Nvidia GeForce GTX 470 review
in Graphics cards
Verdict
Ludicrous power draw wipes out the small advantage gained in gaming tests, and the price needs to drop
Review Date: 26 Mar 2010
Reviewed By: David Bayon
Price when reviewed: £255 (£299 inc VAT)
Features & Design
![]()
Value for Money
![]()
Performance
![]()
Just beneath the flagship GTX 480 sits what Nvidia hopes will be its new upper-mainstream challenger, the GTX 470 - but it's not going to come cheap. Whether it can justify the price will depend largely on what you intend to do with it.
The GeForce GTX 470 has 448 stream processors, along with a 607MHz core clock and 1.25GB of GDDR5 memory with a 320-bit memory interface. It's similar in size and performance to ATI's Radeon HD 5850.
As with the GTX 480, the new Fermi card does have the slight edge over its nearest ATI rival in games. In Crysis at 1,920 x 1,200 and Very High settings, the GTX 470 averaged 33fps to the HD 5850's 32fps; higher settings saw similar margins. World in Conflict had the Nvidia card consistently ahead by just under 20%, and in Stalker: Call of Pripyat that margin was around 5%. Other games had ATI's card ahead by a whisker, and if we average all the results, Nvidia's edge looks to be between 5% and 10%.
(Sample framerates at 1,920 x 1,200 at Very High settings.)
Like the GTX 480, power consumption is a real concern. Nvidia quotes a TDP of 215W, and with it installed in our test rig the whole system idled at 131W and peaked at a massive 380W. By comparison, with the HD 5850 installed in the same rig we measured 122W when idle but a much lower 231W at full load - that's an unbelievable 60% less than with the Fermi card installed.
Nvidia Fermi full review
Click here to read the in-depth verdict on Fermi's architecture and its GPGPU capabilities.As we cover in the full Fermi review, GPGPU applications are also a focus. Using CyberLink PowerDirector 8 we rendered out a 1080p AVCHD file with a single accelerated filter effect applied to it, and found the GTX 470 was 35% faster with acceleration than without. The ATI card was boosted by a similar percentage using ATI Stream; given Nvidia's hype over the technology it isn't the result we were hoping for.
But it's gaming performance that most readers will want from a GeForce card, and although the GTX 470 does best its ATI rival by a small amount on average, its £299 inc VAT expected price puts it a good £70 above the HD 5850. It'll take a significant price drop before we'd consider opting for such a noisy, power-hungry card over the existing competition.
Author: David Bayon
From around the web
On the Back Foot
I was hoping that the Fermi cards would make a better fist of competing on Price per /watt /processing power than this with ATI.Not that i'm a fanboy as such.It's just that i would like some pressure put on ATI to bring it's 57xx /58xx prices down a bit.No chance of that now looking at these cards.If anything they may raise them !!!
By Jaberwocky on 27 Mar 2010 ![]()
I think this is good. They've had a good run of domination and now ATI has had a good 6 - 12 months up the top. A dud from Nvidia at this point may now make them go away and create something special.
By TimoGunt on 28 Mar 2010 ![]()
Slack reporting!
I quote...
" In Crysis at 1,920 x 1,200 and Very High settings, the GTX 470 averaged 33fps to the HD 5850's 32fps; higher settings saw similar margins. World in Conflict had the Nvidia card consistently ahead by just under 20%, and in Stalker: Call of Pripyat that margin was around 5%"
Since when did a 1 fps, of 33 advantage over 32, equate to a 20% (twenty percent) advantage as a whole, in MOST scenarios?
In your World in Conflict graph the 470 has an approx. 8fps advantage. That equals 13.5% but bears no relation to the Crysis figure. In Stalker it appears to be 55 v 60 or 8%. Neither are 20% or even support a blanket 10% claim of faster 'cos none of your figures even peak at plus 10%, if a one frame advantage in Crysis equals +/- 3%!
***kin' poor.
As is plastered all over the net, Nvidia's latest creatures run hot, suck up electricity like it was the 1950's, do a vasty good impression of a jet engine, cost the Earth (and yes that is also a reference to their TDP) and offer only a 5-10% performance improvement, at SOME but NOT ALL resolutions, to their Ati equivalents.
Your summarization is way too lite, in accurate and undeserving of a publication of such a previously held, high reputation!
Again, ***kin' poor!
By fingerbob69 on 31 Mar 2010 ![]()
Benchmarks
@fingerbob69
I appreciate your point, but the graphs on this review are simply a single comparison sample from each game, not our entire set of results.
We actually ran each game test at between five and seven different resolutions, and the most useful figures - i.e. not including the triple-figure scores at low resolutions - averaged out to the figures stated in the review. Some of our tests had Nvidia ahead by up to 20%, others had no lead at all.
With regards to your final points, I'm not sure what you're getting at. We agreed with most of the net in saying 5-10% on the whole.
By DavidBayon on 31 Mar 2010 ![]()
its all about info
give us more info and allow us the buyers to make our informed minds up . what about adding aa 4x does this impact on the performace of these cards the gtx 470 is faster in most games with aa on than the 5870 and if your paying 300 quid or more you should be applying aa at every resolution
By j7dunwiddie on 3 Jul 2010 ![]()
advertisement
- Google legal chief: privacy laws too hard on SMBs
- No free Visual Studio for Windows 8 desktop developers
- Facebook spends $1bn on Instagram... then launches its own Camera app
- Who sends Google the most takedown notices? Microsoft
- Microsoft wins text patent battle against Motorola
- Watchdog fines firm £50,000 over Android malware
- Intel to test smartcity future on London
- June decision on Microsoft's billion-dollar EU fine
- Yahoo browser launch marred by security flaw
- Autonomy management walk out over HP bureaucracy
- Laptop bag reviews: nine tested
- Sony VAIO T Series Ultrabook review: first look
- Revealed: the military standards and robots HP uses to test its laptops
- Windows 8: multi-monitors and double standards?
- Why is TalkTalk's year-old porn filter suddenly big news?
- Why are laptop screens so far behind mobiles?
- HP EliteBook Folio review: first look
- The shoebox-sized all-in-one printer
- Forget the Ultrabook: here comes the HP Sleekbook
- HP Spectre XT review: first look
- Can you buy technology with a clean conscience?
- The death of email
- How to use Windows 8 Metro
- 30 best features of Windows 8
- How to become a cyberspy
- Create your own smart home
- Install a custom ROM on your smartphone
- Can the Raspberry Pi save computing?
- Google: the pirates' best friend?
- Backups: ten tips to keep your data safe
- Why you have to be left in the dark on OS patches
- Is Microsoft mismanaging Windows on ARM?
- Dealing with spam surrogates
- Why 3G broadband can be better and cheaper than ADSL
- Is Twitter bad for business?
- Publishing your email address isn't a security disaster
- Why you'll need a fax machine to develop iOS apps
- Learning to adapt to the mobile web
- Why you shouldn't use WPS on your Wi-Fi network
- Disabled users suffer when software breaks the rules
advertisement







