Associate
- Joined
- 4 Dec 2009
- Posts
- 780
Anyone got experience with this? 
Is the scaling finally working properly or is it still rubbish?
Is the scaling finally working properly or is it still rubbish?
-
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.
GTX 470/480 Tri/Quad SLi
- Thread starter Prof. Cheese
- Start date
Associate
- Joined
- 20 Dec 2003
- Posts
- 557
- Location
- Leeds, UK.
This review is from Monday:
http://www.tomshardware.co.uk/geforce-gtx-480-3-way-sli-crossfire,review-31886.html
"But in order to get enough airflow through the card and out its rear-facing exhaust, it has to spin those Delta blowers incredibly fast—fast enough to generate the sort of noise I can guarantee you don’t want to hear."
http://www.tomshardware.co.uk/geforce-gtx-480-3-way-sli-crossfire,review-31886.html
"But in order to get enough airflow through the card and out its rear-facing exhaust, it has to spin those Delta blowers incredibly fast—fast enough to generate the sort of noise I can guarantee you don’t want to hear."
This review is from Monday:
http://www.tomshardware.co.uk/geforce-gtx-480-3-way-sli-crossfire,review-31886.html
"But in order to get enough airflow through the card and out its rear-facing exhaust, it has to spin those Delta blowers incredibly fast—fast enough to generate the sort of noise I can guarantee you don’t want to hear."
This is why I reccomend the GTX470 in SLI over 5850CF/5970 even tho the single GTX470 is hard to justify.
It seems a lot depends on the res. In tom hardware review tri sli had worst framerates than dual sli in some games
How does a review of gtx480 in sli allow you to justify that gtx470 in sli are so good?
I;m not saying that they are not a potent force in sli, it's just that review results are lacking.
From what I can see, xfire and sli scales pretty much evenly depending on the game.
Therefore if a 5850 beats a gtx470, two 5850's in xfire will beat two gtx470's in SLI and price difference is even bigger.
The gtx 470 is overpriced, fact. Buying two doesn't make that statement any less valid.
It seems a lot depends on the res. In tom hardware review tri sli had worst framerates than dual sli in some games
How does a review of gtx480 in sli allow you to justify that gtx470 in sli are so good?
I;m not saying that they are not a potent force in sli, it's just that review results are lacking.
From what I can see, xfire and sli scales pretty much evenly depending on the game.
Therefore if a 5850 beats a gtx470, two 5850's in xfire will beat two gtx470's in SLI and price difference is even bigger.
The gtx 470 is overpriced, fact. Buying two doesn't make that statement any less valid.
According to the write up at the end
http://www.tomshardware.co.uk/geforce-gtx-480-3-way-sli-crossfire,review-31886-13.html
CF scaling is all over the place.
Last edited:
Yeah, in one game which is a Nvidia title
Apart from that xfire is around 70% as well.
So buy gtx 470 in SLI if you want the best performance out of Just Cause 2. Other than that, gtx 470 SLI will be slower than 5850 xfire and cost more.
I want to see proper gtx 470 SLI results. I fail to see how you can so recommend 100% as the solution to everything. Do you even own two gtx 470?
Shame. There does seem to be a complete absence of decent gtx470/5850 sli/xfire with latest drivers review.
I am actually quite tempted to buy one gtx 470 with the intenetion of another one later when I go i7.
LoL toms can't even bench crysis with xfire at max rez when every other site can, says it all really, toms suck and not to be trusted.
http://www.techpowerup.com/reviews/ATI/Radeon_HD_5870_CrossFire/7.html
And now toms results..lol, I can post up many more reviews showing xfire 5870 scaling as above.
http://www.techpowerup.com/reviews/ATI/Radeon_HD_5870_CrossFire/7.html
And now toms results..lol, I can post up many more reviews showing xfire 5870 scaling as above.
Last edited:
I'd love to do a review, but I'm not paying for a pair of 5850s that I'd never use myself :S
Admittedly I've only benched RE5, DMC4 and crysis but I'm convinced from what I've seen that its the superior multi GPU setup without the silly cost of a pair of 480s especially with the 256+ nVidia drivers increasing SLI performance and compatibility.
Admittedly I've only benched RE5, DMC4 and crysis but I'm convinced from what I've seen that its the superior multi GPU setup without the silly cost of a pair of 480s especially with the 256+ nVidia drivers increasing SLI performance and compatibility.
Well I'm not a great fan of toms - but without knowing the details of the settings its hard to draw a comparision they may have been attempting to run higher/different settings one may even be DX9 v DX10 settings.
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
Yes, thats fantastic, except Tom clearly had problems with Crysis, and he choose the worst resolution at 1920x1200 Just cause shows the same scaling between Sli and xfire, multiple other reviews show NO issues with crossfire in Crysis, Tom was just being a douchebag, 20 websites can make it work, one website can't make it work......... justifiably crap in Crysis or Tom being incompetant, is it really hard to draw the correct conclusion?
So by choose the 5870's worst resolution(because of memory) and comparing it to Nvidia's best(because of memory) and including some results that simply aren't matched by the majority of websites they show SLI has amazing scaling and Xfire to not have such significant scaling.
But this is the problem a 470GTX IS NOT a 480gtx, the 480gtx has a 15% clock speed advantage, higher memory bus and MORE MEMORY.
So Nvidia's biggest wins due to memory scaling at the highest res possible, simply won't be comparable to a 470gtx with less memory, less bandwidth and less clock speeds.
Using 480gtx sli scaling and results in extreme best case scenario is NOT remotely representative of what 470gtx sli can do.
Firstly the best case scenario SCALING will be worse because the scaling came in the worst memory situations Metro results being absolutely ridiculous are purely a memory problem, nothing more or less, several other games at 2560x1600 + 4xaa will be memory limited.
SO Scaling will be worse, on top of circa 20% worse performance in single AND SLI modes, because both cards are slower. Woo, how great are the 470gtx's.
Considering how god damned good 5870's are in xfire at a much more used 1920x1200, and how much worse SLI scales comparitively at non memory limited resolutions, the 5870cf becomes EASILY the better choice, and even better is the 5850cf which get painfully close to 5870 performance when both overclocked at an even cheaper price.
480gtx sli actually scales very nicely at high res, but is less good at 1920x1200. 470gtx sli will NOT scale the same, and pretending it will is, daft.
I'd point out that, if 480gtx's cost £300 each, I'd be incredibly tempted by 480gtx sli(or a single card), if the 470gtx cost £200, I'd be very very tempted by SLI, but cherry picking sli results to make it seem better than it is, or making xfire seem worse than it is, is simply unhelpful to people asking serious questions and decided to spend money based on your advice.
470gtx sli and 480gtx sli is great, but its NOT significantly better than xfire and not to a point that spending an extra £140 gives youy £140 higher performance.
Infact overclocked everything to within an inch of their lives, even with worse scaling 5850cf(the right ones, which admitedly are harder to come by, especially cheaply) can beat the 470gtx sli because each card is faster overclocked and though it gain less in xfire. If you increase from 100fps by 60%, you'll get better performance than two cards doing 80fps that scale at 85%, 160fps vs 148fps.
Its not about scaling %, 5850 cf offers by FAR the best bang for buck and the single most important question you need to ask from that point on is............ do 5850's in xfire provide less performance than I need, if yes, do they still provide less than you need overclocked, if the answer is yes, by all means look at faster and more expensive cards. If the answer is no, if you don't need to go from 100fps to 120fps in Metro, then why spend more than you need to?
If the situations were reversed and Nvidia were slower, but signifcantly cheaper and offered the performance you NEED for less money, I'd recommend them in a heartbeat, the only problem is, they don't.
This is why I've also mentioned a X2 type card on GF104 could enable a 480gtx performance card for a LOT cheaper and thats incredibly interesting, its all about paying for what you need. If Nvidia brings prices down so they are more expensive by the same % they are faster, they'll start being worth recommending, if they come down in price further than that, they'll be excellent buys.
Now the big advantage(and bad thing for Nvidia) is stock on shelves, few people buying and losing money, meaning Nvidia has a reason to drop prices to at least sell the things. If we all ignore the cards for another month they'll drop in price again, significantly, they might actually become a decent buy eventually.
Last edited:
What impartial reviews? the closest I can think of is anandtech and they don't really do comprehensive enough results for my tastes... TPU is certainly not impartial.
http://www.anandtech.com/show/3745/nvidias-geforce-gtx-465/
Good set of benchmarks there, unfortunatly no 470 SLI but some interesting results none the less.
Lesser cards in SLI will typically (appear to) scale far better due to being more GPU limited than CPU limited, which the top end cards often bump into with multi GPU scaling. Personally I ignore 2560x results anyway as not many people run that kinda setup and theres way too many variables such as VRAM limitations.
