I'm confused...HD4890 or gtx285

touchdowntexas13

Distinguished
Apr 13, 2009
759
0
19,010
Ok so I am looking at making a new system build here in a month or two with a core i7 920 at its heart. I was also planning on using a GTX285 for my GPU, but i am reading in some places that the 4890 can outpace the gtx285.

So first of all, i would like to put price/performance aside. I know that the hd4890 is a great deal, and would allow me to crossfire sooner than sli'ing a gtx285 (budgetly speaking).

What I want to know:
Does a reference-clocked gtx285 offer "significant" (i know that can mean a range of things) gains over a reference-clocked hd4890? Any legit links comparing the two would be appreciated.

Does an overclocked gtx285 offer "significant" gains over an overclocked hd4890? -same thing about the links

THEN:

AFTER answering those two questions, can someone tell me whether the performance difference (if there is one) is worth the difference in price?

and what would you suggest buying?

The reason i ask to keep price/performance separate is because actual performance difference is more or less factual, while people have their own opinions on what price is worth a certain amount of performance. I would like to get the facts straight first, and the opinions second.

If it makes a difference at all, my preferred mobo at this point is the p6t deluxe(are a lot of these boards coming shipped dead?). I should also get good cooling out of the full tower coolermaster haf 932 for overclocking. I can give any other information on my proposed build if it helps at all.

Thanks everyone.
 
Solution
You will get the same time out of all the cards I discussed. It comes down to what games you play and how each card handles those games. If mostly what I played was crysis and games written with the nvidia in mind I would stick with nvidia as you will get better performance out of a card the game was optimized for and vice verse with ati.

chef7734

Distinguished
Mar 3, 2009
856
0
19,010
Ok here it is. If you get a good 4890 you can oc it to 800 to 1000mhz and get performance equal to the gtx 285 at stock speeds. It is a good price/ performance, but you can still get a gtx 285 to perform faster at a higher price. If you want the fastest single gpu solution the gtx 285 is still the fastest, but some are not willing to pay the difference for a small percentage. now if we are talking stock speeds the gtx 285 is quite a bit faster.
To me if you have a crossfire board then it is not worth the price difference as you can always add a second one. With an sli board it is a tough decision. I went 2 285 in sli for my main rig and my secondary rig has crossfire.
 

touchdowntexas13

Distinguished
Apr 13, 2009
759
0
19,010
Well the p6t deluxe is supposed to be both sli/crossfire compatible i think.

Ideally, when i purchase, i would like the best single gpu configuration. I would consider sli/crossfire only after at least a year (unless there were crazy price drops). So when i purchase, i want the single gpu configuration that will give me the best performance for at least a year.

Does anyone have any links to an overclocked hd4890 benchmark? overclocked gtx285 benchmark?

I plan on playing at 1920x1080, so when you say small percentage, do you mean like a 3 fps difference, 5 fps difference, or a 10 fps difference? For example in games like Crysis or FC2.

Ahhhhhhh im stuck between the two! I tend to like nvidia too...
no really good reason though
 

chef7734

Distinguished
Mar 3, 2009
856
0
19,010

Never said it didn't. The 285 is still faster and if you have a sli mobo then crossfire is a mot point as is sli on a crossfire board. As I stated the 4890 is a great price/performance and something to think about.
 

chef7734

Distinguished
Mar 3, 2009
856
0
19,010

In a situation like this it is a really tough decision. Like I said the 4890 when oced to 900 to 1000mhz you will get pretty much stock 285 performance. This is of course all dependent on what games you play. Both would be a great choice, but 4890 would save money.
 

touchdowntexas13

Distinguished
Apr 13, 2009
759
0
19,010
I just looked at that link from jaydeejohn. It seems that in game benchmarks, the two are pretty much neck and neck in cf/sli. However the gtx285 sli gets much better scores in 3dmark. I wonder why this is?

My problem with that link is that it took cards that came factory overclocked. I would like to see performance benchmarks on where someone takes their gtx285 and oc's it to the max, and where someone takes their hd4890 and oc's it to the max. Preferably non-sli/cf, because like i said, i would like to be able to refrain from a multi gpu configuration for as long as i can.

Another reason i would not like to compare sli/cf configurations is because there is another factor in that equation. It could be that ATI just does a better job right now with cf than nvidia does with sli. However, in the future i would expect nvidia to better optimize sli so that it is not a factor anymore.

Thanks for everyone's input. I can't really say i am too much closer to deciding, but at least i am learning more about the two cards. Any more input would be appreciated.
 

touchdowntexas13

Distinguished
Apr 13, 2009
759
0
19,010


Well I play mostly FPS, but if i stopped console gaming, then i would play racing/sports games. I am also open to mmorpg's, though i have never done that much with them. I don't know if that narrows it down at all...probably not.
 
To me, if both cards are maxxed out, its so close, the extra money isnt worth it. I did see a review with them oceed, but cant remember where.... like has been said, tough decision if money isnt a huge factor, as theyre close in perf, but IMHLO, it still isnt worth the extra monies, in single card setup
 

touchdowntexas13

Distinguished
Apr 13, 2009
759
0
19,010


I agree. But could it be that the ati card they used was better oc'ed than the gtx285? That's my main beef with those numbers, on top of it being a cf/sli comparison.

This is all so strange to me though, because when the hd4890 came out, i thought it was a known fact that the 4890 wouldn't perform like a gtx285. In fact, I thought the gtx275 was supposed to be right with the hd4890.
 

touchdowntexas13

Distinguished
Apr 13, 2009
759
0
19,010


Thanks for your input. I will search the internet for user-based oc's to see if i can find a maxxed 285 and a maxxed 4890 comparison. The money isn't exactly a huge factor, because like i said i want the single gpu setup that will last the longest. This decision might get all the more easier if either card takes a drop in price. However i would expect this of the gtx285, not the hd4890.
 

touchdowntexas13

Distinguished
Apr 13, 2009
759
0
19,010


This is why I am confused. In that first link, the stock gtx275 performs better than the stock hd4890. So my assumption from that is that an oc'ed 4890 couldnt touch an oc'ed gtx285. But this is not so?

I can't really compare the other two links to the ones jaydeejohn posted because the games are tested at different settings, but i am bewildered by that first link in accordance to the consensus that a heavy oc'ed 4890 is almost just as good as a heavy oc'ed 285.
 

touchdowntexas13

Distinguished
Apr 13, 2009
759
0
19,010
Going back to jaydeejohns link, i misread. All three gpu's were user oc'ed. I thought they just took the factory oc's and left them there. However, that site claims that they got a rather poor oc on both of the gtx285's. I have read everywhere that the gtx285 is supposed to oc extremely well, so could it be that they just got a bad batch of cards?

I should compare the oc's some people reach with their 285's on the internet with the oc's reached on that link. Then i would know if the cards they got just happened to not be as good at ocing.
 

touchdowntexas13

Distinguished
Apr 13, 2009
759
0
19,010


Wow that was a pretty good link. The stock 285 really only outshines the 4890 when there is anti-aliasing enabled. But even then the difference is only like 4 frames at most. But this isn't consistent with the other link you gave me chef (the one where even a stock 275 beats the stock 4890).

There must be other factors here, such as how well either card works with the other hardware, or that between those two links they used different drivers. Or that there is a difference in performance between the different brands used.

That last link you sent chef really made me really want to consider the 4890. However, with the inconsistencies between the benchmarks in different links, I have to say that the mystery continues. If i can't find a perfect oc'ing comparison, it might just come down to my gut feeling and or price drop of the 285.

Thanks guys, and if you find anything else, you know i really appreciate it!
 

chef7734

Distinguished
Mar 3, 2009
856
0
19,010
Problem i find with review sites is they are never consistent. You will get different results depending where you go. Visually and real world you will not notice a difference. Only when you run fraps and benchmarks will it really see a difference. You can not visually see a difference between 60fps and higher., especially if you are running a lcd at 60hz. If you are on the fence I would buy the 4890 and save some money.
 

touchdowntexas13

Distinguished
Apr 13, 2009
759
0
19,010


I guess you are right about the difference in review sites. I am just surprised at how off they can be.

The hd4890 is definitely a card that i am considering a lot more now then when i first made this post. For me, only time will tell (prices, special deals, future reviews).
Another selling point on the 4890 is that it is dx 10.1 compatible, though that was never a major selling point for me anyway. the 285 has physx and cuda so i guess they even out.
 

belial2k

Distinguished
Feb 16, 2009
1,043
0
19,310
I know this wasn't part of your op, but is there a reason you haven't thought of using any of the dual gpu cards? The 4850x2 slightly outperforms both the gtx285 and the 4890, and currently cost about $70 less...and your board supports quad crossfire, so you could still add another one when you wanted to upgrade. So to answer your question about price to performance ratio of single card solution, its neither card you are considering, its one that is not yet been mentioned.