Single 4870 vs. Crossfire 4850s, brand?

T

Tabs

Limp Gawd
Joined
Aug 24, 2005
Messages
147
I'm gonna upgrade my single 8800GT this week to either a single 4870 or twin 4850s. Is it generally correct that two 4850s are faster than a single 4870?

Is the ASUS one pretty much the best because it can control the fan independently of hacking CCC profiles etc?

Thanks,
Ryan
 
if you play crysis or quake wars, 4850 xfire typically scales for crap (at least when i had them). if either of those are your bread and butter, go with the 4870. far cry 2 and crysis warhead shouldn't have issues, but the possibility is there.

xfire 4850's are great as long as you are willing to put up with xfire/sli side effects of poor/non-existent scaling and possible micro-stutter. i personally can't stand these things, so i prefer a single gpu, but that varies person to person.

also consider your case air flow. if you have great airflow, it is a non-factor. if you have restricted airflow (small case, rat's nest wires, etc.) avoid two gpu's. they will generate too much heat.

either way, you should get great bang for your $.

i would recommend either asus for fan control or visiontek for lifetime warranty.
 
I think I'm gonna go with the single Asus 4870 now after reading some more... Is the "TOP" model that's OCed with the fan already working better actually available anywhere yet?
 
Gone are the days of micro stuttering and problems when it comes to crossfire. I run 4850cf and couldn't be happier so far. Crazy frame rates in all games including crysis. I run crysis at 1920x1200 with all settings on high and 2 settings at very high and scored 45fps in the crysis gpu test. Gameplay is a solid 30fps or higher in most areas with on occassional dip into the 20's in fierce combat situations. I get at least 40% scaling in crossfire with crysis which is pretty good. I get 60% in cod4 and cod2. The average scaling seems to be in the neighborhood of 25-50% in most games for the extra card which is well worth the extra 179 bucks i paid for it. A 4850cf solution will walk away from a single 4870 and only require two 6 pin pcie power plugs to do it. I recommend the 4850cf over the 4870 any day as long as you have a good x38/x48 motherboard. Even on older boards it still rocks but it is really bottlenecked unless you have the dual pcie2.0 x16 slots. The new x48 board I bought to replace my old 975x board is night and day faster in crossfire than the 975x board was and even the 975x board's performance was no slouch.
 
Astrodave said:
Gone are the days of micro stuttering and problems when it comes to crossfire. I run 4850cf and couldn't be happier so far. Crazy frame rates in all games including crysis. I run crysis at 1920x1200 with all settings on high and 2 settings at very high and scored 45fps in the crysis gpu test. Gameplay is a solid 30fps or higher in most areas with on occassional dip into the 20's in fierce combat situations. I get at least 40% scaling in crossfire with crysis which is pretty good. I get 60% in cod4 and cod2. The average scaling seems to be in the neighborhood of 25-50% in most games for the extra card which is well worth the extra 179 bucks i paid for it. A 4850cf solution will walk away from a single 4870 and only require two 6 pin pcie power plugs to do it. I recommend the 4850cf over the 4870 any day as long as you have a good x38/x48 motherboard. Even on older boards it still rocks but it is really bottlenecked unless you have the dual pcie2.0 x16 slots. The new x48 board I bought to replace my old 975x board is night and day faster in crossfire than the 975x board was and even the 975x board's performance was no slouch.
Click to expand...


I disagree. I had two 4850's in xfire on the system in my sig. I experienced very evident microstuttering in The Witcher with settings maxed out. The value of a 4850 xfire system is incredible, but if you have the cash, I still recommend a stronger single gpu solution over a budget xfire setup.
 
I'm gonna wait for Newegg to get the ASUS 4870TOP in... I've read good things about them fixing the fan issues etc.
 
I haven't noticed anything wrong as far as stuttering, but then I'm mainly playing COD4, which CF 4850 does well with.
 
With the 4870X2 (single-board Crossfire) being ATIs flagship product now, I think the Catalyst team are going to spend alot of time improving Crossfire scaling in as many games as they can, to make their flagship card look as good as possible. This will benefit users of 4850/4870 CF as well. So scaling will only improve over time, IMO.
 
Any stuttering that people are getting with 4800 series cards in crossfire is most likely a function of the health of their pc and not the cards. I get no stuttering in my current box or my previous one that is now my htpc.
 
Astrodave said:
Any stuttering that people are getting with 4800 series cards in crossfire is most likely a function of the health of their pc and not the cards. I get no stuttering in my current box or my previous one that is now my htpc.
Click to expand...

or you aren't playing anything that can get them to dip below 30fps, where microstuttering is most notable.
 
Admittedly I've never had a crossfire or SLI system, but from the stuff I've read and the graphs I've seen, microstuttering manifests itself as giving an effective framerate lower than what is actually being displayed by whatever you're using to monitor framerate. This is because frames are rendered fast, slow, fast slow, fast slow, etc, alternating between fast and slow between each frame. That means you'll get a framerate of the average of the fast frame and the slow frame, however effectively you're ACTUALLY seeing is closer to the slowest frame.

That means the program could be telling you 90fps, but what you're actually seeing is 60-70fps if there were no microstutter.

Based off that, you clearly wont notice microstutter until the displayed framerate is around 30fps, because what you're effectively seeing is more like 20-25fps. Above 30fps you probably wont notice the decrease.

Just as a side, "micro" generally means something too small to see (macro meaning something you can see), or scientifically it means x 10^-6. So by definition, when you have "microstuttering", you dont actually see stuttering, otherwise it'd be called macrostuttering :p What you see with "microstutter" isn't stutter, its an effective framerate which is lower than what's displayed. If you do actually see stuttering, its probably because the framerate is so low and inconsistant that you'd see stuttering regardles of whether it was a single card or a dual card set up.
 
You must log in or register to reply here.
Back
Top