler.

I recently bought 2 Jetway x1950pro 512mb GDDR3 pci-e cards to go in crossfire with my Asrock AliveXFire-Esata2 (http://www.asrock.com/mb/overview.asp?Model=ALIVEXFIRE-ESATA2) mobo, I looked up a fair few x1950pro's and ended up on the jetway's for price and availablilty as well as the reviews I read on them seemed allright :).

Now I finaly get all my new stuff together and booted and crossfire wouldn't work did all the usual things like switching the CF dongles etc and nothing :confused:. Did a bios flash and enabled all 4 monitor outputs (IBM P260 DVI-I & VGA plus a Dell on a KVM switch) and crossfire then enabled, although on the vga output (on the suposed "master" card) by default (a lot of stuffing about with disabling, unpluging and primary montior selection jugling to get it default on the DVI) it doesn't seem to mind running on either card as master.

So now I bench the cards in single and CF, get around 16 - 16.5k on each card in 3Dmk03 and around 26.5k in CF mode. Now this is a HUGE improvement on my 4-5k with my old 9600xt eXtreme 128 :laugh: , In 3Dmk06 I get about 5k CF. As is usualy the case after 3Dmarking and seeing how high others are and their clock speeds, it's off to the gym to work out for a while and try again ;).

3 days and ALOT of driver, tool and hardware stuffing around I end up here :D

I'll post all the relavent technical details once I go back through all the progs and save the data and screencaps

[VeDz]

You get 5k in 3dmark06? Seems low. It should be around 10k to be honest.

5k is reasonable with a single x1950 pro though.

[Kasamax]

Been back through AtiTool (26+hacked & 27B2), AtiTray tool, RivaTuner, Powerstrip and a few bios progs and then RaBiT.

The following is both my cards tech specs details...

[cdawall]

maybe some ocing to get the rig to run faster but other than that cool :cool: xfire is an amazing performance booster :D

[Kasamax]

Yeah I checked my 3Dmk06 jpg and it was about 8k not 5, close enough :)

[Kasamax]

Now I'm more than happy with the performance of the cards and in crossfire, I wasn't expecting them to be :respect: .

I've gone through the net about OCing the buggers and I'll do some more work on that later today, one thing has got me rather perplexed and worried though...

In AtiTrayTool, the benchmark spinning ship at full screen on the main card starts at around 800-900 FPS and then slowly drops and raises in increments from 800-750-790-810-730-710-650-610-530-570- around 515-560 and stays there untill the first "speed" bench finishes and then on the second "quaility" bench it starts at about 400-500 and drops down to a lowest of around 200-300.

Now the second card when set to primary display on the second monitor starts at around 800ish and stays up around 700-900 the whole test, both "speed" and "quality" tests.

If I run the test on the second card (main output) and monitor without setting it to primary display it tries to run through the main card to the second card and then to the monitor droping the FPS to around 100ish. This is without crossfire enabled or cables on.

My main worry is that something is wrong for the "main" (primary slot mobo book says is the main card) for it to be doing that, either that or it's something to do with my 21" monitor on the main card and the 17" dell and kvm switch I have on the second cards outputs.

:banghead: :confused:

[Kasamax]

From what I remember of looking at the cards before I put them in the box the bottom "main" card has a GPU data of

215PADAKA12EG - NA7356.00 (NA5947.01) - 0643SSP - TAIWAN

and has all (I think all the chips were the same) hynix memory of

HYNIX KOR - HY5RS123235 - FP-14 - 645A

The top "second" card has the same GPU chip BUT, from memory it has a few samsung or sony ram chips with a couple of different numbers and another bunch of chips that I think started with a K, kinex, kline or something like that ( I knew at the time not writing ALL of the ram data down on the second card was going to come back and bite me in the preverbial :banghead:).

So whether that could be a cause is another story, what I find interesting though is that both cards when run individualy on 3Dmk03 get a very close and similar 16k score and it's only the AtiTrayTool benchmark that does it to the main card.

I would just take the cards back out and look at them again but..... I'd have to buy another bottle of coolant :p

[Kasamax]

Now I have both cards doing the FPS drop on AtiTrayTool, but the top card still gets 11k and the bottom "main" card still only get 6k, they're BOTH doing the weird FPS droping on the benchmark but stil letting the same results as before when only one was doing it.

wondering whether it's a catalyst option/problem or whether it could just be the position of the cards on the mobo or the mobo's weird throughput rate or the fact that AtiTrayTool is still a bit off with x1950Pro's.

These cards must have either been put together a few months ago or just bios flashed with a 21/05/2007 bios, I had to wait 3 weeks for the wholesaler to get them from the jetway distributor.

The most interesting thing is that second card with all those different memory modules of the different types on it instead of just one brand, haven't encountered that before.

Guess I'll just have to bite the bullet and drain the coolent and rip the cards out and take another look at them and put them back in the other way around lol.

[Ketxxx]

5k in 3dm06 with a CF setup is slow.. I can crank out 6.1k in 06 with my single 1950Pro.

[Kasamax]

Yeah bit of a miss quote there with the 5k, was actual 8k ish.

[Ketxxx]

Still seems a bit low imo. What drivers you using?

[Kasamax]

Here's some pics of the rig...

The case I've had for about 7 years or so and I've tried to replace it with a better looking one but it ends up back in charge, just like my chair I've had for about the same time and can't find a replacement lol.

Have been wanting to go water cooling for many years and now since it's cheaper these days and while I had the money and was upgrading my rig I thought "why the hell not"

[Kasamax]

Using the WarCat 7.4 Pure edition drivers (cat 7.8 based), tried the ones on the jetway site and they got me 13k in 3dmk03 on one card so ditched them and went back to the warcat with a slight modification of some of the files with some from the ones specific for my card from jetway. Pure warcat 7.4 was what I got the original 3dmk results posted above with, with the "extra" files I get around the same give or take 50-100 points.

I'm in the process of dismantling the vga coolers and am going to switch the cards around and see what happens, then I'm going to flatten the drivers and try the cat 7.9 beta drivers and see how that does. Other than that there's the Omega, Dna and another set of drivers I can try apart from the warcat.

But my main "mission" atm is trying to figure out why one card does 6k and the other does 11k in AtiTrayTool benchmark but get almost identical results in 3dmk, whether its the mobo, the mem on the vid cards or the possibility that I've damaged the card that gets 6k when installing the vga cooler or when putting it in the rig, but with that said it still gets almost identical performance as the other card in other bench's :confused: I really don't like these sort of results/problems/dilemma's, being a techie it's at my core to figure them out.

Which doesn't leave much time to enjoy the damn equipment :banghead:

[Kasamax]

Well that was a job n a half...

Got the cooler drained, cards out, re heat sinked and put back in opposite slots.

It's still doing the same thing so it's not the cards, tried the omega, warcat 7.3 & 7.4 as well as the vanilla 7.8 catalyst drivers and the only difference is a few points here and there and minor nuances. With omega and warcat both cards score singly 16 - 16.3k each in 3dmk03 and on 3dmk06 they get 4989 each, with the 7.8 cat they score on 3dmk03 around 15.3k.

Now I've also noticed that my cpu score has dropped from 1600 down to 1380 ish on all the drivers in 3dmk03 but that's another matter.

I must have been on something good :toast: when I put the heat sinks on these cards cause when I took them both out again they both have ALL hynix mem and no other brand. So their core and mem speeds are 575/608(614)(depending what prog you look at) and bios default is 575/610.
The hynix ram is fp-14 which I found is rated to 700, I never seem to be able to get the cards with the samsung ram :cry:

So by rights I should be able to clock these to a point they can handle then edit the bios to the clocked values and flash it and all should be good?

I should have just waited and gone with two 2600 or 2900's :ohwell:

[Ketxxx]

I havent looked at your mobo specs, but it could be the chipset the board uses has to limit one of the PCIE lanes when dual graphics are used. If not that, its likely a BIOS bug, try updating your mobo BIOS to the latest there is.

[Kasamax]

It's just a really weird situation that's annoying me that I have to figure out lol.

Anyway, I've managed to get some O/Cing done on one of these cards.

using some of binormalkilla's methods using rivatuner for the mem I got it up to around 689 from 608 without artifacts and up to 730ish with artifacts then the system crashed, not sure if it was from heat (got the ram heat sinked but didn't feel that warm), the ram is only rated to 700mhz or whether it was the other parts on the PCB that I have yet to heat sink, or it just needed more ram voltage.

The core I got up to 681.75 from 573.5 with rivatuner and atitraytool stable any higher and then the VPU recover blanks the screen, with atitool the system just locks up trying to do the mem or the core. Now I assume the core isn't over heating cause the max it got to with the probe on the back of the card was 43.9C under load from atitraytool artifact scanning, so that leaves the core is dodgy like the mem or it could do with an over voltage mod, or it's the un-sunk parts on the board again.

When I had the core and ram clocked I tried the atitraytool bench again and instead of 6092 it got 6119 so the clock made a difference but why that pci-e port only gets 6k and the other gets 11k in atitraytool bench but they both get the same score and bench from every other program is what's puzzling, maybe it's just the way atitraytool does it's benching ,fixed with this boards chipset.

Either way it's a mute point now and my focus is now set on these cards clocks and how high I can get them to be stable at.

Is the volt mod a generic process for all the x1950pro's or does it depend on what the PCB setup is like? I'll be getting these cards out again to heat sink all the rest of the parts that need it so I'll take a snap of them and post (really should have done it when I had the mongrels already out :banghead:) and that really depends on if the volt mod will do any good for stabilizing these cards.

I haven't seen much about jetway card modding or anything but reviews, probably cause their the crappy cards but they don't seem to bad stock and for the price it was the "best" option for me at the time.

[Kasamax]

Oh and I was using bios 1.8 for my board, upgraded to 2.0 after checking the cards before clocking them and my cpu temps sky rocketed after the flash but have tapered down to what the used to be after it settled in, I did get quite an interesting set of options after the flash though.

The memory setting in the bios now have all the ...

TWR-TWTR-TRWTTO-TWRRD-TWRWR-TRDRD & TRTP

settings, 5 of those I don't know and can't find anything anywhere as to what to set them to for my ram.

Other than that I haven't noticed much difference with the new bios from the previous one.

Only thing I can think of that could be causing me problems is my mobo's got the 480x CF chipset which equates to the old 1600 CrossFire chipest, Ideally the 3200 would have been better but this board was a steal so :o

[Frenzy]

Which the bios versions can be compatibility with Jetway X1950Pro 256bit 256MB?

Thanks!

[Ketxxx]

Oh and I was using bios 1.8 for my board, upgraded to 2.0 after checking the cards before clocking them and my cpu temps sky rocketed after the flash but have tapered down to what the used to be after it settled in, I did get quite an interesting set of options after the flash though.

The memory setting in the bios now have all the ...

TWR-TWTR-TRWTTO-TWRRD-TWRWR-TRDRD & TRTP

settings, 5 of those I don't know and can't find anything anywhere as to what to set them to for my ram.

Other than that I haven't noticed much difference with the new bios from the previous one.

Only thing I can think of that could be causing me problems is my mobo's got the 480x CF chipset which equates to the old 1600 CrossFire chipest, Ideally the 3200 would have been better but this board was a steal so :o

Make a BIOS dump from one of your Pros and I can tweak it. As for the extra memory options, see link in sig :cool:

[Kasamax]

there should be 2 bios images ziped linked to the 3rd post ^ , I took one with rabit and the other with atitraytool I think, either way what ever prog I've tried it saves the bios the same. have tried a fair few progs to try and edit/view the bios and some won't even open it saying it's not an ATI bios and from the other forum when opened in rabit it says it's an x1650xt bios lol. I haven't had a chance to dos boot and try to save it with atiflash yet so I might get to that shortly.

On the bios note, I looked up sites about and explaining "how to" and tried to get a flash copy and view of my mobo bios and no prog I found and tired could save it or even want to attempt to. It's an AMI bios but I've tried the latest AMI bios editor/saver and it doesn't work so any thoughts you have on that'd be appriciated.

Back to the vid cards... I managed to get some O/C ing done as I said in a previous post, I found out a few "stable" clock mix for the mem and core. First off I got 661.50/634.50 which gave me 17.2k in 3dmk03 and tonight I managed to get 573.50/688.50 for 16.4k at 36.1c/26.7c and then got 654.50/688.50 at 39.1c/27.5c for 17897k and one more test got 654.75/688.50 at 38c/28c for 17908k (core/mem). Temps aren't entirely accurate for the workload as I only let the card cooldown a certain extent before running the next tests.

I can run the mem or the core at the "max" separately but the last test is the highest I have gotten them to run together and not crash, although after running the last test and doing a couple of other things, opening up the atitraytool 3d renderer window crashed it :ohwell:

I have the temp probes for the core sitting on the back of the card in the middle of the back of the GPU (so give or take a few C) and the mem probe I have sitting on the GPU cooler frame lodged on two heatsinks of the ram chips. Since this being the jetway budget, it doesn't have the sensor probes that other x1950pro models have.

Attached is the 3dmk03 screen cap, a base outline of the top of my card from jetway (different GPU heatsink came with mine), a screen cap of the "patched" ati tool v26 memory timings view of my card (think theyre the same as the rabit report says) and a weird ati tool "thing" that happened after a reboot while I had it running with atitraytool and rivatuner and the VPU recovery kicked in and rebooted, it came up twice and hasn't come up since.

Anywho that's me done for the nights hypo :rockout:

P.S... Found an actual image of my cards (heat sink with graphic but base outline is the same), think there's a pic of the back of it on one of the review sites, easier to find than taking them out of the rig again if I need to find out where to volt mod :)

[theonetruewill]

Well I have to say it; Thanks for not being one of those n00bs who don't detail anything and expect instant feedback with no pics/system specs etc. I found your posts to be informative and exceptionally detailed while still being clear. Even your attachments are in my, personally preferred, unobtrusive manner. The 8K is a tad low as Ketxxx says but I wouldn't worry about it at the moment- as you say you're going to tweak them.

:)

[Kasamax]

:D Thanks for your kind words

Though everyone is a n00b at one point or another, except for *insert generic great 1337 persons name here* :respect:

As for me I tend to give TOO much information when I'm looking for help with something, I'm one of those personality types that needs to know everything about a situation/problem when trying to solve it (well as much info as possible) and tend to lump all the useless info together with the relevent stuff. For this "adventure" I felt like delving back into an online community of wealth and knowledge and portraying my achievments and failures to the great beyond.

As Vanilla Ice sung "If there was a problem YO I'll solve it" :o

God I need to get some sleep lol (Damn AU timezone compared to the rest of the world)