AMD accused of "Golden Sample" on 290X given to reviewers, retail bought cards throttling

[ihatelag]

Actually, thank you Andy for bringing this to our attention.

Sorry that these people are pointing fingers at you - making us aware of those articles *from two different sites* was really appreciated.


I hope that we can see the performance gains now that AMD have worked on it, I think in time the results should bring the performance back on par to press sample. Don't like that the cost of that is higher fan speed but to be honest, I am thinking of getting two and throwing it under water and that'll sort it out!

 

[mmj_uk]

New article from Toms on the issue:

http://www.tomshardware.co.uk/radeon-r9-290-driver-fix,review-32821.html

It's been fixed but the average fanspeed has gone up.

It's not fixed just alleviated at the cost of noise, just look at Toms results the retail samples still consistently lag about 40-80mhz or so behind the press sample, they can't fix variation completely unless they run the fan at a speed which prevents throttling altogether (in the poorest quality/leakiest cards) and that wouldn't be quiet. Or better still, design a cooler that's up to the job and go back to guaranteeing a minimum base clock.

They even explain that it still exists:

By default, 190 MHz separated the average core frequencies of our fastest and slowest samples. After AMD's newest driver build, that number shrinks to 81 MHz.

So it's either just a coincidence that Toms got a great press sample or AMD are indeed cherry picking cards for reviewers.

 

[flopper]

They even explain that:



So it's either just a coincidence that Toms got a great press sample or AMD are indeed cherry picking cards for reviewers.

might just be amd sending worse cards to some reviewers..

 

[Make]

Might just be them not testing the card's (Reviewing them themselves in effect) before sending them? Silicon has differences. Some cards run hotter at given voltage and Mhz than others.

Hence the 40% fanspeed cannot give the same performance to all cards if they are running near throttling limit already.

So their fix works, but increases fanspeed/noise a little bit. On some cards it gives more performance, on some it prevents throttling

Again, no-one is disputing the stock cooler is crap. But AMD is not sending out golden samples, they had a "fault" with the fanspeed that has now been fixed, Silicon differences cannot be fixed. Perhaps they do advertise a bit too high "boost clock" for the card in quiet mode, but they do not guarantee it. Dodgy, but not cheating.

 

[Addicted]

Might just be them not testing the card's (Reviewing them themselves in effect) before sending them? Silicon has differences. Some cards run hotter at given voltage and Mhz than others.

Hence the 40% fanspeed cannot give the same performance to all cards if they are running near throttling limit already.

So their fix works, but increases fanspeed/noise a little bit. On some cards it gives more performance, on some it prevents throttling

Again, no-one is disputing the stock cooler is crap. But AMD is not sending out golden samples, they had a "fault" with the fanspeed that has now been fixed, Silicon differences cannot be fixed. Perhaps they do advertise a bit too high "boost clock" for the card in quiet mode, but they do not guarantee it. Dodgy, but not cheating.

Pretty much summed it up imo.

 

[GoogalyMoogaly]

The 560gtx used the same pcb as the 460gtx, it was all but the same core with minimal changes, same pin out and no significant changes, it was clocked a bit higher.

But you said:

No card I can recall and certainly no big new release has ever had custom cooling on launch

"No card" not "no card using a new PCB design".

 

[andybird123]

Actually, thank you Andy for bringing this to our attention.

Sorry that these people are pointing fingers at you - making us aware of those articles *from two different sites* was really appreciated.


I hope that we can see the performance gains now that AMD have worked on it, I think in time the results should bring the performance back on par to press sample. Don't like that the cost of that is higher fan speed but to be honest, I am thinking of getting two and throwing it under water and that'll sort it out!

you are welcome, if this has been in any way useful to people in knowing a potential pitfall before buying then it is worth it, I wish someone had done the same thing for me before I bought a 7970m

the same problem with the 670 reference was well documented so a bit surprised that people have been so hostile in getting the same news about the 290/X
edit; wrong choice of words, not surprised, par for the course on this forum

I'm quite sure that the custom cooler cards won't have this issue, or as you say, just throw them under water and problem solved