Slashdot Mirror
MSI and ASUS Accused of Sending Reviewers Overpowered Graphics Cards (theverge.com)
An anonymous reader writes from a report via The Verge: TechPowerUp discovered that the MSI GeForce GTX 1080 Gaming X card they were sent for review was running at faster GPU and memory clock speeds than the retail version. This was because the review card was set to operate in the OC (overclocking) mode out of the box, whereas the retail card runs in the more regular Gaming mode out of the box. This may result in an unobservant reviewer accidentally misrepresenting the OC performance numbers as the stock results from the card, lending MSI's product an unearned helping hand. The site found this was a recurring pattern with MSI stretching back for years. Fellow Taiwanese manufacturer ASUS, in spite of having better global name recognition and reputation, has also show itself guilty of preprogramming review cards with an extra overclocking boost. Needless to say, the only goal of such actions is to deceive -- both the consumer and the reviewer -- though perhaps some companies have felt compelled to follow suit after the trend was identified among competitors. The Verge notes that TechPowerUp revealed its finding on Thursday of last week, and has not received any official response from either MSI or ASUS. They did update their story to note that MSI addressed the matter, in a comment provided to HardOCP Editor-in-Chief Kyle Bennett, back in 2014.
I've seen this endlessly excused on sites like reddit by people claiming that it just saves the reviewers time since we'd all go into the drivers and overclock them anyways and other variations of that. I don't buy that one bit personally.
On the Oregon Cost born and raised, On the beach is where I spent most of my days
A good reviewer should buy things anonymously for their review, lest they tempt the producer to send them a higher quality product. Of course, it's harder to buy things anonymously when you want the producer to gift it to you. This has been obvious for a long time.
And thou shalt take no gift: for the gift blindeth the wise, and perverteth the words of the righteous.
Don't waste your vote! Vote for whoever you want, unless you live in a swing state it won't matter anyways
It's not "sites like reddit" claiming it.
It started with techpowerup article, mentioning exactly what was wrong: cards running at higher frequencies than normal retail versions: Here is it:
MSI and ASUS have been sending us review samples for their graphics cards with higher clock speeds out of the box, than what consumers get out of the box. The cards TechPowerUp has been receiving run at a higher software-defined clock speed profile than what consumers get out of the box. Consumers have access to the higher clock speed profile, too, but only if they install a custom app by the companies, and enable that profile. This, we feel, is not 100% representative of retail cards, and is questionable tactics by the two companies. This BIOS tweaking could also open the door to more elaborate changes like a quieter fan profile or different power management.
http://www.techpowerup.com/for...
These two respectful companies have no need to cheat, their records stand for themselves.
Time is what keeps everything from happening all at once.
just wait for the DMCA to shut down bad reviews
Wait... why do these cards have a "good mode" and a "shitty mode" in the first place, and why is the shitty one called "gaming mode"?
I'm going to guess that for power and thermal reasons that die yield of the chips support a performance envelope range, where one cards "good mode" can never be relied upon to be the same as another cards "good mode", but based on minimum burn in acceptance criteria, all cards are capable of operating in "shitty mode".
And further: that at the bottom of the barrel, "good mode" asymptotically approaches "shitty mode".
The fix would therefore seem to be: disallow overclocking.
(now listen to the gamers scream louder than they were screaming about "overclocked by default" in the original article...)
What kind of money do you think those review sites make per review? It is going to be a real problem staying in business if every review is predicated on being able to purchase one (or more) of a piece of high end hardware that can cost in the real of $700 in this particular case. It might be a nice idea to think they'd do it out of the goodness of their hearts and just spend their own money to help people but that isn't how things work. They need to get paid and they all have to cope with the rise of adblockers dropping revenue. So ya, review samples are pretty important.
Also in some cases samples are sent prior to public launch, so reviewers have to to get their review ready for when the public can purchase the hardware, so they don't need to sit around waiting for a review or go in blind. These things take time to do right, getting them hardware early is the only way to make that happen.
If you have a solution where people can still get the hardware early enough time to meet release day reviews and can do it cheaply enough to be able to pay their people to work on reviews for a living, I'd love to hear it. However the reality is I don't think there's a way.
Dunno if you've ever played with overclocking but the relation between speed increases past a chip['s normal limits and heat increases are not linear. So if you don't need/want the extra power, it is not a good idea to run your card harder than needed, particularly since those fans can get noisy when they spin up.
Support and stability are another reason. If a company rates a chip to a given level, they'll support/replace it there. Past that, maybe there's problems, maybe there's failures.
There can be a difference between what you normally design for and what something can actually do. Like my processor is spec'd to 140 watts TDP. What that means is Intel certifies that at standard operating frequency and voltage it will never dissipate more than that amount of heat, so if my thermal solution can handle that, I'm good. However the chip itself can survive more power, if there's a good enough solution. I can push it past that limit, if I want. If I do though, it is on me with regards to cooling and stability. If the chip messes up, they won't replace it.
"though perhaps some companies have felt compelled to follow suit after the trend was identified among competitors."
Strange none of the competitors may have felt compelled to make the whole thing public after seeing the first competitor trying to deceive the market.
Sounds a bit like, it would be acceptable for anyone to steal/rape/kill/..., provided they saw someone else do it before them.
It's standard practice that hardware reviewers are sent 'golden samples', or chips determined to be the most stable and functional. They tend to be the most overclockable, thus leading to exaggerations of how overclockable the chips are in general. Can't find it now but I recall a site having an image macro of Fry with the caption "Not sure if golden sample or super-fast video card".
A few years ago, AMD was found to be sending out video cards (the 290x IIRC) that ran hot and were heavily affected by thermal throttling, yet sent cards with modified firmware to hardware reviewers. This modified firmware caused the fans to spin at a faster RPM than consumer versions of the card, which made the cards cooler (and louder) and thus throttle less due to overheating.
Any reviewer worth their salt will know the stock and boost clocks for the part they're testing, and use software to check current clock speed. Pre-overclocked hardware is only a problem for lazy reviewers who didn't check.
Corruption is convincing someone that the selfless ideal is the same as their selfish ideal.
Although most TVs in stores are very badly set up, it's almost universal now that they have a "shop mode" that makes the colours more vivid and the picture brighter. This would simply seem to be an extension of that practice.
politicians are like babies' nappies: they should both be changed regularly and for the same reasons
I remember the days of yesteryear, when Asus put wallhack in their drivers, not once, but twice. Not that those are the previous incidents Asus has been involved with, but those are the ones that made me never forgive Asus.
And MSI has been making crap from day one.
The review of a laptop will say something like "HP Mobile 2300" but there'll be about 200 variations encompassing cpu/gpu/ram/display tech/hard drive size-speed-architecture/northbridge/etc/etc/etc. You'll never find the one which was reviewed for sale in your country and that's assuming the review actually mentioned the various product id which would enable you to even look for it. So yeah, great, thanks for the 3dbench scores and "this was faster than that pc" but I have no idea what you just reviewed.
Exactly my thought. Lazy people at MSI/Asus not resetting/reflashing the card between reviewers sounds like a more likely explanation.
They were driving a diesel Volkswagen to the job so it seems to be par for the course....
CR should start reviewing video cards (in case you don't know, they anonymously retail-buy everything they test).
It would also probably drop their demographic age by about 10 years in one fell swoop.
Yes, a lot of the auto "journalists" get specially modified models of the cars they review.
How is this different than when car companies provide review vehicles with all the bells and whistles pre-installed? Most will purchase a mid range trim level vehicle so the review may not adequately reflect what the average car buyer will experience.
"A person is smart. People are dumb, panicky dangerous animals and you know it." - K
So MSI (not that one) sent overclocked GPUs to reviewers whose job is to publish performance ratings. That's a lot like showing up with an RPG to a shooting competition and asking the judges, "Is this okay?". At what point did they think this would be an 'advantage'? It's not like getting the best dish in a restaurant because you're the critic, there is no incentive here. This must be an advertising campaign.
unless there was a woman giving a blowjob in exchange for a positive review, in which case make a fuss about it. everyone should have to pay for positive reviews with cold, hard, equitable cash. this is america.
"They were pure niggers." – Noam Chomsky
The clocks vary by 1-2% between the review and retail products. I can think of far more important issues to froth over on the internet.
The retail products even have an option to kick up to that performance level. Granted, it's not the default, but it's supported and very easy to change.
Manufacturers make far more outlandish claims by carefully selecting benchmarks, and no one even pays attention to that anymore.
Is it bullshit on some fundamental level? Yeah, maybe it's dishonest, and I'm not going to give corporations the benefit of the doubt when it comes to their motives. But it's barely a blip on the bullshit meter.
If the user didn't have the ability to adjust the clocks, then I could understand getting angry. At a 1-2% clock difference, it would still be a minor point. We can adjust the clocks though, so we get the same level of performance at home.
I'm in the market for a video card during this product cycle (waiting for Vega details), and this just doesn't matter enough to affect my decision at all.
---
According to the latest ruleset, this post should be modded as Vorpal Flamebait +5.
How about instead of sending cards, the manufacturers send rebate forms (worth 100% of the total price) to the professional reviewers. Assumedly the reviewers are getting some revenue, so they'd only have to cover the gap between card purchase and rebate submission.
That or have a buy-back program for cards used in reviews where the cards can be returned.