Jump to content

Multi-GPU Performance Evaluation


Nemo

Recommended Posts

I completely understand why he would prefer the green team over the red team (driver issues alone). It's not that ATi engineers and manufacturing facilities produce inferior products, it's just that it takes them so long to release suitable drivers for their hardware.

 

Those of us that live in the real world don't need driver updates every 3 days. If you're a Hardcore gamer or Benchmark Junkie then yes I can see it, but this is what, 1 or 2% of the population? ATI nor any other company is going to cater to this. Ok so maybe nVidia does by releasing drivers every few hours but it's certainly the exception. Obviously the GTX 295 setup was king of the hill but I think ATI hangs fairly well in the Cost per FPS segment which is a very accurate gauge of what you can expect performance wise per $$ spent. I love both ATI AND nVidia and keeping the competition up will only be good for the consumer in the end. And yes I will agree the review was not biased in any way, just Bosco that bleeds green. B:)

 

 

 

p.s. I know that was fairly "in your face" for my first post. Does anyone remember a lil' ol' website by the name of OCAddiction? ;)

Share this post


Link to post
Share on other sites

  • Replies 88
  • Created
  • Last Reply

Top Posters In This Topic

Those of us that live in the real world don't need driver updates every 3 days. If you're a Hardcore gamer or Benchmark Junkie then yes I can see it, but this is what, 1 or 2% of the population? ATI nor any other company is going to cater to this. Ok so maybe nVidia does by releasing drivers every few hours but it's certainly the exception. Obviously the GTX 295 setup was king of the hill but I think ATI hangs fairly well in the Cost per FPS segment which is a very accurate gauge of what you can expect performance wise per $$ spent. I love both ATI AND nVidia and keeping the competition up will only be good for the consumer in the end. And yes I will agree the review was not biased in any way, just Bosco that bleeds green. B:)

 

 

 

p.s. I know that was fairly "in your face" for my first post. Does anyone remember a lil' ol' website by the name of OCAddiction? ;)

Valid points now look at it from this point of few. If ATI spent a little more effort on their drivers maybe they would have a better Cost per FPS, maybe not if they decide to increase their pricing who knows. I just think if there was more of an effort made on the driver side of things they would have more success.

 

If you have a company A doing better then you in something I would think you as company B would want to do the samething no???? I dunno maybe its just me but personally from where I stand I don't think ATI gives a flying tit about customer's wants or needs.

Share this post


Link to post
Share on other sites

I noticed that the two HD4850 X2s in CrossFire owned two HD4870 X2s in CrossFire (Crysis), how is that possible rly? :O

But the review is very very helpful and i like it rly much. Id say it's the most helpful review about multi-gpus in the whole wide world in like last 3 years...

Then again... it is OCC and OCC is the best... makes sense imo. :D

 

I was surprised as well with the results but sometimes things just don't make sense. I ran this test 5 times and varied by 1 fps so I let it stand. The 4850x2 is an awesome combination. If you look at the results at 2560x1600 it fall back into what would be considered normal.

 

 

While the 4850 crossfire setup was clearly trumped by every other crossfire/sli combination tested, we have to keep the test results in perspective. Not everyone is going to (or can afford to) splurge on a 30" monitor in order to realize the monsterous horsepower of some of these cards.

 

For many casual gamers who are still playing on a 20-24" monitor a 4850 crossfire setup is still powerful enough to play most of todays games at the LCD's native resolution and medium-high quality settings. And if you can pick up two 4850's for a little over $200 that's not too bad a deal IMHO.

 

Personally I don't have a grand to spend on video cards and another grand for a monitor :(

 

Point taken! But you have to figure not everyone buys a new rig every 3 to 6 months either and many people save for the system upgrade and buy then. At least Im sure normal people do. :rolleyes:

 

 

This is a really cool article. Nice work guys!

 

 

This is an excellent point! When you figure in both price of the cards and price of the monitor needed to use the cards, the lower ranking options still make a lot of sense, if not more. The higher end solutions put out some great bench numbers, but is that worth $1,100 to most of the users here? Probably not.

 

You can't let yourself get sucked in by high scores alone. Sure some people have the connections/cash for 2x295s, but for the rest of us the name of the game is still performance per dollar.

 

Thats where the GTX 260 combo came in as the price per FPS leader at 3.44 bucks per fps at 1920x1200 just ahead of the HD 4850s!

 

I enjoyed the review but Bosco's Thoughts were fairly biased. Why would I want to read blatantly biased reviews? Especially from someone who cares more about folding than what the graphics cards were really meant to do?

 

Bosco likes Nvidia what can i say! :thumbs-up: While yes the ATI cards do fold they just don't do it as well as the Nvidia cards. Thats one comparison I did not run since you cannot predict which work units you get and means it would take even longer than it did to run the testing to compare results. PPD is not sn accurate comparison when it is based off the work units you complete. If they all had the same point value then that could be done but since they don't i lft well enough alone.

Share this post


Link to post
Share on other sites

Sorry my bad english guys.

 

The review it is ok.You forgot GTX275.How do you calculate the price-performance ratio.Because i see the HD4890 CF has a better ratio than GTX295 Quad.

If the guy ho write the CONCLUSION is not the same guy ho made the reviews his credibility is 0.

Share this post


Link to post
Share on other sites

I was surprised as well with the results but sometimes things just don't make sense. I ran this test 5 times and varied by 1 fps so I let it stand. The 4850x2 is an awesome combination. If you look at the results at 2560x1600 it fall back into what would be considered normal.

 

 

 

 

Point taken! But you have to figure not everyone buys a new rig every 3 to 6 months either and many people save for the system upgrade and buy then. At least Im sure normal people do. :rolleyes:

 

 

 

 

Thats where the GTX 260 combo came in as the price per FPS leader at 3.44 bucks per fps at 1920x1200 just ahead of the HD 4850s!

 

 

 

Bosco likes Nvidia what can i say! :thumbs-up: While yes the ATI cards do fold they just don't do it as well as the Nvidia cards. Thats one comparison I did not run since you cannot predict which work units you get and means it would take even longer than it did to run the testing to compare results. PPD is not sn accurate comparison when it is based off the work units you complete. If they all had the same point value then that could be done but since they don't i lft well enough alone.

 

All bs aside guys this was a damn great article. I'm just kinda miffed because I ordered my parts to build a new system late last week and ordered a 4890 planning to do Crossfire once the 4890 cards come down a little more in price. I'm pretty sure I'd be happy with either setup and chose not to get a GTX 280, 285 or 295 because they were significantly more and in a single card environment didnt' seem to perform that much (if any) better. Kinda wished I had seen this article first but that's the breaks. However I plan to do ZERO folding so I should be good. ;)

Share this post


Link to post
Share on other sites

Sorry my bad english guys.

 

The review it is ok.You forgot GTX275.How do you calculate the price-performance ratio.Because i see the HD4890 CF has a better ratio than GTX295 Quad.

If the guy ho write the CONCLUSION is not the same guy ho made the reviews his credibility is 0.

 

I tested the cards first then Frank tested them so as for my credibility its intact thank you.

 

As this page clearly states http://www.overclockersclub.com/reviews/cr...re_vs_sli/2.htm the 275's did not make testing for this round.

 

As for the how the performance ratio was done read here

 

http://www.overclockersclub.com/reviews/cr...e_vs_sli/24.htm

Share this post


Link to post
Share on other sites

Sorry my bad english guys.

 

The review it is ok.You forgot GTX275.How do you calculate the price-performance ratio.Because i see the HD4890 CF has a better ratio than GTX295 Quad.

If the guy ho write the CONCLUSION is not the same guy ho made the reviews his credibility is 0.

 

 

I stated the reason for the 275 not being included was that i did not have 2 at my disposal in time to get them in the testing. Although you could make a case for the 295 being 275 in SLI because that in essence is what it is.

Share this post


Link to post
Share on other sites

very nice review guys...

 

very informative and the facts stand where they are...

 

I have always had nvidia cards but I too currently enjoy some ATI cards...

 

I would definitely love to play with 1 or 2 of those Sapphire 4850X2 2gigabyte cards

 

I have been looking at them for over 2 months but just cant bring myself to get one but I would still love to try atleast one...

 

very thorough...thanks guys

Share this post


Link to post
Share on other sites

First I want to thank those involved in writing this article. I find these kind of comparisons fascinating and helpful.

 

But, keeping in mind I'm not an expert, I had a few problems with it. If I'm confused or incorrect about something, please let me know.

 

The conclusion to the article was very disappointing in that it seemed very biased. All the talk about being on some "team" or being a "fanboy", or bleeding a certain color, are things I would expect to see in a discussion group, not in a review that people may be relying on to help them make a purchase. I would hope for a bit more impartiality.

 

Also, the emphasis on 2560X1600 resolution is perplexing since I've read that 10% or less of the people have such monitors. I would also bet that the majority is still gaming at 1680x150. I know I can't afford one, which brings us to the question of expense.

 

I have Gigabyte UD3P motherboard and two HD 4830 in crossfire. The costs where $116 for the MB and $85 for each card, which is the reason I went with ATI/crossfire. It was a lot cheaper than buying a (not x38 or x48, my mistake) 780i or 790i motherboard and going SLI. And from what I've read and now experienced, the difference between PCI-E x16 and x8 is not that huge. The cost of each type of platform would also be helpful people trying to figure out which way to go.

 

In addition, it seems that some consideration should be made for how many FPS do people really need. If 60 to 70 FPS is the level at which smooth play is reached, how far above that does one really need to go. Below I have a table with your figures for cost and cost per FPS at 1920x1200 (which is what will affect most people), and included the the actual FPS for each that wasn't provided. This indicates that all these graphic card setups would be more than adequate (in fact, way more than adequate) for most people. So the cheapest solution, $260 for 2 HD 4850s, would be good enough and is also the the lowest power user. The closes is the $509 GTX 295. Of couse if you need 100 FPS or more for some reason, then the GTX 260 SLI is tempting, but I still have my two questions. Do I really need it and how much to set it up?

 

 

______________________Price_$______$_Per_FPS______FPS

GTX_295_Quad_SLI________1120________7.6________147.37

HD_4870x2_Crossfire_X_____800________7.3________109.59

HD_4850x2_Crossfire_X_____520________5.0________104.00

GTX_295__________________509________4.6________110.65

HD_4870x2________________400________4.7_________85.11

HD_4850x2________________260________3.8_________68.42

GTX_285_SLI_______________680________5.4________125.93

HD_4890_Crossfire__________530________5.3________100.00

HD_4870_1GB_Crossfire______440________4.7_________93.62

GTX_260_SLI_______________360________3.4________105.88

HD_4850_Crossfire__________260________3.5_________74.29

 

Of course, which games are desired can be a determining factor. But I don't think anyone needs to spend $400-$600 for graphics cards. Doesn't make sense to me. Most reviews I've read lead me to believe that 3-4 card configs just aren't worth it.

 

Anyway, I'm very happy with my little crossfire system. It plays Mass Effect and Hell's Highway very well at 1920x1200. And it seems to Fold pretty good too. I was not considering Folding@Home when building this system, and I don't thing most people are. They are thinking about playing games the best they can afford, and I think you know that. The emphasis on other things that graphics cards can do, while interesting, is not the huge factor that you are trying to make it out to be.

Edited by PopcornMachine

Share this post


Link to post
Share on other sites

@ Popcorn

 

I am sorry you feel that it was biased. It was written that way because it's the truth. I don't sugercoat anything if I don't like something I will tell you why. A year ago I was a huge ATI fan but I just got frustrated with the same bs over and over again.

 

As for only 10% have 30" LCD's we have no way of really actually knowing to be honest. The idea of testing at that resolution is to see what the cards can do.

 

FYI X38 and X48 boards do not have SLI so I am not sure if that is a typo but I get what you are saying.

 

60 or 70 FPS is a good range to play at no doubt about it. Just remember the higher resolution you go the more the cards take a hit. 4850's in CF would not power a 30" LCD properly for a good gaming rig hence the reason we showed a few different combo's to give people and idea where things fall in.

 

Everyone has different needs and wants for what they want for their system :)

Share this post


Link to post
Share on other sites

man you two really rustled some feathers with the fanboys :D. I also just realized that a good cheaper nvidia card that could have been used in the review could be the 9800gtx+/250 1GB, while I'd never buy one myself, it'd be interesting to include. But great review guys, you deserve ALOT of credit for what must have been a grueling review to put together (though I'd take your jobs anytime ;), or maybe one or both of those 4850x2's :lol:)

Share this post


Link to post
Share on other sites

man you two really rustled some feathers with the fanboys :D. I also just realized that a good cheaper nvidia card that could have been used in the review could be the 9800gtx+/250 1GB, while I'd never buy one myself, it'd be interesting to include. But great review guys, you deserve ALOT of credit for what must have been a grueling review to put together (though I'd take your jobs anytime ;), or maybe one or both of those 4850x2's :lol:)

That was the whole point I wanted people to see and look at it. There has been so much smack talk I just wanted to get this out there so people can have some idea's.

 

I think we might do another one of these articles at a lesser price point, I can hear Frank already saying I am going on Vacation lol.

Share this post


Link to post
Share on other sites


×
×
  • Create New...