AMD: Volcanic Islands R1100/1200 (8***/9*** series) Speculation/ Rumour Thread

[RedVi]

The only thing abstracted in the new AMD control panel is the power limit setting. Everything else is completely transparent and available for tweaking, including maximum fan speed - which is kind of the point of these dynamic boost states to begin with, getting the most performance out of a capped noise level.

I guess directly linking performance to noise is a new way of looking at it. Noise is minimised by allowing the temperature threshold to be reached before ramping up the fan, and then performance is kept constant as long as the fan is still able to keep that temperature stable. This wouldn't have been possible with older hardware, the regulation hardware on chip couldn't keep up with changes that quickly so the fan would need to always spin a bit faster in a pre-emptive fashion in case power use changed quickly.

Nvidia don't need to change much in their control panel to match AMD here. It's merely a question of whether the hardware sensors and power regulation is up to the task of setting the target temp that high, and even then, whether they are able to delay ramping up the fan in the same way without it then having to spin even faster to make up for the delay. I guess they don't really have to match it for the GK110 based cards, having better coolers and a larger, partially de-activated die should certainly make staying up to 10c cooler not too hard of a task.

 

[Dave Baumann]

The only thing abstracted in the new AMD control panel is the power limit setting.

First time I sat down and gamed with 290X I ran with CCC open on a second screen and watched the activities; took a while to sink in and understand what is actually going on but when I did I came to the conclusion that the actual power reading is the missing point from the CCC readout. Having that visible would make things click with people a lot quicker.

 

[gkar1]

Wow so now that AMD has a similar temperature based boost some websites suddenly decide to change their testing procedures and heat up video cards before actually benchmarking them...

 

[entity279]

Yes, because now the websites weren't taken by surprise anymore. No need to read anything malevolent into it.

 

[xDxD]

OMG, why Amd always fails to do silent coolers? http://www.youtube.com/watch?v=QQhqOKKAq7o

 

[hoom]

Why is everyone making a big hoohah about 100% fan being loud?
100% fan has always been very loud but no cards have actually used that for years if ever.

Giving an easy option to mess with the max fan setting that will be used doesn't mean 'this card is designed for 100% fan' []

 

[Gipsel]

OMG, why Amd always fails to do silent coolers? http://www.youtube.com/watch?v=QQhqOKKAq7o

You can do the exact same video with each and every card with a direct heat exhaust cooler.

 

[lanek]

I dont understand what they do there, they compare the fans Db with the same % set.... 30% - 40% -55% ( why not 50% ..? )...etc... lol.

 

[CarstenS]

Was there a specific slide that gave a geometry front end + rasterizer + CU group + ROP and said that was a pixel pipe?

No, not explicitly.

 

[CarstenS]

You can do the exact same video with each and every card with a direct heat exhaust cooler.

Agreed, 100% alone is pretty pointless, except to show how much potential the cooler's got left for massive air-cooled OC. The Problem is, that 290X at 55% is also so loud, you'd probably want to wear headphones. In fact, everyone who has a reference 7970 can check for themselves, because the noise levels are pretty much identical.

 

[trinibwoy]

Does NV have actually voltage controls in their own control panel, in addition to temp & power controls?
AMD offers temp and power in the control panel + those 2 profiles (and manual max fan speed control too)

nVidia hasn't had overclocking or monitoring functionality in their control panel for several years now. Afterburner, PrecisionX etc have taken up that charge.

Both companies give the end user control over temperature, fan speed, voltage and power consumption targets to varying degrees. The difference seems to be in how far they allow you to push it.

 

[Psycho]

I dont understand what they do there, they compare the fans Db with the same % set.... 30% - 40% -55% ( why not 50% ..? )...etc... lol.

20% - idle speed
40% - max speed of quiet mode (ie where it will usually end)
55% - max speed of uber mode
100% - headroom

 

[Dave Baumann]

In fact, everyone who has a reference 7970 can check for themselves, because the noise levels are pretty much identical.

The 55% limit was taken from the 7970 GHz Ed. However, 7970 GHz is a fan table will oscillate around this point, while the fuzzy logic fan controller will be fairly steady around any set point. People usually find the oscillation more annoying.

 

[xDxD]

Why is everyone making a big hoohah about 100% fan being loud?
100% fan has always been very loud but no cards have actually used that for years if ever.

Giving an easy option to mess with the max fan setting that will be used doesn't mean 'this card is designed for 100% fan' []

You can do the exact same video with each and every card with a direct heat exhaust cooler.

Agreed, 100% alone is pretty pointless, except to show how much potential the cooler's got left for massive air-cooled OC. The Problem is, that 290X at 55% is also so loud, you'd probably want to wear headphones. In fact, everyone who has a reference 7970 can check for themselves, because the noise levels are pretty much identical.

This.

 

[jimbo75]

Wow so now that AMD has a similar temperature based boost some websites suddenly decide to change their testing procedures and heat up video cards before actually benchmarking them...

Yes, because now the websites weren't taken by surprise anymore. No need to read anything malevolent into it.

Boost has been around since the 680 well over a year and a half ago. To only now pre-heat cards and measure the drop-off in performance is a bad joke.

 

[trinibwoy]

I still don't get why they didn't just put a better cooler on it and ask $600. Fan noise has to be one of the most aggravating things to live with. Granted, I used to care much less in my younger days when 60mm deltas were all the rage

 

[trinibwoy]

Boost has been around since the 680 well over a year and a half ago. To only now pre-heat cards and measure the drop-off in performance is a bad joke.

Boost 1.0 doesn't look at temperature. That was introduced with Titan. Also nVidia's stock cooler is pretty good so even when you set target temperature to say 95 degrees it never actually gets that hot before hitting the tdp limit.

i.e warming up nVidia cards is not necessary.

 

[Trawler]

My bet is that AMD is keeping a 'just good enough' cooler on these cards to keep their hardware vendors happy. The hvs can differentiate and add value with custom coolers.

Looking forward to picking up a 290 with a decent cooler and I doubt I'll be the only one.

 

[DuckThor Evil]

Boost 1.0 doesn't look at temperature.

It does look at it a little bit. 680 drops one boost bin at 70c and another at 80c, still this equates to a drop of something like 26Mhz in total, so a pretty meaningless impact on performance. Titan is more restricted at 80c, but not at 94c.

 

[jimbo75]

Boost 1.0 doesn't look at temperature. That was introduced with Titan. Also nVidia's stock cooler is pretty good so even when you set target temperature to say 95 degrees it never actually gets that hot before hitting the tdp limit.

i.e warming up nVidia cards is not necessary.

Hardware.fr found differently on the reference coolers for the refresh.

http://translate.googleusercontent....s.html&usg=ALkJrhg3mcvTSGAXtv_IjnwqBhJy6VJVbA

http://www.hardware.fr/articles/896-22/recapitulatif-performances.html