I think I read on [H] that dynamic clocking is standard practice in games consoles, as is adaptive Vsync. So I'm sure that AMD will provide a similar solution shortly.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
GeForce 680 & 670 vs. Radeon 7970 & 7950 Gaming Performance @ [H]
- Thread starter FrgMstr
- Start date
I think I read on [H] that dynamic clocking is standard practice in games consoles, as is adaptive Vsync. So I'm sure that AMD will provide a similar solution shortly.
Yes I realize that they personally used 7970s first. And I also agree that nVidia came out with some great and amazing cards and I'm looking at switching too, from a 5870 to a 670. But it is there ever so slightly in the voice of the writing. I'm not discrediting them for anything and even complimented them on their approach to new products and their testing methods. I always look forward to their reviews. I think its one of those things that when you favor something, you own it, and you pit it up against something else, it's very difficult to not even have the slightest sway, you might not even know you're doing it. You just can't remove that part of information in your brain thats telling you that you have that product.
They are obviously enthusiasts, so it's probably very very difficult for them to not be enthused about the products they are reviewing. And a little bit of that enthusiasm comes out. I value their opinion and obviously so do you, otherwise you wouldn't be here. And they give their opinions in these reviews, which is what we want. Otherwise they might as well take out all writing as just post graphs and tables. It would be impossible to write a piece right down the middle. Does that make sense? IDK, maybe its just me. I wasn't trying to start anything, just backing up someone else's opinion when he got jumped on for something that I felt as well. So it wasn't just him.
Last edited:
MavericK
Zero Cool
- Joined
- Sep 2, 2004
- Messages
- 31,944
I get what you're saying, and I agree mostly. However, I think the term "biased" is being misused here. When I think of "bias" I think of crediting a product over another where credit is not due. As I stated and as you stated as well, both products are good, and the overall gameplay experience seemed to be better with the 680. Both products received Gold awards. Of course you are going to lean toward the product that has given you the better experience and that is what they have done here. However, that is not "bias", that is choosing the product that gave you the better experience.
Yeah I would agree. I wouldn't call it bias either, more like the flavor of the writing. I believe Solomutt was basically getting at the same thing. So we can move on now and get this thread back on topic. Yay for nVidia!!
Pyroja
Gawd
- Joined
- Jan 4, 2005
- Messages
- 594
I really find the assertion that HardOCP is somehow biased against AMD completely laughable.
I'm sorry if the numbers don't make you feel good about your choice of GPU, but the facts are all clearly laid out and explained well. Any bias you're seeing is the bias in your own head.
I'm sorry if the numbers don't make you feel good about your choice of GPU, but the facts are all clearly laid out and explained well. Any bias you're seeing is the bias in your own head.
Wow dude, way to come to the conversation late and make sure you read everyone's post in detail with good reading comprehension before trying to start a flame war. No one said they were bias. geezus
MavericK
Zero Cool
- Joined
- Sep 2, 2004
- Messages
- 31,944
Solomutt actually did pretty plainly say that, but I don't want to drag this off-topic again.
jhatfie
[H]ard|Gawd
- Joined
- Mar 19, 2003
- Messages
- 1,636
Great review [H]! I only wish your 7950 was a better clocker for the OC'd results. 1050Mhz seems to be on the low end of the 7950 OC spectrum. I also would like to see maybe one or two more games thrown in to add some additional perspective, but good stuff otherwise.
I am very pleased with my 7950, but I like what Nvidia has delivered this round. Not enough to sell my 7950 but am interested in maybe a 660 ti (if they are indeed releasing one) to replace my 6950 in my HTPC to reduce some heat and power consumption.
I am very pleased with my 7950, but I like what Nvidia has delivered this round. Not enough to sell my 7950 but am interested in maybe a 660 ti (if they are indeed releasing one) to replace my 6950 in my HTPC to reduce some heat and power consumption.
Pyroja
Gawd
- Joined
- Jan 4, 2005
- Messages
- 594
Emphasis mine.
Now then... Ace review. If I hand't had JUST jumped on a good deal on a 6950, I'd have my eyes set on a 670 for sure. Though, to be honest, I think my system would bottleneck it, which is why I didn't feel bad getting the 6950 in the wake of all these new, fantastic cards being released. I guess I better focus upgrading my CPU before I keep drooling over all these new GPUs
Trepidati0n
[H]F Junkie
- Joined
- Oct 26, 2004
- Messages
- 9,269
This is a good point. What I find amusing is that we are fundamentally arguing about is card X is better than card Y "if" you can get a good overclock. Right now at stock clocks..the 670/680 are the clear leaders. However, when OC'd the line "blurs" quite a bit. Howver when you need to balance a "given" (stock) with an "unknown" (potential OC) when chosing a video card...you really shouldn't put all your eggs in the OC basket unless you are that fuckwad that keeps buying and returing cards until they get the OC they want fucking up the prices for the rest of us.
Either way, the consumer is "winning" this battle more than nVidia and ATI are which is fine by me.
SonDa5
Supreme [H]ardness
- Joined
- Aug 20, 2008
- Messages
- 7,437
Thanks for the review.
Seems like your HD7950 was kind of slow at over clocking.
Average over clock for HD7950 on air is listed as 1161/1621MHZ at HWbot.org
http://hwbot.org/hardware/videocard/radeon_hd_7950/
Not sure if that would change the results much with the games that were looked at.
On water cooling HD7950 average over clock is listed as 1246/1776MHZ at HWbot.
Seems like your HD7950 was kind of slow at over clocking.
Average over clock for HD7950 on air is listed as 1161/1621MHZ at HWbot.org
http://hwbot.org/hardware/videocard/radeon_hd_7950/
Not sure if that would change the results much with the games that were looked at.
On water cooling HD7950 average over clock is listed as 1246/1776MHZ at HWbot.
Don't forget all of the cards in this review are reference models. I am pretty sure that this database includes all cards, including overclocked models with fancy heatsinks. Naturally the non-reference models overclock higher. Bottom line is that the overclock they got here is a reasonable one.
The most interesting thing about this is the comment at the end... that you guys are "going Green". I guess that's because you have multi-monitor dual-GPU rigs, and you've made it clear in your evals that nVidia smokes AMD in those departments.
But for us mortals who have a single-monitor rig, obviously the most powerful card in this evaluation is the overclocked 7970. I've been on the fence about which card to get, but it seems like there's a good case to be made for the 7970 given its penchant for super-high clock speeds and the obvious supply problems that nVidia is having, which don't show any signs of being resolved in the coming weeks.
But for us mortals who have a single-monitor rig, obviously the most powerful card in this evaluation is the overclocked 7970. I've been on the fence about which card to get, but it seems like there's a good case to be made for the 7970 given its penchant for super-high clock speeds and the obvious supply problems that nVidia is having, which don't show any signs of being resolved in the coming weeks.
7950 OC's were extremely low - people are getting 1200+ on air these days. Would like to have seen the SAPPHIRE 100352OCSR Radeon HD 7950 which is stock @ 950Mhz and surely able to get 1200+... Which should even the score like 7970 did and may advance ahead of 670.
But OCing is never guaranteed that's why 6xx is better atm and gets praise it does.
But OCing is never guaranteed that's why 6xx is better atm and gets praise it does.
Brent_Justice
Moderator
- Joined
- Apr 17, 2000
- Messages
- 17,755
I actually game on a single display, 2560x1600.
Why does this conclusion only talk about 670? Does 680 not exist? Pretty sure the article is is titled with 680 as well. Why does the conclusion nowhere mention that max OC in the 4 games tested, 7970 beats 680 for $50 less? Why doesn't it say "at these prices 680 cant compete in the market" like it does about 7950, when it's in almost the same position 7950 is vs 670?
Your exact wording:
Hmm, last I checked 580>6970. Hmm, last I checked 480>5870. Hmm last I checked 280>4870. Hell I cant even remember the last time AMD had the fastest card, maybe X1950XT? So why does the conclusion say AMD has had the fastest cards?
The fact is 7970 is closer to Nvidia's best (if not equal) than AMD has been in generations. 7970 when fully Oc'd is almost equal to 680, much closer than 6970 was to 580.
To say nothing of the fact you ignore in your conclusion 680 isnt available and hasnt been for months (easily literally the worst paper launch of all time). And 670 is barely available. While AMD has a full lineup like 7870, 7850 etc.
Yet again you paint it as "oh boy Nvidia killing AMD"? Sure Nvidia has been doing well but they've had tons of flaws, like I dont know, you cant buy any of their new cards half the time. They dont have anything from this gen out under $400. You know, just tiny irrelevant things like that.
Then you top it all off with the usual driver stuff, all I can say is if Nvidia drivers are as good as you claim, why does it seem like 95% of all forum driver problems are Nvidia cards? Why has my 4890 had zero problems ever with drivers, where I go on forums and everybody says Nvidia drivers are better, but then I look and see countless posts with Nvidia driver problems and I'm literally scared to go back to Nvidia? It
Too be fair you mention crossfire, I understand AMD drivers are bad in that and I dont have experience there. but I will say I'm pretty 80% of the people saying Nvidia drivers are better than AMD on forums are Nvidia fanboys and are not being truthful.
Am I saying AMD has handled things perfect, or has no flaws? Of course not. 7950 is a joke compared to 670, and why is AMD so slow on needed price cuts (in this example, to the 7950)? But I am saying the bias seems present again here.
Last edited:
If you read the quote again, you'll see it says they have the most power efficient, fast and affordable cards - it doesn't say they have the fastest. They have the best combination of those three things and they've always had the most power efficient cards - but nowhere did the quote you posted say they had the fastest cards. I know it is semantics, but your whole rant is kind of based on missing that distinction.
And again, minor distinction, but the 7970 and 680 tie in BF3, they tie in Batman (1% margin), the GTX 680 wins Skyrim, and the 7970 wins Duex Ex. I don't see the 7970 winning all 4. You can argue that it is better price/performance, but it doesn't win outright.
I agree that it seems weird that they sort of left out that the oc'd 7970 was the fastest card, but there aren't any guarantees that a reference card will clock that high, so it's kind of thorny to use that as the basis for a recommendation when at stock speeds nVidia has the advantage. They pointed out that the 7970 is limited at stock frequencies, which is a reasonable criticism.
Having said that, it seems like most people are getting very high oc'd frequencies on AMD cards, and AMD has the rather decisive advantage of actually having lots of cards in stock. Case in point, me: the 670 whet my appetite for a new card, but I went for a nice oc'd 7970 with a non-reference cooler instead because a) it's probably going to end up just as fast if not faster than a 680, and b) I can actually buy it, while the 600 series cards have been out of stock since they launched. nVidia snoozes, they lose...es.
Performance wise when OC the results are so close it is just splitting hairs and not very important. Believe Kyle runs dual cards and 3 screens - if CFX and/or Eyefinity is causing issues and the Nvidia solution does not or has less issues even if slower from a FPS point then why not switch? Driver stuff, well HardOCP has been talking for awhile about driver stuff too from lack of game support on new games, CFX etc. That is also part of the review and performance of a given card. So the poorer FPS card (in this case insignificantly slower) beats the AMD solution overall from their view not limited to just FPS. I am very happy with my 7970, awesome card for my setup for someone elses setup the 680 can be the better choice.
Do wish IQ was talked about more, hit on performance when different combinations are used. Probably more appropriate on a separate review. Plus thank god I didn't see the Adative SSAA used (which does not exist in the drivers as far as I can tell
). Anyways SSAA on the 6xx series is showing better performance then the 7xxx series, my guess is fillrate of the 6xx is much higher then the 7xxx which shows up with 4x SSAA and high resolutions.Lord_Exodia
Supreme [H]ardness
- Joined
- Sep 29, 2005
- Messages
- 7,008
@sharky and anyone else referencing the GTX 680 vs 7970 overclocking results. The 7970 is cheaper and in this review at least slightly faster when overclocked, however keep in mind that this is nowhere near a final conclusion. There are many factors that come into play. I may buy a 7970 and only get 1180mhz out of the core then where does that leave me? On top of that they need to be at least 25% faster in my opinion to justify the shitty drivers people are given with those cards.
Aside from all of that, time and time again it's been shown that AMD royally sucks in multicard with the GCN architecture. BSOD's with anything but the release drivers, 3 cards are totally fubar'd choppy performance. SLi although not the same fps is seeming to give a smoother experience. That's with early drivers. I have more faith that in the coming months that nvidia will improve performance with the drivers even beyond what it is now.
Finally if you look at the results in the overclocking section, where amd barely beats nvidia in a few of those games in the apples to apples settings, nvidia usually has better minimum fps which matters more if you ask me. At best those few games show a draw with a very (ever so slight) win for AMD. This could have easily changed had hardocp either had a better clocking GTX 680 like mine, or a worse clocking 7970 which can likely happen.
Single card vs single card I'd say at $50 cheaper the 7970 is still at a disadvantage vs the GTX 680. I'd even go as far as saying that at the end of this generation nvidia will outsell 7970s 2 to 1 with GTX 680s once the dust settles.
You need to keep everything in perspective.
Aside from all of that, time and time again it's been shown that AMD royally sucks in multicard with the GCN architecture. BSOD's with anything but the release drivers, 3 cards are totally fubar'd choppy performance. SLi although not the same fps is seeming to give a smoother experience. That's with early drivers. I have more faith that in the coming months that nvidia will improve performance with the drivers even beyond what it is now.
Finally if you look at the results in the overclocking section, where amd barely beats nvidia in a few of those games in the apples to apples settings, nvidia usually has better minimum fps which matters more if you ask me. At best those few games show a draw with a very (ever so slight) win for AMD. This could have easily changed had hardocp either had a better clocking GTX 680 like mine, or a worse clocking 7970 which can likely happen.
Single card vs single card I'd say at $50 cheaper the 7970 is still at a disadvantage vs the GTX 680. I'd even go as far as saying that at the end of this generation nvidia will outsell 7970s 2 to 1 with GTX 680s once the dust settles.
You need to keep everything in perspective.
Call me a fanboy with my 7950 that I paid $349 for two weeks ago but I think the comparison is slightly skewed here. I can still find reference 7950's for 369 with 3 free games right now at ncix.ca and I can find OC aftermarket Sapphire and Gigabyte 7950's once again with 3 free games at $399, the same price as the reference 670.
Firstly, If you are going to call the reference 7950 that I paid $349 or I can find in 5 minutes of searching for $369 the same price as the $399 670 then why not at least compare the OC versions from sapphire and gigabyte that really are $399?
Secondly, looking at the OC results when you add up and average the overall performance increase of the 670 over the 7950 is less than 6% and the difference between the 670 and the 7970 in OC results averages 12.5% across the 4 tests.
So when I see this statement:
"GeForce GTX 670 to the Radeon HD 7950, the value delta is tremendous; it is simply an embarrassing scenario for the Radeon HD 7950. The GeForce GTX 670 clearly offers much better gaming performance at the same price. At the current prices the Radeon HD 7950 cannot compete in the market."
Isn't that just a bit of a stretch? 5.x% overall performance difference in the OC results and it can't compete in the market? Priced the same?? Yes the stock clock results are a greater difference but the OC results are really the bottom line IMO. Below is my suggestion to be more fair about price and performance comparison.
Overall I'd like to see the 670 compared to the OC versions of the 7950 at the same price because the prices will be 100% fair at that point ($399), the non OC results will be much closer with the stock clocks on some of these 7950 cards around 900-950 rather than 800mhz, and the additional cooling I suspect would provide enough headroom to overcome the whopping 5.x% OC lead the 670 currently holds. THAT test would be fair.
Others have already mentioned the fact that the 680 availability isn't really being called out, the 7970 OC provides way better value/performance ratio beyond what you are slamming the 7950 compared to the 670 for as well. Call me a fanboy but if you tested fairly as I suggested at the exact same pricepoint with OC cards then I wouldn't be annoyed when you call them the same price or say only the non OC results are forming your opinion when the 79xx series is obviously underclocked at reference and the OC results are VERY close, certainly competitive in the market unlike your statement.
I don't see the insistence of testing reference vs reference and overlooking the $20-50 price savings when you can simply test an OC version of the card that actually is the same price and let the FPS speak for themselves. That's my rant of the year from a long time HardOCP fan that has a lot of respect for your work and wouldn't bother ranting if I didn't.
Firstly, If you are going to call the reference 7950 that I paid $349 or I can find in 5 minutes of searching for $369 the same price as the $399 670 then why not at least compare the OC versions from sapphire and gigabyte that really are $399?
Secondly, looking at the OC results when you add up and average the overall performance increase of the 670 over the 7950 is less than 6% and the difference between the 670 and the 7970 in OC results averages 12.5% across the 4 tests.
So when I see this statement:
"GeForce GTX 670 to the Radeon HD 7950, the value delta is tremendous; it is simply an embarrassing scenario for the Radeon HD 7950. The GeForce GTX 670 clearly offers much better gaming performance at the same price. At the current prices the Radeon HD 7950 cannot compete in the market."
Isn't that just a bit of a stretch? 5.x% overall performance difference in the OC results and it can't compete in the market? Priced the same?? Yes the stock clock results are a greater difference but the OC results are really the bottom line IMO. Below is my suggestion to be more fair about price and performance comparison.
Overall I'd like to see the 670 compared to the OC versions of the 7950 at the same price because the prices will be 100% fair at that point ($399), the non OC results will be much closer with the stock clocks on some of these 7950 cards around 900-950 rather than 800mhz, and the additional cooling I suspect would provide enough headroom to overcome the whopping 5.x% OC lead the 670 currently holds. THAT test would be fair.
Others have already mentioned the fact that the 680 availability isn't really being called out, the 7970 OC provides way better value/performance ratio beyond what you are slamming the 7950 compared to the 670 for as well. Call me a fanboy but if you tested fairly as I suggested at the exact same pricepoint with OC cards then I wouldn't be annoyed when you call them the same price or say only the non OC results are forming your opinion when the 79xx series is obviously underclocked at reference and the OC results are VERY close, certainly competitive in the market unlike your statement.
I don't see the insistence of testing reference vs reference and overlooking the $20-50 price savings when you can simply test an OC version of the card that actually is the same price and let the FPS speak for themselves. That's my rant of the year from a long time HardOCP fan that has a lot of respect for your work and wouldn't bother ranting if I didn't.
Last edited:
Then you'd have to bring in OC versions of the 670 which are also $399...
Elox, I agree that the 7970 performs great when overclocked but I think you're overlooking the point of the review. In fact, having used both 7970CF and 680 SLI myself I can absolutely state that *I think* the 7970 scales much better with clockspeed than the 680...and once you get into the 1150-1300 range...it outperforms the 680 in many games. However, the out of the box experience of the GTX 670 is better than the 7950. Just like the out of box experience with the 680 is better than the 7970.....remember , overclocks aren't guaranteed and shouldn't be the basis for reviews. There is nothing wrong with what is stated in the review - by default the 670 is a better value and better for the money at 400$ - i'd have a hard time recommending 7950 over a 670 to a new buyer at current prices. Thats not to say that the 7950 is a bad card, it is very good, i'm sure you're enjoying it.
Last edited:
Yes, why not if they are indeed $399 and in stock somewhere then it would be fair to compare with those.
I think you missed the point of my post. The whole point of my post was to show why it makes more sense to compare a OC 7950 in the core clock range of 900-950 that costs $399 with a reference 670 that costs the same for a fair comparison.
I disagree that there was nothing wrong stated in the review. They made a conclusion based on the price of the 670 and 7950 being the same price which is wrong and if they want to compare as I suggested two cards that are in fact the same price then I am willing to bet the stock and OC results are going to be much much closer to the point that the 7950 is certainly competitive, unlike their conclusion.
You're telling me that I missed the point of the article in saying the stock clock of the 670 (with dynamic boost) outperforms the stock clock of a reference 7950 but I didn't miss that point at all, I simply disagree with the validity of the comparison and presented what I feel would be more accurate. Hopefully now you understand the point of my post.
Ok Elox I'll bite.
The only widely available overclocked 7950 for $400 is the Gigabyte model. It features a 900 MHz core clock vs 800MHz stock. Memory clock is not OC'ed.
This means theoretically, it is AT MOST 12.5% faster than a stock 7950.
Assuming PERFECT scaling from [H]'s apples to apples reviews, this how it would stack up against a STOCK 670 (Min - Max - Avg):
31.5 - 78.8 - 48.7 vs 30 - 84 - 51.5 in BF3
20.3 - 74.3 - 41.3 vs 28 - 73 - 44.7 in Batman AC
32.6 - 92.3 - 59.1 vs 31 - 146 - 68.6 in Skyrim
22.5 - 87.8 - 63.8 vs 36 - 78 - 58.9 in DX HR
The only widely available overclocked 7950 for $400 is the Gigabyte model. It features a 900 MHz core clock vs 800MHz stock. Memory clock is not OC'ed.
This means theoretically, it is AT MOST 12.5% faster than a stock 7950.
Assuming PERFECT scaling from [H]'s apples to apples reviews, this how it would stack up against a STOCK 670 (Min - Max - Avg):
31.5 - 78.8 - 48.7 vs 30 - 84 - 51.5 in BF3
20.3 - 74.3 - 41.3 vs 28 - 73 - 44.7 in Batman AC
32.6 - 92.3 - 59.1 vs 31 - 146 - 68.6 in Skyrim
22.5 - 87.8 - 63.8 vs 36 - 78 - 58.9 in DX HR
Not planning on ANY oc is similar to planning a long driving trip and only expecting to go the speed limit. You're going to go 10 over, and more if you can get away with it -- same with overclocking. You sure as hell aren't going to leave the card at stock, so those numbers are a baseline, but mostly useless when considered in the real world.
This article was excellent and has further reinforced my idea to grab a 670 now and something bigger and better with more ram later for triple 30" gaming.
Bought a Gigabyte 670 OC (non-reference PCB, extra power phase, 6+8 pin, Windforce cooler) from NCIX on launch day for $399CAD, clocks 980/6000GHz (reference is 915MHz). Boosts to 1176MHz in BF3 without touching anything.
When I was buying last week, I looked at the 7950's from NCIX, but couldn't find a good non-reference dual fan style cooler for under $400 and I wanted a cool/quiet card not a leaf blower and not questionable brands with foreign RMA (Sapphire, HIS etc).
Ya I picked up my reference 7950 for $349 from NCIX as well two weeks ago. I see they are OOS on the model you purchased but that definitely seems like the best buy at $399.
wabbitseason
[H]ard|Gawd
- Joined
- Jun 16, 2010
- Messages
- 1,511
I find myself asking the same question. It seems everyone wants Nvidia to "win" this round, which is baffling. I would think that the card that did the best when pushed to its limits would be hailed as a champion here, at [H].
IdiotInCharge
NVIDIA SHILL
- Joined
- Jun 13, 2003
- Messages
- 14,675
We don't have to want it- we just have to explain it to others. Nvidia has already 'won'; and they did it with Kepler vs. GCN. Smaller, lower power draw, cooler and quieter, while being as fast or faster depending on the clocking scenario.
wabbitseason
[H]ard|Gawd
- Joined
- Jun 16, 2010
- Messages
- 1,511
Power and temperature differences are negligible. I'm sure you recall Fermi vs the 5870. Smaller? Wow you're reaching. As fast or faster is a flat-out lie, as the 7970 OC outperforms the 680 OC while being $50 cheaper.
The fact is people are coming out of the woodwork, rekindling the positive feelings they want to have about Nvidia with the 6xx-series (it's actually competitive on price/performance! it doesn't use twice the power of AMD's offering! that means its great!). And Nvidia has had a great history. But the company today is a corporate cesspool, a bunch of cheats and liars riding the coattails of a fine engineering team.
The fact is people are coming out of the woodwork, rekindling the positive feelings they want to have about Nvidia with the 6xx-series (it's actually competitive on price/performance! it doesn't use twice the power of AMD's offering! that means its great!). And Nvidia has had a great history. But the company today is a corporate cesspool, a bunch of cheats and liars riding the coattails of a fine engineering team.
Definitely puts them closer at stock speeds. Certainly competitive in my opinion. Now the other aspect of it was the overclock #'s which we get a max overclock Hard was able to achieve on this card http://hardocp.com/article/2012/03/01/xfx_radeon_hd_7950_black_edition_video_card_review/7
Which was 1195Mhz.
This is a 49% clock increase. Based on the ratio of FPS increase seen going from 800 to 1050 and applying that to the 1195 would get more like a 41% actual FPS increase rather than 49%.
That would bring the avg FPS for these titles to about:
61.1FPS BF3
51.7 FPS Batman
74.0 FPS Skyrim
80.0FPS Deus Ex
As a reference here's the 670 OC #'s from the article:
58.1 BF3
49.7 Batman
74.5 Skyrim
67.4 Deus Ex
Now that particular card tested by Hardocp to get the 1195Mhz I can find in a few mins of searching for $419 here: http://www.memoryexpress.com/Products/MX37656
No idea if 1195 is typical for these or other 7950 cards with better cooling than reference but it's the best sample I could find pulling data from hardocp's articles.
These #'s are not perfect, but based on the information I could pull here I would draw the conclusion that the 670 will still offer better out of the box performance as Xeth noted, but the performance would be fairly close. In addition the 7950 OC FPS would pull away higher than the reference 670 by about the same margin.
The dynamic clocking is a nice feature and still gives the 670 the out of the box advantage over the 7950 at 900Mhz (which cost the same) but at max OC the better cooled non reference 7950 holds just as great an advantage when each card is pushed to the limits.
I think we can at least agree that the 7950's that actually cost the same $399 are actually competitive depending on your use. (Little less or little more depending on if you OC or not).
I would still like to see it actually tested rather than my napkin math, but I've fulfilled my curiosity on the matter as far as my resources allow.
Personally I think the 7970 and 680 are equivalent in price/performance give or take. I just prefer nVidia due to better experiences on SLI vs CFX and the much better acoustics on reference nVidia cards. My overall decision probably is weighed something like this:
Raw price/performance: 50%
SLI/CFX compatibility: 30%
Acoustics/Efficiency: 15%
Manufacturer support quality: 5%
I'm sure most people have some variation of this in their head when choosing a video card. Almost nobody puts 100% on raw price/performance - that'd be like buying a car just for the price per horsepower ratio. Dismissing all those secondary factors is simply unrealistic. After all, a 7850 @ $250 beats all of the cards reviewed in price performance.
Raw price/performance: 50%
SLI/CFX compatibility: 30%
Acoustics/Efficiency: 15%
Manufacturer support quality: 5%
I'm sure most people have some variation of this in their head when choosing a video card. Almost nobody puts 100% on raw price/performance - that'd be like buying a car just for the price per horsepower ratio. Dismissing all those secondary factors is simply unrealistic. After all, a 7850 @ $250 beats all of the cards reviewed in price performance.
Lord_Exodia
Supreme [H]ardness
- Joined
- Sep 29, 2005
- Messages
- 7,008
You draw this conclusion based on one review with a limited # of games. If the [H] used one of my GTx 680s which does 1332mhz boost with 6950mhz on the vram this review could have gone differently. Same could be said if they got a better clocking 7970 or a worse clocking one etc... The scenario is definitely easily changable as no overclocking is guaranteed. The only guarantee is that you are going to get your out of box performance. That said, many people here overclock, but because hardocp's samples reached the clocks they did shouldn't draw a end all conclusion of what hardware is faster.
Looking at the games in the apples to apples oc review'
Skyrim: Nvidia wins slightly, aside from that one down spike where nv went to 28 and amd to 31 nvidia is more consistently over the 60fps line and has higher avg and max fps
battlefield 3: is tied with a microscopic nod to nvidia for higher minimum fps,
Batman AC: another one for nvidia, with higher minimum and smoother gameplay as per the review
Deus Ex: AMD wins with better average and Max but nvidia wins with better minimum fps. Depending on someone's view on this it could be another win for nvidia as you can see a big difference in minimums here which can be felt more than max or average to some people
What is a fact is that AMD's 7970 is cheaper by $50 and would be considered an excellent value had their drivers not suck as they have for GCN architecture. I dunno maybe I read this review different than most /shrug
Now you're breaking your own "rules" and assuming everyone OC's their cards beyond the manufacturer clocks, which has been argued to death here so I'm not going to restart it.
Marketing is geared for the lowest levels of society, and NVIDIA is great at marketing: the rest is self-explanatory
.Anyway, the [H] article is a joke; did this become tweaktown when I wasn't looking? I thought [H] meant taking hardware to its maximum as well as appreciating the nuances of the technology. Now I guess it means "let's test 4 games with heavy manufacturer bias and lol about how easy it was to put the card in the slot." Disappointing.
This is pretty much the way I see it, having used both. I think they're both awesome cards honestly, but a lot of people don't give the 7970 a fair shake. It really does scream when its overclocked, but I concede that the 680 is a better out of the box experience and a better value for a new buyer. My upgrade path was from 7970CF to 680 SLI and I actually noticed a few games being noticably slower - alan wake, metro 2033 and crysis 2 among them, of course the 7970s were heavily OCed.
All that said, there's no getting around the fact that nvidia won this round. 670 is probably giving them a fit.....I really hope AMD steps it up with Sea Islands. ATI released a lot of monsters in years past, maybe they can have a return to the glory days at some point. They also need to increase the size of the software development team, by a lot - nvidia throws a lot of money into software development for their cards and it usually shows. AMD needs to do this - nvidia has raised the bar in this respect and status quo simply won't cut it for AMD.
Marketing is geared for the lowest levels of society, and NVIDIA is great at marketing, the rest is self-explanatory
Anyway, the [H] article is a joke, did this become tweaktown when I wasn't looking? I thought [H] meant taking hardware to its maximum as well as appreciating the nuances of the technology. Now I guess it means "let's test 4 games with heavy manufacturer bias and lol about how easy it was to put the card in the slot." Disappointing.
LOL if [H] was so biased why did they give a GOLD award to the 7970 when it came out?
Marketing is geared for the lowest levels of society, and NVIDIA is great at marketing: the rest is self-explanatory
Anyway, the [H] article is a joke; did this become tweaktown when I wasn't looking? I thought [H] meant taking hardware to its maximum as well as appreciating the nuances of the technology. Now I guess it means "let's test 4 games with heavy manufacturer bias and lol about how easy it was to put the card in the slot." Disappointing.
Its not bias, its about out of the box experience and the 680 wins. Sure you can put a 7970 under water, overclock to 1300, and it will probably beat a 680 easily, sure....but thats not a proper guage for a review. I say that as someone who also thinks the 7970 is a great card.
As far as the games tested, I go back and forth on this. Its hard to deny that Batman: AC and Skyrim favor nvidia cards for some reason, while other untested games may favor AMD. Shrug.