4870X2 or GTX280 OC

R

Rico1234

Weaksauce
Joined
Oct 25, 2006
Messages
79
Hi just wondering what is the better card to go with? I read that the dual GPU isn't as good as single GPU but I am not sure so figured I would ask on here because many people know a lot about these cards here. I haven't used ATI for a while so not sure if they have problems or if they are making good cards again with this one?
Appreciate any help, thx

Rico
 
ATi if you have the money... it's faster in most games besides Crysis where dual chips don't scale well.
 
aZn_plyR said:
ATi if you have the money... it's faster in most games besides Crysis where dual chips don't scale well.
Click to expand...

I've been reading around a bit...how are the drivers for that card doing these days?

I had seen some complaints about "immature drivers" and older games not working well and things of that nature.

What's the word on that?
 
right now 4870X2 for 450AR > GTX280 OC at 385AR. This is coming form a happy owner of a GTX280. If I was buying then it would be an easy choice for me.
 
Ya not sure about the drivers, I was reading things like that myself so that is what started me wondering which way to go etc. Should I be just getting something like an 8800 Ultra that was a good card that i never heard many bad things about, for less or is the difference between the GTX280 and the 4870x2 worth getting the new ones?.
 
Rico get the new ones.

They are both awsome cards.

If you have good ventilation in your case go with the 4870x2.

If you dont, or if heat bothers you, then go with the GTX 280.

/opinion

Off course, I assume you are playing at retty high resolution.
 
The GTX 280 OC runs pretty hot too. They both expel the hot air out the back vent anyway. At 1680 or lower res, go for the 280. At 1920 or higher, go for the X2. That may be over simplifying, but for the most part its true. It also depends on what games you plan to play too.
 
Does the OC really make that much of a difference? I just found an EVGA GTX 280, not OC with cashback and mail in rebate coming to 285 shipped. Isn't that on pretty close to on par with the x2? and for nearly $100 less...
 
By the time you get your cash back from Live.com, you could probably snag a GTX280 new (and with a an actual proper warranty) for about that much :p
 
Eggothewaffle said:
By the time you get your cash back from Live.com, you could probably snag a GTX280 new (and with a an actual proper warranty) for about that much :p
Click to expand...

Buying it off the mwave ebay store front so i don't need to deal with the bs fake cards or doa stuff. Plus i figure its probably about the same as mir's but alittle less hassle.
 
From what I have read, the 4870x2 does whoop on the GTX 280, but only if there is plenty of CPU power to allow the two GPU's to communicate well.

If your not swingin' mandingo CPU power, the GTX 280 will likely outperform the 4870x2.
 
if only [H]ardocp.com would put out a review to comprehensivly cover these kinds of questions. :(
 
vengence said:
if only [H]ardocp.com would put out a review to comprehensivly cover these kinds of questions. :(
Click to expand...


Please tell me exactly what questions you would like answered.

I always think, "Damn, I wish our readers would be specific when they are sad so we can try to help them out, instead of giving me generalized statements."
 
vengence said:
Sorry, you obviously took my post completely wrong. The sarcasim didn't come through. [H]ard has great reviews that cover EXACTLY this, with much more detail than any other site. There are pages and pages of dicussions on "4870x2 vs 280". I was suggesting he go look at a review which would answer this exactly. Or just search the forums.

Of course now I'm wondering if I've missed the sarcasim in your post.
Click to expand...

Hehe, sorry. I read your post more than I did the thread. :) Going to have to teach you to us this ----> :rolleyes: :D
 
Here are the two reviews I am talking about:

http://www.legionhardware.com/document.php?id=775&p=16

http://www.legionhardware.com/document.php?id=770&p=12

It seems like if you have anything AMD for the cpu, the GTX 280 is a better match because the 4870x2's require more CPU power. I would imagine that this means anything in the intel camp that is a Q6600 or below (dual core) might benefit more from the GTX 280 as well.

From the article:
"Even more interesting is the fact that when paired with the GeForce GTX 280, the Athlon64 X2 6400+ delivered the same performance at 1920x1200 as the Core 2 Quad Q6600. This is interesting because when testing with the Radeon HD 4870 X2, the Q6600 was 50% faster than the 6400+, which is incredible.

This was no anomaly, as similar results were seen in a number of games. For example, the Core 2 Duo Q6600 was just 2% faster than the Athlon64 X2 6400+ in Unreal Tournament 3 when testing with the GeForce GTX 280. However when testing with the Radeon HD 4870 X2, the Q6600 became 62% faster than the 6400+, which was again quite amazing."

So if your CPU is on the weak side... I would stick to the GTX280. If you have mandingo cpu, the 4870x2...
 
4870x2 would just annihilate the gtx 280 in the majority of demanding games out today except for twimbp crysis :D

Too bad both Crysis its expansion Crysis: Jason Statham are both $!%ing awesome (Except for ultra horrible I want to commit hari-kiri Core/Ascension Crysis levels.)
 
DeadSkull said:
4870x2 would just annihilate the gtx 280 in the majority of demanding games out today except for twimbp crysis :D

Too bad both Crysis its expansion Crysis: Jason Statham are both $!%ing awesome (Except for ultra horrible I want to commit hari-kiri Core/Ascension Crysis levels.)
Click to expand...

What I was pointing out in my posts was that this may not be the case. If the OP doesn't have at least a Q6600, or has an AMD proc, the GTX 280 might be a better choice.
 
AbsintheCommando said:
Kyle, here's what I would like answered in video card reviews: how does each card perform given a consistent group of settings across several relevant resolutions? I have seen something similar to this in some reviews from [H]ardOCP listed as "apples to apples" comparisons and they are by far the most useful part of the review, when they are present. I would like to see comparisons at standard resolutions (1280x1024, 1600x1200, 1680x1050, 1920x1200) at a standardized (per game) level of in-game quality settings. The current review system is not very useful to me. Being able to see at a glance what card runs fastest at each resolution given fixed settings tells so much more than "card A runs best 1600x1200 with 2AA8AF and medium settings while Card B runs best at 1920x1200 with 0AA0AF at low settings except physics and object detail which are at high." It doesn't even matter what the settings are, as long as they are high enough to challenge the cards in question.

I feel this would be an improvement because a lot of people game on LCDs and lots of those folks don't have LCDs that scale well to different resolutions. In addition, it's just good research design to change only one variable during a test, and for the reason I just listed, I feel that resolution is the best variable to change. I realize that people can currently infer (sometimes) what card will treat them best, but it's not always intuitive or easy, and I assume that one of [H]ardOCP's goals is concise, clear presentation of data to enable readers to make use of your hard work. I have read every current and previous generation video card review that [H]ardOCP has written, and like the people who ask "4870 or GTX260?," I am not sure what card to get for my resolution. In my opinion, this is the sort of question that video card reviews should answer very clearly.

Things I do like about the review system currently in place include the subjective experiences of the reviewer, such as telling us about odd drops in framerate, showing graphs to display how cards compare at similar points in a game, and reporting the high/low/mean FPS scores of a card. I also like it when reviewers include screenshots to compare picture quality or to demonstrate with a screenshot the real difference (or lack thereof) between a given level of settings.
Click to expand...

Thanks for the feedback, I am sorry you do not find our current reviews very useful to you. I will say right now that there are tons of other sites that likely have information that would be better to your liking. Benchmark sites will give you what you are looking for in a more “scientific” method. We evaluate what gameplay the hardware produces, not what metrics we can make with it.
 
AbsintheCommando said:
Kyle, here's what I would like answered in video card reviews: how does each card perform given a consistent group of settings across several relevant resolutions? I have seen something similar to this in some reviews from [H]ardOCP listed as "apples to apples" comparisons and they are by far the most useful part of the review, when they are present. I would like to see comparisons at standard resolutions (1280x1024, 1600x1200, 1680x1050, 1920x1200) at a standardized (per game) level of in-game quality settings. The current review system is not very useful to me. Being able to see at a glance what card runs fastest at each resolution given fixed settings tells so much more than "card A runs best 1600x1200 with 2AA8AF and medium settings while Card B runs best at 1920x1200 with 0AA0AF at low settings except physics and object detail which are at high." It doesn't even matter what the settings are, as long as they are high enough to challenge the cards in question.

I feel this would be an improvement because a lot of people game on LCDs and lots of those folks don't have LCDs that scale well to different resolutions. In addition, it's just good research design to change only one variable during a test, and for the reason I just listed, I feel that resolution is the best variable to change. I realize that people can currently infer (sometimes) what card will treat them best, but it's not always intuitive or easy, and I assume that one of [H]ardOCP's goals is concise, clear presentation of data to enable readers to make use of your hard work. I have read every current and previous generation video card review that [H]ardOCP has written, and like the people who ask "4870 or GTX260?," I am not sure what card to get for my resolution. In my opinion, this is the sort of question that video card reviews should answer very clearly.

Things I do like about the review system currently in place include the subjective experiences of the reviewer, such as telling us about odd drops in framerate, showing graphs to display how cards compare at similar points in a game, and reporting the high/low/mean FPS scores of a card. I also like it when reviewers include screenshots to compare picture quality or to demonstrate with a screenshot the real difference (or lack thereof) between a given level of settings.
Click to expand...


well as far as the resolution is concerned I've brought this up before as being a limitation of the H reviews, but I don't know how they could practically address it given the way they test. The amount of work to find the highest playable settings on different monitors would by itself be a huge increase in their workload. So if they did that then they would also do it for different hard ware types for reasons mentioned above. This is the price they pay for not using canned bench marks.

And to top it off you still have the problem that "highest playable settings" is pretty subjective. For instance,the H reviewers will generally trade eye candy for resolution long before I will. Some people will agree others will not. As it is a personal preference how could they match it?

H reviews are pretty unique and I put a lot of weight into them. A lot of things that do not show up in bench marks show up in their testing. it may not be perfect but it is surely a better metric for gamers to go by then regular benchmarks. and really no need to use just one site to make buying decisions on, even [H].
 
undertheradar said:
"Even more interesting is the fact that when paired with the GeForce GTX 280, the Athlon64 X2 6400+ delivered the same performance at 1920x1200 as the Core 2 Quad Q6600. This is interesting because when testing with the Radeon HD 4870 X2, the Q6600 was 50% faster than the 6400+, which is incredible.

This was no anomaly, as similar results were seen in a number of games. For example, the Core 2 Duo Q6600 was just 2% faster than the Athlon64 X2 6400+ in Unreal Tournament 3 when testing with the GeForce GTX 280. However when testing with the Radeon HD 4870 X2, the Q6600 became 62% faster than the 6400+, which was again quite amazing."

So if your CPU is on the weak side... I would stick to the GTX280. If you have mandingo cpu, the 4870x2...
Click to expand...

can someone explain why this is the case?
 
Valset said:
Possibly the overhead of using dual GPU?
Click to expand...

The physics of motion are mostly still calculated on the CPU along with all the other backend stuff in games such as location tracking, quest logs, buffs, raid party info etc. Yes the graphics card will display all those moving people but your cpu is telling the gpu where to put them.


You can get crap fps on the 4870x2 by having a low power CPU, that is what he was talking about. IMO there is nothing out there that will GPU bind the 4870x2 as long as the game wasn't coded crysis level retarded.

I have noticed that the avg fps on my rig has made two big jumps up when I switched from the P5k-e mb to a newer P5Q-Pro, and when I OC'd my Q6600 from the stock 2.4 to 3.2ghz(400x8, fsb 1:1 with ram). I do not think the p5k-e truly ran at pci-e 16x, more like 8x with 16x plug, thus caused a choak in textures and what not loading in.

If you play games like Vanguard, Warhammer, or UT3 then you can run them at 1920x1200 on a 24" monitor with all settings at max and you should see 50+ fps avg*.


*some hiccup's from saves, zoning/chunking, etc. will happen, that is mostly backend stuff not GUI glitches, nothing you can do about it on any graphics card set.
 
Nanan said:
You can get crap fps on the 4870x2 by having a low power CPU, that is what he was talking about. IMO there is nothing out there that will GPU bind the 4870x2 as long as the game wasn't coded crysis level retarded.
Click to expand...
Does crysis load the entire level as soon as you start it up or does it constantly load data from the hd therefore sometimes inflicting high fps drops because of slow data transfer?
 
JohnM said:
4870x2 if you can afford it.
Click to expand...

And have a beefy PSU as it consumes a little more than a GTX280, and if you can fit a X2 in your case (haha!)...

My GTX280 was CPU limited by my GTX280. OCing it has given me some pretty sizeable gains in COD4 & Warhead...I did not test STALKER Clear Sky on my pre-OCed Q6600 (sadly)...but once I got my Q6600 to 2.7 GHz, I really saw my Warhead performance jump (@ 1680x1050, 0xAA I jumped something like 15 FPS in some cases)

Great to see you on the forums getting feedback Kyle. :D
 
DeadSkull said:
Does crysis load the entire level as soon as you start it up or does it constantly load data from the hd therefore sometimes inflicting high fps drops because of slow data transfer?
Click to expand...

Crysis only loads the textures and environmental meshes in a limited range around you, most of the bad fps issues are due to it. To totally load a crysis map into ram and g-ram you would need prob 6-7gb of ram and 10-12gb of g-ram.

Crysis does not use a very computer memory friendly system for packing its files and has to access a large file to take out one or two items every time you move a few feet.
 
Nanan said:
Crysis only loads the textures and environmental meshes in a limited range around you, most of the bad fps issues are due to it. To totally load a crysis map into ram and g-ram you would need prob 6-7gb of ram and 10-12gb of g-ram.

Crysis does not use a very computer memory friendly system for packing its files and has to access a large file to take out one or two items every time you move a few feet.
Click to expand...

Ok I know what I need to buy for Christmas now. Two SSDs in RAID0, here I come!

Btw I wonder if each level has its own textures/environmental mesh load script. Seems like some Korean levels before the aliean core level really suffer from the fps drop due to slow data load from hd.
 
DeadSkull said:
Ok I know what I need to buy for Christmas now. Two SSDs in RAID0, here I come!

Btw I wonder if each level has its own textures/environmental mesh load script. Seems like some Korean levels before the aliean core level really suffer from the fps drop due to slow data load from hd.
Click to expand...

Isn't there a config setting to turn streaming off?

I doubt textures would require 6 GB...like I said, I think you can turn streaming off, which loads all textures into RAM...and you're better just putting them all in RAM and letting the GFX card grab them on demand, rather than have the GFX card look for them in RAM, not find them, cause a page fault, load them from the HD...etc.

Feel free to correct me if I am wrong, but look on tweakguides for all the settings...or dig up some Crysis custom config files.
 
If you got the cash, go with a 4870x2. If you can afford a 50% performance decrease in games, gtx280 OC is the way to go. It's as simple as that.
 
You must log in or register to reply here.
Back
Top