-
We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.
Another GTX 260 vs 4870 review
Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
1. You're using the old drivers, not the latest ones that many sites have shown greatly improve the performance of the NVIDIA cards. (of course the other side it's a 4850 vs a 260)
2. I'm not even going to load up ancient relics like "Doom3" to verify your results, but for example, on the OpenGL page the only 8XAA difference that matters at all is the Quake Wars where the 260s advantage got it over 30fps average. (and even that is pretty irrelevant because you wouldn't want to play Quake Wars at 25 or 30fps- you'd be toast! The two 4850 "victories" have the 260 at 74 and 79fps average- so who cares? )
3. This is like Kyle Bennett moderating here and then saying "go to my site instead of this one for your hardware info!", fairly unique in the forum world!
Anyway, it a nice looking review BFG. Personally I wouldn't have wasted my time on some of those old games because I think you're the only one who plays them, but it's a good looking site and I don't doubt your results. Looks well done, grats.
BFG10K
Lifer
That's quite true, but then would you expect any of those older titles to show performance gains with Big Bang? I wouldn?t, but I sure hope so, and it'll certainly be interesting to find out. 😉
You're missing the point. The specs of the 4850 should not allow it to get anywhere near the GTX260+, yet it's administering a beating in some situations. In Quake 4 for example the 260+ is almost 19% slower than the lowly 4850 when running 8xAA.
Not at all. I'm not moderating anyone. Other reviewers have posted figures and I posted mine. I simply added more figures in the mix and it's up to the reader to draw their own conclusions.
This is just the beginning. 🙂
error8
Diamond Member
I don't trust those benchmarks, that's all. I'm not an ATI fanboy and I know that after the "big bang" drivers, the core 216 is the faster card, overall. If a site like Anandtech will provide me some benchmarks, clearly showing that the GTX is faster then the 4870, when 8X AA is used, then I'll believe it. Or if someone on this forums will make some benchmarks at 8XAA, someone except you of course 😉 .
A person needs to remember that 8XMSAA is not the only possibility for greater then 4X MSAA with NVIDIA products.'
NVIDIA's CSAA modes offer slightly less precise 8X and 16X AA method, and their transparency modes are unrivalled for image quality AFAIK.
At 8X CSAA the NVIDIA part may well offer IQ that isn't readily discernible from 8X MSAA in many situations, yet offers superior performance in all due to the halved Z and color samples, but identical texture and coverage sampling.
IIRC, ATi's coverage sample method is the ghastly blur inducing tent mode.
Aaaiiiiieeeeee- too many embedded quotes! It burns! It burns!
The problem is Anandtech seems to be the only site on the web saying the GTX260s are not improved with the new drivers, and faster than the 4870s now. Whereas I and others have linked to lots of sites that show the GTX260C216 as faster.
I mean no disrespect to Derek as I've always enjoyed the reviews here, but in this case I'm going to believe the rest of the web over Derek/AT. I'd say the same if the situation was that Derek was the only guy saying the NVIDIA card was faster.
The only thing I can think of is Derek's benches happened to hit on areas of the game that aren't improved by whatever optomizations NVIDIA did. When I look at nV's benching recommendations with things like "set it all to highest, put it in god mode, run around in the big open area getting shot and then set off some huge explosions at spot X" I don't think they're trying hide anything, it just looks like they want you to replicate game play/stess the gpu.
The problem is Anandtech seems to be the only site on the web saying the GTX260s are not improved with the new drivers, and faster than the 4870s now. Whereas I and others have linked to lots of sites that show the GTX260C216 as faster.
I mean no disrespect to Derek as I've always enjoyed the reviews here, but in this case I'm going to believe the rest of the web over Derek/AT. I'd say the same if the situation was that Derek was the only guy saying the NVIDIA card was faster.
The only thing I can think of is Derek's benches happened to hit on areas of the game that aren't improved by whatever optomizations NVIDIA did. When I look at nV's benching recommendations with things like "set it all to highest, put it in god mode, run around in the big open area getting shot and then set off some huge explosions at spot X" I don't think they're trying hide anything, it just looks like they want you to replicate game play/stess the gpu.
While its legitimate to discount those results because they're using old drivers irrelevant to this discussion, mocking them because they're not done by "pros" isn't. Not sure how you expect anyone to take your home-brewed results seriously if you're going to be so quick to mock others. I mean lets be honest here, it doesn't take any special degree or even proper grammer to benchmark video cards and write reviews, the only difference between those kids goofing around and you is that they probably had to pay for their video cards. 😉
So why even bother injecting old results into discussions that clearly focus on improvements with Release 180.48? Let's stick to the relevant reviews/results instead of trying to prove points no one is disputing.
Here's some relevant results:
PCGH 178.24 to 180.48
Shows there are gains in both old and new titles, but not every title. AC and GRiD show nice gains for older titles, L4D and FC2 for newer ones.
HWC 180.43 to 180.48 and 8.10 to 8.11
Shows significant gains even from 180.43 to 180.48 in FC2, which shows results that claim they used Release 180 drivers but not 180.48 are not fully valid either.
What we have so far is all review sites using 180.48 Drivers showing marginal to significant gains in numerous, but not all titles. And then we have AT, who doesn't show any gains, but does not disclose what drivers were used, what test bench was used amongst many other problems in their recent testing.
BenSkywalker
Diamond Member
Nice review BFG, I would take exceptions to a couple of things in there though-
Emphasis mine, don't you mean only? We can get into current releases but let's say ATi is having a very rough time with their drivers and new games at the moment. Derek had to give up even more journalistic integrity to protect ATi in his latest blog about FC2 performance due to ATi driver issues. Catalyst AI, on or off btw? I saw you forced HQ in nV's control panel, was that just for mips or for all settings? I ask mainly as having HQ forced in nV drivers forces off base filtering optimizations while ATi will still have them enabled with Cat AI on AFAIK(you can't turn off the AF hack, but it used to at least clear up some visible mip transitions with trilinear on).
Also, the legacy OS thing 😉 DX9 is quaint man, it really is, but unless you want to focus ONLY on legacy gaming then the impact of testing second/third generation DX10 parts on an OS that can't even handle it is going to negatively impact how much importance it has to potential consumers.
Minor quibbles though, obviously your testing methodology utterly trounces most web sites as usual and of course the ever important hands on feedback is paramount- Kick ass job man 😀
Emphasis mine, don't you mean only? We can get into current releases but let's say ATi is having a very rough time with their drivers and new games at the moment. Derek had to give up even more journalistic integrity to protect ATi in his latest blog about FC2 performance due to ATi driver issues. Catalyst AI, on or off btw? I saw you forced HQ in nV's control panel, was that just for mips or for all settings? I ask mainly as having HQ forced in nV drivers forces off base filtering optimizations while ATi will still have them enabled with Cat AI on AFAIK(you can't turn off the AF hack, but it used to at least clear up some visible mip transitions with trilinear on).
Also, the legacy OS thing 😉 DX9 is quaint man, it really is, but unless you want to focus ONLY on legacy gaming then the impact of testing second/third generation DX10 parts on an OS that can't even handle it is going to negatively impact how much importance it has to potential consumers.
Minor quibbles though, obviously your testing methodology utterly trounces most web sites as usual and of course the ever important hands on feedback is paramount- Kick ass job man 😀
toyota
Lifer
how did they get such low results in Far Cry2 with the GTX260 core 216? I have an E8500 just like they used and a regular 192sp GTX260 and at 1680x1050 on Ultra with 2x AA I got 51fps in the benchmark. thats 8 fps more than they got even though they used the better 216sp card and no AA. 😕
Are you using the exact same section of the game they are?
"In Far Cry 2, part of the jeep ride during the introduction, where the player is taken to the hotel in Pala, was used. Ultra-high graphics were enabled and tests were run under Windows XP (DirectX 9) and Windows Vista (DirectX 10)."
If you're using Ranch Small or something, you'll get different results.
BFG10K
Lifer
Because the Bing Bang driver isn?t a magical cure-all for the 8xMSAA issue, unlike nRollo?s implication that the difference is a non-issue with the advent of said drivers.
That was the point of discussing the likes of Quake 4. Bing Bang is unlikely to fix that so it?s far from a non-issue.
I was never disputing that; what I was disputing was that the Bing Bang drivers make 8xMSAA a non-issue. They don?t. They just accelerate a handful of newer titles when running AA.
BFG10K
Lifer
Yes, I know all about the latest FC2 issues with ATi and about Derek?s issues, but I also know certain past TWIMTBP titles have required immediate driver hot-fixes from nVidia as the last WHQL driver rendered the game unplayable.
Regardless, most of my comments come from personal experience and I don't have FC2 at the moment. However the modern titles I do have like Stalker Clear Sky ran fine on my 4850.
Though I must admit the gap in driver quality has narrowed significantly with 178.24 as it's a damn good driver. I'm almost scared of getting the Big Bang driver in case it breaks something that was working previously.
If nVidia could only fix my Red Faction and Vampire Bloodlines issues, I'd be happy.
Also nVidia need a constant release schedule, if not monthly then every 2-3 months. Customers need to know where they stand with GPU drivers. This business of ?release whenever we feel like it? is unacceptable, especially for people who pay a lot of money for SLI rigs.
AFAIK HQ automatically disables all optimizations and clamps the LOD but in case it doesn?t, I do it manually. That?s how I run nVidia?s DX10 hardware when I play games. With such a miniscule performance hit and such superb image quality, it?s criminal not to.
It?s not really feasible to disable Catalyst AI as it disables application specific optimizations and compatibility which can cause games to not work properly and/or run slower than they should. To get the equivalent on nVidia you?d need to delete all profiles and also rename all of the executables, and nobody runs a gaming rig like that.
However I make it quite clear that nVidia?s anisotropic filtering is visibly superior to ATi?s during gaming, and I am currently investigating anti-aliasing quality.
True, but to be perfectly honest I?d voluntarily be running DX9 paths in most modern games even if I had Vista as performance is sluggish enough just with DX9.
Also this is my rig (not just a test bench) so Vista has to be good enough for me, but issues like file copying still exist. I regularly backup almost 180 GB worth of games and reports are still coming in that Vista sometimes fails outright in such conditions, but the same rig running XP doesn?t have such issues. An OS that can?t guarantee file copying is simply unacceptable and should have never shipped. You can see from Microsoft?s blogs they butchered the file copying in RTM (so called ?experts? decided it needed ?fixing?) and then SP1 tried to bandage it back to behaving like XP, but it?s still broken.
With Windows 7?s improvements already visible in the beta, there?s no doubt that Microsoft dropped the ball with Vista so there?s really no point in getting it. I?ll be skipping to Windows 7 for my DX10 needs.
Kind words ? thanks. 🙂
The performance improvements extend beyond a handful of newer titles, unless you're referring to GRiD and AC as newer titles as well. Considering how significant the gains are in settings and resolutions that previously favored the 4870, I wouldn't be surprised at all to see AA performance increases even in older titles.
Anyways, here's a nother 4870 vs GTX 260 comparison:
GTX 260 vs 4870
Done mainly at 1920 and 2560 with AA, 180.47 and 8.11 Hot Fix drivers. Another convincing win for the GTX 260.
evolucion8
Platinum Member
Wow, now suddently the GTX 260 216 became much faster than the HD 4870 1GB, that's what I call the results fiddling magical finger!! A card such as the HD 4870 512MB was consistently faster than the old GTX 260, now with 24 additional stream processors which barely improved the performance (between 2 to 8%) against the old GTX 260 is now much faster than the HD 4870 1GB which proved itself to be between 4% to 35% faster than the HD 4870 512MB in high resolutions with lots of FSAA. Such magical 24 stream processors.
Wreckage
Banned
It's not suddenly, it's been that way for awhile.
http://www.techreport.com/articles.x/15651
The new drivers just made the gap a bit wider.
deerhunter716
Member
It is a shame though they have to use an overclocked 260 216 vs. STOCK 4870 1GB. Let me know when you see some reviews of STOCK vs. STOCK. I love how all the reviews have to use overclocked GTX vs. stock 4870's. MAYBE the stock does still come out on top of some games; but I'd like to see it on even ground when quoting benchmarks.
deerhunter716
Member
I mean hell there are so many conflicting reviews out there: http://www.anandtech.com/video/showdoc.aspx?i=3415&p=10
This Anandtech review has the 4870 beating the 260 216 and states in conclusion the 4870 is the better buy.
Another review also --> And I quote: "In our testing though it seems there are mixed results, depending on the game. In Age of Conan the GTX 260 offered a better experience, but in GRID the 1GB 4870 offered the better experience. If we look toward the newer games Crysis: Warhead and Stalker: Clear Sky, there weren?t really any gameplay differences at all." http://enthusiast.hardocp.com/...wxMCwsaGVudGh1c2lhc3Q=
So stop witht he 260 216 rules, lol In 2 minutes I found 2 reviews where one says the 4870 is still supreme and another where they are dead even. And yes of course the fanboys can find reviews where the 260 rules OR the 4870 rules. Bottom line is both seem about dead even.
This Anandtech review has the 4870 beating the 260 216 and states in conclusion the 4870 is the better buy.
Another review also --> And I quote: "In our testing though it seems there are mixed results, depending on the game. In Age of Conan the GTX 260 offered a better experience, but in GRID the 1GB 4870 offered the better experience. If we look toward the newer games Crysis: Warhead and Stalker: Clear Sky, there weren?t really any gameplay differences at all." http://enthusiast.hardocp.com/...wxMCwsaGVudGh1c2lhc3Q=
So stop witht he 260 216 rules, lol In 2 minutes I found 2 reviews where one says the 4870 is still supreme and another where they are dead even. And yes of course the fanboys can find reviews where the 260 rules OR the 4870 rules. Bottom line is both seem about dead even.
Keysplayr
Elite Member
It's a shame? When you can buy an overclocked GTX216 for 10.00 less than the best 4870 1GB there is to offer?
Take a look, and tell me what is shameful about offering an excellent value.
GTX260 216 OC 650 core 2100 mem 239.99 AR
4870 1GB 244.99 AR
Even OC'd, the GTX216 can be had even cheaper than a stock clocked 4870 1GB.
You love how all the reviews "have" to use overclocked GTX vs. stock 4870's?
Of course they don't have to. They do because they can, and for cheaper. And I really don't believe you love it at all. 😉
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 25K
-
-
-
