AMD: Southern Islands (7*** series) Speculation/ Rumour Thread

[imaxx]

at contrario

I'm not english-speaking, yet it sounded clear to me. Just in case, "al contrario" translates better as "on the other hand"

 

[Kaotik]

TPU does change drivers quite rarely, sadly, but considering they actually do bench all of the last 2 gen cards when they do, it's quite understandable that they don't jump to every new driver set

 

[UniversalTruth]

TPU does change drivers quite rarely, sadly, but considering they actually do bench all of the last 2 gen cards when they do, it's quite understandable that they don't jump to every new driver set

I think it would be far more fair if they review every single new card with the newest drivers available at that moment. They don't have to retest every single of the old ones with the new driver set. This excuse sounds ridiculous.

So, please TPU! I hope you are reading and can make reasonable decisions.

 

[Kaotik]

I think it would be far more fair if they review every single new card with the newest drivers available at that moment. They don't have to retest every single of the old ones with the new driver set. This excuse sounds ridiculous.

So, please TPU! I hope you are reading and can make reasonable decisions.

They do on new models, like 7970 GHz Edition, it has 12.7 betas like it should, but when it's just AIB customized model of something older, they don't

 

[jimbo75]

http://www.xbitlabs.com/articles/graphics/display/radeon-hd-7970-ghz-edition_9.html#sect0

It wouldn’t be correct to claim that the AMD Radeon HD 7970 GHz Edition is the fastest single-GPU graphics card available today because our tests do not show it to be overwhelmingly better than the reference GeForce GTX 680

Looks like being only 26% faster isn't good enough any more. Maybe Lost Planet 2 carries special weight that negates the 60-70% "non overwhelming" results from other games?

There has to be some reason right?

 

[kalelovil]

Its Xbitlabs, they have a history doing stuff like that.

 

[lanek]

Well Sniper Elite and Showdown are yet really faster on AMD cards, but if we look results with AA, im even suprised to see the AMD one so clearly in front. I ask me if AMD had release a GHZ edition just before the 680 release, how will have look the review of the GHZ. ( i dont want say for the 680, as some mhz more on the turbo clock should have have do the job )

 

[Kaotik]

On the talk about TPU couple posts back, they've updated their drivers now, 12.7 betas for AMD and 304.79 beta

 

[neliz]

Its Xbitlabs, they have a history doing stuff like that.

We literally have to face this a lot with xbit.

Take an XL-ATX mainboard. we get minus points in a review because the size doesn't fit a lot of cases (honest truth)
Our competitors with an XL-ATX mainboard get positive points because they have "spacious design"

some sites just need a big "BIAS" slider at the bottom so you can get a nice realistic review.

 

[gkar1]

http://www.techpowerup.com/170436/AMD-Updates-Radeon-HD-7950-to-Thwart-GeForce-GTX-660-Ti.html

AMD updating 7950 specs, clocks bumped to 850 base and 925Mhz boost. The BIOS is also available for download to anyone that wants to use it.
Pretty creative leveraging of their dual BIOS feature to counter the competition (it's almost as if they had planned this all along )

 

[Psycho]

http://www.xbitlabs.com/articles/graphics/display/radeon-hd-7970-ghz-edition_9.html#sect0

Also, why is the 7970GE labeled as 1050 mhz, while the mostly ~1100 mhz 680 is labeled 1006mhz?

 

[jimbo75]

http://www.techpowerup.com/170436/AMD-Updates-Radeon-HD-7950-to-Thwart-GeForce-GTX-660-Ti.html

AMD updating 7950 specs, clocks bumped to 850 base and 925Mhz boost. The BIOS is also available for download to anyone that wants to use it.
Pretty creative leveraging of their dual BIOS feature to counter the competition (it's almost as if they had planned this all along )

Yep that's pretty interesting.

I wonder how many of the tech press have swallowed this one in time for the 660 Ti reviews. This move looks like AMD's safety net and I can't even see the 660 Ti beating it at Anandtech/Hardware Canucks now.

edit - http://www.hardwarecanucks.com/news...cts-existing-customers-to-look-the-other-way/

Well no suprises that SKYMTL opted out, I wonder if Anand/Smith will use the same excuse or have their own.

 

[gkar1]

Wow, that "article" bends over backwards trying to turn a free upgrade for all 7950 owners into something negative. Pretty pathetic.

 

[jimbo75]

Looks like AT might go with it - at least Smith is benchmarking the bios update now -

http://forums.anandtech.com/showpost.php?p=33827549&postcount=46

We'll see if it is included in the Ti's review however. Also note that Shrout did a mini review yesterday at Pcper - http://www.pcper.com/reviews/Graphics-Cards/AMD-Radeon-HD-7950-3GB-PowerTune-Boost-Review

His results seem to indicate a much higher performance gain compared to Smiths at Anandtech. I don't really get how the AT ones are so low actually, with 6.25% gains being the low mark and 15% being the expected high mark...AT being below the low point of what would be expected by just the flat 50 MHz increase looks a bit off. Shrout's at PCper is 8%.

 

[Ryan Smith]

His results seem to indicate a much higher performance gain compared to Smiths at Anandtech. I don't really get how the AT ones are so low actually, with 6.25% gains being the low mark and 15% being the expected high mark...AT being below the low point of what would be expected by just the flat 50 MHz increase looks a bit off. Shrout's at PCper is 8%.

http://www.anandtech.com/show/6152/amd-announces-new-radeon-hd-7950-with-boost

You're assuming we get perfect scaling with GPU clockspeed increases. That has never been the case, especially since some of our tests are more sensitive to memory bandwidth than they are GPU clockspeed. 6% is rather consistent with what we've seen from manually overclocking the 7950 in the past. Otherwise the 7950B never gets to boost for an extended period of time (every time it reads as 850MHz, it's at or below 850MHz), so the boost clock is almost worthless here.

 

[jimbo75]

Well I actually compared your results to PCPer's and the same games you tested, you had an average of 4.6% compared to their 8.4%.

PCPER BF3 - 7% increase
AT BF 3 - 7% increase

PCPER Skyrim - 5% increase
AT Skyrim - 0% increase

PCPER Dirt 3 - 17% increase
AT Dirt 3 - 6% increase

PCPER Batman - 12% increase
AT Batman - 5% increase

PCPER Metro 2033 - 1% increase
AT Metro 2033 - 5% increase


Granted you are benching at 1920x1200 instead of 1080p, so that might account for a couple of percent. Your results seem to indicate that you are barely getting any boost at all most of the time though, but who knows that could just be down to card and conditions variability.

 

[Alexko]

Well I actually compared your results to PCPer's and the same games you tested, you had an average of 4.6% compared to their 8.4%.

PCPER BF3 - 7% increase
AT BF 3 - 7% increase

PCPER Skyrim - 5% increase
AT Skyrim - 0% increase

PCPER Dirt 3 - 17% increase
AT Dirt 3 - 6% increase

PCPER Batman - 12% increase
AT Batman - 5% increase

PCPER Metro 2033 - 1% increase
AT Metro 2033 - 5% increase


Granted you are benching at 1920x1200 instead of 1080p, so that might account for a couple of percent. Your results seem to indicate that you are barely getting any boost at all most of the time though, but who knows that could just be down to card and conditions variability.

There shouldn't be any variability from card to card, because PowerTune is deterministic.

The benchmarks themselves could be different, though.

 

[jimbo75]

Techpowerup got an average of 6.3% at 1920x1200 - http://www.techpowerup.com/reviews/AMD/HD_7950_Boost_BIOS_Upgrade/25.html

 

[Kaotik]

Techpowerup got an average of 6.3% at 1920x1200 - http://www.techpowerup.com/reviews/AMD/HD_7950_Boost_BIOS_Upgrade/25.html

6.4%
Or at least last time I checked .38 is rounded to .4 rather than .3

 

[A1xLLcqAgt0qc2RyMz0y]

Wow, that "article" bends over backwards trying to turn a free upgrade for all 7950 owners into something negative. Pretty pathetic.

ALL 7950 Owners!! The article clearly shows that is not the case.

While AMD’s public affairs corps want this to be seen as being inline with industry standards, the optics of AMD’s play read as a bait and switch as, reportedly, these clock speeds aren’t even available on all products (let alone the ones already in the channel) and AMD itself doesn’t guarantee that it will work. AMD is cautioning users to do the update only if the card has a dual BIOS switch.

Hardware Canucks has reached out to five of AMD’s major North American board partners, which have all stated that any customer flashing this BIOS to their HD 7950 will immediately void their warranty should the BIOS switch be incorrectly set during the update. As such, AMD and by extension their board partners are recommending that existing customers not use this update. As per AMD, new versions of the 7950 should be hitting shelves with this BIOS update — and thus a putative speed increase of approximately 15% — as of this week pending the support from AIBs. The caveat? These new cards will have no new rebranding. Customer confusion is bound to ensure.