AMD Radeon HD 6970 and 6950 Video Card Review @ [H]

ihira said:
I'm not being skeptical but I'm sure Im not the only one wondering why [H]s results differ so much from other sites.
Click to expand...

Likely because they actually played the games, rather than posted some bullshit 3D mark scores.
 
ihira said:
I'm not being skeptical but I'm sure Im not the only one wondering why [H]s results differ so much from other sites.
Click to expand...

It's probably because all the [H] comparisons were done at 2560x1600, which is where the HD 6900's excel when compared to NVIDIA's offerings.
 
MACMAC said:
It's probably because all the [H] comparisons were done at 2560x1600, which is where the HD 6900's excel when compared to NVIDIA's offerings.
Click to expand...

A good point as well... it is only ~12% slower across all games at 2560x1600 per some other reviews, inconsistencies included in the averages (i.e. high-performing games and low-performing included). This card suffers at 1920x1200/below :(.
 
MACMAC said:
It's probably because all the [H] comparisons were done at 2560x1600, which is where the HD 6900's excel when compared to NVIDIA's offerings.
Click to expand...

I think it would be safe to say that using eyefinity this thing would really shine. We have no reviews showing that side of the story yet, but it sure would be interesting to see compared to HD6850 crossfire.
 
GoldenTiger said:
Partially their review methodology... and probably partially driver versions too. The games HardOCP used happen to be ones other reviews show doing well for the 6970 too, but even in the [H] review BFBC2 minimum fps is horrible compared to the GTX 580, for example. I'd say they're actually fairly in-line with other reviews, all things considered.... the 6970 looks to be a very inconsistent card across various games, some it does near-580 in, while the rest it bounces between 5870 performance and GTX 570... :confused:.
Click to expand...


I'm sure that some of the inconsistencies will be hammered out over the next few weeks. They just all won't exist in the same hotfix.
 
96redformula said:
I think it would be safe to say that using eyefinity this thing would really shine. We have no reviews showing that side of the story yet, but it sure would be interesting to see compared to HD6850 crossfire.
Click to expand...

Eyefinity and 30" with AA cranked is the sweet spot. I think people forget these are 2 GB cards as well, not the 1gb offerings of old.
 
Well Hard review says 6970 is great for price, BUT for me in CA, cheapest I can get the 6970 today is $415. I can get the GTX570 for $350.

What's the better choice? Playing at 25x16 btw.
 
So for a guy going from a single GTX 480 SC to perhaps Dual GTX 580's, or Dual 6970's.. i know which is the obvious better buy.. but which is the better performer all around?

ninja edit - i play at 1920x1200.
 
BeavermanA said:
Well Hard review says 6970 is great for price, BUT for me in CA, cheapest I can get the 6970 today is $415. I can get the GTX570 for $350.

What's the better choice? Playing at 25x16 btw.
Click to expand...

Probably the GTX570 if I had to choose between those two prices.
 
ihira said:
I'm not being skeptical but I'm sure Im not the only one wondering why [H]s results differ so much from other sites.
Click to expand...


because [H] reviewers actually play the game, unlike most review sites that use generic benchmarks app's that come with the games or use generic levels on a map and run around the map doing random ass shit. they also use generic graphic settings instead of basing those settings on actual playable performance. because theres a few times where the gtx 580 can actually run a game at a higher setting but that doesnt mean its actually playable.

GoldenTiger said:
Partially their review methodology... and probably partially driver versions too. The games HardOCP used happen to be ones other reviews show doing well for the 6970 too, but even in the [H] review BFBC2 minimum fps is horrible compared to the GTX 580, for example. I'd say they're actually fairly in-line with other reviews, all things considered.... the 6970 looks to be a very inconsistent card across various games, some it does near-580 in, while the rest it bounces between 5870 performance and GTX 570... :confused:.
Click to expand...

i'd chalk up the inconsistency to drivers. AMD has had a hell of a time trying to keep up with BFBC2 and this last game update seems to of broken everything AMD did to fix it before. so we'll see how long it takes them to get a hotfix out for the 10.12's to fix the performance. they may be waiting for BC2 vietnam to release before releasing the hotfix.


96redformula said:
Probably the GTX570 if I had to choose between those two prices.
Click to expand...


at that resolution i'd go gtx 580 or HD6970. i wouldnt bother with the 570. but thats just my opinion. if you were playing at 1920x1200 then the 570 or 6950 are a much easier choice.

DarkStryke said:
Eyefinity and 30" with AA cranked is the sweet spot. I think people forget these are 2 GB cards as well, not the 1gb offerings of old.
Click to expand...

they definitely aren't gpu bottle necked like the 5870's where thats for sure. looks like the 2GB of vram already has a much bigger performance improvement over the 2GB 5870's.
 
Last edited:
I am still waiting for Asus HD6950 card to appear. I am also considering just going HD6850 crossfire, but I want the one that will be better at eyefinity resolutions in the long run.
 
sirmonkey1985 said:
because [H] reviewers actually play the game, unlike most review sites that use generic benchmarks app's that come with the games or use generic levels on a map and run around the map doing random ass shit. they also use generic graphic settings instead of basing those settings on actual playable performance. because theres a few times where the gtx 580 can actually run a game at a higher setting but that doesnt mean its actually playable.
Click to expand...

It's a preference thing, personally I like apples-to-apples by far, the best is when it's apple-to-apple and in-game run-throughs. I feel it is a much more accurate picture, and minimum FPS reflect a much more important picture than avg. in most cases: who cares if it averages 40 but min's at 10 often, when the competitor is 35 avg but 20fps min providing a far superior actual experience?
 
Wow, awesome review guys! The reason why I always read HardOCP's reviews on new tech first is because you guys cut through the BS and show how it really is. You don't overwhelm us with mountains of charts that get very confusing, and keep the comparisons neat and fair. Seeing those mountains of charts is useless when they won't tell you like HardOCP's reviews about what a particular GPU will ACTUALLY bring to the table in terms of gaming experience. I know that your review style requires an incredible amount of work, but I do wish you guys would add an additional game or two during the reviews though.

As for the Cayman parts go, what really impresses me is the value of the 6950 as it is going to prove itself to be the 5850 of this current Southern Islands generation of GPU's. I hope you guys will do a CFX review at some point in the future, as it looks very promising what it could be capable of. That said, it looks like I will be skipping this generation since my two 5850's in CFX are still doing just fine. I am a bit bummed that ATI doesn't have a GTX580 killer at this time, but they are clearly doing what AMD does best which is offering an impressive price/performance ratio.
 
Good Review overall, and well worded!

I actually starting [H] reviews because their benchmark scores seem to be the closest to my own experiences on the video cards I have used over the years. Other sites... just seem way out there, I can't replicate what they do. +/- 5 FPS isn't a game changer for me.

I'm starting to sorely miss multi-monitor testing though. 2GB seems to be begging for it. Follow-up article?

No upgrades planned for me right now.. but who knows... I really like the fact these new cards are Quiet though! :)


Y.
 
Last edited:
Here is a question I have for HardOCP editors: What fraction of your audience do you think:

1) Games at 2560x1600

AND

2) Considers average framerates of 30~40 fps acceptable?

The HardOCP style of GPU review has always been an interesting contrast to other sites, but in this case I think you chose your "playable settings" poorly. Like what I assume is a much larger margin of the audience, I'm more interested in knowing how these cards do at 1920x1200, where they may be closer to 60 fps.
 
I think in the past they've said this is [H]ard|OCP, where the big boys play at 2560 x 1600 or higher (Multi-Monitor), and the cards are shown stressed to whatever breaking point performance plateaus there are at higher resolutions & AA levels. If a game is cranked at 30" + and getting 40FPS, do you really need to know that it gets 70FPS by dropping down 1-2 tiers of resolution?
 
LeninGHOLA said:
Well, $369 is something I can stomach with this year's bonus. I ran dual 5850's until my then GF needed another card, so I ran a single 5850. I had a tiny bit of trouble with crossfire in some games, but overall it was a great experience. Doubtful I will ever get the other 5850 back, but one might be going in FS/FT to help fund a 6970.

The 580 and 570 both impressed me, but AMD has a way with pricing their cards at launch...
Click to expand...


Apart from the price / perfomance, even in terms of power usage and heat both the cards are much better than 570 / 580, which yealds to better efficiency although i don''t know how 6970 reached 89 degree (but an aftermark coolar should resolve that problem).

But the fact still remains that 6970 consumes 88W less power than 580, that means it has more potential to OverClock.

If I am not wrong, a superclocked 6970 with new catalyst drivers (Catalyst 10.12) should outperform 580. Same may apply in case of 6950 vs 570.

I was impressed with GTX 570. But these 2 cards are one step ahead will perfect suit for my upcoming build in which I am planning to Game @ 25X16 resolution
 
LeninGHOLA said:
Oh lord, you again? ;)

Which other sites do non bench/FRAPS partial runs? Asking out of curiosity now
Click to expand...

I can think of a couple off the top of my head... hardwarecanucks does some when appropriate, and there's another one starting with a "G" that I don't visit much. Pretty sure bit-tech does some too. I'd also note that everyone uses FRAPS, how do you think [H] records their FPS #'s? Your argument here is in-game run-throughs vs. bench tools: many sites do in-game areas, especially on-rails areas for consistency, nowadays, instead of running the bench tool at default settings and calling it a day :). As to whether in-game manual run-throughs of non-scripted areas are superior to consistently testable, repeatable results... that's a whole 'nother ballgame that isn't really fitting for this thread.
 
Ahh, this looks win!! Looks like 2x 6950 for me :D

It's nice to know that in real gameplay the 6950 is very impressive against the 570. Also, looks like AMD at least solved most of their minimum framerates issues. Makes me feel Cypress was really unbalanced.

Oh and thanks for the review Kyle and Brent!
 
tony-kun said:
1) Games at 2560x1600
Click to expand...
If you don't game at 25x16, or higher, then make sure you check other reviews before you buy.
That's just the message I've gotten tonight after scrolling through about 5 different websites.

The performance at 2560x1600 is not indicative of overall performance.
 
jimbean21 said:
So for a guy going from a single GTX 480 SC to perhaps Dual GTX 580's, or Dual 6970's.. i know which is the obvious better buy.. but which is the better performer all around?

ninja edit - i play at 1920x1200.
Click to expand...

THe cheapest i can find the gtx 580s for in the states is $520 so your looking at $1040 vs $740

Is the extra performance from the gtx 580s worth the $300 ?

Also not for nothing but cayman seems to have alot of changes from cypress, so we may see more performance gains in the future from new drivers than the gtx 580 which is basicly a fully working gf100 chip. So the gap between the cards could shrink in the future.

Personaly , I'd save the $300
 
DarkStryke said:
I think in the past they've said this is [H]ard|OCP, where the big boys play at 2560 x 1600 or higher (Multi-Monitor), and the cards are shown stressed to whatever breaking point performance plateaus there are at higher resolutions & AA levels. If a game is cranked at 30" + and getting 40FPS, do you really need to know that it gets 70FPS by dropping down 1-2 tiers of resolution?
Click to expand...

I'd just like to know who is getting 60+ fps on current gen games at 2560x1600 @ 8xmsaa +. Like, put me on your christmas list, assholes. :D
 
The HD cards are priced to kill, why does Nvidia even bother?
 
BeavermanA said:
$36.02 tax, $8.50 shipping for Sapphire. I already ordered it, debating whether to cancel and get a 570 tax free free shipping at Amazon.
Click to expand...

Ah i see.

Up to you. The 6970 is faster than the gtx 570 from what i can see. Also remember cayman has made alot of changes compared ot cypress vs the changes from the gf110 to the gf100 . So we may see further gains with newer drivers. The 6970 could grow on the 570.

Really up to what you perfer tho , nvidia or amd .
 
ihira said:
I'm not being skeptical but I'm sure Im not the only one wondering why [H]s results differ so much from other sites.
Click to expand...

tony-kun said:
Here is a question I have for HardOCP editors: What fraction of your audience do you think:

1) Games at 2560x1600

AND

2) Considers average framerates of 30~40 fps acceptable?

The HardOCP style of GPU review has always been an interesting contrast to other sites, but in this case I think you chose your "playable settings" poorly. Like what I assume is a much larger margin of the audience, I'm more interested in knowing how these cards do at 1920x1200, where they may be closer to 60 fps.
Click to expand...


Thanks for the feedback. The red X in the corner of the screen will solve your issues.
 
BeavermanA said:
$36.02 tax, $8.50 shipping for Sapphire. I already ordered it, debating whether to cancel and get a 570 tax free free shipping at Amazon.
Click to expand...

Keep in mind that's the Canadian Newegg site, not the US one. Obviously you're in the states as our Amazon.ca doesn't have video cards (unless you count the Amazon resellers). That's probably why you're being hit with taxes.
 
all depends on the game Tony. personally i dont care about low frame rates in single player. its multiplayer where minimum's matter. but theres no possible way to benchmark multiplayer games. theres just to many variables in it. so you give the absolute maximum performance in the review and the reader then uses that brain of theirs to figure out what the performance would be for them..

for example i game at 1920x1200. so in game it benchmarks at 2560x1600 @ mid 40fps range with say 8x AA/16x AF. then i know at 1920x1200 i should be in the high 50's average at 8x AA/16x AF and probably be able to increase the AA settings and keep a 40fps average. if the game benchmarks at 4x AA/16x AF at 2560x1600 then i know i'll probably be able to run 8x AA at my resolution.

its really not that hard to figure out, just requires a little common sense.
 
GoldenTiger said:
I can think of a couple off the top of my head... hardwarecanucks does some when appropriate, and there's another one starting with a "G" that I don't visit much. Pretty sure bit-tech does some too. I'd also note that everyone uses FRAPS, how do you think [H] records their FPS #'s? Your argument here is in-game run-throughs vs. bench tools: many sites do in-game areas, especially on-rails areas for consistency, nowadays, instead of running the bench tool at default settings and calling it a day :). As to whether in-game manual run-throughs of non-scripted areas are superior to consistently testable, repeatable results... that's a whole 'nother ballgame that isn't really fitting for this thread.
Click to expand...

What argument? I was just asking you a question. Dunno what axe you're trying to grind, but it seems to be the results the [H] came up with.
 
LeninGHOLA said:
What argument? I was just asking you a question. Dunno what axe you're trying to grind, but it seems to be the results the [H] came up with.
Click to expand...

You were clearly being sarcastic with the "Oh lord, you again" comment, so I said "if we were making an argument about it, it doesn't belong here." No axe being grinded by me.
 
GoldenTiger said:
Nope, ATI has already said they're raising prices in January, so it won't last. This is according to the owner of OcUK, an online shop.
Click to expand...

This makes sense. With the launch this late into the holiday shopping season, they'd want to sell as many as they can. I'm sure the 2GB of 6gps DDR5 is costing them a pretty penny. These prices are much better than I expected. The 6950 is a steal @ $299! Backplate, vapor chamber cooling, dual bios, 2gb DDR5.
 
You must log in or register to reply here.
Back
Top