[Legitreviews] R9-290X Press vs retail performance variances/anomalies still exist

X

xoleras

2[H]4U
Joined
Oct 11, 2011
Messages
3,551
http://www.legitreviews.com/amd-radeon-r9-290x-press-sample-versus-retail_129583

http://www.legitreviews.com/wp-content/uploads/2013/11/retail-card-hot-fps-645x492.jpg[/i

[img]http://www.legitreviews.com/wp-content/uploads/2013/11/bf4-retail-hot-clocks-645x491.jpg[/i

[quote]So in our worst case scenario the AMD Radeon R9 290X press card and retail card showed clear signs of throttling when things got warmer, but they weren’t close to being the same. Our test results from the chart above showed the press card averaged 880.86MHz over the course of the benchmark run versus just 752.95MHz on the retail card. This makes the retail cards clock speed roughly 14.5% lower than the press card in the same system with the same ambient temperature.
So, there is a clear difference between the press card that we were sent and the retail card when it comes to real world performance after the card heats up and is run in a warm system. Both cards are virtually identical, so this could mean that AMD cherry picked press sample boards or that there is a huge variance between cards. It was then that we noticed both cards were running different BIOS versions.

Read more at [url]http://www.legitreviews.com/amd-radeon-r9-290x-press-sample-versus-retail_129583/2#3AzRZOuJjHFhmm3s.99[/url][/quote]

This, is why AMD should be kicking themselves for putting such a shit cooler on an otherwise capable chip.
 
Last edited by a moderator:
The big issue here is the BIOS. We know AMD put out a last minute update for the 290s after Nvidia dropped prices and started competing. Which means the 290X is definitely being held back in 2 areas.
 
Notice how they don't show you all the data at once, which is what you need to make any sort of conclusions from their tests. They don't want you to form your own conclusion...
They fake you out with those charts on the last page, like they mean something.
 
This should properly get investigated whether AMD cheated or retail just had a bad batch of cards.
 
Couldn't this just be silicon lottery? I am not sure I buy this argument.
 
they only tried 1 sapphire card. That's in no way a large enough sample base to say that all 290x behave the same way.
Amd thought it was a rpm issue, but if u look at the 3rd page , retail card with press bios = press bios performance. So which sellers have an older bios? This article make no effort into finding out.

Only affects quiet mode zzzzZZZ.

Even with the amd drivers for the increased rpm, sapphire released a performance bios for the 290 (increase oc ability) .

http://www.sapphiretech.com/presentation/downloads/?pid=2086&psn=0006&lid=1&os=0
(290 only)
 
madgun said:
This should properly get investigated whether AMD cheated or retail just had a bad batch of cards.
Click to expand...
It already was properly investigated. The article is making a whole lot of something out of nothing by misrepresenting things. Like LordEC911 said, read the whole article, especially the last page.

tl&dr: they bump the fan cap a bit and the clock throttling issue goes away
 
Sounds like 290X issues are software related, GPUs don't tend to perform at their best until about 6 months later, has been like this for ages now.
 
mesyn191 said:
It already was properly investigated. The article is making a whole lot of something out of nothing by misrepresenting things. Like LordEC911 said, read the whole article, especially the last page.

tl&dr: they bump the fan cap a bit and the clock throttling issue goes away
Click to expand...

AMD can solve the entire situation by having a minimum base clock. That way detractors won't be able to label something like this as cheating in the benchmarks. Their current implementation does seem to raise eye brows every now and then especially if a minimum performance metric isn't guaranteed within the same set of chips.
 
mesyn191 said:
It already was properly investigated. The article is making a whole lot of something out of nothing by misrepresenting things.
Click to expand...

Whatev in regards to the article, but the comment section is entertaining me though. "AMD should be sued for this!" or "BIOS isn't software you idiot"! LOL.

I see this article doing nothing more than inciting the fanboy ramble. At least for a few days.
 
heflys20 said:
Whatev in regards to the article, but the comment section is entertaining me though. "AMD should be sued for this!" or "BIOS isn't software you idiot"! LOL.

I see this article doing nothing more than inciting the fanboy ramble. At least for a few days.
Click to expand...



Yup, same thing happened over @ PCPer in the comments section. For some reason this issue is really pissing off Red Team despite everyone saying the chip is great, the cooling is not.
 
I bet ASIC quality has something to do with it. Haven't seen any review list the ASIC quality score.
 
SonDa5 said:
I bet ASIC quality has something to do with it. Haven't seen any review list the ASIC quality score.
Click to expand...

That still shouldn't make that big of a difference in a review. I mean if you buy a 290x, you shouldnt have to worry about ASIC quality. It should run just like the review samples on websites.
 
DASHlT said:
That still shouldn't make that big of a difference in a review. I mean if you buy a 290x, you shouldnt have to worry about ASIC quality. It should run just like the review samples on websites.
Click to expand...

It does when you use the 40% fan cap. Different bins have different voltages (anywhere from 1.11v to 1.17v), and with such a low fan speed the higher voltage binned chips will have to throttle more. If the "Uber" BIOS is used the fan is allowed to spin fast enough that even with different voltages it's kept cool enough to run at max boost.
 
Quiet mode seems pointless. That's why there's a need to define a minimum base frequency. Why have a mode that will take the card to the brink of being mediocre.
 
DASHlT said:
That still shouldn't make that big of a difference in a review. I mean if you buy a 290x, you shouldnt have to worry about ASIC quality. It should run just like the review samples on websites.
Click to expand...


Legit Review did state the leakage quality factor.
 
This is why AMD needs to go away. They constantly release shoddy equipment and even worse drivers.
 
images
 
Sprayingmango said:
This is why AMD needs to go away. They constantly release shoddy equipment and even worse drivers.
Click to expand...

They can't go away. They're excellent competition for NVIDIA, and keep prices down for discrete graphics solutions. Who would prevent a monopoly and raised prices if they left? Matrox? S3? Not a chance. (Speaking of, I'd really like to see a third major player in this industry. Would be really interesting, I think.)
 
Intel's Iris GPUs are pretty good, as APUs are now looking like the future I think we will se Intel invest more and more into GPUs.
 
GameDebate (which isn't quite the best source, I know) posits the Iris Pro 5200 around the level of the GT 640/HD 7670.

The Iris is promising, but APU's will be almost certainly cheaper and I do believe that Kaveri will handily outperform them, not to mention how expensive iX-4xxxR processors are compared to the supposed price and performance of the Kaveri APU chips, which have been estimated to have performance roughly equal to an underclocked HD 7790 with DDR3 RAM, which isn't too shabby by any stretch. And, you can add a Radeon dGPU to expand upon it, which is also nice.

I suppose Iris has its comfy place in Macbooks, since price/performance doesn't seem to be a major limiting factor for people who buy them anyway, and the Iris has very low power consumption and small form factor for a graphics processor that pulls some weight, making it a bit nicer for laptops than AMD's design in terms of power and cooling, which the MacBook (or most laptops, for that matter) doesn't have all of the room in the world for.

Personally, though, I'd rather have a Kaveri laptop than one with Iris, since I play plugged in when I do play on my laptop, and I don't really want to spend extra for a MacBook. Not that I won't keep the laptop I have now for a while yet, as suits my gaming needs just fine.
 
Last edited:
mesyn191 said:
Detractors will always find something to bitch about, that is why they're detractors.
Click to expand...

I mean that's completely status quo and fine per some AMD posters here, right? And the people calling BS are just..."detractors". Whatever you say.

Look, just tell AMD to fix their shit so that none of this circus bullshit is never brought up in the press; this is AMD's fault, they are doing a disservice to customers who are buying cards that aren't performing as they should. People are going to be detractors when a product has performance variance from one card to the next of up to 15%. That type of performance variance is bullshit. Don't blame the users for pointing it out, blame AMD for releasing their card in a semi-broken state - clearly the Hawaii chip is very powerful but AMD doesn't know what the hell to do with it. Again, AMD needs to fix their shit. And no, not everyone is going to run uber dust buster fan, and not everyone will know that "quiet mode" has severe performance variances.
 
Spazturtle said:
Intel's Iris GPUs are pretty good, as APUs are now looking like the future I think we will se Intel invest more and more into GPUs.
Click to expand...
I wouldn't say they're pretty good. 650M performance in a iGPU is impressive but is nowhere near a 7970, 7970GE, etc. They're also kind've expensive since the cheapest CPU's will the Iris Pro 5200 run over $400 MSRP and are still seem to be mostly for laptops only. Otherwise you're stuck buying ultra-compact micro-ITX systems like the BRIX with little to no upgrade options.

It would really suck if that was essentially the only other option to nVidia. Especially since both Intel and nV would start dragging their feet around and milk their customers even more by releasing products at a even slower pace and charging more money too.
 
xoleras said:
Oh yeah, someone buying a GPU from retail won't mind a 15% performance variance, not at all.
Click to expand...
Not when its easily and trivially "fixable" by bumping the fan cap 5% or so. No one sane or who has bought the card gives a crap about that. And many of the people who have bought the card haven't had an issue at all. You can act hysterical about that all you like but you just come off looking ridiculous.
 
xoleras said:
I mean that's completely status quo and fine per some AMD posters here, right? And the people calling BS are just..."detractors". Whatever you say.

Look, just tell AMD to fix their shit so that none of this circus bullshit is never brought up in the press; this is AMD's fault, they are doing a disservice to customers who are buying cards that aren't performing as they should. People are going to be detractors when a product has performance variance from one card to the next of up to 15%. That type of performance variance is bullshit. Don't blame the users for pointing it out, blame AMD for releasing their card in a semi-broken state - clearly the Hawaii chip is very powerful but AMD doesn't know what the hell to do with it. Again, AMD needs to fix their shit. And no, not everyone is going to run uber dust buster fan, and not everyone will know that "quiet mode" has severe performance variances.
Click to expand...

No, the problem is people like you who keep bashing AMD because they tea bagged Nvidia with their newest cards. It has been explained to you over and over again by owners of the new cards that you have been and still are basically 100% with your bullshit.
 
SonDa5 said:
Yes I do know.
Click to expand...
Linking the feature list to the software doesn't tell you anything about how it determines ASIC quality which as near as anyone can tell is pure guess work and arbitrary.

The only people who can actually tell you how "good" a given GPU is would be either AMD or TSMC.
 
mesyn191 said:
Linking the feature list to the software doesn't tell you anything about how it determines ASIC quality which as near as anyone can tell is pure guess work and arbitrary.

The only people who can actually tell you how "good" a given GPU is would be either AMD or TSMC.
Click to expand...


Do your research. AMD works with W1zard to make the software work. W1zard is legit.
 
SonDa5 said:
Do your research. AMD works with W1zard to make the software work. W1zard is legit.
Click to expand...

yes and no.. GPU-z simply reads a "register" value from the GPU, the "Value" associated to that varies, particualry with regard to the 290(x) the "Asic" quality means very little
 
SonDa5 said:
Do your research. AMD works with W1zard to make the software work. W1zard is legit.
Click to expand...
What Framebuffer said.

AMD and others work with him to help identify their hardware properly past that its all w1zard. So is he a smart guy who means well? Absolutely. Does that mean everything he does is correct? Nope.

All the foundries and design houses spend heaps of cash on chip validation to determine which chips are good or not for a reason: its really damn hard to do and requires lots of specialized equipment and know how. Some software that just looks at the chip's voltage use, clocks, and heat isn't going to cut it.
 
mesyn191 said:
What Framebuffer said.

AMD and others work with him to help identify their hardware properly past that its all w1zard. So is he a smart guy who means well? Absolutely. Does that mean everything he does is correct? Nope.

All the foundries and design houses spend heaps of cash on chip validation to determine which chips are good or not for a reason: its really damn hard to do and requires lots of specialized equipment and know how. Some software that just looks at the chip's voltage use, clocks, and heat isn't going to cut it.
Click to expand...


Once the electrical data required can be controlled based on electrical theory W1zard should be able to make the software work. It seems to have worked in 79xx series. Big difference with new 290 series is it is using voltage driver made by International Rectifier. I think 290 series has alot of potential for improved operation by voltage control but it takes time for W1zard to figure it out. I think he can do it and I believe the ASIC score to be very important. From more research it looks like more work is needed.


Looking forward to improved BIOS and betterviltage control.
 
darkstarcow said:
No, the problem is people like you who keep bashing AMD because they tea bagged Nvidia with their newest cards. It has been explained to you over and over again by owners of the new cards that you have been and still are basically 100% with your bullshit.
Click to expand...

the funny part is the R9 290 users are themselves happy running the card at 55% and enjoying the outstanding bang for buck. in fact there are users who are ok to the loud noise and overclock the R9 290 to 1100 mhz and run at 75% fan speed. its the people who don't have the card who are ranting forever. if you want quiet operation install a accelero extreme iii or wait for custom cooler cards like asus dc2 , his iceq x2 etc. once the custom cooler R9 290 cards are out and Mantle BF4 is out there is going to be a lot more crying and cribbing on the forums. :D
 
You must log in or register to reply here.
Back
Top