AMD playing dirty tricks with 6850 samples?

E

eddieck

Gawd
Joined
Dec 13, 2009
Messages
1,010
Check this out at TPU. AMD seems to be playing games with at least some of the 6850 samples and sending them out with 1120 shaders enabled.
 
Waitwaitwait, how did he do this?:

The HD 6850 specification requires 960 shaders. Unfortunately neither HIS, nor AMD could provide help with getting the board fixed so I had to reconfigure the shaders on my own.
Click to expand...

He should share! :D
 
"I know of one other site who received a Sapphire HD 6850 with 1120 shaders enabled, so the problem might be more widespread."


problem.. i dont see a problem.. :D possible chance of getting a 6850 with 6870 performance for way less :D

interesting problem though.. looks like some one screwed up labeling the cards.. or the problem is the AIB's since they said they came from HIS and sapphire.. most engie samples come directly from AMD..
 
sirmonkey1985 said:
problem.. i dont see a problem.. :D possible chance of getting a 6850 with 6870 performance for way less :D
Click to expand...

That's nice and all, but it's misleading to buyers who may be reading reviews showing the 6850 as performing better than most cards in the market actually will.
 
Well, if most reviews used the 960SP version while people buying actual cards occasionally got 1120SP variants, that'd be pretty cool lol

I agree the other way around wouldn't be quite so good though.
 
eddieck said:
That's nice and all, but it's misleading to buyers who may be reading reviews showing the 6850 as performing better than most cards in the market actually will.
Click to expand...


completely agree but then the problem lies on the reviewers not doing their jobs and checking the hardware they receive before doing the review..
 
eddieck said:
That's nice and all, but it's misleading to buyers who may be reading reviews showing the 6850 as performing better than most cards in the market actually will.
Click to expand...
I highly doubt that is intentional as even a cursory check by most reviewers would catch this. Who doesn't at least do a quick check with GPU-Z? Reviewers who don't catch this aren't doing their job.

If anything, it might spell a potential unlock for buyers, which would make the 6850 a simply incredibly deal.
 
sirmonkey1985 said:
completely agree but then the problem lies on the reviewers not doing their jobs and checking the hardware they receive before doing the review..
Click to expand...

That's somewhat debatable IMO. I'm sure after this incident reviewers will check their cards, but generally when you get a review sample of a product that has "Radeon HD 6850" stamped on it, you assume it's a.... Radeon HD 6850!
 
eddieck said:
That's somewhat debatable IMO. I'm sure after this incident reviewers will check their cards, but generally when you get a review sample of a product that has "Radeon HD 6850" stamped on it, you assume it's a.... Radeon HD 6850!
Click to expand...
And you don't check with GPU-Z to verify the clock speeds, shaders, memory, BIOS version,etc? Poor work, in my opinion. Especially if you get suspicious results.
 
that might explain the differing reviews going around. lol I would love to see the comments if these were nvidia cards

Hopefully this was just a mistake and not AMD trying to skew reviews
 
Rizen said:
And you don't check with GPU-Z to verify the clock speeds, shaders, memory, BIOS version,etc? Poor work, in my opinion. Especially if you get suspicious results.
Click to expand...


agree, should be the first thing done.. hell i do it with every card i buy..
 
Rizen said:
And you don't check with GPU-Z to verify the clock speeds, shaders, memory, BIOS version,etc? Poor work, in my opinion. Especially if you get suspicious results.
Click to expand...

I'm not a reviewer myself. I can say when I get a new card I don't check it with GPU-Z, but I may do differently if I was reviewing cards. (I certainly would after this.)
 
eddieck said:
That's somewhat debatable IMO. I'm sure after this incident reviewers will check their cards, but generally when you get a review sample of a product that has "Radeon HD 6850" stamped on it, you assume it's a.... Radeon HD 6850!
Click to expand...

The reviewer, W1zzard, at techpowerup! wrote:

"One major issue with the sample I received was that it came with 1120 shaders enabled. The HD 6850 specification requires 960 shaders. Unfortunately neither HIS, nor AMD could provide help with getting the board fixed so I had to reconfigure the shaders on my own. If you see other HD 6850 reviews on the web that show surprisingly high performance, please ask the reviewer to check fillrate or using GPU-Z. I know of one other site who received a Sapphire HD 6850 with 1120 shaders enabled, so the problem might be more widespread."

So, I assume the results are still accurate.
 
eddieck said:
That's nice and all, but it's misleading to buyers who may be reading reviews showing the 6850 as performing better than most cards in the market actually will.
Click to expand...

I totally agree. They should get new cards to test, so it won't be anything misleading!

Would be nice to know how he managed to change the amount of shaders active though :D

A bit OT: I'd also wish that Wizzard would spend this much effort in checking what the newest driver revision is. Love his GFX noise tests in the reviews, but would have preferred he didn't use drivers which are several revisions old.
 
heflys20 said:
The reviewer, W1zzard, at techpowerup! wrote:

"One major issue with the sample I received was that it came with 1120 shaders enabled. The HD 6850 specification requires 960 shaders. Unfortunately neither HIS, nor AMD could provide help with getting the board fixed so I had to reconfigure the shaders on my own. If you see other HD 6850 reviews on the web that show surprisingly high performance, please ask the reviewer to check fillrate or using GPU-Z. I know of one other site who received a Sapphire HD 6850 with 1120 shaders enabled, so the problem might be more widespread."

So, I assume the results are still accurate.
Click to expand...

Correct, but what about other reviewers who may not have known about this bullshit? Whether reviewers should check GPU-Z first or not is an entirely different issue.
 
eddieck said:
Correct, but what about other reviewers who may not have known about this bullshit? Whether reviewers should check GPU-Z first or not is an entirely different issue.
Click to expand...

I'm sure they know now. Also, imagine if you could unlock 1120 shaders?
 
Maybe the shaders are only disabled and all that is needed to enable them is a BIOS flash?
 
H-street said:
TP said:
* No support for CUDA / PhysX
Click to expand...
I like this in the Cons section of their review.... don't know why but made me laugh.
Click to expand...

I don't really care about the Physx, but really do wish ATI cards and Nvidia cards could run a common api that is actually used. Unfortunately, CUDA is what is used and so I need a video card that does CUDA.

That and I have never been fond of getting ATI drivers to work right in linux.
 
PersonalJ said:
Maybe the shaders are only disabled and all that is needed to enable them is a BIOS flash?
Click to expand...
That's what is being reported. If that's true and the shaders are not physically cut on retail channel boards that would be pretty awesome.
 
Tamlin_WSGF said:
Thats what I am hoping. Remember earlier, when you could unlock GPU's via software, before they started to physically disable them.
Click to expand...

I like how the vibe of the topic went from..........."Amd might be cheating!," to, "Wow, can all 6850' unlock like this"?

LOL.
 
heflys20 said:
I like how the vibe of the topic went from..........."Amd might be cheating!," to, "Wow, can all 6850' unlock like this"?

LOL.
Click to expand...

Lol! Got some good vibes from the past when I saw he altered the shader count himself. :D

Its an important thing though to get out that some of the cards might have had too many shaders active. We buy our cards often based upon review results. Kudos to OP for bringing this up!


(and W1zzard@TPU, using old drivers on AMD cards and Nvidia cards have the same effect, so please use recent drivers in your reviews!)

NVIDIA: 258.96
ATI: Catalyst 10.7
HD 6800: Catalyst 10.10 Oct 12
Click to expand...
http://www.techpowerup.com/reviews/HIS/Radeon_HD_6850/5.html
 
Leave it to Nvidia's damage control bullshit train to turn good news for consumers into something negative. If a 6850 can have more shaders enabled that is a good thing. It means yields are good and a cheaper card can be turned into a more expensive one without much effort. It's the same reason why the Phenom II x2 Black Edition CPUs that can unlock 4 cores are so popular.
 
Lorien said:
Leave it to Nvidia's damage control bullshit train to turn good news for consumers into something negative.
Click to expand...

You don't seem to get it, do you? Reviews of a new card should be performed using an item that would be equivalent to what a regular consumer could get from Newegg.

This could be a positive thing for consumers indeed, but it is also misleading. Most consumers choose which card to purchase based on reviews. If those reviews are using a card that is faster than what you or me could get right now from Newegg, that's misleading.

I probably shouldn't have worded it as I did in the title, but innocent or not, this certainly deserves some explanation from AMD.
 
Ordered two Sapphire 6850 from Newegg last night... 50/50 chance at least one has 1120 shaders when they arrive? haha
 
Lorien said:
Leave it to Nvidia's damage control bullshit train to turn good news for consumers into something negative. If a 6850 can have more shaders enabled that is a good thing. It means yields are good and a cheaper card can be turned into a more expensive one without much effort. It's the same reason why the Phenom II x2 Black Edition CPUs that can unlock 4 cores are so popular.
Click to expand...

Wait what? AMD (and/or AMD partners) sent out some video cards for reviewers that were over spec and somehow this is Nvidia's fault?!? People use the reviews to help them spend money, the review cards need to be a retail product that people can buy, not a card that is much faster than the actual product.

Also since these are review cards, AMD partners are doubly responsible for sending out a legit product to testers. Random cards with too many shaders should not just slip out as an "oops" moment.
 
Lorien said:
Leave it to Nvidia's damage control bullshit train to turn good news for consumers into something negative. If a 6850 can have more shaders enabled that is a good thing. It means yields are good and a cheaper card can be turned into a more expensive one without much effort. It's the same reason why the Phenom II x2 Black Edition CPUs that can unlock 4 cores are so popular.
Click to expand...

Hmm let me put it this way, I race my car and post the time from 0 - 60 in 5 seconds. In a 4 cylinder car forum only.

Someone in the forum buys the exact same car thinking he is going to get the exact same results only to find out it takes his car 7 seconds.

Only to find out my car had a turbo on closer inspection and I forgot to post that too.

You need to think what a consumer would buy in 6 months down the road from now. If those cards shaders will come at the correct number or not.

Another perspective, when the 465 came out, people realize they had to just re-flash them to 470 and it will work. They did not come that way from the factory, but consumers found out about it and we bought cards knowing there is a chance it could be flashed to a 470. Now none of those cards got benched at 470 but at 465 because thats what the specs said on it even though we could easily flash them to 470. Now help me find a 465 that could be flash to 470. Very hard to find because it was only at the beginning. This card is spec and should be benched at spec.
 
eddieck said:
I probably shouldn't have worded it as I did in the title, but innocent or not, this certainly deserves some explanation from AMD.
Click to expand...

Its important to bring up, though you might have jumped the gun a bit on your thread title. :)

W1zzard explains it like this:

but this does not happen for retail boards anymore, for years all amd gpus have been fused at the asic level and not via bios. as mentioned in the article, my gpu is marked as an engineering sample and amd probably forgot to fuse it correctly. since his did expect the gpus to be fused they did not change the config via bios.
Click to expand...
http://forums.techpowerup.com/showpost.php?p=2064560&postcount=20

And, he unfortunately also says that you cannot unlock shaders on retail boards. :(
 
ocellaris said:
Wait what? AMD (and/or AMD partners) sent out some video cards for reviewers that were over spec and somehow this is Nvidia's fault?!? People use the reviews to help them spend money, the review cards need to be a retail product that people can buy, not a card that is much faster than the actual product.

Also since these are review cards, AMD partners are doubly responsible for sending out a legit product to testers. Random cards with too many shaders should not just slip out as an "oops" moment.
Click to expand...

Seems to me the reviewer should NOT have reviewed a knowingly suspect product. To me THAT is fishy in itself. You can blame AMD or the card manufacturer but the card should have been sent back without a review.
 
KWH said:
Seems to me the reviewer should NOT have reviewed a knowingly suspect product. To me THAT is fishy in itself. You can blame AMD or the card manufacturer but the card should have been sent back without a review.
Click to expand...

AND when he got a proper card, he should have used 10.9a drivers from last month with updated application profiles and not 10.7 drivers from July... :p
 
Tamlin_WSGF said:
AND when he got a proper card, he should have used 10.9a drivers from last month with updated application profiles and not 10.7 drivers from July... :p
Click to expand...


I would have used current drivers for everything I reviewed. If that driver has issues, use a different driver while making clear in the review what and why you are doing.
 
eddieck said:
You don't seem to get it, do you? Reviews of a new card should be performed using an item that would be equivalent to what a regular consumer could get from Newegg.

This could be a positive thing for consumers indeed, but it is also misleading. Most consumers choose which card to purchase based on reviews. If those reviews are using a card that is faster than what you or me could get right now from Newegg, that's misleading.

I probably shouldn't have worded it as I did in the title, but innocent or not, this certainly deserves some explanation from AMD.
Click to expand...

Chill the fuck out, you're taking this very personal for no apparent reason.

I say, bring on the release of cards that can be modded to enable more cores than what they are meant to have. We didn't hear all this bitching about the Phenom II's core unlock. Why are we bitching so adamantly about an issue that only benefits the consumer.

As you can see, there are plenty of reviews out there showing the 6850 with 960 outperforming the 1gb 460 in 90% of titles. The fact that they can be unlocked is awesome fucking news for us.
 
edifyingkarma said:
I say, bring on the release of cards that can be modded to enable more cores than what they are meant to have. We didn't hear all this bitching about the Phenom II's core unlock. Why are we bitching so adamantly about an issue that only benefits the consumer.

As you can see, there are plenty of reviews out there showing the 6850 with 960 outperforming the 1gb 460 in 90% of titles. The fact that they can be unlocked is awesome fucking news for us.
Click to expand...

you wont be able to unlock those cores.

they learned that lesson a few generations ago.
 
KWH said:
Seems to me the reviewer should NOT have reviewed a knowingly suspect product. To me THAT is fishy in itself. You can blame AMD or the card manufacturer but the card should have been sent back without a review.
Click to expand...

some of these sites use ad revunue to keep food on their tables. no one wants to be the one site that posts a review two weeks late and misses out on a huge number of page hits during a new product launch. they came clean about the card oddities and reviewed it, not much more they could do.
Posted via [H] Mobile Device
 
so lets see, Nvidia creates controversy with some typical Nvidia bullshit, and now someone is posting about some new and diabolically evil AMD bullshit.


I am sure it totally a random coincidence. Just totally random and a coincidence, yup, it sure is.

nice inflammatory title too, really grabs the attention, almost like you did that on purpose.

but I bet you were just a journalism major, and you write everything like that, coincidences out the butt on this one.
 
edifyingkarma said:
As you can see, there are plenty of reviews out there showing the 6850 with 960 outperforming the 1gb 460 in 90% of titles. The fact that they can be unlocked is awesome fucking news for us.
Click to expand...

From what I heard all of the reviews of the 6850 are with the higher shader count.
 
You must log in or register to reply here.
Back
Top