Spin Off: AT's Testing Methods & Uber Mode

[3DVagabond]

How about my standard as a consumer? You and I both know that the typical reader for GPU reviews look at FPS charts and nothing but. Seeing uber mode numbers without knowing the context behind it will lead folks to buy a GPU and not understanding that they lose 20-25% performance for using a quiet mode fan profile. This is nonsense for a product to have such a deficiency. NONSENSE. You guys need to stop defending AMD on this crap and demand a better product. This BIOS switch SHOULD NOT EXIST. It should have a shroud that doesn't necessitate a noise versus throttling compromise. Period.

If AMD wants the 290X to be benchmarked in uber mode by every review website, they should make 55% fan the default speed with no quiet mode option. OR, they should put a label on their cards warning consumers that they lose 20-25% performance by using any sort of quiet mode - that's what really annoys me about this shroud. Noise is part of it, but with the prior generation ATI Cards you could easily fix this with custom fan profiles. With the 290X? Sorry you're screwed, you want a quiet card - you lose 20% performance.

OK, you've posted this 20%, 25%, 30%, sp much now that you need to post sources. I say this because it's simply not true. Look at all of the reviews. The difference between quiet mode and uber mode is only a few percent.

I've already made a lengthy post with results from Hardware.fr and they say basically the same thing as TPU. Please stop posting this FUD about 20% and 30% throttling. It's as bad as the crap that was posted about hearing loss from the cooler.

 

[HurleyBird]

In terms of resolutions, I think a fair amount of it comes down to common sense. Ultra low end and integrated graphics shouldn't be tested higher than 1080p, and by the same token it probably doesn't make sense to test ultra high end any lower than 1440p -- yes, there are some 120Hz 1080p screens that could end up being paired with high end graphics cards, but these will be phased out very quickly the moment higher resolution 120Hz displays start appearing.

Unlike some other forum posters I don't have any issue with testing 4K resolution in every game. First, it only makes sense to test out an ultra high end card with ultra high end settings. Second, 4K is the defacto standard that the industry is moving to -- even if market penetration is currently low there's no telling how many consumers are going to upgrade to a 4K display somewhere down the line while using one of these cards. Lastly, it makes an excellent proxy for multi-monitor resolutions such as 5760x1200.

In a perfect world, I would rather have AT test multi-monitor resolutions instead (they are more common, and make a decent proxy for 4K in the meantime) and then switch to 4K after it picks up more steam. I understand that AT is already testing 4K and switching to multi-monitors could amount to a lot of work though, so I'm happy enough with what we have.

 

[3DVagabond]

The heatsink dissipates watts, not degrees Celsius. Don't confuse heat and temperature. In terms of heat output the 780 Ti and 290X are very close, with review sites split between which consumes more power. The fact that the 290X is running 12 degrees hotter means that the ASIC is losing more power to leakage and helps the AMD reference cooler to achieve a higher performance and lower noise rating than it would be able to if AMD kept the maximum temperature at 83 degrees like the 780 Ti.

So yes, given that the 780 Ti heatsink is much quieter despite the fact that power dissipation is comparable between both products (and is being handicapped with the 83 degree temperature target to boot) it's fair to say that Nvidia has the far, far superior reference cooler.



The 5870 had a different cooler with no vapor chamber. From 6970 onward the design has stayed fairly constant.

What? The higher temp is because the cooler isn't as efficient. All of the power any chip uses gets converted to heat. They aren't actually doing any physical work.

 

[HurleyBird]

What? The higher temp is because the cooler isn't as efficient. All of the power any chip uses gets converted to heat. They aren't actually doing any physical work.

The hotter you run an IC, the higher the power consumption. Leakage increases as temperature does. If that sounds confusing, have a look at this thread:

http://forums.anandtech.com/showthread.php?t=2200205

 

[Keysplayr]

What? The higher temp is because the cooler isn't as efficient. All of the power any chip uses gets converted to heat. They aren't actually doing any physical work.

These little gems just keep getting better and better. The chips don't actually do any work. The graphics just appear on screen by magic. These semiconductors are 0% efficient. We just run electricity through them because it's trendy.

 

[Z15CAM]

You shouldn't be affected at all.

And if it does I'll blame it all you GaiaHunter but I thank you for your encouraging response - LOL

Guess I can UpDate my Platform Info as follows:

------------------------------------

i7 2700k/ASUS P8Z68-V Pro Gen3/Corsair H110 AIO (running @ .996v/1600Mhz to 1.376v/4600Mhz 24/7 between 36 to 67C), 4 x's 4GB sticks of Samsung MV-3V4G3D-US DDR3 running at 1.34v/1866Mhz 9-9-9-24 1T with 4GB's assigned to a RAMDisk drive to handle Win7 64 Bit Sluff and negate writes to the SSD, Samsung 840 Pro 256 SSD, 2 x's WD5001AALS HDD's in Raid-0, 1 x's WD1002FAEX 1TB, ASUS DRW-24B1ST DVDRW, Hauppauge WinTV HVR-1250 PCIe, 2nd Intel NIC for Dual Networking, XFX 850W Pro Black Black Edition modular PSU, AMD Ref Gigabyte 1000Mhz R9 290X/Accelero Xtreme III, eVGA GTX 280 in reserve for Physic's-X support, Fractal Design ARC Midi R2 case, QX2510 Samsung PLS 2560x1440 res display at 120Hz.

 

[Keysplayr]

And if it does I'll blame it all you GaiaHunter but I thank you for your encouraging response - LOL

Are there any reviews with the Accelero Xtreme III cooler on the 290x?
I'd wait for them if possible.

 

[3DVagabond]

The hotter you run an IC, the higher the power consumption. Leakage increases as temperature does. If that sounds confusing, have a look at this thread:

http://forums.anandtech.com/showthread.php?t=2200205

The heat increases because the power usage increases. The energy has to go somewhere, so it's dissipated as heat.

 

[SiliconWars]

Blackened those graphs show an average drop of 13%, not 20%.

 

[3DVagabond]

These little gems just keep getting better and better. The chips don't actually do any work. The graphics just appear on screen by magic. These semiconductors are 0% efficient. We just run electricity through them because it's trendy.

Without the hyperbole, pretty much accurate.

 

[R0H1T]

Are there any reviews with the Accelero Xtreme III cooler on the 290x?
I'd wait for them if possible.

I think there are, though not as reviews on different sites, but I understand that a number of users on various forums have posted their positive experience & something like 30~35 degrees drop in temp while stress testing the GPU.

 

[Z15CAM]

Are there any reviews with the Accelero Xtreme III cooler on the 290x?

Tom's Hardware has one on a R9 290 and I'm impressed enough.

http://www.tomshardware.com/reviews/radeon-r9-290-review-benchmark,3659-19.html

 

[Keysplayr]

I think there are, though not as reviews on different sites, but I understand that a number of users on various forums have posted their positive experience & something like 30~35 degrees drop in temp while stress testing the GPU.

Links please. Thanks.

 

[Keysplayr]

Tom's Hardware has one on a R9 290 and I'm impressed enough.

http://www.tomshardware.com/reviews/radeon-r9-290-review-benchmark,3659-19.html

Looks very nice. I'd go for it.

 

[Keysplayr]

As for testing methods, I feel that UBER mode should have been included alongside Quiet mode marks. We are all able to understand the differences and should have been given all information.

 

[SiliconWars]

Links please. Thanks.

http://translate.google.com/transla...celero_xtreme_iii_meets_amd_r9_290/index6.php

Almost 40 degrees and 50 Watts lower.

 

[Z15CAM]

If we can just keep Hawaii running at 1+ GHz without creating a racket, it’d have another shot at the high-end crown.

 

[HurleyBird]

The heat increases because the power usage increases. The energy has to go somewhere, so it's dissipated as heat.

Yes, and as you run a transistor at higher temperatures it consumes more power (again, because the transistor becomes more leaky). It's what you call a positive feedback.

 

[R0H1T]

Links please. Thanks.

I'll have to scan through a bunch of'em but if you need evidence then just click any of the links referencing 290/x ~

www.google.co.in/search?q="Accelero+Xtreme+III"&client=opera&tbm=dsc&source=lnt&tbs=qdr:m&sa=X&ei=sZN_UrSaAoHEtQbCpYDIAQ&ved=0CBwQpwUoBA

 

[Keysplayr]

I'll have to scan through a bunch of'em but if you need evidence then just click any of the links referencing 290/x ~

www.google.co.in/search?q="Accelero+Xtreme+III"&client=opera&tbm=dsc&source=lnt&tbs=qdr:m&sa=X&ei=sZN_UrSaAoHEtQbCpYDIAQ&ved=0CBwQpwUoBA

Thanks, but the Toms article was quite enough for me. It looks really good.

 

[Keysplayr]

"If AMD could just keep Hawaii running at 1 GHz without creating a racket, it’d have another shot at the high-end crown."

You can probably do that easily with that Accelero Xtreme III cooler. And quietly to boot.

 

[BallaTheFeared]

Tom's Hardware has one on a R9 290 and I'm impressed enough.

http://www.tomshardware.com/reviews/radeon-r9-290-review-benchmark,3659-19.html

1,150MHz with a cooler that is arguably as good as any aftermarket cooler shipped?

That was all they could get with an overclock?

 

[Gikaseixas]

Stop him from what? Talking about something that you don't want discussed?

Stop him from derailing this thread that's what i meant. You're seeing things that don't exist dude, take a chill pill.
I'm all open for discussion on the 290 series cooler.

 

[blackened23]

OK, you've posted this 20%, 25%, 30%, sp much now that you need to post sources. I say this because it's simply not true. Look at all of the reviews. The difference between quiet mode and uber mode is only a few percent.

It's only a few percent on a cold run. What happens on a "warm" run? Well let's see here, the GTX cards throttle by 1 bin in a warmed up run while the 290 throttles by 200mhz+. See this chart? Dipping to 757mhz.

http://www.anandtech.com/show/7481/the-amd-radeon-r9-290-review/15

FUD? :whiste:

To get down to the 34%-38% fan speed range, the 290 has to shed an average of 22% of its performance, peaking under a few titles at 25%. To be sure this makes the card much quieter – though not as quiet as a GTX 780 – but it also sacrifices much of the 290’s performance advantage in the process. At this point we’ve essentially reduced it to a 280X.

http://images.anandtech.com/graphs/graph7481/59582.png

Essentially, the 290 series of cards is going to require new benchmarking philosophy in "quiet" mode because the performance lowers over time as the card gets hotter. And make no mistake, they do get hotter with that crap shroud that AMD uses. So benchmark sites that are doing a single run may get OKAY results from quiet mode, but more sites should be doing 15-20 minute "warmed up" runs with a 22-25% loss in performance for real world data.

I'll be more than happy to show you more evidence of 20% loss in performance from quiet mode from "warmed up" runs. There is a clear difference between "first benchmark run" and subsequent benchmark runs due to lowering performance over time. DRASTICALLY lower performance over time due to the faulty cooler on the 290 series of cards. Again, i'll be happy to provide that data if you'd like. This cooling situation is not acceptable because any quiet fan profile reduces performance over time - when you're using a high GPU load for 15+ minutes, the performance is okay for the first couple of minutes and then tanks from excessive throttling.

From a consumer point of view this is not acceptable. The 7970 series of cards could overclock ridiculously high and did not ever, ever require compromises for quiet custom fan profiles. The fact that 290(X) performance tanks as a "quiet" trade-off is a design decision that AMD should regret and fix with a "B" revision card. To be clear, this situation with throttling and uber vs silent mode could have COMPLETELY been prevented with better engineering by AMD. I wanted to like the 290 series of cards, but this throttling situation in quiet mode is simply not acceptable. Again, I loved my 5870, 6970 and 7970 cards (ALL REFERENCE BTW) - I could use those cards, overclock, *AND* maintain relatively quiet custom fan profiles. Sadly, I would not be able to do this with a 290 without losing a TON of performance.

 

[Keysplayr]

It's only a few percent on a cold run. What happens on a "warm" run? Well let's see here, the GTX cards throttle by 1 bin in a warmed up run while the 290 throttles by 200mhz+. See this chart? Dipping to 757mhz. Ouch.

http://www.anandtech.com/show/7481/the-amd-radeon-r9-290-review/15

FUD? :whiste:





Essentially, the 290 series of cards is going to require new benchmarking philosophy because the performance lowers over time as the card gets hotter. And make no mistake, they do get hotter with that crap shroud that AMD uses. So benchmark sites that are doing a single run may get OKAY results from quiet mode, but more sites should be doing 15-20 minute "warmed up" runs with a 22-25% loss in performance for real world data.

I'll be more than happy to show you more evidence of 20% loss in performance from quiet mode from "warmed up" runs. There is a clear difference between "first benchmark run" and subsequent benchmark runs due to lowering performance over time. DRASTICALLY lower performance over time due to the faulty cooler on the 290 series of cards. Again, i'll be happy to provide that data if you'd like.

WOW!!!