7850 vs current nvidia offerings

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Ieat

Senior member
Jan 18, 2012
260
0
76
The 7850 is slightly slower than the GTX570 as per the Anandtech review, Losing in Dirt 3, BF3 and Batman while wining in Civ 5, rest is about even. The GTX560ti 448 performs worse than the GTX570. AS for why the total score is so low in techpowerup's review, check the Dragon Age 2 benches. I'm guessing it's a driver issue.

I think you read the Civ 5 chart wrong. Gtx 570 handily beats the 7850 in that game as well. You also left out Portal 2. So out of 10 games the 7850 is noticeably slower in half of them while roughly tying the other 5. The gtx 560 448 is only 3% to 4% slower then the gtx 570. Given those results I would say its safe to assume the 7850 is slower at stock on average.
 

aaksheytalwar

Diamond Member
Feb 17, 2012
3,389
0
76
Stock to stock both are equal, 5-10% give or take depends on the game and is irrelevant.

Period.

Once Oced, 7850 is better. Period. Anybody who says otherwise is biased. Once overclocked you are nearly equal or better than a stock 7950 game dependent, on air and using the stock cooler

2gb VRAM is important. Period. Earlier cards with just 512-768 mb VRAM suck in modern games and that is what will happen to 1-1.5gb cards in 1-2 years max, maybe sooner. No arguing on that.

For everything else the 7850 wins as well.

Only place where 560 ti 448 wins is Cuda, physx, 3d, and slightly lower price.

And noticeably slower needs to mean at least 15-20%. Anything about 10% is not noticeable and irrelevant.
 

jacktesterson

Diamond Member
Sep 28, 2001
5,493
3
81
I'd take a 7850 anyday over a 570. I've owned both cards.

I go through video cards like candy, always buying and playing, then selling for something new.

I owned both a EVGA 570 and 580, and I prefer the 7850. Runs cooler and quieter highly o/c'd compared to the GTX's while sipping power.

Maybe one of the best value cards ever made. Its that good of a card.
 

aaksheytalwar

Diamond Member
Feb 17, 2012
3,389
0
76
I would take a 7850 over a 570 but not over a 580 because once the 580 is overclocked it will perform about as fast as a 7850 oc while giving faster stock performance. However due to power, noise, oc, memory etc if 7850 and 580 are both $250, then it is a debatable choice and not everybody would prefer the 580 other than nvidia fans and some others.
 

Ieat

Senior member
Jan 18, 2012
260
0
76
Stock to stock both are equal, 5-10% give or take depends on the game and is irrelevant.

Period.

Once Oced, 7850 is better. Period. Anybody who says otherwise is biased. Once overclocked you are nearly equal or better than a stock 7950 game dependent, on air and using the stock cooler

2gb VRAM is important. Period. Earlier cards with just 512-768 mb VRAM suck in modern games and that is what will happen to 1-1.5gb cards in 1-2 years max, maybe sooner. No arguing on that.

For everything else the 7850 wins as well.

Only place where 560 ti 448 wins is Cuda, physx, 3d, and slightly lower price.

And noticeably slower needs to mean at least 15-20%. Anything about 10% is not noticeable and irrelevant.

Uh the 5 games that it was slower at was 15%-20% or more. Also If 15% to 20% is your threshold then the 2 cards are equal even when overclocked... Gtx 560 448 is 8% faster on average at stock. It can overclock about 25% which means a boost of about 33% over a stock 7850. A 7850 overclocks about 40%. So that equals a 7% or so advantage for the 7850 overclocked. So how can it be better. Period. When it doesn't meet your threshold? Your statement seems wildly inconsistent. I mean anything under 10% is irrelevant right?

Also all of these responses are pointless. I never said the gtx 560 448 was a better overall product. The guy I was responding to said the 7850 was faster at stock. I was just pointing out that he is flat out wrong. It is not faster. It is not equal. It is slower. That is my argument. Do factual statements no longer matter here? If someone wants to dispute that then go ahead. Anything else is just going off on tangents.
 
Last edited:

SickBeast

Lifer
Jul 21, 2000
14,377
19
81
Wow, people are saying that the 7850 is better than a Gtx 580. I suppose it has more vram and runs cooler. I'm really looking forward to mine. If I needed the power I would buy two of them for crossfire.
 

stormkroe

Golden Member
May 28, 2011
1,550
97
91
Don't let power draw costs sway you with these 2 cards, even a 120 watt difference is only $0.0096 more per HOUR of maxed out, bogged-into-minimum-frame-rate-area gaming (which is not most of our game time) at $0.08/kwH. 3 hours per day every day worst case is $0.86 per month extra.
 

aaksheytalwar

Diamond Member
Feb 17, 2012
3,389
0
76
I never said 7850 was faster than 560 TI 448 by a significant measure at this point in time. But 2GB VRAM will make it a lot lot faster in games released after 1-1.5 years and at the moment both are pretty much equal. But in the future 7850 wins without a shadow of doubt. It is obvious which is better.
 
Feb 19, 2009
10,457
10
76
Don't let power draw costs sway you with these 2 cards, even a 120 watt difference is only $0.0096 more per HOUR of maxed out, bogged-into-minimum-frame-rate-area gaming (which is not most of our game time) at $0.08/kwH. 3 hours per day every day worst case is $0.86 per month extra.

What about $0.3/kwH and 4 hrs gaming per day (including weekends average).

2 years typical GPU usage lifespan.
 

3DVagabond

Lifer
Aug 10, 2009
11,951
204
106
Don't let power draw costs sway you with these 2 cards, even a 120 watt difference is only $0.0096 more per HOUR of maxed out, bogged-into-minimum-frame-rate-area gaming (which is not most of our game time) at $0.08/kwH. 3 hours per day every day worst case is $0.86 per month extra.

This is irrelevant. Just because you can afford to pay doesn't justify the wasted energy. Step out of the 19th century. Wasted resources, excess pollution, etc... Higher efficiency is a good thing. It's a metric that everything in the world that uses energy is striving to improve.
 

Jaydip

Diamond Member
Mar 29, 2010
3,691
21
81
7970 has 1 GB more vram and look how that turned out.680 even beats it in triple monitor gaming.Having more Vram is irrelevant unless u can use it optimally.Now a days where console ports are abundant a 5870 is still relevant for 1080p gaming.I will at-least try to compare with nv's offering before taking the plunge.But 7850 is a nice card and will be sufficient for 1080p gaming for quite some time.
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
7970 has 1 GB more vram and look how that turned out.680 even beats it in triple monitor gaming.Having more Vram is irrelevant unless u can use it optimally.Now a days where console ports are abundant a 5870 is still relevant for 1080p gaming.I will at-least try to compare with nv's offering before taking the plunge.But 7850 is a nice card and will be sufficient for 1080p gaming for quite some time.

I agree. 1080p with moderate AA can run on 1GB just fine.

Yeah there are high-rez packs and monitors with 2560x1440 or higher resolution, and Eyefinity users. For them, more than 1GB is great; for everyone else, it is not absolutely necessary.

1. Cross-platform development has really reigned in PC game hardware minimums, especially ports that hardly touch graphics. Some devs at least use higher-res textures for the PC versions, but clearly they are not trying to push the envelope in the way that, say, Crysis did.

2. In many cases a card will become obsolete faster due to its GPU (e.g., can't support DirectX 12 or is simply too slow) or power/noise/heat rather than its VRAM. In this case, 1GB VRAM is not enough, cards with 1GB VRAM will be obsolete for reasons other than VRAM.

3. The reason why we needed more and more VRAM in the past several years also has to do with monitor resolution growth, not just because of more demanding games. But monitor resolutions have stopped growing so much.. we're flattening out at 1080p now. That means that the minimum amount of VRAM we need has also stopped growing as fast.

I'm not going to name names but some people on this forum are always like "get 16GB of RAM and a 7970 3GB and the fastest Intel CPU you can afford or you can't get playable framerates" or whatever. That is simply not true. PC gaming doesn't have to cost a fortune if you spend wisely and can live with not maxing out every little setting, especially settings that hardly improve gameplay or visuals.
 
Last edited:

Jaydip

Diamond Member
Mar 29, 2010
3,691
21
81
Yeah absolutely.I will take ubersampling from Witcher 2 as an example.It adds virtually no noticeable IQ with a hefty performance cost.My 580 can't run it on 1080P.I have seen screenshots of ubersampling and i'm not impressed but the opinions may vary of course.This year is going to be disappointment for PC gamers.Only MP3 can be a saving grace.
 

stormkroe

Golden Member
May 28, 2011
1,550
97
91
What about $0.3/kwH and 4 hrs gaming per day (including weekends average).

2 years typical GPU usage lifespan.

Then it's $4.32 per month more if your gaming session is like 4 hours of looping furmark. Nevertheless, my recommendation was for the OP in Bakersfield, CA, which is $0.14/kwh, I don't know how much he actually plays games so I just had to guess. Obviously if you're power is $23 per kw minute you might have to consider power draw.

I also love the bandwagon passion of people getting indignant about 120 watts of power for a $300 PURE LUXURY item. You think I'm living in the 19th century? Go power factor correct the inductive loads in your house, like your fridge and hvac blower motor, you'll probably save over 600 watts just from that. And turn the brightness down on your monitor. And quit buying luxury items like video cards and plastic.

Ahh perspective, you elusive lady.
 

Concillian

Diamond Member
May 26, 2004
3,751
8
81
Then it's $4.32 per month more if your gaming session is like 4 hours of looping furmark. Nevertheless, my recommendation was for the OP in Bakersfield, CA, which is $0.14/kwh,

Cali has tiered power. Average may be 14 cents, but most people are living in the tier that is ~30 or even in the highest tier, which is -40 cents a kWh. Especially in Bake - a - field in the summer (A/C).

This means that if you "save" power by using a lower consumption card, it comes out of the high tier.
 

stormkroe

Golden Member
May 28, 2011
1,550
97
91
Cali has tiered power. Average may be 14 cents, but most people are living in the tier that is ~30 or even in the highest tier, which is -40 cents a kWh. Especially in Bake - a - field in the summer (A/C).

This means that if you "save" power by using a lower consumption card, it comes out of the high tier.

Yeah I saw that. When I posted, I just googled electricity rates for bakersfield and it said 14.5 cents for people who use 550kwh or less per month. I guess just find your current rates and extrapolate, decide if the price difference is a clencher for you.
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
Then it's $4.32 per month more if your gaming session is like 4 hours of looping furmark. Nevertheless, my recommendation was for the OP in Bakersfield, CA, which is $0.14/kwh, I don't know how much he actually plays games so I just had to guess. Obviously if you're power is $23 per kw minute you might have to consider power draw.

I also love the bandwagon passion of people getting indignant about 120 watts of power for a $300 PURE LUXURY item. You think I'm living in the 19th century? Go power factor correct the inductive loads in your house, like your fridge and hvac blower motor, you'll probably save over 600 watts just from that. And turn the brightness down on your monitor. And quit buying luxury items like video cards and plastic.

Ahh perspective, you elusive lady.

As someone else said, CA has tiered rates and especially in the summer it can get brutal... 30+ cents/kWh.

Furthermore, I have said this over and over again on these forums, but idle power matters more than load power if you leave your PC on 24/7 or for much longer periods of time than you game. I crunched numbers recently and it's not a huge number, but it is recurring.. like $25-40 more than if you bought a 40nm card that offers equivalent performance in games, per year, every year, assuming 10 cents/kWh and a typical load profile w/ 80% efficient PSU.

Your other "point" is a false argument because you can easily buy a more power efficient card AND do all the other things you said. Or just not game at all or with with even lower-powered card, but I'm talking about efficiency purely, not conservation. It also takes very little effort to buy more efficient cards than to buy a new fridge, for instance.

Power draw also has additional effects such as possibly louder fans, and if you care, CO2 emissions and other pollutants from most power plants.

So yes power matters, especially to those who have high usage rates or high power rates or both. Like a car. If you drive a little, a cheap car may be a better option than a hybrid. But if you drive a lot, or if you pay $10/gal. for gasoline, or both, then a hybrid is probably more cost effective.

One size doesn't fit all. I think most people would agree with that.
 
Last edited:

aaksheytalwar

Diamond Member
Feb 17, 2012
3,389
0
76
IMO power draw is irrelevant except for needing. Better psu. But, when you notch up the Rez and not play at 16x10 the 570 is about. Equal and at 25x14 7850 is mostly faster. ATM u can't tell the difference btw a 7850 and 570 in a machine in most cases, but in the future 2gig will come in handy like where at skyrim 25x14 ultra 4x msaa 570 scores 39fps while 7850 57fps. While at 16x10 570 is faster and that is irrelevant

Because games 2 years hence will require more than 1.25 gig on 16x10 and then the 7850 will best the crap outta 570 even at that rez

And 2 years is just a figure of speech, it could even mean 6 months for a few games
 
Last edited:
Feb 19, 2009
10,457
10
76
So over a 2yr lifespan of typical GPUs, a 120W difference could end up saving a gamer ~$100 in electricity bill or more (if it makes the room hotter requiring more AC)?

Takes pricing in a whole different perspective doesn't it, when you nitpick over $20 or $50 differences.

Efficiency matters to those who understand maths and those who pay household bills. If you're neither of those, I guess it doesn't matter.
 

Arzachel

Senior member
Apr 7, 2011
903
76
91
I think you read the Civ 5 chart wrong. Gtx 570 handily beats the 7850 in that game as well. You also left out Portal 2. So out of 10 games the 7850 is noticeably slower in half of them while roughly tying the other 5. The gtx 560 448 is only 3% to 4% slower then the gtx 570. Given those results I would say its safe to assume the 7850 is slower at stock on average.

Yeah, you're right. Still, I wouldn't quite call Dirt 3 or Portal 2 a loss for the 7850, because both it and the GTX 570 easily hit over 60 fps at 1920x1200 on those games.
 

aaksheytalwar

Diamond Member
Feb 17, 2012
3,389
0
76
Yeah, you're right. Still, I wouldn't quite call Dirt 3 or Portal 2 a loss for the 7850, because both it and the GTX 570 easily hit over 60 fps at 1920x1200 on those games.

This.

1. There is no point comparing settings less than 1920x1080 maxed out or higher than that, AA may or may not be included depending on the case, the only exception being point 2

2. You need to compare only when one of the cards goes below 60 FPS or the other has a lead of at least 30-50% which too mostly matters only for 120 Hz.

3. When you take the above cases, especially at 25x14 especially with MSAA, for modern/intesive games a 7850 stock >= 570 stock.
 

kidsafe

Senior member
Jan 5, 2003
283
0
0
Yeah go for 7850 now or wait for the NV refresh.But it can be a long wait.I like the ASUS Direct CU II cards but that's a personal preference.
I don't see a point in waiting for the GTX 660 (GK106). If the following leaked specs are to be believed, then it won't be much better than a GTX 560 Ti.

28 nm, around 210 mm² die-area, Kepler architecture
Two Graphics Processing Clusters (GPCs), four Streaming Multiprocessors (SMXs)
768 CUDA cores
64 TMUs, 24 ROPs
192-bit wide GDDR5 memory interface, holding 1.5 GB or 2 GB memory
 

bigsnyder

Golden Member
Nov 4, 2004
1,568
2
81
I don't see a point in waiting for the GTX 660 (GK106). If the following leaked specs are to be believed, then it won't be much better than a GTX 560 Ti.

28 nm, around 210 mm² die-area, Kepler architecture
Two Graphics Processing Clusters (GPCs), four Streaming Multiprocessors (SMXs)
768 CUDA cores
64 TMUs, 24 ROPs
192-bit wide GDDR5 memory interface, holding 1.5 GB or 2 GB memory

Probably, but to play "devil's advocate" the 7800 specs did not inspire confidence that it would perform on par with the 6900 cards. Well we obviously know how that turned out ;)
 

aaksheytalwar

Diamond Member
Feb 17, 2012
3,389
0
76
A 7850 stock is like 10% faster than 6950 and about 5-10% slower than 6970. A 7850 OCED with stock volts is faster than 6970 and nearly as fast as 580. A highly oced 7850 on stock cooling is faster than a 580 and sometimes nearly as fast as a 7950.

A stock 7870 is about as fast as a stock 580 and once overclocked competes with stock 7950 or at times better.

The 7800 series isn't even a replacement of the 6900 series, yet it does a great job blowing them through the water. The only point you can make is that 6950 was like $250 about a year back and the improvement in performance considering overclocks has been about 25-35% for the same $. But this is still pretty good and considering AMD was literally throwing away their cards in the past, there is no reason to complain.
 

Concillian

Diamond Member
May 26, 2004
3,751
8
81
Probably, but to play "devil's advocate" the 7800 specs did not inspire confidence that it would perform on par with the 6900 cards. Well we obviously know how that turned out ;)

It didn't? Based on 79xx performance, the leaks allowed us to expect ~=69xx performance stock and huge overclocks.

The only surprise was that it ended up quite significantly better than 79xx performance per shader, but it also included less shaders than the leaks. Leaks had it at 1280 for the 7850 instead of 1024.

The GK106 is pretty efficient per shader and I don't expect the same surprise in efficiency per shader from Kepler. GK106 drops ROPs where Pitcairn did not. I find it very difficult to believe that GK106 can be competitive with 7850 clock for clock.

7970 vs. GTX680
32 ROP / 2048 shaders @ 925 vs. 32 ROP / 1536 shaders @ 1GHz

When clocked equal they're near equal with a slight advantage to GTX680.

7850 vs. leaked GK106
32 ROP / 1024 shaders @ 860 vs. 24 ROP / 768 shaders @ ?MHz

Exact same ratio of shaders at the two tiers... but we know that the 7850 is more efficient per shader (assumed due to ROP difference, possibly a cache architecture difference). The ROP advantage give the 7850 a leg up. Both are expected to OC similarly, so regardless of stock clocks, an enthusiast OCing will be using them at about the same clocks. AMD has majorly sandbagged the 7850 on stock clocks, so I'm sure nVidia will clock the GK106 where it needs to be in order to win stock vs. stock, but I think there's little chance the 7850 won't be the faster card when comparing OC vs. OC. At that point it comes down to price... like Ryan likes to say in his articles, there's no bad video cards, only bad prices.

I do expect the GK106 will come in at better power consumption than the 7850 though, and that could make the comparison more interesting.
 
Last edited: