Official AMD Radeon 6000 Series Discussion Thread

[CDdude55]

Holy crapsauce!

I was expecting the 6970 to push ahead with ease, but apparently not according to what has been said, and if all that's true, then that's fail.:shadedshu

Hopefully they're priced good and still bring something to the table though.

 

[figuretti]

Holy crapsauce!

I was expecting the 6970 to push ahead with ease, but apparently not according to what has been said, and if all that's true, then that's fail.:shadedshu

Hopefully they're priced good and still bring something to the table though.

well, if i understand correctly, the 6950 is about 30~40% faster than the 6870... thats about 20~30% faster than the 5870, if im correct

Tomorrow they will test the 6970... and the specs floating around put the 6950 aprox. 20% behind the 6970 (someone on XtremeSystems do the math, and he puts the 6950 is aprox. 1.32x 6870, and the 6970 aprox. 1.59x 6870)

 

[bear jesus]

well, if i understand correctly, the 6950 is about 30~40% faster than the 6870... thats about 20~30% faster than the 5870, if im correct

Tomorrow they will test the 6970... and the specs floating around put the 6950 aprox. 20% behind the 6970 (someone on XtremeSystems do the math, and he puts the 6950 is aprox. 1.32x 6870, and the 6970 aprox. 1.59x 6870)

If that's right i will be pretty happy, until Nvidia cards gain the ability to run 3 monitors with a single card/chip my only option is with AMD and 6870's in crossfire did not really suit my needs so I'm down to a single one of them and 50% faster would give me the kind of gains needed for newer games at 5040x1050.

 

[figuretti]

Some pr0n
http://www.hardwareluxx.de/community/15863940-post170.html

GPU-Z Screenshot
http://img196.imageshack.us/i/gpuzko.jpg/

 

[the54thvoid]

Rapid calculation = On Metro 2033 at 2560 res the 580 is 1.7 x better at fps than the 6870.

So, if the case is as you are quoting through the other site and the maths is correct and thi slooks like best case scenario given all the negative vibes coming out recently, then the 6970 would in fact be behind the 580.

Although it's only 1.4 x better at fps in Battlefield BC at 2560 res.

So, maybe, again if those numbers are correct, it's better and worse, game dependant?

 

[bear jesus]

Some pr0n
GPU-Z Screenshot
http://img196.imageshack.us/i/gpuzko.jpg/

Why would a 69xx card have a lower pixel fillrate than a 68xx card?

The link shows 28.2gp/s yet my 6870 has 28.8gp/s

 

[EastCoasthandle]

Why would a 69xx card have a lower pixel fillrate than a 68xx card?

The link shows 28.2gp/s yet my 6870 has 28.8gp/s

GPUz has not been updated for the 6900 series yet.

 

[bear jesus]

GPUz has not been updated for the 6900 series yet.

I would assume then some of the other information could be incorrect as well.

 

[HTC]

Holy crapsauce!

I was expecting the 6970 to push ahead with ease, but apparently not according to what has been said, and if all that's true, then that's fail.:shadedshu

Hopefully they're priced good and still bring something to the table though.

Not necessarily so: if the power usage is lower then the competing card (apparently, that would be the 570) and / or priced better, that will still make it the better deal, no? Granted: power usage in a high end card ain't much of an argument ...

Still, i too was expecting the 6970 to beat the 580 (barely) and am somewhat disappointed by these news, if true


In any case, it may actually be better for ATI to lose this round because it will only drive them harder in their quest to beat the competition, just as nVidia apparently did in this round.

If ATI passes nVidia by a bit, that's fine. If ATI trails nVidia by a bit, that's fine too. If either company gets a big gap against the other, that's BAD ... BAD, i tell you ...

 

[figuretti]

Rapid calculation = On Metro 2033 at 2560 res the 580 is 1.7 x better at fps than the 6870.

So, if the case is as you are quoting through the other site and the maths is correct and thi slooks like best case scenario given all the negative vibes coming out recently, then the 6970 would in fact be behind the 580.

Although it's only 1.4 x better at fps in Battlefield BC at 2560 res.

So, maybe, again if those numbers are correct, it's better and worse, game dependant?

yeah... i think is game dependant

the 3d mark score 11 is 4770 -> http://img340.imageshack.us/f/69703dmark11.jpg/
a GTX580 with i7 marks 5330 aprox
someone linked new drivers to test and the score is up... 4939 -> http://img262.imageshack.us/i/69703dmark11p.jpg/

edit: Heaven -> http://img834.imageshack.us/i/6970heavenbench.jpg/

nobody notice if the test has been done in "gaming mode" (190w TDP) of the new power tune, or in "maximun performance" (250w TDP)
i think because the drivers is beta, the VGA's runs in gaming mode... otherwise... :S

 

[cadaveca]

yi think because the drivers is beta, the VGA's runs in gaming mode... otherwise... :S

If these cards are VLIW4, then yes, driver affecting performance is going to be an issue of sorts for some time.

Add in the new profiling side of the driver, hardware-based power and frequency balancing, and a limited silicon process, these cards are nothing more than a working driver testbed for the next generation.

The fact that the lower-performing cards, which might have actual usable benefits running in tri and quad configs, are not functionally capable of such, and that the high-performing cards, which are far less likely to need such configs, do have the functionality, anyone should have a very nice picture of what these cards are, and what to expect.

Why do the top-teir cards have dual crossfire fingers? Why can't we run 6870 in quadfire?

 

[erocker]

Hmm.. I think it's time I pick up a GTX 580.

 

[cadaveca]

Hmm.. I think it's time I pick up a GTX 580.

I think I might get 2.

Even so, I only see 6870 Crossfire as a viable purchase, but performance isn't there. AMD drivers should be best optimized for that config.

Moving from VLIW5 to VLIW4 is gonna hurt.

 

[erocker]

Just bought me an evga GTX 580 SC. I should of done it a while ago. Hey, if I don't like it, I'll sell it and pick up something else.

 

[bear jesus]

Just bought me an evga GTX 580 SC. I should of done it a while ago. Hey, if I don't like it, I'll sell it and pick up something else.

Dam you, yes that's right dam you

Yes i mean dam not damn

I would love to give nvidia some of my money but i don't want to have to shell out for something like a matrox triple head just to be able to carry on with 3 screens.

 

[KainXS]

Why would a 69xx card have a lower pixel fillrate than a 68xx card?

The link shows 28.2gp/s yet my 6870 has 28.8gp/s

its right

pixel fill rate on modern cards is ropsXcore so the fill rate for the 6970 is 28160 and for the 6870 its 28800
but so a good bit has changed between the 2 cores so you can't compare the two'

so that switch is on the cards, i wonder what it does

 

[figuretti]

the test with F1 2010 gave him more FPS than a 580

about the powertune, the gaming mode or the maximum performance mode, the guy responds

Get the CCC only installed, nothing changed ... setting is on "quality"

(by google translate)

Neliz@B3D:

You need something better than the latest publicly released 10.11 beta's.

The problem here is this... is the GPU Working in Gaming Mode (190w TDP with clocks & voltage automatically regultated to fit the TDP assigned)? or is working in Maximum Performance Mode (250w TDP, everything max)

 

[figuretti]

so that switch is on the cards, i wonder what it does

dual bios change switch, like the MSI GTX480 Lightning (tip by Neliz@B3D)

 

[figuretti]

Crysis: http://www.abload.de/image.php?img=6970-crysis4mny.jpg#

 

[Atom_Anti]

3DMark11;
GTX480 versus HD6970

 

[pantherx12]

3DMark11;
GTX480 versus HD6970

I'm fairly sure that's a 6950, I'm fairly certain the 6970 will be 2gb on referance models.

 

[bear jesus]

3DMark11;
GTX480 versus HD6970

I find this a little odd
"PhysicsTest 580 15.4FPS 69xx 19.38 FPS"

I thought the physics test was on the cpu? and both results are listed with the same cpu at the same speed although the 69xx result only has half the ram for some reason.

 

[erocker]

I find this a little odd
"PhysicsTest 580 15.4FPS 69xx 19.38 FPS"

I thought the physics test was on the cpu?

Could be latency is playing a factor. The 480 system is using 8gb of RAM while the 69XX system is using 4gb.

 

[bear jesus]

Could be latency is playing a factor. The 480 system is using 8gb of RAM while the 69XX system is using 4gb.

If that was the cause then i would be very unimpressed with the 8gb kit.

 

[erocker]

If that was the cause then i would be very unimpressed with the 8gb kit.

Add more ram it adds latency. I don't know why it would be different considering all the other specs are the same.