Nvidia GTX 580 Reviews

[safan80]

GeForce GTX 580 3-way SLI gets tested

http://www.guru3d.com/article/geforc...ay-sli-review/


in the conclusion.

So then, in conclusion as always, this is x-factor stuff, extreme, nearly silly stuff to do and show. But people out there do purchase setups like shown today. At least you now know the pros and cons. Our recommendation is simple, if you have the hots for a multi-GPU setup based on the GeForce GTX 580, we recommend a 2-way SLI setup at best. With 3-way you'll run into heaps of curiosities like power consumption, noise levels and then the CPU bottlenecks - that last one is just sheer dreadful.

Flipside of the coin, 3-way SLI can be really helpful with 3D Stereoscopic gaming and a triple monitor gaming setup. The drivers, as always, are done well, we did not have any issues significant enough to mention, so from NVIDIA's side it's all good. But on a single monitor we do think that the infrastructure and sheer cash needed outweighs the extra performance you'll get for it in return. But that doesn't mean it's not tremendously cool stuff to play around with though.

Oh and 4-way SLI you wonder? Worse... that's 2-3% performance increase if you are lucky, with 3DMark Vantage as the exception. So yeah, stick to 2-way SLI we say -- unless you need to break 3DMark records of course

[Monstru]

On to your point. Have you seen my sig? Why would i go out and buy 2 Nvidia cards when ATI's new cards (Cayman) are just around the corner and I would be able to make a clear decision. I have brains to not go out and waste my money. Who in their right mind would paid for a ud9 and 4 580GTX when SLI does not scale well beyond 2 cards. You basically did the same thing that guru3d did 9 days before you with there 580GTX review
http://www.guru3d.com/article/geforce-gtx-580-review/15

honestly there was no new information in your review.
+1 exactly power consumption being one of those.

Sorry, what is the connection between your sig, 4WAY GTX 580 and my gaming article? I fail to see that connection. And what wasn't clear from the phrase "gamer's review".

Simply put, I have my launch article in the exact same day with Guru3D, thank you, and our overclocking study also in the same day, thank you again.... This is just a study for gamers featuring some of the latest games. Now sorry if I don't see COD Black OPS, Medal of Honor, NFS Hot Pursuit, Darksiders or any other new title in Guru3D's review. I am not saying they should be there, they are not in my launch article either. But in my gaming article I tested for the gamers, and strictly for the gamers (to make this clear, before I hear anything more about your sig, gamers with one VGA and a 24" display, which is most relevant to my readers) . So how can you compare two completely different things?

[Leeghoofd]

There is a huge difference between 1600x900 and 2650x1600 (about 2,656,000 pixels) also 30" displays have pretty much always been around $1200. Unless you're thinking about 1600x1200 which is pretty much a dead resolution.

darn I'm just testing cardsfor a review at 1280 x 1024 and 1600 x 1200 res Snooped around at a lan party and most "only" had a 19-24 inch monitor max

Will do the high end cards then at 1920 x 1200 too then... (max my 24" supports at home)

Pretty hard Safan80 to be original in reviews these days... we at shrimps are always at least a month later then the rest due to 1) lack of samples 2) taking our time when testing stuff ( at least motherboards ) and doing more debugging then reviewing p

[safan80]

Sorry, what is the connection between your sig, 4WAY GTX 580 and my gaming article? I fail to see that connection. And what wasn't clear from the phrase "gamer's review".


.... in my gaming article I tested for the gamers, and strictly for the gamers (to make this clear, before I hear anything more about your sig, gamers with one VGA and a 24" display, which is most relevant to my readers) . So how can you compare two completely different things?

Hardly any game scales in 4way SLI, only benchers use that (for 3dmark, vantage, etc). I was responding to your comment

Now as far as my Analitycs satistics tell me, and as far as Steam statistics tell me, very few people have 30" monitors.

I thought about it like this:

1920x1200 with everything maxed out, UD7 + i950, HD 5970 or GTX 580 = tipical high-end machine

2560x1600, UD9, i980X, 2xGTX 580 or more = extreme enthusiast machine

That is the connection that you've failed to make. Not every gamer uses steam. Most gamers will not pay for something they will not even use like a UD9 because it is not practical. There are a lot of people who game at 30" and they will buy the game disc rather than buy it off of steam. you can check out places like here or even better http://www.widescreengamingforum.com/. You created two categories by which might be true somewhere, but it most cases are not. the second category is a typical of a bencher setup minus the high resolution. Benchers in most cases do not use a 30" monitor in their systems they tend to use a max of 1920x1200. I find it insulting because The article is listed as a gamers test, however, rather than be thorough and test both of the 2 high end resolutions you just dismiss 2560x1600 altogether by making up a category to suit your reasoning..

from the link that you posted here

Certainly there are cases where we have only a graphics card with 6-7 days before launch.

In that time you could easily tested at 2560x1600 and 1920x1200. It is not like you need to change the monitor to run benchmarks. If you were smart you would of already done all the benchmarks with the 5970 before hand or you could of just waited and ran all the benchmarks on the GTX 580 and tested the 5970 after you had to give the GTX580 up. All in all it article was not released at launch so you could of taken a little more time. You released this article 9 days after launch. Your 3 page conclusion does not even touch on the impact the difference in video memory between the two cards. You would noticed that had you tested 2560x1600, anandtech managed to in their 580GTX SLI review. All in all I think it would be best if you worked on time management and thoroughly investigated as much as you can because their are cases where certain games are clearly cpu limited and the game had nothing to with the video card used. it should be about quality not quantity. It does n

Pretty hard Safan80 to be original in reviews these days... we at shrimps are always at least a month later then the rest due to 1) lack of samples 2) taking our time when testing stuff ( at least motherboards ) and doing more debugging then reviewing p

and do less vantage and 3dmark Seriously reviewers have it their heads that they need to rush with the numbers rather than be thorough and it shows. I don't even look at a review any more that is missing 2560x1600 because 1920x1200 is cpu limited in most cases.

[Leeghoofd]

That's one of the reasons, due to my monitor limit, that I have withdrawn myself from multiple GPU setups ( besides 4870X2 and GTX 295 ) CPu power is there, but can't get over 1920 x 1200 res...

[Iconyu]

darn I'm just testing cardsfor a review at 1280 x 1024 and 1600 x 1200 res Snooped around at a lan party and most "only" had a 19-24 inch monitor max

Will do the high end cards then at 1920 x 1200 too then... (max my 24" supports at home)

Pretty hard Safan80 to be original in reviews these days... we at shrimps are always at least a month later then the rest due to 1) lack of samples 2) taking our time when testing stuff ( at least motherboards ) and doing more debugging then reviewing p

1280x1024 is too low, a few games run at that resolution on a high end GPU will probably run into a CPU wall. 1080p monitors are everywhere, and 1920x1200 are more the choice of the professional user as when reading documents you really miss those extra pixels.

I think more of the older PC gamers that went from 1600x1200 to 1920x1200, before 1080p became so popular, are slowly getting more of the old 16:10 monitors back into circulation, by voting with their wallet. But 4:3 monitors are really hard to find, any reviews done on them would be appreciated by a small audience.

[cold2010]

xbitlabs Review

http://www.xbitlabs.com/articles/vid...e-gtx-580.html

[Leeghoofd]

1280x1024 is too low, a few games run at that resolution on a high end GPU will probably run into a CPU wall. 1080p monitors are everywhere, and 1920x1200 are more the choice of the professional user as when reading documents you really miss those extra pixels.

I think more of the older PC gamers that went from 1600x1200 to 1920x1200, before 1080p became so popular, are slowly getting more of the old 16:10 monitors back into circulation, by voting with their wallet. But 4:3 monitors are really hard to find, any reviews done on them would be appreciated by a small audience.

Thx for the input, I'm mainly doing CPU scaling tests 3 - 3.6 - 4.2 and I needed to retest some older cards to to get som more data ( GTX 260 is the lowest ) , even the 5770 struggles already at 1600 x 1200 res... Really fun to see how DX10 made a difference in some games, DX11 brings most cards on their knees...

The AA war and co is done by the main editor of the website...

Loads of GPU's inbound

[Monstru]

Hardly any game scales in 4way SLI, only benchers use that (for 3dmark, vantage, etc). I was responding to your comment

This is totally true, I didn't say otherwise.

That is the connection that you've failed to make. Not every gamer uses steam. Most gamers will not pay for something they will not even use like a UD9 because it is not practical. There are a lot of people who game at 30" and they will buy the game disc rather than buy it off of steam. [...] I find it insulting because The article is listed as a gamers test, however, rather than be thorough and test both of the 2 high end resolutions you just dismiss 2560x1600 altogether by making up a category to suit your reasoning..

No I do not fail to make, I just don't see the point. You have a 30" monitor. I do too. So what? That makes us a majority or a special category? Unfortunately, based on solid statistics, that makes us a special category, not a majority. No, there are not a lot of people using a 30" display. In case you didn't realise, Stam statistics don't mean people who buy their game off Steam, but people who have games that require Steam as authentification. 30 milion of them. Not too many 30" display there. Sorry, I rather base on solid statistics than a few opinions on a forum. I also take into account my Analytics stats, which show me exactly what my readers are using.

I do not make 2 categories to better suit me, you have to make some sort of categories in order to find the best testing method for a typical crowd. 30" is not a typical crowd, even if you feel insulted byt this or not.

In that time you could easily tested at 2560x1600 and 1920x1200. It is not like you need to change the monitor to run benchmarks. If you were smart you would of already done all the benchmarks with the 5970 before hand or you could of just waited and ran all the benchmarks on the GTX 580 and tested the 5970 after you had to give the GTX580 up. All in all it article was not released at launch so you could of taken a little more time. You released this article 9 days after launch. Your 3 page conclusion does not even touch on the impact the difference in video memory between the two cards. You would noticed that had you tested 2560x1600, anandtech managed to in their 580GTX SLI review. All in all I think it would be best if you worked on time management and thoroughly investigated as much as you can because their are cases where certain games are clearly cpu limited and the game had nothing to with the video card used. it should be about quality not quantity. It does n

I am very, very sorry that you are totally MISSING THE POINT. I had 2 launch articles, with all the resolutions you need, with higher levels of overclocking then anyone else, at the launch day, in the same time frame as everybody, with the latest drivers, with 2 AA settings, with benchmarks for overclockers also, not just gamers, and so on. This is a different article with a different target. No, the vast majority of the gamers do not own a 30" display. I base this on statistics, you base the opposite on your opinion. Well, you are entitled to your opinion, but that does not change the way I work.

Seriously reviewers have it their heads that they need to rush with the numbers rather than be thorough and it shows.

If this is what you understand by comparing several artciles over the web, including ours, then I am sorry that we cannot futther communicate. Yes some reviewers do, and some don't , and the differences are obvious. What are you looking for, and where you are looking for it, it is something that I am obviously missing, because your conclusion are totally opposite from what feedback I am usually getting. If you are unable to comprehend the differences between different articles done with a different purpose in a different timeline, and the only thing you can come back to m with over and over again is the lack of your resolution, from a test that obviously has something to do with a typical crowd, not you, and the fact that this test, unlike the launch article, was published 9 days later, then what can I say, sorry for not testing for your specific configuration.

[Monstru]

Here, I took some time to dig some data from Analitycs. These are the guys reading my website, and they are prety much the enthusiast cream aroung here (GTX 480 SLi, GTX 580 SLI, i980X, etc). Of course, I will not test in 1280x1024 just because more people use that res. But I will not consider 2560x1600 as beeing relevant for a specific scenario gaming test, when only 0.26% of my readers use that res. And yes, I think that it is very important to use that res in the launch article for high-end cards. Now, you can say whatever you want, but untill I do not see the same graph from TechpowerUP, Guru3D, Anandtech or any other hardware testing site, this is the crowd I am writing for.

[W1zzard]

here's ours

[Manicdan]

the steam survey dosnt provide the different resolutions for a single game does it? like we couldnt see what res people played Black Ops at? i think that would be most useful. i know alot of people probably read these reviews while at work or school or on their laptop and not their gaming desktop.

[tajoh111]

Here, I took some time to dig some data from Analitycs. These are the guys reading my website, and they are prety much the enthusiast cream aroung here (GTX 480 SLi, GTX 580 SLI, i980X, etc). Of course, I will not test in 1280x1024 just because more people use that res. But I will not consider 2560x1600 as beeing relevant for a specific scenario gaming test, when only 0.26% of my readers use that res. And yes, I think that it is very important to use that res in the launch article for high-end cards. Now, you can say whatever you want, but untill I do not see the same graph from TechpowerUP, Guru3D, Anandtech or any other hardware testing site, this is the crowd I am writing for.

http://www.anandtech.com/show/4025/h...ilders-guide/8

Hardware Component Price
Processor Intel Core i7-950 Bloomfield 45nm
(Quad-core + HTT, 3.06 to 3.33 GHz, 8MB L3, 130W) $295
Motherboard EVGA X58 SLI3 (Intel X58+ICH10R, LGA1366) $210
Video EVGA GeForce GTX 580 1.5GB $510
Alt. Video Sapphire Radeon HD 5970 2GB $460
Memory Corsair XMS3 6GB (3x2GB) DDR3 1600
(TR3X6G1600C9) $120
Hard Drive Western Digital Caviar Black 1.5TB (WD1501FASS) $120
SSD OCZ Vertex 2 120GB (-$30 MIR) $205
Optical Drive LG 22X DVDRW (GH22LS50) $17
Sound Card Creative SoundBlaster X-Fi Titanium $80
Case Antec P183 Black Aluminum $155
HSF Tuniq Tower 120 Extreme $65
Power Supply Antec TPQ-850 (850W) $150
Monitor HP ZR24W (24" 1920x1200, S-IPS) $400
Alt. Monitor ASUS VG236H w/3D Kit (23" 1920x1080 TN, 120Hz) $475
Total System Price (Range based on GPU and LCD choice) $2277-$2402

While this is a high-end system, 30" monitors are in a price category of their own so that leaves us with 24" LCDs.

[Monstru]

here's ours

Very usefull info Wizzard, thank you for sharing. You have a very good penetration of the 1920x1080 monitors (which is how it should be, if you ask me, this type of panels probably have a very competitive price/quality ratio) and the old 1280x1024 is slowly backing off. 1920x1200 and 1024x768 are very similar between our readers I see, and your 2560x1600 numbers are almost double, meaning 0.51% compared to my 0.29%, so still a very, very small crowd. Well, if any other sites would be willing to share, I guess we could get to a very nice statistic after all.

While this is a high-end system, 30" monitors are in a price category of their own so that leaves us with 24" LCDs.

You are totally right, and that is exactly what I am trying to say. That is why I was thinking about that type of categories above. Many people may be willing to spend their money on one GTX 580, but only a few of those would spend their money on a 30" display. On the other hand, he who pays for a GTX580 will most likely have a 24" monitor. The same ideea applies to the 2560x1600 crowd. If you spend your money on 2 or 3 GTX 580, you can afford a 30" panel and you most likely should have that in order get the best bang for your bucks.

Are there exceptions? Hell yeah, but they don't make the rule, and unfortunately, you can never have a review that makes absolutley everybody happy.

[safan80]

Here, I took some time to dig some data from Analitycs. These are the guys reading my website, and they are prety much the enthusiast cream aroung here (GTX 480 SLi, GTX 580 SLI, i980X, etc). Of course, I will not test in 1280x1024 just because more people use that res. But I will not consider 2560x1600 as beeing relevant for a specific scenario gaming test, when only 0.26% of my readers use that res. And yes, I think that it is very important to use that res in the launch article for high-end cards. Now, you can say whatever you want, but untill I do not see the same graph from TechpowerUP, Guru3D, Anandtech or any other hardware testing site, this is the crowd I am writing for.

Monstru
You've made your point. Stats always win me over It looks like more people browse your site with their HDTVs than anything else. I should thank you for not testing in 1080p. Maybe you could use high levels of AA with edge detect. most cards can do 4xAA without any problem now a days. I think I've seen something like this from Nvnews.

here's ours

Thank you guys for this.. I think I'm most shock by the amount of readers using 1920x1080 and I've noticed a few review sites using this res. 1080p is double the 1920x1200 userbase This points to most users going cheap on their monitors. I'm not really counting the lower resolutions in these because they are probably school computers, laptops, 1024x600 is netbooks, and other mobile devices.

[Monstru]

You've made your point. Stats always win me over It looks like more people browse your site with their HDTVs than anything else.

I don't think they are really HDTV, but mostly 1920x1080 monitors. Thing is, a normal person, like you or me, would not pay much attention to this. And then let's say you want to buy a new monitor, let's say some 22-24" regular stuff. And...maybe you want 1920x1200 (I feel very comfortable working on this setting, for instance). Than you have a surprise...a ton of 1920x1080 monitors starting from a very low price-point. The 1920x1200 kick in only after a price limit, and probably most of the guys go for the cheap stuff. I know the prices from my country, or from EU space are not relevant for US users, but in my country, the cheapest 1920x1080 monitor costs ~ 200$. The cheapest 1920x1200 monitor costs ~300$. Our local "NewEgg kind of etailer" has 121 1080 models in the offer, and 14 1200 models in the same offer. And...don't shoot me for saying this, I know this is an abomination, but the cheapest 2560x1600 (LG-W3000H-BN) is....1650$. If I would tell you how much is the Dell we use for testing, in my country, I would even consider shooting myself

[Sam_oslo]

I don't think they are really HDTV, but mostly 1920x1080 monitors. Thing is, a normal person, like you or me, would not pay much attention to this. And then let's say you want to buy a new monitor, let's say some 22-24" regular stuff. And...maybe you want 1920x1200 (I feel very comfortable working on this setting, for instance). Than you have a surprise...a ton of 1920x1080 monitors starting from a very low price-point. The 1920x1200 kick in only after a price limit, and probably most of the guys go for the cheap stuff. I know the prices from my country, or from EU space are not relevant for US users, but in my country, the cheapest 1920x1080 monitor costs ~ 200$. The cheapest 1920x1200 monitor costs ~300$. Our local "NewEgg kind of etailer" has 121 1080 models in the offer, and 14 1200 models in the same offer. And...don't shoot me for saying this, I know this is an abomination, but the cheapest 2560x1600 (LG-W3000H-BN) is....1650$. If I would tell you how much is the Dell we use for testing, in my country, I would even consider shooting myself

Yep, they are 1920x1080 monitors.

I've been looking for a new monitor recently, and did a lot of research actually. Almost all new monitors are 1920x1080 these days. So is my new NEC EA231WMi that I finally got. BTW, it got a great ISP-panel with a really good price.

[duploxxx]

Yep, they are 1920x1080 monitors.

I've been looking for a new monitor recently, and did a lot of research actually. Almost all new monitors are 1920x1080 these days. So is my new NEC EA231WMi that I finally got. BTW, it got a great ISP-panel with a really good price.

yeah and i don't like that. both my laptops i have 1920*1200 and my game rig is 1600*1200. I really need the 1200 base for the remote desktops, that way I can rdp any server with 1280*1024res. chocked by the fact that my game rig res is only used by 1.1% of people.

high amount of 1280*1024 is still because of the time the first tft were introduced on 15" and 17" many used that resolution and it was hot/cool to replace the huge screens at that point in time.

[safan80]

I don't think they are really HDTV, but mostly 1920x1080 monitors. Thing is, a normal person, like you or me, would not pay much attention to this. And then let's say you want to buy a new monitor, let's say some 22-24" regular stuff. And...maybe you want 1920x1200 (I feel very comfortable working on this setting, for instance). Than you have a surprise...a ton of 1920x1080 monitors starting from a very low price-point. The 1920x1200 kick in only after a price limit, and probably most of the guys go for the cheap stuff. I know the prices from my country, or from EU space are not relevant for US users, but in my country, the cheapest 1920x1080 monitor costs ~ 200$. The cheapest 1920x1200 monitor costs ~300$. Our local "NewEgg kind of etailer" has 121 1080 models in the offer, and 14 1200 models in the same offer. And...don't shoot me for saying this, I know this is an abomination, but the cheapest 2560x1600 (LG-W3000H-BN) is....1650$. If I would tell you how much is the Dell we use for testing, in my country, I would even consider shooting myself

Nah ... I only would of resorted to shooting had you tested 1280x1024

Since you've had your review up, how much of your traffic has been local v global? I've just been a little pissed when it has comes to these new 1080p computer monitors coming out, it feels like a step backwards going from 4:3 -->16:10 and now --> 16:9. One site that I don't like anymore and is suppose to be from gamers, Firingsquad, removed 1920x1200 and put in 1920x1080 in its place in their review. I'm still holding out for the 3840x2400, but it looks like it might be replaced by 3840x2160 before it makes it to the masses.

[Monstru]

The traffic is balanced, but of course mostly local, since the language is not English.

Here is my last GTX 580 article. Oh boy, I am finnaly done with GTX 580!

Asus GTX 580 - GTX 580 vs SLI vs 3SLI vs 4WaySLI | lab501 | Google Translate

[safan80]

The traffic is balanced, but of course mostly local, since the language is not English.

Here is my last GTX 580 article. Oh boy, I am finnaly done with GTX 580!

Asus GTX 580 - GTX 580 vs SLI vs 3SLI vs 4WaySLI | lab501 | Google Translate

Very cool, just one problem... The graphs have too similar of colors. Could you please change those? I can't tell which is which

[Monstru]

Ah, this time I thought it would not be a problem if I use the same colour, since you have 1,2,3,4 cards in the same order

[ajaidev]

Interesting bit:





http://en.hardspell.com/doc/enshowco...38&pageid=7656

6850 CF/5970 < GTX 580 "Loaded" and 6850 CF < GTX 580 "Stand by"