Radeon HD 6970 Launch May Face Delays

[cadaveca]

They already said "Ooops" during the 6870/6850 launch. Just none but me heard them.

And I agree, both have failed hard, but really, It's not thier fault, except when they lie to cover up the problems. Honesty wins all, with me. They can make mistakes, that's no big deal...it's the lying about it that gets me going.

 

[bogie]

ok any final details about it
i remember this
850 core , 1600 memory , 256 bit , 32 rops

I heard it was gonna be 900 or 950 on the core! Almost guaranteed 1GHZ o/c! Yay!

I would not believe these false reports about being late, pure FUD!

 

[the54thvoid]

They already said "Ooops" during the 6870/6850 launch. Just none but me heard them.

And I agree, both have failed hard, but really, It's not thier fault, except when they lie to cover up the problems. Honesty wins all, with me. They can make mistakes, that's no big deal...it's the lying about it that gets me going.

I'm assuming you're implying TSMC. Many people think it stands for Taiwanese Semiconductor Manufacturing Company but it does in fact stand for "Take Shit and Make Crap"

 

[Steevo]

Perhaps the issue with the cursor is specific combination of hardware, software, drivers, OS.....etc

All things beyond their control. If one piece of software for playing DVD's causes a fast switch issue and thus the large cursor, is it ATI's fault, or the software?


I enabled things like geometry instacing, a few different patterns of AA, AF and the ability to force DXVA and WMV hardware accleration. It has gotten me a couple times of video hangs due to codec problems, but once I sorted them it works great.


Take the average user that tries to enable it with a incompatible codec or one known to cause issue and they will cry like a little bitch about how horrible ATI is with their drivers.


Issues I have are currently.

Even with a custom BIOS the fast switch hang and video corruption still occurs with the codecs i have to use with my HG20 camcorder. I can't blame ATI for Pixelas shitty software though.

Occasional hang and green flash video. Voltage, GPU speed, memory speed, multi-monitor, Windows 7 and Vista. No difference, so I tend to blame Adobe for flash, as silverlight with hardware acceleration on Netflix is flawless.

--------------------------------------------------------------------------

http://www.bit-tech.net/hardware/graphics/2006/04/07/transparency_adaptive_aa_explained/2

For ATI cards the ability to use reg hacks to enable things that are no officially supported has been around for a long time.

"It's also possible to get Adaptive AA working on any GPU that is newer than Radeon 9700 via a registry hack. This is not officially supported by ATI, and performance implications are quite large. However, if you're interested in giving it a try for yourself, please read our news story covering the topic."


Nvidia does some things differently, however I have used alot of third party tools on their cards to enable features.

 

[cadaveca]

Perhaps the issue with the cursor is specific combination of hardware, software, drivers, OS.....etc

I have eliminated all those possibilities. Differnt rig, board, mem, vgas, HDD's, psu...It's been over a year, and I hafvee eve ntried both AMd and Intel...and EVERY USER IN THE EYEFINITY THREAD HAS CORRUPTION. EVERY ONE. THe only commonality is using a DP connection on AMd cards.

I know what the problem is, exactly, with the cursor. It will basically NEVER be fixed on 5-series cards, 100% for sure...it's been 13 months now!

 

[Steevo]

Are you using adapters?

It occurred on my new 5870 but as soon as I flashed my custom BIOS on it it has never happened sense.


If we could find WHY it happens, or WHEN it happens. Is it a scaling issue, does changing to a different pointer fix it?

 

[cadaveca]

timings for DVI and DP are different. AMD hardware has issues translating that. It's hardware-based...no driver will fix it. Been saying this for months now. Adapters don't make any difference. Driver's only change the end result, but do not prevent corruption.

Likewise, a bios will not fix it. I am running the most recent bios from AMD.


I have tried both reference and non-reference cards.

 

[wolf]

reminds me of when I bought my 5870 at launch and there was driver issues with a fair few games, and a lot of cloudyness surrounding eyefinity.

In fact I remember the 5870 product page on the official AMD/ATI website at the time, had a disclaimer at the time to the effect of;

*driver support for 5800 series cards sheduled for 2010

 

[Steevo]

Have other users replicated it on DP monitors, with the same result?

 

[cadaveca]

NO offense, Steevo, but what part of ALL EYEFINITY USERS ON THIS FORUM(and others) HAVE THE PROBLEM do you not understand?

It's not software, it's the display output of the 5-series cards that causes the issue. 6-series cards have DP1.2, to fix the problem, but it only works with either the multistream boxes, or daisy-chained monitors. Without either of those, even the 6-series cards ahve the problem, as the DP and DVI outputs have different timings and protocols.

To eliminate the corruption, you must use the same type of output for all monitors, however, as it stands today, this is not possible, as the hardware is not on the market, and will not be, until next year.

And with that said, Eyefinity isn't even Beta...it's more like "alpha"...bare functionality, but broken as hell.

 

[yogurt_21]

It's not as big as some of you may think.
HD 5870
http://i403.photobucket.com/albums/pp112/erocker414/5870.jpg
HD6970
http://i403.photobucket.com/albums/pp112/erocker414/6970.jpg

that's what she said!

didn't expect it to be much larger, even despite being stuck on 40nm it doesn't seem like the core is goign to get that much larger and 256-bit mem says they don't have to add too many chips onthe board either.

 

[bear jesus]

I have to chime in on this, i started getting the corrupt cursor yesterday (no idea why it took so long to start happening) not on the monitor with the display port adapter but one that's on a dvi port. i was really not looking forward to getting this problem as i knew before buying that basically everyone else running eyefinity was but now i see it as a very minor issue that is more than made up for with the extra screen space.

But i accept that it is an old issue that can be very annoying for some people but if it is a hardware limitation thus unfixable why does ati keep adding "fixed corrupt cursor" to the driver fixes list?

 

[cadaveca]

Becuase I bitch about it so much, they now admit it's an issue, at least, and maybe they can pull some driver magic to fix it, but I highly doubt that they will ahve any success implementing a driver fix. They've listed it as fixed in the release notes, several times now, only to have it appear in the next month's notes as "a known issue".

You, bear, will be able to switch to a multi-stream box, when they come out, because you have 6-series, and this will eliminate the problem. So all is not lost for you. 5-series owners, unfortunately, do not have this option. I bought 5-series cards specifically for Eyefinity, so yeah, it's a sore point for me.

It's doesn't happen on the DP monitor for me, most often, either...it's on the DVI monitors that the cursor shows up corrupt most often. But give it some time, and you'll see it on the DP monitor as well.

 

[bear jesus]

You, bear, will be able to switch to a multi-stream box, when they come out, because you have 6-series, and this will eliminate the problem. So all is nto lost for you. 5-series owners, unfortunately, do not have this option. I bought 5-series cards specifically for Eyefinity, so yeah, it's a sore point for me.

It's doesn't happen on the DP monitor for me, most often, either...it's on the DVI monitors that the cursor shows up corrupt most often.

for me so far it's been a non issue so i doubt i would want to change form my current setup, but i admit one of the reasons i was happy waiting to upgrade is so that ati or nvidia worked out the bugs with triple monitor setups... it did not work out so well but i admit if i had got the 5xxx cards and been dealing with this issue for the past year while expecting a fix i would be pretty pissed by now.

 

[lashton]

Ain't that the truth

I'd prefer for the card to come out asap, so the price war begins.
Plus, I can't wait to see how 6970 stacks up against gtx 580 that comes out this Tuesday!

And with the green team you will be buying new cards when tour cards fan stops

 

[cadaveca]

for me so far it's been a non issue so i doubt i would want to change form my current setup, but i admit one of the reasons i was happy waiting to upgrade is so that ati or nvidia worked out the bugs with triple monitor setups... it did not work out so well but i admit if i had got the 5xxx cards and been dealing with this issue for the past year while expecting a fix i would be pretty pissed by now.

Well, it wouldn't really be a config change for you...the multi-stream box would connect to one single DP output, and then that box connects to all monitors. It would just be a simple addition, and would eliminate 2 cables coming out from the pc. That single cable, could, as well, also connect USB to the monitors, as well, further eliminating cable clutter. But i do not think that the AMD cards offer the proper hardware for USB over DP1.2...time will tell...

 

[newtekie1]

So a 6-pin and an 8-pin means it is breaking the 300w barrier. I wonder by how much...

 

[the54thvoid]

Is a 6 pin not 75W and an 8 pin 150W and the PCI express 75W? That makes 300W tops.

Would you not need 2 x 8 pins to crash over 300W?

And even if i'm right? i'm sure it's not that straight forward...

 

[bear jesus]

Well, it wouldn't really be a config change for you...the multi-stream box would connect to one single DP output, and then that box connects to all monitors. It would just be a simple addition, and would eliminate 2 cables coming out from the pc. That single cable, could, as well, also connect USB to the monitors, as well, further eliminating cable clutter. But i do not think that the AMD cards offer the proper hardware for USB over DP1.2...time will tell...

I'm using 3 dvi monitors with one on an active display port adapter so it would involve buying 2 more adapters the multi stream box and all for a problem that i don't really see often and does not have a negative effect on my usage, to be honest i would rather keep that money to go towards a 6970 or 580

But i admit if i was using display port monitors i would grab the box so i had the ability to add more monitors.

 

[cadaveca]

nah, the multistream boxes will do DVI-D/HDMi/VGA/DP out, just DP in. I'm hoping they'll sell for about $69.99.

See page 7:

http://www.amd.com/us/Documents/AMD_Radeon_Display_Technologies_WP.pdf

 

[bear jesus]

nah, the multistream boxes will do DVi-D out, just DP in. I'm hoping they'll sell for about $69.99.

Oh i assumed it was display port in and out, if your right about the rough price that would make it a pretty awesome thing to have, i will have to keep an eye out for one.

 

[TRIPTEX_CAN]

LOL driver problems.. I'm pretty surprised.


"I accidentally . the driver"

When AMD is apparently months ahead of the competition to release these products (5000/6000) I would think taking a month extra to pound out driver conflicts before launch would be a good thing.. NOPE 10.10y hotfix is on the way.

 

[cadaveca]

Basically, it's exactly as that pic shows.

Except the TH2G only does max 1680x1050 per screen, when using 3. AMD's solution supports HD res, plus 3D, etc.

 

[overclocking101]

so it looks like the 6970 is only 1 inch to 2 inches longer then the 5870 comparing the photos of the backside, that is if the core is located on the same plece on the pcb which it looks to be so. also, no bios OR drivers?? this close to launch?? one would think with all the hype this stuff would have been sorted, also with leaked benches etc and people getting cards already one would think they have it sorted, but who knows now a days

 

[newtekie1]

Is a 6 pin not 75W and an 8 pin 150W and the PCI express 75W? That makes 300W tops.

Would you not need 2 x 8 pins to crash over 300W?

And even if i'm right? i'm sure it's not that straight forward...

Nah, PCI-E 2.0 does 150w, IIRC.

That is why the GTX470 pulling 232w can make do with two 6-pin connectors or the HD5970 pulling 304w needs a 6-pin and an 8-pin. At least that is the way I've always understood it.