• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

5850,5850 in CF vs GTX 480 SOC Results

OP is doing it right, links DM posted are canned warhead benchmarks, it looks like the OP is taking his results from actual gameplay which is what counts. OP's 480 is at 820 core and not 700 core as it is in the linked benchmarks.

I said they'd be higher due to higher clocks, but keep in mind he said Crysis, not warhead and AMD are AHEAD of Nvidia in Crysis, and behind in warhead, so if he meant crysis, then his results are even more out there. The point being 5850xf is OVER 20% ahead in Warhead(would be even further in Crysis vanilla) and a 20% overclock, unless Raven is doing it, rarely gains you 30% performance.

Theres also the little issue of 5850xf's, well I've literaly never, ever seen a 5850 that can't do 850Mhz, if its reference or has a voltage control non reference, frankly I've never seen a 5850 unable to get past 900Mhz, which is a 20-25% overclock, and mine overclocked over 40%.

Hi guys,

All my benchmarks with playing games was me playing the game. So on different maps the fps was different, I played all map with both cards and took the avg, min and max fps from that.
When I set up games my setting are set to max.

With my 5850 in crossfire my mobo it is set up as 8x 8x pci-e.

Also remember when the other benchmarks what have been posted on they use an i7 950 with 6gb or ram.

These benchmarks are from my Pc spec so they will be different.

Cpu won't make much of a difference, marginal difference, peak max might be higher, but for a couple seconds in an entire level that will make marginal difference. Anything from a tricore 3Ghz to a 17Ghz hexcore will make little difference to overall results.

The main question would be, how did you get the average, did you run fraps to keep track of the fps.

Basically the benchmark is canned, and won't give identical results as in game all the time and different levels perform differently, some will favour Nvidia some AMD, the benchmark though gives a rough idea how fast a setup is. Basically 5850xf's should average 30-35% faster, they overclock further than 480gtx's, so both setups at max overclocks should extend the 5850's lead.

unfortunately in "real gaming" things like min/max fps will be fairly different to benchmarking, though shouldn't ultimately make a huge difference. Sometimes you just get a load that takes ages, or a background task that causes a 1fps min for 2 seconds in a 5 hour gaming session, it doesn't change how good or bad a card performs but, there is a reason people try to use standardised benchmarks as it minimises the differences.

As for 460gtx sli being on par with 5850, no, its not, in any way. Also those are sonic platinum versions or, 17.5% overclocked at standard, and a "standard" 460gtx 1gb vs a standard 5850 according to anandtech, are over 10% slower than 5850's, sli/xfire that might gain a bit, but go past, no.


Basically as I suggested earlier, your single card performance looks woeful frankly, while the dual seems to scale well vs the single card, but overall performance is just not where it should be.

Basically you'd need a suprememly excellently clocked 480gtx to match a 5850xf in the vast majority of games, and really 5850xf's are very much underclocked cores to start with and easily do 5870 speeds, and more.


480gtx's aren't remotely "bad" cards, they just aren't great, even at £200 they aren't brilliant, just insanely better value than they were. A 6950 costs the same, is basically a 6970 and is just plain faster, not by miles but less power, faster. The Gigabyte editions have great cooling, which is a HUGE factor for me as anyone reading my posts would know, stock cooling from both companies is a joke, not horrible just, not great and both companies could EASILY release silent cards for stock speeds.


I also won't suggest you don't get a 480gtx, it will be a downgrade in most things but, plenty of people play a single game, or two, consistantly, if Nvidia has a huge advantage in the game you happen to play all the time going Nvidia is a great option, the reverse is true aswell.
 
In regards to PSUs unfortunatly there are a lot of sub-par PSUs out there just doesn't come to light as people invariably buy a PSU with rated power specs 5x what they actually need... when something comes along that actually puts a real draw on the PSU it quickly seperates the wheat from the chaff.

Sadly its not uncommon for high quality parts to be used on the first batch/review versions then quietly downgraded later as no one really notices normally.

ERrm, its actually fairly rare for components to be completely changed, validation for a psu is expensive and takes time, if you change a SINGLE component inside the law requires the whole validation process to be completely redone. PSU's do change, its usually to do with a model no longer being produced so switching to an alternate, sometimes worse, sometimes better, often getting a new model launch with a (not always) appropriate price change, IE replace with newer better psu's, price increase on new "better" range, or a new cheaper range.

Without a link to the thread or a mention of what PSU's its pretty hard to speculate in this thread.

However in general any cards will have some fail, thats life, when a "good deal" is on and more people buy one product, and also have a thread about it, more problems get seen in a shorter space of time and more publically. Overclocked models DO tend to have higher DOA/problems because, no matter what people think 99% of companies put ZERO effort into speed binning cores to get the best ones.

Cores improve, manufacturing improves, at some point Gigabyte realise by testing a couple cores in every batch every couple months that now 80% are doing 820Mhz with ease, so they make a new SOC model and put cores in them, they don't check they do 820mhz individually. Because its closer to the edge, a few more will fail to make the grade, thats life. RMA< get a new one, chances of two bad in a row is incredibly small but still will happen once in a while, its not a huge deal at all.
 
^^ mmm not good reading - I'm havingto send mine back unfortunately - tried everything last night / this morning to no avail - stock volts @ 760 mhz seems to be the only stable clock speed (stock volts is 1.038)

I thought if it really can do 820 and its my PSU at fault (700W Tagan -= note when it was tested online - when 100% load it was drawing 840W from the wall according to review - in my system in Furmark - max I've seen with the 480 is 470, and with CPU overclock removed it was down to 350)

anyway - tried dropping core volts down 2 notches to 1.01V and 700mhz 480 stock - so my SOC is 820/1.038 I figured it should at least do stock speed with only a smal bump down in volts - nope loads and loads of errors in OCCT

if I can be guaranteed that its my PSU at fault - I'd buy a new one tomrorow - but I can't therefore I'm just having to send the card back under DSR

note 2 people having problems had 1000W PSUs ...
 
ERrm, its actually fairly rare for components to be completely changed, validation for a psu is expensive and takes time, if you change a SINGLE component inside the law requires the whole validation process to be completely redone. PSU's do change, its usually to do with a model no longer being produced so switching to an alternate, sometimes worse, sometimes better, often getting a new model launch with a (not always) appropriate price change, IE replace with newer better psu's, price increase on new "better" range, or a new cheaper range.

Can't think of a specific example to link to off the top of my head but been several instances of it especially with OCZ PSUs. What they seem to do is switch the entire internals from a certified PSU to another certified model while still keeping the outward brand/model i.e. the OCZ GameXStream IIRC originally had high quality Mushkin made internals, then swapped to high quality FSP model, but later quietly changed to a lower quality/cheaper FSP model (which wasn't half as good as the first 2 versions but was still sold as the same GameXStream models).
 
Last edited:
In defense of the Op, I did some quick stability tests on each of my GTX 480 SoC individually (only 1 installed in machine at one time) and got similar results to his on the opening level of Crysis.

His HD 5850 x-fire results are way off though. I was getting circa 65fps average in the same area I was getting circa 50fps average on the GTX 480 SoC (max settings, No AA, not sure on AF - not forced through control panel).

In same area, adding 4xAA dropped frames to circa 45fps average.

This is while standing still, moving will cause the frames to drop around 5fps (same or more on X-fired HD 5850).
 
hi,

Just got in from and ready to do some more benchmarks with the cards to find that when ever i put my system under load
my PSU makes a horrible whiney noise waht sounds like a baby crying!.
now i am sure my psu is on it's way out, so by the looks of things and how i've got to be tight with my money cuz of
my wedding, i am going DSR the 480 and get a new PSU (850w)
 
squealing PSU is normal (ish) for most modern day graphics cards and the load on the PSU - its nothing (in general) to worry about

my PSU squeeled with my 8800GTX, 4890, 570 and interestingly is quietest (although some squeeling still) with the 480 ! lol
 
squealing PSU is normal (ish) for most modern day graphics cards and the load on the PSU - its nothing (in general) to worry about

my PSU squeeled with my 8800GTX, 4890, 570 and interestingly is quietest (although some squeeling still) with the 480 ! lol

It just dont sound rite lol i rather not take that risk, I dont want to wake up in the morning to fun my pc in melt down mode But the noise is only when i run 3d mark software! :S..
 
Back
Top Bottom