Fallout New Vegas - Contemporary Graphics Cards Compared

[RussianSensation]

"With the highest level of detail and turned-on antialiasing the game runs perfectly well even on the rather inexpensive Radeon HD 6870 which is also the most economical card in this category."

"Radeon HD 6850 is let down by its cut-down GPU and reduced GPU clock rate. The GeForce GTX 460 1GB that wins in the performance/ mainstream category. If you want to save some money, you may prefer the GeForce GTX 460 768MB which is but slightly inferior to its 1GB cousin in most games. It is good enough for playing Fallout: New Vegas with all possible comfort."

"The Radeon HD 5700 series triumphs over the GeForce GTS 450, the senior Radeon HD 5770 model delivering much higher frame rates at resolutions of 1920x1080 and above and the junior Radeon HD 5750 being just as good as the Nvidia card in terms of bottom speed."

Full the article, including image quality comparisons, click here.

 

[hawtdawg]

That's a pretty poor showing by the 6850.

 

[Lonyo]

That's a pretty poor showing by the 6850.

Correction: pretty WEIRD showing by the 6850.

 

[exar333]

Correction: pretty WEIRD showing by the 6850.

Agreed. Driver issue possibly?

 

[exar333]

Great post OP. I wish more reviews had this depth showing what new and old cards can do. Thanks for posting this.

 

[Lonyo]

Agreed. Driver issue possibly?

Or they made a big mistake and tested the HD5830 twice by mistake.

http://techgage.com/article/amd_radeon_hd_6870_hd_6850/11

 

[blastingcap]

http://techgage.com/article/amd_radeon_hd_6870_hd_6850/11

I could just be seeing things but it seems as though this site disagrees with the other one on 6850 performance.

 

[nitromullet]

I think it's less an issue with the review and testing, than it is with the game itself. New Vegas is known to be somewhat buggy in general, and in my experience the frame rate feels pretty erratic. I could swear sometimes it hits the single digits, especially when exiting a dialog with an NPC or VATS, for no apparent reason. I imagine it's pretty difficult to get a consistent picture of performance in this game.

 

[blastingcap]

Or they made a big mistake and tested the HD5830 twice by mistake.

http://techgage.com/article/amd_radeon_hd_6870_hd_6850/11

You beat me to it (posting the link to techgage). Yeah I think xbit might have tested the 5830 twice, since those numbers are so close together. Haha.

 

[cusideabelincoln]

Those 6850 numbers... don't make any sense.

The 6870 is shown to be 60% faster than the 6850 when no other game shows that discrepancy and the average is only to be 15% faster:

Sheet 3
https://spreadsheets.google.com/ccc...zN6RUR3TkNRWTNzeE5CZkE&hl=en&authkey=COPFrYoD

I also would have expected the HD 5670 to do better in relation to the HD 5750 and GT 240.

 

[blastingcap]

Omg, did you put that spreadsheet together yourself? Haha.

I think it's clear that xbit messed up and benched the 5830 twice when it made these charts.

 

[hans007]

anyone else feel like new vegas is slower than fallout 3 by a lot? they would seem to be basically teh same engine, and i actaully have a slightly faster cpu than when i played fallout 3 (i had an athlon ii x2 250 then, and have a phenom ii x4 910 now) with a 4850, and fallout new vegas seems like its more sluggish, or at lesat in certain instances is.

 

[happy medium]

Gota love them 5970 scores, CF at its best again.

 

[GaiaHunter]

http://www.xbitlabs.com/articles/video/display/fallout-new-vegas_3.html#sect0

Performance was measured with a game save atop of a fully built Zerg base with a hundred plus army of Protoss opponent arriving to present their proposition.

Fallout always surprise me.

 

[Outrage]

Gota love them 5970 scores, CF at its best again.

cpu bottleneck....... cf works perfectly fine with this game.

 

[cusideabelincoln]

cpu bottleneck....... cf works perfectly fine with this game.

His bias prevents him from seeing what's really going on [he's looking for negatives to CF].

Most of the cards hit a wall at the mid 70 fps. I mean it's quite obvious as framerates don't change with resolution.

 

[Aristotelian]

His bias prevents him from seeing what's really going on [he's looking for negatives to CF].

Most of the cards hit a wall at the mid 70 fps. I mean it's quite obvious as framerates don't change with resolution.

Precisely, since the 5970 is on the whole faster than the 580.

 

[happy medium]

His bias prevents him from seeing what's really going on [he's looking for negatives to CF].

Most of the cards hit a wall at the mid 70 fps. I mean it's quite obvious as framerates don't change with resolution.

You could have just corrected me and exsplained to me that this game is easy on graphic cards and theres a cpu bottleneck.
No need for the personal attack.

 

[nyker96]

http://www.xbitlabs.com/articles/video/display/fallout-new-vegas_3.html#sect0



Fallout always surprise me.

yeah I saw that when I was reading the test spec. I think the writer just cut and pasted from Another Famous Game. :]

 

[RussianSensation]

Or they made a big mistake and tested the HD5830 twice by mistake.

http://techgage.com/article/amd_radeon_hd_6870_hd_6850/11

I also wondered if their HD6850 results are way off. I looked at TechGage's review and GTX460/470 actually have higher minimum framerates at 2560x1600 than at 1680x1050. Fallout NV testing produces some strange results.

Also, HD6870 is producing 49 fps min in the above review vs. 38 fps min at Xbitlabs. Most likely, the 2 benchmarks were ran at completely different settings.

Techgage: "but as a general rule we don't benchmark with AA if we have to force it through the driver panel, to avoid any inconsistent results in case any GPU vendor wants to get a little crafty."

So it's probably impossible to compare the results.

Overall, both HD5770 and HD6870 win their respective price/performance positions.

 

[cusideabelincoln]

I also wondered if their HD6850 results are way off. I looked at TechGage's review and GTX460/470 actually have higher minimum framerates at 2560x1600 than at 1680x1050. Fallout NV testing produces some strange results.

Also, HD6870 is producing 49 fps min in the above review vs. 38 fps min at Xbitlabs. Most likely, the 2 benchmarks were ran at completely different settings.

Techgage: "but as a general rule we don't benchmark with AA if we have to force it through the driver panel, to avoid any inconsistent results in case any GPU vendor wants to get a little crafty."

So it's probably impossible to compare the results.

Overall, both HD5770 and HD6870 win their respective price/performance positions.

The game looks CPU-heavy. Hence why, in the Techage article, at 1680 and 2560 all the cards practically get the same framerate. There's a spread from 63 fps to 69 fps from the slowest card at 2560 to the fastest card at 1680.

The minimums are also similar per card when switching resolutions. If this game is hard to bench then these numbers are probably within margin of error.

 

[emilyek]

Did they say anything about running d3d9 fix for the facegen problems?

I also noticed they ran Water Multisampling at something other than "Low" for the higher quality tests.

The engine is bugged-- you need the d3d9 fix so that the faces don't run in DX10, which will bring the game to a crawl if more than a couple are on-screen, and Water Multisampling set at something besides Low eats up fps all out of proportion to what it should be-- so something else is wrong.

My 8800GTX runs 2xAA and 15AF @ 1920 x1080 with all draw distances set to max and I probably get close to the same frames as any card in the linked article.

 

[Highmodulus]

Confirmed from personal experence- with the rig below I am running New Vegas 1920x1200 ultra with zero issues. But the Vegas code is still pretty buggy (big patch is in the works) and the 6800 series seems to have plenty of driver headroom based on all the hotflixes released already. Be interesting to see what the next "full" catalyst drivers provide.

 

[Rhezuss]

Correction: pretty WEIRD showing by the 6850.

That

 

[nemesismk2]

Did anyone see what the specs for the GeForce GTS 450 are in that review?

Are they basic 783mhz Core and 3600Mhz Memory or top of the range ie 930Mhz Core and 4000Mhz memory like what I have?