GTX580 reviews thread

Page 17 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Skurge

Diamond Member
Aug 17, 2009
5,195
1
71
Um, you have the HAWX2 and Uniengine included... Not to be mean but once you remove those two benchmarks, you receive a drastically different number.

Yeah, might as well thow Final Fantasy XI in there where the 480 gets less then 2FPS too. Or FFXIV where its slower than a 5850.

I see what you are trying to say, realistically the 5870 isnt 40% slower than the 580. Because of ONE game and ONE benchmark, it heavily skews the results. But look at it how you like.
 
Last edited:

cusideabelincoln

Diamond Member
Aug 3, 2008
3,275
46
91
Um, you have the HAWX2 and Uniengine included... Not to be mean but once you remove those two benchmarks, you receive a drastically different number.

I knew this would be an issue, but I decided to keep the results anyway. Including them gives you the "worst case scenario" for the HD 5870 when compared to the 580, and as such I think it gives us a good estimate to hope what the 6970 can do. If the 6970 is indeed 38% faster than the 5870 across various games while being about 50% faster in DX11 subset of games, then the majority of the time the 6970 should be faster than the 580, since the extreme HAWX2 numbers and the vast majority of Unigine numbers (since almost every site tested it) do skew the average in favor of the GTX 580. Without those numbers the 580 would have a smaller performance advantage, generally-speaking.

Also when looking through the numbers it's easy to spot the outliers. You can download this spreadsheet for yourself and simply delete those outliers to find.

I can see the problem with including Unigine scores, since it isn't actually a game that you can play while everything else is. So for your consideration sheet 3 will throw out those results (just give me a second to do it). However HAWX 2 will stay, since it is a game, and there's only a couple of results which really won't affect the average that much.
 
Last edited:

bryanW1995

Lifer
May 22, 2007
11,144
32
91
This is what happens when a review actually includes demanding games instead of Medal of Honor and F1 (HardOCP) or Call of Juarez 2, Dawn of War 2, Unreal Tournament 3 (TPU). Computerbase arrives at GTX480 and GTX580 as 36% and 58% faster, respectively, than the HD5870 at 2560x1600 8AA. I am sure Kyle Benett doesn't agree with that review. Computerbase shows DX9 vs. DX10 vs. DX11 performance averages too.

it will be interesting to see what 6970 looks like at 25x16 if it is really 2gb. 1 gb on a high end card is just not enough any more.
 

bryanW1995

Lifer
May 22, 2007
11,144
32
91
It is. HardOCP / Brent Justice are the real deal. I would honestly bet it's the other sites with malevolent intentions, rather than [H]. There are certainly obviously biased sites out there, both ways, than HardOCP.

1. AT
2. Computerbase
3. Hardware Canucks


show me bias in any of those.

It's much better to look at many review sites instead of claiming "XYZ is the BESTEST evAr!!!"
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
1. AT
2. Computerbase
3. Hardware Canucks


show me bias in any of those.

It's much better to look at many review sites instead of claiming "XYZ is the BESTEST evAr!!!"

Most review sites aren't very good. It's a common distribution in life: a few elite, some good, and a TON of mediocre or worse. No sense in wallowing through the crap, just go straight for the best.
 

bryanW1995

Lifer
May 22, 2007
11,144
32
91
He was right about JHH lying to the public.
He was right about card not arriving in the Fall.
He was right about JHH showing a fake wooden card to fool investors.
He was right about He was right about Fermi not arriving in December.
He was right about the multiple, endless respins.
He was right about the design defects that caused the long-long delay.
He was right about Fermi not arriving in January-February.
He was right about Fermi not coming in 512 cores flavor.
He was right about Fermi being unmanufacturable with original clocks and cores.
He was right about Fermi being late, even beyond March.
He was right about Fermi being out-of spec both in terms of power and thermal with its original clocks and specs.
He was right about having horrible yields.
He was right about Fermi being cut down to 4xx cores to make it manufacturable.
He was right about lowered clocks to make it manufacturable and standard-compliant -> marketable whatosever.
He was right about the final shipping date.

Did I miss something..?
I predict that the temperature will be >= 70f tomorrow.

I predict that the temperature will be <70f tomorrow.

I sure hope I'm right!
 

cusideabelincoln

Diamond Member
Aug 3, 2008
3,275
46
91
it will be interesting to see what 6970 looks like at 25x16 if it is really 2gb. 1 gb on a high end card is just not enough any more.

It definitely needs more than 1GB at that resolution. Maybe AMD will release two versions, like the 4870, but they definitely need a +1GB card for anything faster than an HD5870.

Here are a few examples of 1GB not being enough:
Starcraft 2: 2560x1600 8xAA
Crysis Warhead: 1920x1200 8xAA, 2560x1600 4xAA and 8xAA modes
F1 2010: 2560x1600 8xAA
Metro: 1920x1200 8xAA

Crysis, Starcraft 2, and F1 2010 aren't just vain victories. The 5870 GPU has enough horsepower for these games, but at 8xAA the framebuffer becomes a problem. So it's reasonable to assume a faster GPU will need an appropriate framebuffer to make sure it doesn't hit a framerate limit unnecessarily.
 
Feb 19, 2009
10,457
10
76
Does MLAA need the extra vram like 8xAA does at high res?

From what i see, the image quality is pretty good *when it works*.
 
Feb 19, 2009
10,457
10
76
Nothing wrong with including a popular game even if its skewed. IF AMD couldn't be arsed to payout developers to push their tech, then they deserve to suffer looking silly when these games are released.

Non-gaming benchmarks however, aren't very relevant.
 

Slaimus

Senior member
Sep 24, 2000
985
0
76
What really surprises me is that the GF110 is A1 revision. Usually graphics cards get released at A2 revision (or A3 in the case of the problematic GF100).

It looks like they got it right ahead of time, which explains the earlier than expected launch.
 

Paratus

Lifer
Jun 4, 2004
17,408
15,253
146
My understanding of MLAA is that since it's shader basef you need very little VRAM.
 

Obsoleet

Platinum Member
Oct 2, 2007
2,181
1
0
You can call me whatever you want, seeing as how when I first joined the forums I was accused of being an NV fanboy, then an AMD fanboy, now an NV fanboy again apparently. Whatever, I think I am one of the more impartial and consistent people around here.

I don't think the HDMI thing is a big deal because who actually uses a GTX580 for HTPC usage, anyway? And if it's not going into an HTPC then you have other ways to get that audio.

I would use one for my HTPC.. if you think about it, you can use your main PC as a HTPC.
I only have to maintain 1 computer for my HTPC + gaming + office work. I do this by running a 50' HDMI cable from my PC to my TV, and using media streamers like a $99 AppleTV when I don't need to dedicate the PC (change the display output to the TV) to a task on the TV (iTunes library which is videos/mp3s). Using something like an AppleTV, someone can be gaming in the office, while someone else streams movies from the PC.. mine is plenty powerful for that and effects nothing.

I've played Starcraft 2 in 1920x1200 with AA/AF, had a fullscreen 1080P movie on my 2nd monitor.. and streamed a movie to my TV at the same time.. (the 5870 does very well with the latest GPU-accelerated VLC Player, making it silky smooth even with a game playing).

When I'm in the mood, I'll play a game on my TV with my "HTPC" remaining in the office, using Microsoft Xbox 360 pads.. they have 30' range as they use Bluetooth. Great for games like Overlord, Dirt2, whatever.

Dirt2 on a 50" plasma in 1080P is not even comparable to a PS3 or Xbox360. Keeping 1 desktop PC in the house like this, allows you to justify stuff you would never buy otherwise if you had multiple rigs.

Using my PC in the office + HTPC is how I personally justified my SSD, 5870 ect. I would have never bought that kind of stuff before I had this justification. The PC gets used far more than most people who have a HTPC + desktop (good return on investment).. and I end up with a fairly powerful HTPC (the SSD helps tremendously on a HTPC), and a powerful PC (in it's day, it's still pretty decent in my book or I'd upgrade it).

I wanted the 4800 series, as I had a 8800GTX at the time.. I was well-aware that the 4870/4890 was the BEST product on the market at the time. I was just waiting to see what was next and then came the 5870.. which was very.. very obviously a knockout punch to NV's glass jaw. Sites that didn't acknowledge it should be shutdown.

My 5870 with it's HDMI audio has been a HUGE boost to my entertainment setup.

The bigger issue imho is lack of single-GPU Surround. It was the biggest reason why I switched my order from a GTX460 to a HD6850 after my initial GTX4604-768MB order arrived defective (damaged PCB).

NV's Surround software also sucks compared to AMD's hotkey system where it's easy as pie to switch modes. This is a big deal to me because of all the switching I do:

a) single-monitor centered (side panels disabled)
b) tri-monitor Extended Desktop centered
c) tri-monitor Eyefinity
d) tri-monitor Eyefinity with bezel correction

I understand that there are a few hacks like Zotac's 460 that does triple-monitors up to 1600x1200 but that's just a hack. Hopefully Kepler will ship with full single-GPU Surround support, full HDMI 1.4 support, and Surround software with hotkey profiles.

I didn't mean anything to be directed towards you, it was really just a statement in your direction. No offense intended!
Yeah, Eyefinity is a HUGE feature+.

That's the thing.
An unbiased person would admit from the 5870 launch that it was ****ing phenomenal.
20 watt idle, low loads, triple monitor on 1 card, dx11, audio over HDMI, audio decoding widgets (I don't use the 7.1 stuff), and it was/is FAST... for $380 shipped on launch day!

hating on the 5xxxx's is essentially incomprehensible. And going for Nvidia over them is simply retarded. In the GTX285's day, sure, that was a tough choice over a 4870/4890 but those days are over, and have been over.
It's just a PR campaign to stem the tide towards AMD, until NV releases a definite superior product.

I just shake my head, because you really did and do have to be a diehard NV fanboy to resist the 5xxxx's and the new 6xxxx series. They truly are the best product on the market. Toss in NV's corporate shenanigans that have left a very poor impression with me (Intel's Insides et all ect ad nauseum) and I'm not a huge fan unless they have a head-and-shoulders superior product.

They do not. Not even close. They run fast, but beyond that it's all AMD.. and AMD's stuff is just as fast!
There are a few people on this forum who have vested interests in NV, some are affiliated with Nvidia, others do it for free (as far as I can tell) and there's a massive propaganda campaign against the truth on what the best product currently really is.
It's AMD since the 4800s (possibly debatable), lets quit kidding around. No doubt since Evergreen and Fermi what was best. Fermi is a friggin mess. All people rant about is their tessellation which helps them with nothing in games, what we use these for.

Epitome of a fanboy post, guarantee a product not even released!.....You do have a point regarding no proper audio over HDMI though....what is the point of the HDMI port if its only good for video?

The 6800s are an evolution of the 5800s, of course they are going to be as amazing and more than the 5800s. At very least, any educated consumer would know to not buy a 580 and wait for the 6970 launch. You CANT mess up something as good as the 5870. If they do, I'll eat a hat.. but the 6870 itself is pretty good already. 6970 is going to destroy.

Again, the 580 doesn't have audio over HDMI??? I thought NV fixed that finally with the 470/480 launch. That, would be simply unacceptable and a big shame.

Then I heard ATIs don't get along with Onkyo, which I just bought a set...

Sad to hear that man. I've learned my lesson on things like this long ago, it takes so much retardedly meticulous research when piecing together such equipment, it's maddening.. but it's required beforehand to avoid that sort of situation. All products have their own little mishaps/flaws.

1. AT
2. Computerbase
3. Hardware Canucks


show me bias in any of those.

It's much better to look at many review sites instead of claiming "XYZ is the BESTEST evAr!!!"

AT I've seen a ton of stuff in the past. I don't read them as much anymore because of it. I'll look at their conclusions, but I don't take them as seriously as others. I dont have a bunch of stuff onhand, as I don't document bias, but when I see it I believe it and will point out the next thing I read from the listed sites that I find fishy.

For one, a lot of sites downplayed the 5800 launch a little much for my taste.
For my $380 (shipped) on launch, my 5870 has been the best feature-packed and reliable video card I've ever owned.. and my first 3d card was a 3dfx Voodoo1 so I've been around.
 

formeremployee

Junior Member
Nov 10, 2010
3
0
0
1. AT
2. Computerbase
3. Hardware Canucks


show me bias in any of those.

It's much better to look at many review sites instead of claiming "XYZ is the BESTEST evAr!!!"


Hardware Canucks was founded by the company behind NCIX/NCIXUS/DirectCanada. Their primary reviewer, SKYMTL, was doing reviews for the NCIX forums before being offered the position with HWC. Originally, most of the hardware tested by SKYMTL came from NCIX.

When there's a direct relationship between an e-retailer and review site, impartiality should be questioned as the ultimate goal is to drive hardware sales.

NCIX also has a very personal relationship with NVIDIA and Samsung, and to a lesser extent, Intel.
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
Hardware Canucks was founded by the company behind NCIX/NCIXUS/DirectCanada. Their primary reviewer, SKYMTL, was doing reviews for the NCIX forums before being offered the position with HWC. Originally, most of the hardware tested by SKYMTL came from NCIX.

When there's a direct relationship between an e-retailer and review site, impartiality should be questioned as the ultimate goal is to drive hardware sales.

NCIX also has a very personal relationship with NVIDIA and Samsung, and to a lesser extent, Intel.

This doesn't surprise me as I've followed HWC. I don't think he's necessarily intentionally biased, but the possibility for subconscious bias is there. Also, HWC fails my quick and dirty check of a site's integrity--which is to see if the site included Far Cry 2 when benching GTS 450. (The test is not an absolute test and returns many false positives, as I'll explain below.)

If no, that site is *definitely* legit. (The short list includes AT, HardOCP, and Bit-Tech. Tech Report too but I despise their inconsistent and sometimes amateurish reviews where they do some pretty weird stuff like only benching at one setting at one resolution, etc.) AT and HardOCP in particular have demonstrated repeatedly that they can't be bought.

If yes, the site could still be legit (e.g., they just happen to test a ton of games, including FC2; or maybe the site is just plain old, like TPU benching some downright ancient games in its suite).

However, FC2 is THE most NV-skewed game and isn't even played much (it didn't rank high in AT and HardOCP reader polls of games that readers wanted tested), and there are so many newer and better games to bench than that, it makes me wonder. FC2 seems to ever get benched when NV releases cards, apparently because it's recommended in the reviewer guide. If a site is so out of touch or plain biased as to include FC2, especially in a small group of games where even one outlier can really skew the average, then there is at least the possibility of bias.

This is especially true if the review site is smaller, which means it has less clout and relies more on the vendor supplying equipment to review in the future. Someone on here talked about this effect in the bike industry, where smaller publications soft-pedal negatives more, whereas the larger publications could afford to be more blunt in their reviews. (Back in the IT realm, HardOCP has popularity in part because can be downright antagonistic in reviews, giving them more street cred and readers.)

It appears however that NV has given up pushing FC2 because I don't see it benched in the GTX580 reviews as much. I think this is because NV knows that Metro 2033 is also skewed but a MUCH better and defensible choice as it is a new game using DX11, and the skew is much more legitimate (highlighting their tessellation strengths). Thus I have no problem with Metro 2033 used as a bench despite the skew.

So I will need to find another quick and dirty test now. Unfortunately there hasn't been a game out yet with the same combination of skew + outdatedness that FC2 had. If there were an outdated game that skewed heavily towards AMD hardware, for instance, that'd be a great test, in my eyes.
 
Last edited:

Will Robinson

Golden Member
Dec 19, 2009
1,408
0
0
Does MLAA need the extra vram like 8xAA does at high res?

From what i see, the image quality is pretty good *when it works*.
I think they should fix the blurring problem that MLAA has with text and numbers before they trumpet it as a new feature.Until that's fixed it's not ready for prime time.
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
I think they should fix the blurring problem that MLAA has with text and numbers before they trumpet it as a new feature.Until that's fixed it's not ready for prime time.

From what I've seen, MLAA might be good for some games but makes others look awful, and in any case I'm sticking with good ol' MSAA. They should have redirected whatever resources they put into MLAA into something more useful.
 

Borealis7

Platinum Member
Oct 19, 2006
2,901
205
106
TechPowerUp SLI review fail. GTX580 SLI vs only single cards (including 5970).
and of course the SLI wins everything except one game where SLI must be broken.
 

Aristotelian

Golden Member
Jan 30, 2010
1,246
11
76
Hardware Canucks was founded by the company behind NCIX/NCIXUS/DirectCanada. Their primary reviewer, SKYMTL, was doing reviews for the NCIX forums before being offered the position with HWC. Originally, most of the hardware tested by SKYMTL came from NCIX.

When there's a direct relationship between an e-retailer and review site, impartiality should be questioned as the ultimate goal is to drive hardware sales.

NCIX also has a very personal relationship with NVIDIA and Samsung, and to a lesser extent, Intel.

Welcome to the forums, formeremployee! (Yoink, Idontcare!)

It's interesting to see a post like yours that looks at the relationships between review sites and retailers. Maybe some further analysis here will explain disparities in quality and numbers between benchmarking sites.
 

Dribble

Platinum Member
Aug 9, 2005
2,076
611
136
Careful with using Computerbase.de.

They are using Catalyst AI OFF for the 5870. They use Catalyst AI: Standard for all the other RV8x0 though.

That will be a few percent off compared to other reviews.

Additionally they are using the texture detail level for the NVIDIA cards in Quality as opposed to High Quality some other reviewers use.

I'm not sure what AT uses. I guess it might be driver default.

That's because computerbase did more detailed image analysis then anandtech or [H] or a number of other reputable review sites and worked out that ati had nerfed image quality in their latest drivers to put scores up. Hence they adjusted so the radeons had the same image quality as the geforces.

Their scores are more accurate, not less.
 

Arkadrel

Diamond Member
Oct 19, 2010
3,681
2
0
I think they should fix the blurring problem that MLAA has with text and numbers before they trumpet it as a new feature.Until that's fixed it's not ready for prime time.


You have to remember that in many cases MLAA is only used when you cant use anything else.

check this out:
http://www.iryokufx.com/mlaa/gallery/index.html

You take your mouse cusor over the "no AA" and the "MLAA" and compair.

it has pictures with NO AA and with MLAA from games like:


exsample:

No AA:


With MLAA:
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |