ATI Radeon HD 4890 vs. Nvidia GeForce GTX 275

Julio Franco

Posts: 8,664   +1,538
Staff member
Gamers looking to spend some $250 on a brand new graphics card were given two new options last month when ATI unleashed their Radeon HD 4890 and Nvidia countered with the GeForce GTX 275.

Today we will put both cards to the test and find out which is faster in a wide range of games using the latest drivers. We will also be looking at their maximum overclocked performance to help you decide which is the best value option for enthusiasts and PC gamers alike.

Read the full article at:
https://www.techspot.com/review/164-radeon-4890-vs-geforce-gtx275/

Please leave your feedback here. Thanks!
 

Captain828

Posts: 297   +22
Was really looking forward to a comparison review of those two; Great job and terrific article! :grinthumb

But I have to ask, why did you use Dead Space for the stress test? everyone knows Furmark can truly stress a GPU at 100% and it's also more consistent as well.
 

Steve

Posts: 2,651   +2,592
Staff member
Thanks for the feedback!

Dead Space or any first person shooter for that matter will keep the GPU at 100% load constantly. 3Dmark drops off between scenes.
 

Captain828

Posts: 297   +22
I said FurMark... it has nothing to do at all with Futuremark's benchmarking software, even though it has the 'Mark' moniker.

Here's some quick info about FurMark:
FurMark is a very intensive OpenGL benchmark that uses fur rendering algorithms to measure the performance of the graphics card. Fur rendering is especially adapted to overheat the GPU and that's why FurMark is also a perfect stability and stress test tool (also called GPU burner) for the graphics card.

The benchmark offers several options allowing the user to tweak the rendering: fullscreen / windowed mode, MSAA selection, window size, duration.
And... no; only Crysis manages to get near the 100% load, and that using extreme settings. I know there was an article somewhere about this, but I forgot where and I'm kinda lazy now.
Dead Space is probably one of the worst examples since it doesn't even need strong GPUs @ maxed settings.
 

Steve

Posts: 2,651   +2,592
Staff member
Ahh sorry I miss read. Yes Furmark is very good and we do also use it. That said we have also noticed that games tend to get the same power load results and the same operating temperatures and since most gamers play games and not Furmark we like to use the games. That said the results are cross referenced with tools like Furmark and ATItools for example.
 
G

Guest

alrighty, im curious to see where in the hell you are comming up with these benchmarks for crysis warhead. i have a amd phenom II 920 with 4gigs of ddr2 1066 ram and a ati hd4830. i run crysis warhead on all gamer settings, aka high settings for crysis. and i average 30fps in 1920 x 1080. So now lets get back to where your hd 4870x2 is only running at 32fps. hmmmmmmmm and your 4890 is only 28fps hmmmmmm once again. Are you trying to run the cards on a nvidia north bridge with no drivers installed or what? Once again terrible reviews of these cards. now if you are meaning enthuesist settings when you say high for warhead. then its alittle closer to reality. Is your test platform a pentium 4? but pls you can let me know how you came up with those benchmarks. oh and my system is all stock no overclock. i over clocked my card to 690 and 1020 and i averaged almost 40fps in all gamer with half of my settings on enthuesist. Im just so mind boggled i guess i have a unique 4830 that out performs 4870x2's and stuff. Have a nice day and maybe try and get it right next time.
 
G

Guest

Oh and i forgot to add that i am running windows vista 64bit with dx10
 

Captain828

Posts: 297   +22
As of when you've posted, the latest ATi Drivers are 9.6 and the latest nVidia ones are 186.18 WHQL. The ones used in the review were both Beta drivers, so any anomaly that occurs with them is strictly because they're Beta.

If you would've paid more attention, Crysis, which uses the same engine as Warhead, has different performance numbers. Also, for Crysis, the graph shows 4xAA and the HD4890 wins, so it's pretty clear that there's something wrong with the Catalyst driver.

Why do you feel like bashing? the specs were clearly stated... did reading this review hurt your fan boy ego too much?
 

Steve

Posts: 2,651   +2,592
Staff member
Why do you feel like bashing? the specs were clearly stated... did reading this review hurt your fan boy ego too much?
Well said Captain, its sad when people tarnish the guest account like that ;)

Still if I was going to make such an outrageous comment I would want to do it anonymously as well.

Anyway it’s fun to compare oranges and apples.
 
G

Guest

What card was the GTX 260? Was that a core 216? The stock original nVidia GTX 260? Need more info!
 

Steve

Posts: 2,651   +2,592
Staff member
We dropped the original GTX 260 as soon as the 216SP version was introduced.
 
G

Guest

Well said Captain, its sad when people tarnish the guest account like that ;)

Still if I was going to make such an outrageous comment I would want to do it anonymously as well.

Anyway it’s fun to compare oranges and apples.
Well this guest would just like to thank you for posting a review that compared green and red, and had the common sense and fairness to actually OVERCLOCK BOTH OF THE CARDS !

Yes, I know it sounds completely obvious, but you'd be surprised ( or maybe you wouldn't be) at the number of highly poignant review sites that wind up doing dirty things like testing a new red card with the triple xxx vapor cooler in an overclocked mode then comparing it to green stock and declaring the red rooster card the big winner.

So anyway thanks for having the sense to overclock both cards - it's amazing what CRAP exists elsewhere on the net.
 

CMH

Posts: 2,050   +14
Just a small note: Make the graphs much easier to read by highlighting the main cards. Put the bars in a different color or something....


Other than that, I think I read this review a little late....
 

Steve

Posts: 2,651   +2,592
Staff member
Just a small note: Make the graphs much easier to read by highlighting the main cards. Put the bars in a different color or something....


Other than that, I think I read this review a little late....
Were you trying to be funny? The green team card was highlighted green and the red team, red!
 

CMH

Posts: 2,050   +14
Oh wait, it was.

However, not for all the graphs (check power consumption graphs)
 

Steve

Posts: 2,651   +2,592
Staff member
Your right 2 of the 47 graphs didn't feature the colored labels ;)
 

CMH

Posts: 2,050   +14
I was happily reading it till that last 2 graphs, and realised I had to slowly scrutinize the graphs to see which ones they were, and jumped here, thats why I kinda didn't realise the rest were already coloured :D:D:D
 
G

Guest

Well said Captain, its sad when people tarnish the guest account like that ;)

Still if I was going to make such an outrageous comment I would want to do it anonymously as well.

Anyway it’s fun to compare oranges and apples.

Yeah...and also after doing your homework properly ;) Thanks for the review. It was very helpful. If you could let me know of the performance gap along with the 1792 evga GTX 275
 
G

Guest

Thanks for the review. I am an avid NVidia consumer and have been considering checking out ATI for the first time, but now I see I made the right choice all along. I have a 1GB ASUS GTS250 and runs GTA IV and Battlefield:BC2 at max settings no problem. I was anxious for the FERMI cards from NVidia and sounds really cool, when you gonna test the GTX480 (I think) that was recently released? Looking forward to a review.