AMD’s Radeon HD 5770 & 5750: DirectX 11 for the Mainstream Crowd
by Ryan Smith on October 13, 2009 12:00 AM EST- Posted in
- GPUs
Crysis: Warhead
Kicking things off, we’ll start with Crysis: Warhead. Warhead is still the single most demanding game in our arsenal, with cards continuing to struggle to put out a playable frame rate with everything turned up.
Update: As a few of you pointed out, there was something a bit off with our Crysis results; we had a Radeon 4850 beating the 5770. As it turns out we wrote down the maximum framerate for the 4850 instead of the average framerate. None of the other results were affected, and this has been corrected. Sorry, folks.
There are a few different situations we’re going to be interested in. The first is the matchup between the 5770, the 4870, and the GTX 260. The second is the matchup between the 5750, the 4850, and the GTS 250. The third is the 5770 as compared to the 5800 series, in order to see what another $100 or $200 is buying you in the Evergreen family.
Unfortunate for the 5770, this is not a game that treats it well. In spite of the clock speed advantage over the 4870, and the architectural advantages (extra caches and what-not), it underperforms the 4870 by about 15% here. AMD had once told us that they believed that they weren’t memory bandwidth constrained on the 4870/4890, but when that’s the only significant difference between the 5770 and the 4870 that would explain the performance difference (certainly Juniper wouldn’t be slower than RV770), we are beginning to doubt that. Meanwhile the GTX 260 outscores the 5770 here too.
Looking at the 5770 compared to the 5850, $100 buys you roughly 50% more performance.
The 5750 fares much better here. It beats the 4850 by 10%-20%, and beats the GTS 250 by a similar margin.
Facebook
Twitter
RSS
117 Comments
View All Comments
strikeback03 - Tuesday, October 13, 2009 - link
10 hrs a day is modest? That seems high to me, unless you are doing work that pays on this, I would think most people don't have 10hrs a day for recreational computing.Mint - Tuesday, October 13, 2009 - link
We're not talking about most people, we're talking about people who bother to get a 5770 instead of living with IGPs. Many people leave their computer on 24/7 to download torrents or fold or act as a file server (it's nice to access it from work) or whatever. I think 10 hours is a reasonable average for the target audience.Even if you reduce it to 5 hours a day, though, that's still $8/year. I like to keep video cards for a long time (usu. 2 years or more), and even when I upgrade, the old one is usually handed down.
My point is that it's not something to ignore when comparing to the 4870. It was much less relevent for $300 cards with a 20W-30W difference (4870 vs GTX260 at launch), but now it's a 50W difference for $150 cards.
UNHchabo - Wednesday, October 14, 2009 - link
Personally, I wish that the 4770 had been included in the power charts. It may be a largely irrelevant card for price/performance, but it's still the cheapest 40nm card that AMD makes.Zingam - Tuesday, October 13, 2009 - link
Real competition does wonderful things! If NVIDIA hasn't done it so great with 8800, we would never had these great prices by ATI today!Unfortunately there is nothing like that on the CPU side. :(
MadMan007 - Tuesday, October 13, 2009 - link
Is the GTS 250 512MB or 1GB? It's not even stated in the test setup notes.Ryan Smith - Tuesday, October 13, 2009 - link
1GB.Adul - Tuesday, October 13, 2009 - link
http://www.monoprice.com/products/product.asp?c_id...">http://www.monoprice.com/products/produ...1&p_...As long as the video card supports outputting hdmi through the display port this will do. So the question is does it support hdmi signals through the display port?
Ryan Smith - Tuesday, October 13, 2009 - link
Passive dongles are not supported on the 5000 series. It has to be an active dongle.danielkza - Tuesday, October 13, 2009 - link
There's a typo in page 5, I think you meant 'GTS 250' instead of 'GTX 250' (1st paragraph after the charts)Skiprudder - Tuesday, October 13, 2009 - link
Thanks for the review!I guess I'm rather surprised at the 5770 results being consistently lower than the 4870 as well, and would be interested in a a bit more hypothesizing as to why exactly this is the case when the stats on the cards suggest they should be at minimum roughly equivalent. Is this situation the sort of thing that might see large changes with updated versions of Catalyst?