Two HD 5770 should be cheaper and easier to find, and performance wise they're are pretty similar. I made a small review of my rig running 2x HD 5770 in CrossfireX and compared results to same system running a GTX 295, the review is in portuguese but you'll be able to see what level of performance you can expect from them by looking at graphs. http://adrenaline.com.br/forum/5333895-post2.html
HD 5770 Xfire - http://guru3d.com/article/radeon-hd-5770-in-3way-crossfirex-review-test/1 HD 5870 - http://www.guru3d.com/article/radeon-hd-5870-review-test/
Why did 3 5770's achieve only 30fps while regular crossfire got 38fps ? I don't understand that at all. Secondly, the prices for 5770's are coming down nicely, especially some of the 'no name' brands out there compared to the likes of Asus..etc. I have an Asus 5770 card right now. Could I add in a second 5770 from a different brand, like Force3d ? It's a lot cheaper so I was wondering if I could do that without getting problems. Or do I absolutely need to have identical brands ?
ExoriaX, do you say you got 40fps. Please tell me what graphic settings, resolution etc you used. Also, did you use a specific benchmark program ?
dont 5770 have a smaller memory pipe anyways? 128bit? higher res with AA,AF will cause that card to fall on its face.
ive gone from 5770 crossfire to a single xfx 5850 xxx edition (waiting for my second one to arrive). the 5770s were much faster. @Olorin - 3 graphics cards requires significantly more processing power from the cpu to handle having to run three drivers and run the same information to all gpus, and it ends up bottlenecking, hence 3 gpus are slower than 2 unless your cpu is sufficient enough to handle it. better to get the asus, it has better warranty as voltage tweak is covered under warranty. on top of that, the bios of the asus has a higher max frequency range. for my xfx 5770s, max core was 960-ish, but the asus can go up to 1050 iirc? also it seems that the cards perform differently with different bios'. @JCazz - AA and AF have no problems. heck, the 128bit memory bus isnt even a bottleneck. if it were a bottleneck, then 2x5770 would be slower than 1x 5850. i can vouch that it is not slower. so far, in mass effect, dmc4, codmw2, fallout3, and the last remnant, my old 2x 5770s have been, on average, 30% faster than my xfx 5850 oc'd edition, so i have no doubt it would have no issues keeping up with a 5870.
Hi Sever, Could you explain a little what you mean by the card performing differently with different bios ? As far as warranty, the thing is, I'm not a hard core overclocker or anything. The only reason why I went for the Asus model was because everything else was out of stock. : ) But besides warranty since I'm not going to oc the units, would a cheaper brand name offering work or do you think there would be stability issues ? Also, sorry so many questions directed towards you. The thing is, with my current cpu which is a major bottleneck I think, ocing the unit would make little sense. But if I don't oc either way, are there any actual benefits to using two of the same brand name units as opposed to different brand names ? I assume they all use the same core chip and board....
Just checking out some benchmarks with twin 5770's and in most tests outperform the 5870...and even rival the 295 ! That's seriously good at the price I think.
Not to discredit those ones having used Crossfired 5770, but i wouldnt put my money on a multi gpu based solution. On top of that, i dont know if this problem still persist, but multi Gpu solutions give a far more jerky playback of games than a single card solution. So i would say rather go with a much more simple solution the 5870.
Sorry, I don't know what terms like micro-stutter even mean. : ) I guess I haven't been participating in enough forums to understand the various terminologies.
Im not sure if what i am refering to is micro stutter. I used to have X1900s in Crossfire. When i enabled crossfire and i had lets say 35fps in FEAR it felt quite jerky. On the other hand when i disabled it and ran only with one GPU at the same 35 fps it felt much smoother.
ive only had microstutter in one game, and it happened to be an x360 port (lol), the last remnant. aside from that, most games run hitch free, even with vsync off. i just prefer having vsync on, as it reduces the stress on my cards, which means they generally run cooler as the gpus do not always run at 100% load. a crossfire pair of 5770s is more than enough for >60fps in most games at 1920x1080 with all details on max. @Olorin, you will still notice an improvement in performance, even with your old cpu without overclocking. but it wont be as much of an improvement if you had a faster cpu. so instead of ~60%, you may get 30-40% improvement. you may find that you would get better performance if you chose to upgrade to an am2+ mobo with an athlon x4 cpu (so you can still use your ddr2, and it will cost less than a 5770). altho an am3 setup would be the better futureproofing, as it am3 is the socket that amd plan to use for the next two years.
Let me address some of the issues with this thread. Two 5770s are as fast as or faster than a 5870 in most cases, crysis being the exception because Crysis has always had poor crossfire scaling, with all cards, for some reason. The memory bandwidth of the 5770 does not at all limit it's performance. In fact, on a % increase basis, most performance benefit from overclocking will come from overclocking the core. For example, a 10% OC on the core might net you 5 FPS in most games, and a 10% OC on the memory might get you 3-4. This indicates the card is not bandwidth starved as most people would have you believe. I can confirm this by saying there isn't a single game (Crysis included) in which I don't play with at least 4x AA. Yes, I play Crysis completely maxed (except slightly lowering the shadow resolution and turning down volumetric effects) at 1680x1050 with 4x AA with no issue. It is perfectly playable and above 25FPS most of the time. The bandwidth argument always comes from the people who don't own the card, and speak only of hardware "theory". If you look at the real numbers, the 5770 falls almost exactly between the 4850 and 4870's memory bandwidth, unoverclocked, and approaches the 4870's once overclocked. So the bandwidth limitation argument against this card simply needs to disappear, it doesn't exist. Micro-stutter is, for the most part, perpetuated by the fact that know one knows what the hell it looks like or what it even is. Typically, it's a poorly coded game, port, or memory/hard drive issues. Many people use the term "micro-stutter" to mean anything from big hitches in the game, to background programs interference, to low FPS slowdowns... In real life, Micro-stutter doesn't exist. No doubt I'll get some arguments here, but I've owned a crossfire setup since the 3k series. I owned a 3870/3850 combo, 4870/4890 combo, and now two 5770s. I have never ever ever experience microstutter. It just doesn't happen anymore. Maybe it did, once, but not anymore.
Settings were all "Very High" and AA off. Video card is not overclocked. Plan to do later. EDIT: Also don't forget about the whole system. Older it is will bottleneck the video card so you might notice very low fps on older system than new. Mine's newer than AMD Athlon 64 6000+ as said on your list.