Dieselgate for GPUs: review-units ship at higher clockspeeds than retail ones


#1

[Read the post]


#2

Clickbait headline for Cory: hed draws facile comparison between two completely different controversies


#3

Somehow i’m unsurprised by this.

What i’d like to know though, is whose idea was this? Nvidia has form for such blatantly anti-consumer practices, but i’m more interested on the manufacturers here…


#4

On a vaguely related matter, please please drop the -gate addition for any issue.

It reached the comically overused state a good 5 years ago at minimum… It’s almost painful to see it reused on EVERYTHING recently…


#5

Seems pretty much a non-issue to me, and a really unfair comparison. Whereas “diesel-gate” was a blatant effort to deceive government dictated emissions tests illegally and (nearly) undetectably, this is just a matter of companies sending their product to reviewers with the better performing mode–available to everyone–toggled to on. No one’s accusing them of tweaking these cards or doing custom over clocks. Just the “turbo” mode button is pressed, in plain sight to the reviewer.

Sure, it should be mentioned in the review–as any good bench marking review would do when listing the conditions of their tests —but I kind of expect benchmarks to show best case available “off the shelf” settings anyway.


#6

Thank you for another excellent and insightful post about the suspicious (at best) consumer practices resulting from intersecting proprietary code, corporate decision-makers and, apparently, one too many tech folks who either don’t understand or don’t care about protecting the public interest.

I’m so glad that EFF works on these issues. They could align more effectively with community groups and other attorneys doing social justice work.

There are days when EFF seems uninterested in the historical importance of aligning with community-based stakeholders (i.e. non-corporate).

The Legal Services Corporation has struggled similarly to favor a community lawyering over an impact litigation practice — or just survive.

OTOH, it’s easy to be a Monday morning quarterback so I’ll move along now. Go, EFF! :smiley_cat:


#7

I’m kind of confused why this is new.

Doesn’t every commercial GPU driver already detect what software is running and then malevolently disable features and change the graphics to pretend to get a few extra scraps of speed?

I was under the impression that this was so widespread that “gamers” would actually whine loudly if the manufacturers display the slightest bit of integrity and make the graphics card actually work correctly.

Like, video game people would get into huge flame wars on the internet because one card is slower at 4x anti aliasing, even though the other card’s driver actually just detected that the game used 4xaa and turned it off.

Never mind that this is transparently evil, impossible to program for, and straight up fraud of course…


#8

There is a less sinister explanation for this:

Notice the different BIOS versions on the review vs. retail cards. Review cards often have a pre-release BIOS so they can get to reviewers quicker. Maybe the older BIOS was more aggressive about clock speeds, but it was turned down a bit for release to avoid overheating?


#9

If Watergate happened now, it would almost certainly be referred to as ‘Watergategate’.


#10

It’s wholly possible that this is a non-story but it sure sounds good in creating outrage.

The review sample has a different (older) BIOS from the retail sample. This could mean many things: different tuning parameters, OC mode may be enabled by default, or something else that would be different in a later revision. Not to create better looking benchmarks, but because that’s what happens when you’re developing new hardware and drivers. It’s not like they juiced the review units with more cores or faster RAM. Doing the math, the difference is a paltry <1% in speeds.

Also understand this is not an integrated graphics card sitting in your grandma’s Gateway 2000. This is a $700 card that will be used only by serious gamers and PC enthusiasts. The kind of people that know their way around overclocking and PC tuning.


#11

My big issue: the reviewers weren’t checking in the first place?
They were relying on review samples only (provided by the manufacturer with knowledge they were for review)?

That’s some lazy work.


#12

That’s kinda catchy.


#13

I’m inclined to agree. Manufacturers list the factory clock speed of cards so surely reviewers should be checking how their sample compares, no?


#14

It looks like you’re trying to write a review of this computer. Can I help you with that?


#15

This topic was automatically closed after 5 days. New replies are no longer allowed.