Spin Off: AT's Testing Methods & Uber Mode

[boxleitnerb]

Nowhere in there does it state there can be only one default.

Yes it does, the "without user intervention" part.

Again, explain SPD on RAM. Are you seriously arguing that only one of the 4+ timing configurations can be considered default settings?

Yes I am. The other comfigurations are presets that are not active by default. I already explained that.

ETA: Uber mode fits that definition. Your issue seems to be the configuration switch. So your position is that the EVGA Super Maxi Ultra 290XXX which ships with the switch on Uber mode would be "default" and quiet would not be tested?

Don't you see why this is a terrible review policy?

My position is that cards should be tested without any changes made to them, software or hardware.
If a card ships with a higher power/temp target or in uber mode, then it should be tested that way. This applies to OC models for example and that would constitute a new product of course.

So you either
a) test default vs default, i.e. quiet mode vs Nvidia standard settings
or
b) uber mode vs Nvidia raised fan speed or raised temperature target

Do you honestly believe a comparison of uber mode vs Nvidia default settings makes sense or is fair? Why should it be allowed to raise the thermal headroom on the AMD card by user intervention but not on the Nvidia card? It doesn't make any sense.

 

[TerryMathews]

Why would you be sad? Just go to TPU or a plethora of other sites that do these extra options..

AT caters to their audience, if their audience doesn't care about it, why should they spend extra resources to test settings which doesn't appeal to the readers?

Either way, as long as AT is upfront that they are benching the new series on the quiet mode for their results, why does it matter? Again, if you want to see the results on uber mode, go to another site...

So again, what if a card ships with the switch in Uber mode?

Or is the policy to only test in Quiet mode?

And to your final point, why are they testing overclocking Nvidia cards?

Look at my join date. I am their audience.

 

[cmdrdredd]

So again, what if a card ships with the switch in Uber mode?

Or is the policy to only test in Quiet mode?

And to your final point, why are they testing overclocking Nvidia cards?

Look at my join date. I am their audience.

Translation: you cheated AMD out of a couple points. Grap your pitchforks.

 

[TerryMathews]

Yes I am. The other comfigurations are presets that are not active by default. I already explained that.



My position is that cards should be tested without any changes made to them, software or hardware.
If a card ships with a higher power/temp target or in uber mode, then it should be tested that way.

So you either
a) test default vs default, i.e. quiet mode vs Nvidia standard settings
or
b) uber mode vs Nvidia raised fan speed or raised temperature target

Do you honestly believe a comparison of uber mode vs Nvidia default settings makes sense or is fair? Why should it be allowed to raise the thermal headroom on the AMD card by user intervention but not on the Nvidia card? It doesn't make any sense.

So which setting on a stick of RAM is default? XMP? JEDEC 1600?

I can't believe you would prefer to stick your head in the sand and let a manufacturer like EVGA sweep the chart here simply by putting the card's switch in the other position before they put it in the box.

To your final point, if Nvidia had some sort of toggle and warrantied both configurations I would have no issues testing both.

I read reviews for data. This isn't a less is more situation.

 

[Silverforce11]

So again, what if a card ships with the switch in Uber mode?

Or is the policy to only test in Quiet mode?

And to your final point, why are they testing overclocking Nvidia cards?

Look at my join date. I am their audience.

Well see, because AT is so objective, they have to factor in the retail version, because retail is default to Quiet Mode, AT would not want readers to be misled thinking they could obtain the extra performance of Uber Mode easily...

wait... but they can.. its just a switch flick.

Apparently PC enthusiasts are too stupid to flick a switch.. because thats the impression I got re-reading Ryan's statement.

Seriously, stop being stupid everyone. Uber Mode is covered under warranty. AMD specifically present it as an alternative mode, more performance but more noise. Don't be lazy and test both like every other good review site...

/stop cat fighting over this lame issue.

 

[TerryMathews]

Translation: you cheated AMD out of a couple points. Grap your pitchforks.

Glad to see your beat, as you had to resort to hyperbole.

I've owned far more Nvidia cards than AMD. I'll buy a good deal from either side of the aisle. This generation, for now, AMD had the better deal at the top.

 

[cmdrdredd]

Well see, because AT is so objective, they have to factor in the retail version, because retail is default to Quiet Mode, AT would not want readers to be misled thinking they could obtain the extra performance of Uber Mode easily...

wait... but they can.. its just a switch flick.

Apparently PC enthusiasts are too stupid to flick a switch.. because thats the impression I got re-reading Ryan's statement.

Seriously, stop being stupid everyone. Uber Mode is covered under warranty. AMD specifically present it as an alternative mode, more performance but more noise. Don't be lazy and test both like every other good review site...

/stop cat fighting over this lame issue.

Then can I show you the door and expect you to simply not come here anymore because of those "good review" sites?

I don't know how it can be any more clearly laid out for you guys to get it. I still call this a witch hunt and a few people feeling slighted because they feel as if Ryan cheated AMD out of a few points on benchmarks.

 

[TerryMathews]

Then can I show you the door and expect you to simply not come here anymore because of those "good review" sites?

I don't know how it can be any more clearly laid out for you guys to get it. I still call this a witch hunt and a few people feeling slighted because they feel as if Ryan cheated AMD out of a few points on benchmarks.

Its simple. We don't want AnandTech's reviewers deciding what configurations are worthy of testing and what aren't. Run both.

Its not a witch hunt over points, as they ran both. Its campaigning over future reviews where they might not.

Another good example I thought of. Some motherboards are shipped with Turbo mode off by default. So should AnandTech only be testing i7s without Turbo on these boards?

 

[Janooo]

Translation: you cheated AMD out of a couple points. Grap your pitchforks.

The review shows NV bias. It hurts AT's name.
People are not dumb. They will put less value to the reviews on this site.

This is as simple feed back as it gets. If AT is smart they will learn from it.

 

[ICDP]

Any device/program cannot have multiple default settings. Per definition, default is one (1) state.

So if Anandtech reviewed televisions they would all get crap reviews because they don't come with the channels programmed out of the box. Or maybe if they reviewed a gaming mouse they would refuse to use the DPI toggle button. I have a cross-platform steering wheel that has a switch that is set by default to PC mode, I have to manually switch it to PS3 or Xbox 360 mode for it to work on a console. I wonder what Anandtech would make of such a device.

I didn't see anyone complaining when HardOCP and other sites tested MSI Lightning GTX680s and 7970s on the LN2 setting. Strange that isn't it, this must only matter when flicking the massively complex switch makes the AMD card look better.

This boils down the fact that Anandtech are not in a position to declare what is and isn't worth testing on consumer PC products. R9 290X cards have a switch that drastically alters how it performs. Any decent review site should test both settings and let the consumer decide what mode is right for them. This goes the same for sites only testing Uber mode. I am not asking for extensive reviews of each setting, even do the quiet mode with a few quiet vs Uber mode comparisons so consumers get a more concise review.

 

[Silverforce11]

Then can I show you the door and expect you to simply not come here anymore because of those "good review" sites?

Given the rampant trolling due to the R290 series launch, no. I rather enjoy being here.

Go on.. complaint about how noisy and hot the R290 series is again, its refreshing.

 

[boxleitnerb]

So which setting on a stick of RAM is default? XMP? JEDEC 1600?

I can't believe you would prefer to stick your head in the sand and let a manufacturer like EVGA sweep the chart here simply by putting the card's switch in the other position before they put it in the box.

To your final point, if Nvidia had some sort of toggle and warrantied both configurations I would have no issues testing both.

I read reviews for data. This isn't a less is more situation.

I don't know, I don't install RAM on a daily basis. But what I know is that you can buy reference 290X all day long and always will the switch be set to quiet mode (unless some dude in a factory in Asia dozed off and made a mistake).

If an AIB selects uber mode, they likely will OC the card as well and charge more for it, making it a new SKU. Your argument is theoretical since it hasn't happened yet. I know of no Nvidia card where the power or temperature target was raised and the card was sold as reference model. Not a single one. They would always come with raised clocks and more memory, better coolers etc.

Why does it have to be a toggle? Slider, switch, nob, hardware, software...it doesn't matter in what form it comes. Warranty is unaffected by temp and power targets btw. In the end I have no problem with testing non-default settings - as long as the tester doesn't mix and match, choosing to change settings on one product but not the other.

I also want more data, but I don't want unfair comparisons.

 

[TerryMathews]

I don't know, I don't install RAM on a daily basis. But what I know is that you can buy reference 290X all day long and always will the switch be set to quiet mode (unless some dude in a factory in Asia dozed off and made a mistake).

If an AIB selects uber mode, they likely will OC the card as well and charge more for it, making it a new SKU. Your argument is theoretical since it hasn't happened yet. I know of no Nvidia card where the power or temperature target was raised and the card was sold as reference model. Not a single one. They would always come with raised clocks and more memory, better coolers etc.

Why does it have to be a toggle? Slider, switch, nob...it doesn't matter in what form it comes. Warranty is unaffected by temp and power targets btw. In the end I have no problem with testing non-default settings - as long as the tester doesn't mix and match, choosing to change settings on one product but not the other.

I also want more data, but I don't want unfair comparisons.

So to follow your hypothetical to it's logical conclusion, assuming that "dude in a factory in Asia" did in fact ship the card with the switch in Uber mode then it shouldn't be modified correct?

I truly don't understand why you can't see that is terrible policy. Test both, chart both, label them appropriately and explain the differences on a page of the review. Hell, even chart the DB differences. Explain how Uber mode is louder and hotter than Nvidia.

I'm not pro- either company. I'm pro-information.

 

[TerryMathews]

I also want more data, but I don't want unfair comparisons.

We are all railing so hard because of this attitude right here. "Uber mode is unfair."

 

[boxleitnerb]

So to follow your hypothetical to it's logical conclusion, assuming that "dude in a factory in Asia" did in fact ship the card with the switch in Uber mode then it shouldn't be modified correct?

I truly don't understand why you can't see that is terrible policy. Test both, chart both, label them appropriately and explain the differences on a page of the review. Hell, even chart the DB differences. Explain how Uber mode is louder and hotter than Nvidia.

I'm not pro- either company. I'm pro-information.

There is a difference if it is a deliberate choice or if it is a mistake. Obviously, the reviewer should be so competent as to notice this if possible. I'm not going to further argue hypothetical scenarios with you. The facts are there - quiet mode is the de facto and intended default factory setting for the reference 290X. If and when that changes, we can revisit this topic.

So where was your outcry when almost nobody touched Keplers temp/power sliders? I have no problem with having both! I have a problem with having both analyzed on product A but only one setting analyzed on product B. That should be sufficiently clear by now, shouldn't it?

So am I. But then be consequent and demand this method is used for all tested products. And criticize if it's not - like I do right now.

We are all railing so hard because of this attitude right here. "Uber mode is unfair."

Nonsense. The testing procedure is unfair if it pits AMDs uber mode vs Nvidias default settings. Quite simple.

 

[TerryMathews]

There is a difference if it is a deliberate choice or if it is a mistake. Obviously, the reviewer should be so competent as to notice this if possible.

So where was your outcry when almost nobody touched Keplers temp/power sliders? I have no problem with having both! I have a problem with having both analyzed on product A but only one setting analyzed on product B. That should be sufficiently clear by now, shouldn't it?

So am I. But then be consequent and demand this method for all tested products. And criticize if it's not - like I do right now.

But that's not the same thing. You can't RMA an Nvidia card if it doesn't run with the slider pushed up.

 

[Leadbox]

So where was your outcry when almost nobody touched Keplers temp/power sliders? I have no problem with having both! I have a problem with having both analyzed on product A but only one setting analyzed on product B. That should be sufficiently clear by now, shouldn't it?

So am I. But then be consequent and demand this method is used for all tested products. And criticize if it's not - like I do right now.

Out of curiosity , does "product B" have a fixed second setting or is it something that we now need to define?

 

[TerryMathews]

Nonsense. The testing procedure is unfair if it pits AMDs uber mode vs Nvidias default settings. Quite simple.

I think that is what I said.

 

[TerryMathews]

Out of curiosity , does "product B" have a fixed second setting or is it something that we now need to define?

You know as well as I do that it does not. They feel that they should be able to overclock the Nvidia side against Uber mode on the ATi side even though one is warrantied and one is not.

Its getting deep in here. I especially like how no one touched he hypocrisy of enabling Turbo mode tech on Intel platforms.

 

[cmdrdredd]

You know as well as I do that it does not. They feel that they should be able to overclock the Nvidia side against Uber mode on the ATi side even though one is warrantied and one is not.

Its getting deep in here. I especially like how no one touched he hypocrisy of enabling Turbo mode tech on Intel platforms.

On Intel CPUs turbo mode is how they work out of the box by default without you doing anything. You don't need to change anything for it to function like this. As I understand, amd CPUs are similar.

 

[TerryMathews]

On Intel CPUs turbo mode is how they work out of the box by default without you doing anything. You don't need to change anything for it to function like this. As I understand, amd CPUs are similar.

Wrong. I own a motherboard and a laptop where the default is Turbo off.

 

[krumme]

It doesn't have two default. The quiet mode is normal mode. It is the standard mode of operation.

That's like saying you should have tested every hd6950 with the switch flipped to unlock the shaders to 6970 which of course wasn't done in reviews because the standard configuration is not with the switch flipped. It is fine to show the difference in initial reviews of the card but it is not how it performs on first boot up.

Blame AMD for making it this way.

All the talk dont change the fact that no one. Absolutely no one. Run the 290x at quiet.
What should the result then be used for?
Nobody but Ryan and NV enthusiast want the number. Its useless because it 100% doesnt reflect behavior. Its only of academic importance. Not for gamers that will use the card.
Beats me why amd left the switch in that position but that is of no excuse to keep it there as no one will do it. Is that so hard to accept?
Such idiotic stubbornness and bias.

 

[blackened23]

Then can I show you the door and expect you to simply not come here anymore because of those "good review" sites?

I don't know how it can be any more clearly laid out for you guys to get it. I still call this a witch hunt and a few people feeling slighted because they feel as if Ryan cheated AMD out of a few points on benchmarks.

Am I crazy or what because I still see uber mode numbers in the benchmarks. In any case, I can see valid points on both sides - I don't think including uber mode benchmarks as being unfair nor do I consider it invalid to include them. If AMD is warrantying uber mode, it's completely up to the reviewer as to include them or not, it is their prerogative. There's no need to get uptight if one reviewer doesn't, because in the end - AMD should not have released the 290 series with that stupid blower of theirs. That is what it boils down to, AMD made a poor design decision by not investing more R+D into a better cooler and hence we have a situation now where reviews and users will bring it up as impacting the user experience.

Again, including uber mode is completely up to the reviewer. But let's look back - how could this situation have been prevented ENTIRELY? The 290X card uses barely more power than Titan in gaming loads, maybe 9-10 watts more at best - therefore it stands to reason that AMD could have made a better solution in terms of cooling performance and acoustic performance. AMD made the wrong decision to NOT do this and we are where we are - had AMD made the right design decision, we could get uber mode performance with quiet mode acoustics. Like I said, that is completely possible given that the Titan is far quieter while consuming only about 5-10W less in gaming loads.

So, I essentially look at this with AMD at fault. With a more versatile cooler, NONE OF THIS would be an issue. Quiet mode throttling? Wouldn't exist with a more versatile cooler. Uber mode noise? Wouldn't exist with a more versatile cooler. AMD is not the victim here. They made a bad design decision and now they have to face the repercussions. Period. I think the 290 and 290X are freaking outstanding performers, you won't find ME saying anything bad about that - but AMD could have prevented all of the complaints of NOISE AND THROTTLING ENTIRELY with a better and more versatile cooler. But they had to cheap out.

I cannot understand the reaction of some users to this. AMD is not a victim. AMD isn't being picked on. This is their fault and they must face the repercussions - how many of the noise and throttling complaints would exist had AMD put a better cooler on their cards? Oh yeah. ZERO. AMD could have provided uber mode performance with quiet mode acoustics and ZERO throttling with a more versatile cooling solution. But no. They went cheap and passed the buck to AIB makers. They thought we wouldn't care. I'm sorry, some people do care about user experience and look at more than just a price tag.

 

[chimaxi83]

snip

You should just add the 290X cooler spiel to your sig... It can save you valuable typing time

So again, since out of the box performance DOES NOT include raising frequency or voltage, why is AT testing that, but can't flip a switch?

Also, this is not even remotely comparable to unlocking 6950 shaders, because that required a BIOS flash and potentially brought out instability. I know, I've done it. Just because MSI released a PDF on how to do it, doesn't mean this was manufacturer supported or "normal" behavior.

 

[KingFatty]

AMD should not have released the 290 series with that stupid blower of theirs. That is what it boils down to, AMD made a poor design decision by not investing more R+D into a better cooler and hence we have a situation now where reviews and users will bring it up as impacting the user experience.

But aren't you inserting your own conclusion here by calling AMD's decision wrong?

To me, it's a trade-off, achieve lower price by using lower priced cooler, get noisier card at lower price.

The issue would be when a reviewer asserts his own personal bias to characterize a given trade-off as unilaterally incorrect, when many readers of that review would just disagree.

This is the crux in my opinion, and I think it's why so many people are put off by the Anandtech review.

For one, the review seems to take a subjective impression, but tells us what is right or wrong for us as though it's objective truth. Like telling me it knows better than I do, and if I disagree I'm just wrong. I dunno, I've replaced the cooler on all my video cards, so I don't think I'm wrong for being OK with a stock cooler that is noisy, especially if I'm planning to replace it and that brings down the price of the card?

For another point, I think given the actual data, reasonable people could come to different conclusions, and good reviews will allow for that reasonable interpretation. Reviews will show people what they can expect from a product. A really good review will let a person imagine they themselves are sitting there testing and using the product, a nice effect where you get into the head of the reviewer and think yes, I'd test that and I'd try that too, go on...

But here, the review seemed to be about explaining why this card was a loser in the competition with another card, how it's the wrong choice and lost the competition. But how does that help me imagine how the product is? If I wanted to buy a car, I don't care how the car loses when compared against a Ferrari, because I'm not going to be racing against Ferraris, I'm going to be driving the car for myself.

This video card review seemed to focus on condemning the card for losing a battle, and missed the opportunity to focus on the features of the card itself.

Anyway, another thing is that if you read a review and see something odd stick out, why isn't that explained. We see very odd situation where the Nvidia cards behave strangely in SLI.

Yet no explanation. We are left to guess. Was the poor SLI showing because the 780 Ti is a slow card? Doubtful, but who knows, it's not discussed. Was it because Nvidia makes poor drivers? Doubtful, but who knows, it's not discussed.

Why didn't the review talk about that, to explain the odd behavior? I can't help but take away the impression that things aren't discussed if they deviate from the pre-selected narrative.

Here, explaining that the 290 outperforms the 780 Ti in dual-card mode might force the review to acknowledge that either the 780 Ti is not as good of a card, or Nvidia doesn't write good software. But that didn't seem to fit the narrative that the review was to select a winner card, instead of getting into explaining things.