• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

AMD officially confirms some 6850 with 1120 Shaders--Blames partners

MentalIlness

MentalIlness

Platinum Member
http://www.fudzilla.com/graphics/item/20609-amd-officially-confirms-some-6850-with-1120-shaders

A very well known graphics overclocker who went to work for AMD, Sami Mäkinen, has sent us some official comments on our “out of spec” 6850 samples story posted here. Sami works for AMD Tech PR these days.

This is what AMD said officially:

“Apparently a small number of the AMD Radeon HD 6850 press samples shipped from AIB partners have a higher-than-expected number of stream processors enabled.

This is because some AIBs used early engineering ASICs intended for board validation on their press samples. The use of these ASICs results in the incorrect number of stream processors. All boards available in the market, as well as AMD-supplied media samples, have production-level GPUs with the correct 960 stream processors."

Of course, the easiest is to blame it on someone else, but we also like to go on conspiracy theory side and believe that AMD set the final shader number after it shipped first 6850 samples, and some simply didn’t get the new bios that will keep the shader number to 960.

Update: AMD has contacted us again: "Your assumption wrt BIOS is simply not accurate. In fact a BIOS update can not fix these early ES ASIC boards to operate according to 6850 spec." So much for special BIOS dreams, it's all in the hardware it seems.
 
But what does this tell us?

If any sites reviewed both the HD6850 AND the 6870, both would then have 1120 shaders, right? Performance would be the same. And it would light a red flag already there.

So. some board partners had the wrong 6850 cards, is that what this is about?

Edit: performance wouldnt be the same ofc since one is clocked at 900mhz while the other is at 775?

Could we have someone confirm which cards Anandtech and Hardocp reviewed? If those are the ones with correct number of shaders, im not giving this any more thought.
 
All you have to do is see if they posted gpu-z screenshots.And any reviewer that didn't run gpu-z after installing should not be reviewing video cards
 
hdfxst said:
All you have to do is see if they posted gpu-z screenshots.And any reviewer that didn't run gpu-z after installing should not be reviewing video cards
Click to expand...

GPU-Z is known to give inaccurate results. Most of its information is pulled from a built-in database, not read from the actual GPU.
 
Scholzpdx said:
I dont see this as an issue. More shaders = better for us.
Click to expand...
Not if they test a unit with more shaders, and then sell standard units to the public based on misleading reviews. But hopefully that's not the case, we just need to see confirmation from the reviewers.
 
so... amd accidentally sends cards with too many shaders, nvidia accidentally leaks info about 580, amd tells people to ignore the tesselation performance gains nvidia has, nvidia pays developers to cripple amd performance...
this is getting to be quite a dirty fight.
 
taltamir said:
so... amd accidentally sends cards with too many shaders, nvidia accidentally leaks info about 580, amd tells people to ignore the tesselation performance gains nvidia has, nvidia pays developers to cripple amd performance...
this is getting to be quite a dirty fight.
Click to expand...

yes it is, and really it always has been.

People do not realize that neither ATI or Nvidia are clean. They both fight dirty and they do it to make their investors money, that's their bottom line.


People get so caught up in who does whatever perceived dirty event at the time and jumps ship to the other camp forgetting about the dirty things that camp have done in the past.

Your bottom line as a customer should always be just buy whatever is the best performance for the amount of money you have to spend at the time.
 
taltamir said:
so... 6850 reviews are incorrectly showing them to be more powerful then they actually are?
Click to expand...

Borealis7 said:
a cheap-shot attempt to make review benchmarks look better?
Click to expand...

taltamir said:
so... amd accidentally sends cards with too many shaders
Click to expand...


To be honest, I think it's unlikely they sent these cards to be reviewed. It would be clear if they did (reviewers should've noticed), and they'd get such a pasting for it, it wouldn't be worth the gain they got for 2 weeks.
 
What company is throwing their partners under the bus ? lol

How can multiple board partners make the same mistake?
Imho, AMD must have marked certain boards as being 'test boards' advised the board partner to put on their own cooling solution/stickers and send those to review sites. Everyone is going to do just that, because money is tight, everyone wants a test sample, and retail cards don't have to be used.

So its AMD's fault imo. Not sure it makes them a bad guy. But either does wood screws or performance claims.
 
This has happened to AMD before with the 4830. http://www.anandtech.com/show/2651 That time the review samples had less SPs enabled than retail cards. I doubt the 6850 with 1120 shaders was malicious. It's impossible to imagine this would go unnoticed so it would be a bad idea to do intentionally. Plus, now they have to throw their AIBs under the bus which I'm sure the AIBs aren't happy with, but their AIBs would probably be livid if there were taking the fall for some scheme cooked up by AMD.
 
JimmiG said:
GPU-Z is known to give inaccurate results. Most of its information is pulled from a built-in database, not read from the actual GPU.
Click to expand...

This. GPU Z is not like CPU Z. It does not read information from the system like CPU Z does. GPU Z sees that your system has X video card then it checks and sees what X video card is supposed to have in a database and it displays that as what you have. It doesn't actually count the number of ROPs and Stream Processors and such.
 
Despite all this hand-wringing, I would happily take a 6850 with 1120 shaders. Not that I'm dropping any hints, AMD board partners.

....

send me one!!!
 
Well they were pretty intent on making this seem like a HARDWARE issue and not a bios issue.

What are the chance's 6850s are 6870s with a bios implementation that limits the number of useable shaders?
 
MentalIlness said:
Update: AMD has contacted us again: "Your assumption wrt BIOS is simply not accurate. In fact a BIOS update can not fix these early ES ASIC boards to operate according to 6850 spec." So much for special BIOS dreams, it's all in the hardware it seems.
Click to expand...

I believe AMD lied to you.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top