What's new
  • Please do not post any links until you have 3 posts as they will automatically be rejected to prevent SPAM. Many words are also blocked due to being used in SPAM Messages. Thanks!

NVIDIA GeForce GTX 470 SLI Review

Status
Not open for further replies.

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
8x MSAA Testing (BF: BC2 / DiRT 2)

8x MSAA Testing (BF: BC2 / DiRT 2)


In this section we take a number of games we have tested previously in this review and bring things to the next level by pushing the in-game MSAA up to 8x. All other methodologies remain the same.

BattleField: Bad Company 2 (DX11)

Note that 8x MSAA is enabled via the game’s config file for the NVIDIA cards since it is not a selectable option within the game menu

GTX470_SLI-42.jpg

While the GTX 470s loose ever so slightly to two HD 5850 cards, NVIDIA’s minimum framerates are a good 10% better which bodes well for a smooth gameplay experience.


DIRT 2 (DX11)

GTX470_SLI-49.jpg

Considering NVIDIA’s extremely good performance in this game, this result is a disappointment to say the least. It seems like GTX 470 simply chokes at this resolution.
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
8x MSAA Testing (Far Cry 2 / Dragon Age)

8x MSAA Testing (Dragon Age / Far Cry 2)


In this section we take a number of games we have tested previously in this review and bring things to the next level by pushing the in-game MSAA up to 8x. All other methodologies remain the same.

Dragon Age: Origins (DX9)

GTX470_SLI-56.jpg

Dragon Age does tend to like ATI cards more than their NVIDIA counterparts and the scores really do show this.


Far Cry 2 (DX10)

GTX470_SLI-69.jpg

What more is there to say? SLI seems to be the way to go for those of you who are still playing Far Cry 2.
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
Comparative Performance Testing

Comparative Performance Testing


To some of you this section may seem to be a bit redundant and it may be just that. However, what we are trying to accomplish here is to give you a quick and easy visual representation of performance across multiple settings without using dozens of charts. In addition, these following graphs can give a user clearer insight about a product’s AA and resolution scaling with just a quick glance. Trust us; there are several interesting and eye-opening charts here that you will want to see.


Aliens Versus Predator

GTX470_SLI-84.jpg


Battlefield: Bad Company 2

GTX470_SLI-85.jpg
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
Comparative Performance Testing (pg.2)

Comparative Performance Testing (pg.2)



DIRT 2

GTX470_SLI-86.jpg


Dragon Age: Origins

GTX470_SLI-83.jpg
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
Comparative Performance Testing (pg.3)

Comparative Performance Testing (page 3)



Far Cry 2

GTX470_SLI-87.jpg


GTX470_SLI-88.jpg


Metro 2033

GTX470_SLI-89.jpg
 
Last edited:

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
Where the Rubber Meets the Road: Multi Card Scaling

Where the Rubber Meets the Road: Multi Card Scaling


Framerate numbers may be great when it comes to overall performance but the bread and butter of any multi card setup is how well a given setup scales when going from one card to two (or more). Unfortunately, spending two times the amount of a single card for a Crossfire or SLI setup will not usually net a corresponding doubling of performance. The closer a given solution gets to that magical “100% scaling” mark, the more value a consumer will get for their money.

Minimum framerates are also important since we have come across many situations where an ultra expensive dual card system won’t necessarily return a better overall gaming experience. On average the increases may be impressive yet both Crossfire and SLI have been known to return larger framerate “dips” than single cards.

In order to determine how well a particular technology is implemented on the driver level, we have taken all of the average and minimum framerate differences between single and dual cards (for the GTX 470 and HD 5850). We were then able to determine how a card will scale at a given resolution and IQ setting. For any card that passed the single card test but failed the dual card benchmark, we used the single card score as both SLI and Crossfire can be easily disabled if a certain game is returning problems. We did this for both Average and Minimum framerates.

GTX470_SLI-73.jpg


GTX470_SLI-92.jpg

Let’s start off with the average framerates. Here, it is quite obvious that in most situations Crossfire is able to scale much better than its SLI counterpart. As we saw in the tests, this allowed the HD 5850 to nearly catch up to the GTX 470 when both were paired up with one of their siblings. There are certain resolutions that SLI is able to surge ahead and it is very important to remember that SLI is still an extremely scalable solution. The only problem NVIDIA has is that their strong performance is simply overshadowed by a strong showing by ATI.

Minimum framerate scaling is about even on average but ATI still has a lead over NVIDIA. This means the GTX 470 was able to hold onto its significant lead in this category for most (but not all) games.

The one important thing to take away from these tests is NVIDIA’s supposed driver immaturity. We expect performance in some of these games to take a (hopefully) leap with upcoming driver releases but since NVIDIA only posts WHQL-certified drivers once every quarter, it may be a while until we see increases for most games.
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
Core Temperature & Acoustics / Power Consumption

Core Temperature & Acoustics


For all temperature testing, the cards were placed on an open test bench with a single 120mm 1200RPM fan placed ~8” away from the heatsink. The ambient temperature was kept at a constant 22°C (+/- 0.5°C). If the ambient temperatures rose above 23°C at any time throughout the test, all benchmarking was stopped. For this test we use the 3DMark Batch Size test at it highest triangle count with 4xAA and 16xAF enabled and looped it for one hour to determine the peak load temperature as measured by GPU-Z.

For Idle tests, we let the system idle at the Vista desktop for 15 minutes and recorded the peak temperature.


GTX470_SLI-76.jpg

Temperatures are about what we would expect considering one of the cards is starved of fresh air while the other one has its airflow partially blocked. This mean the fans of each card have to pull double duty and as such, noise is significantly more than when running a single card. Over the boom and music of your typical action game this won’t cause any problems but it could get annoying if you were playing a slower-paced turn-based strategy game.


System Power Consumption


For this test we hooked up our power supply to a UPM power meter that will log the power consumption of the whole system twice every second. In order to stress the GPU as much as possible we once again use the Batch Render test (at 8xAA) in 3DMark06 and let it run for 30 minutes to determine the peak power consumption while letting the card sit at a stable Windows desktop for 30 minutes to determine the peak idle power consumption. We have also included several other tests as well.

Please note that after extensive testing, we have found that simply plugging in a power meter to a wall outlet or UPS will NOT give you accurate power consumption numbers due to slight changes in the input voltage. Thus we use a Tripp-Lite 1800W line conditioner between the 120V outlet and the power meter.

GTX470_SLI-75.jpg

You can also see that some of the results have increased a bit from past reviews as we slightly adjusted the testing regime for higher-end cards.

Overall efficiency is not one of the GTX 400 series’ strong suits and this chart illustrates it in glowing neon letters. 100W more than two HD 5850s is extreme to say the least considering the slight performance lead the NVIDIA cards have. Idle power consumption on the other hand isn’t bad at all.
 
Last edited:

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
Conclusion

Conclusion


If you take a look into the past, multi card setups have always been a somewhat temperamental solution to the never-ending quest for high performance. We remember a time when just getting SLI and Crossfire to run properly in the vast majority of applications was a lesson in frustration. Well, times have changed. While scaling still isn’t perfect across every game for a number of reasons, both NVIDIA’s and ATI’s technologies have matured to the point where we can consider them a viable option for people who want the best of the best. Naturally, there are still pitfalls as we saw with NVIDIA’s rendering problems in Aliens versus Predator (EDIT: Please note this is a game issue that affects NVIDIA cards and NOT a driver issue. It should be fixed in an upcoming patch.) ) and ATI’s image quality issues in Battlefield: Bad Company 2 but these scenarios are much less prevalent than they were about a year ago. That in itself bodes well for multi card setups.

Back when we first reviewed the GTX 470, we were impressed with its overall performance in comparison to the HD 5850 particularly when image quality settings were increased. Our observations regarding a framerate advantage haven’t changed when another card is added since for the most part two GTX 470s in SLI still routinely beat out a pair of HD 5850 cards.

However, the facts are quite clear: in general, ATI’s cards offer better multi GPU scaling than anything NVIDIA can currently offer. Believe it or not, NVIDIA’s SLI scaling is actually very, very good considering we saw on average a 60% to 81% increase when going from one card to two. We’re just not that excited about it because ATI threw a live hand grenade into the SLI party by posting performance increases that were simply jaw-dropping in nearly every respect. It is because of this dual card scaling that a HD 5850 Crossfire setup can make up a ton of lost ground when compared to ATI’s single GPU scores. On the other hand, the HD 5970 is simply trampled by the GTX 470 SLI setup.

GTX470_SLI-91.jpg

Luckily for NVIDIA, we didn’t stop at the average framerates and included (as usual) minimums as well. In this category the GTX 470 SLI is still head and shoulders above the competition when IQ settings are pushed since it doesn’t exhibit the numerous dips in performance that characterize the Crossfire experience. I don’t know about you but I would much rather have a slightly lower average framerates if it means my system won’t bog down when I’m fighting for my life in a game. Basically, a fluid gaming experience is what SLI is ideal for…or at least until ATI can figure out a way to bring up their minimum framerates in some games.

Considering the drivers we were using for NVIDIA’s cards aren’t exactly what we would call mature, the stability these two GTX 470s exhibited was nothing short of praise-worthy. There was a minor hiccup when it came to rendering some overhead objects in AvP without AA enabled but other than that, sailing was smooth. ATI’s solution on the other hand had numerous issues in Aliens versus Predator, Metro 2033 and Battlefield: Bad Company 2, only of which the latter didn’t affect gameplay too much. There were also problems with the drivers (10.3a and 10.2 WHQL) actually detecting both cards which seems to be a reoccurring frustration with every second driver from ATI’s team. NVIDIA’s average SLI scaling might not be up to the level of Crossfire but its seamless implementation has yet to be equaled by the competition.

While performance is good and minimum framerates are awe-inspiring, our main concerns about a GTX 470 SLI lie in two areas: price and power consumption. Both aspects wouldn’t have been too much of a concern if NVIDIA had managed to carry their average single card performance lead over the HD 5850 into multi card benchmarks. Unfortunately, that didn’t happen so as it stands we feel that spending $700 for two GTX 470 cards is a hard sell when you can buy a similarly performing solution from ATI for more than $100 less. When you add a power consumption gap of more than 100 watts into the equation, it becomes extremely hard to recommend a GTX 470 SLI setup at this point in time for someone who isn’t hell-bent on going with NVIDIA cards.

It is quite evident that while the GTX 470 is a clear winner against a single HD 5850, SLI is currently struggling to mirror the downright amazing scaling that Crossfire offers. Overall framerates for a GTX 470 SLI system are still incredible but the HD 5850 in Crossfire is simply too close for comfort in most cases. It is also important to remember that NVIDIA’s drives are still quite immature and multi card performance is usually one of the first things addressed when new versions are released. As such we intend to take another look at GTX 470 SLI performance in the future. While we can’t outright recommend going SLI in this case, it offers enough convincing performance wins that it should be at least considered by enthusiasts.



 
Last edited:
Status
Not open for further replies.

Latest posts

Top