So, I got my gtx260s today and I thought i'd do some benchmarks and let you know what I think of them.
Thus far i'm very happy with the purchase, but my one and only complaint is the cards are ENORMOUS, but I still managed to fit them in my Antec900 with only having to shuffle forward the fan bays.
Heat production seems par normal at 40% fan speed in evga precision 60-70 degrees - I liked the software package which was included with these cards: Fraps, Precision Utility and 3dMark Vantage Advanced is available as a download from EVGA.com once you register your graphics cards.
A rather major thing I wanted to note about my benchmarking experience in Crysis - I have so far experienced zero visible microstutter which is a typical issue of SLI mode cards. So far, so good; I will note when/if I do have microstutter issues however.
Relevant system specs:
Intel e8400 wolfdale @ 4050mhz, 4gb corsair ddr2, 7200rpm WD SATA hdd, EVGA 750I SLI FTW motherboard, Vista 64 bit
1000 watt thermaltake toughpower cable management, 86A power supply
I'm going to do stock clocked results in Crysis first, overclocked Crysis and stock-clocked vantage results will be posted in a little bit; i'm not going to overclock for vantage just cause it could lead to misleading numbers for something that is not totally representative of true gameplay anyways. I'm also not taking the time to do screenshots, i'm just going to post logs and if verification is really that important, i'll do that up later.
Nvidia control panel settings:
Force 16x Anisotropic filtering, Single-display performance mode, Trilinear optimization, Anisotropic sample optimization all enabled. Vsync forced off, other settings Nvidia defaults.
Stock clocked:
Crysis, 1680*1050; Very High, 0xAA, 16xAF
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 63.18s, Average FPS: 31.66
Min FPS: 19.76 at frame 1335, Max FPS: 41.62 at frame 931
Average Tri/Sec: -10950508, Tri/Frame: -345902
Recorded/Played Tris ratio: -2.65
!TimeDemo Run 1 Finished.
Play Time: 61.33s, Average FPS: 32.61
Min FPS: 19.76 at frame 1335, Max FPS: 42.38 at frame 916
Average Tri/Sec: -10771551, Tri/Frame: -330324
Recorded/Played Tris ratio: -2.78
!TimeDemo Run 2 Finished.
Play Time: 61.08s, Average FPS: 32.75
Min FPS: 19.76 at frame 1335, Max FPS: 44.60 at frame 61
Average Tri/Sec: -10823947, Tri/Frame: -330539
Recorded/Played Tris ratio: -2.77
!TimeDemo Run 3 Finished.
Play Time: 61.40s, Average FPS: 32.57
Min FPS: 19.76 at frame 1335, Max FPS: 44.60 at frame 61
Average Tri/Sec: -10770232, Tri/Frame: -330644
Recorded/Played Tris ratio: -2.77
TimeDemo Play Ended, (4 Runs Performed)
Min -19.76 fps, Max 41-45 fps, Average 32-33fps
Crysis, 1680*1050, Very High, 4xMSAA, 16x AF
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 76.16s, Average FPS: 26.26
Min FPS: 17.10 at frame 694, Max FPS: 34.22 at frame 891
Average Tri/Sec: -8629397, Tri/Frame: -328597
Recorded/Played Tris ratio: -2.79
!TimeDemo Run 1 Finished.
Play Time: 72.61s, Average FPS: 27.54
Min FPS: 17.10 at frame 694, Max FPS: 34.69 at frame 871
Average Tri/Sec: -8633355, Tri/Frame: -313437
Recorded/Played Tris ratio: -2.92
!TimeDemo Run 2 Finished.
Play Time: 72.01s, Average FPS: 27.77
Min FPS: 17.10 at frame 694, Max FPS: 34.74 at frame 76
Average Tri/Sec: -8664793, Tri/Frame: -311996
Recorded/Played Tris ratio: -2.94
!TimeDemo Run 3 Finished.
Play Time: 72.27s, Average FPS: 27.67
Min FPS: 17.10 at frame 694, Max FPS: 34.74 at frame 76
Average Tri/Sec: -8671914, Tri/Frame: -313365
Recorded/Played Tris ratio: -2.93
TimeDemo Play Ended, (4 Runs Performed)
Min -17.10 fps, Max - 34-35 fps, Avg 26-27fps
Crysis, 1680*1050, Very High, 2xMSAA, 16xAF
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 73.93s, Average FPS: 27.05
Min FPS: 17.57 at frame 1334, Max FPS: 35.89 at frame 871
Average Tri/Sec: -9053086, Tri/Frame: -334654
Recorded/Played Tris ratio: -2.74
!TimeDemo Run 1 Finished.
Play Time: 71.52s, Average FPS: 27.97
Min FPS: 15.93 at frame 1910, Max FPS: 35.91 at frame 894
Average Tri/Sec: -8930102, Tri/Frame: -319326
Recorded/Played Tris ratio: -2.87
!TimeDemo Run 2 Finished.
Play Time: 70.91s, Average FPS: 28.20
Min FPS: 15.93 at frame 1910, Max FPS: 36.27 at frame 886
Average Tri/Sec: -9007670, Tri/Frame: -319383
Recorded/Played Tris ratio: -2.87
!TimeDemo Run 3 Finished.
Play Time: 71.38s, Average FPS: 28.02
Min FPS: 14.73 at frame 1938, Max FPS: 36.27 at frame 886
Average Tri/Sec: -9002637, Tri/Frame: -321319
Recorded/Played Tris ratio: -2.85
TimeDemo Play Ended, (4 Runs Performed)
Min - 15-17fps, Max 35-36 fps, Avg 27-28 fps Note: Weird anamoly with 4xAA vs 2xAA - 2xAA min fps was lower by a couple frames than 4xAA min fps; go figure.
Crysis, 1680*1050, High, 4xAA, 16xAF
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 50.93s, Average FPS: 39.27
Min FPS: 22.50 at frame 1365, Max FPS: 52.27 at frame 934
Average Tri/Sec: -25587256, Tri/Frame: -651621
Recorded/Played Tris ratio: -1.41
!TimeDemo Run 1 Finished.
Play Time: 47.64s, Average FPS: 41.98
Min FPS: 22.50 at frame 1365, Max FPS: 60.42 at frame 121
Average Tri/Sec: -27260394, Tri/Frame: -649333
Recorded/Played Tris ratio: -1.41
!TimeDemo Run 2 Finished.
Play Time: 47.25s, Average FPS: 42.33
Min FPS: 22.50 at frame 1365, Max FPS: 61.86 at frame 109
Average Tri/Sec: -27466720, Tri/Frame: -648856
Recorded/Played Tris ratio: -1.41
!TimeDemo Run 3 Finished.
Play Time: 47.24s, Average FPS: 42.34
Min FPS: 22.50 at frame 1365, Max FPS: 61.86 at frame 109
Average Tri/Sec: -27514862, Tri/Frame: -649873
Recorded/Played Tris ratio: -1.41
TimeDemo Play Ended, (4 Runs Performed)
Min fps - 22.50, Max fps 60-61 fps, Avg Fps 41-41 fps Note: another anamoly, in timedemo 1 max fps was 52.27 which brings timedemo 1's average down to 39.27 fps. I would say this is probably not totally unrealistic and that the 62fps maxes are more anamalous than the 52 fps max maybe - anyways, thats the results.
In Crysis, seems the biggest hit to max FPS comes from "Very High" settings, and there is a difference of 5 min fps between 1680*1050 Very High 4xAA and High 4xAA - i'd say between Very high 4xAA and High 4xAA, High 4xAA is more playable by that standard.
First note: temps were good, never went above 70 degrees
1680*1050, Very High, 4xAA, 16xAF
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 74.69s, Average FPS: 26.78
Min FPS: 15.91 at frame 1316, Max FPS: 35.43 at frame 922
Average Tri/Sec: -8788302, Tri/Frame: -328188
Recorded/Played Tris ratio: -2.79
!TimeDemo Run 1 Finished.
Play Time: 71.04s, Average FPS: 28.15
Min FPS: 15.91 at frame 1316, Max FPS: 35.51 at frame 910
Average Tri/Sec: -8853474, Tri/Frame: -314461
Recorded/Played Tris ratio: -2.92
!TimeDemo Run 2 Finished.
Play Time: 70.67s, Average FPS: 28.30
Min FPS: 15.91 at frame 1316, Max FPS: 35.62 at frame 922
Average Tri/Sec: -8862007, Tri/Frame: -313145
Recorded/Played Tris ratio: -2.93
!TimeDemo Run 3 Finished.
Play Time: 70.74s, Average FPS: 28.27
Min FPS: 15.91 at frame 1316, Max FPS: 35.62 at frame 922
Average Tri/Sec: -8856334, Tri/Frame: -313264
Recorded/Played Tris ratio: -2.93
TimeDemo Play Ended, (4 Runs Performed)
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 84.47s, Average FPS: 23.68
Min FPS: 13.57 at frame 142, Max FPS: 34.06 at frame 94
Average Tri/Sec: -7792291, Tri/Frame: -329109
Recorded/Played Tris ratio: -2.79
!TimeDemo Run 1 Finished.
Play Time: 81.10s, Average FPS: 24.66
Min FPS: 13.57 at frame 142, Max FPS: 37.89 at frame 91
Average Tri/Sec: -7761038, Tri/Frame: -314725
Recorded/Played Tris ratio: -2.91
!TimeDemo Run 2 Finished.
Play Time: 81.19s, Average FPS: 24.63
Min FPS: 13.57 at frame 142, Max FPS: 39.96 at frame 80
Average Tri/Sec: -7713197, Tri/Frame: -313114
Recorded/Played Tris ratio: -2.93
!TimeDemo Run 3 Finished.
Play Time: 80.62s, Average FPS: 24.81
Min FPS: 13.57 at frame 142, Max FPS: 39.96 at frame 80
Average Tri/Sec: -7804021, Tri/Frame: -314595
Recorded/Played Tris ratio: -2.91
TimeDemo Play Ended, (4 Runs Performed)
Min 13 fps, Max 34-40 fps, Avg 23- 24 fps
Last test in Very High, I raise the memory clock to 1216mhz - but I leave the core/shader the same as the previous test above. (GTX280 FTW speeds)
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 80.28s, Average FPS: 24.91
Min FPS: 13.24 at frame 147, Max FPS: 33.88 at frame 103
Average Tri/Sec: -8195812, Tri/Frame: -328968
Recorded/Played Tris ratio: -2.79
!TimeDemo Run 1 Finished.
Play Time: 77.84s, Average FPS: 25.69
Min FPS: 13.24 at frame 147, Max FPS: 39.24 at frame 86
Average Tri/Sec: -8060078, Tri/Frame: -313713
Recorded/Played Tris ratio: -2.92
!TimeDemo Run 2 Finished.
Play Time: 77.07s, Average FPS: 25.95
Min FPS: 13.24 at frame 147, Max FPS: 40.49 at frame 77
Average Tri/Sec: -8100151, Tri/Frame: -312144
Recorded/Played Tris ratio: -2.94
!TimeDemo Run 3 Finished.
Play Time: 76.94s, Average FPS: 25.99
Min FPS: 13.24 at frame 147, Max FPS: 40.49 at frame 77
Average Tri/Sec: -8135068, Tri/Frame: -312963
Recorded/Played Tris ratio: -2.93
TimeDemo Play Ended, (4 Runs Performed)
Min - 13 fps, Max - 39-40 fps, Avg 24-25fps
Crysis at Very High settings, 4xMSAA, 16xAF doesn't seem to want to budge its min fps through overclocking - it is nice to see the 6fps Max improvement over the stock clocks though.
Again, temps are very acceptable 65-70 degrees even while overclocked. The gtx260/280 cooler definitely works well, at least in my Antec 900 case.
Message edited by ovaltineplease on 07-04-2008 at 01:06:09 AM
Last Crysis benchmark, I thought i'd do another runthrough on high with my last overclocked settings:
1680*1050, High, 4xAA, 16xAF
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 53.85s, Average FPS: 37.14
Min FPS: 18.82 at frame 1048, Max FPS: 61.42 at frame 117
Average Tri/Sec: -24224660, Tri/Frame: -652201
Recorded/Played Tris ratio: -1.41
!TimeDemo Run 1 Finished.
Play Time: 61.36s, Average FPS: 32.59
Min FPS: 18.82 at frame 1048, Max FPS: 66.20 at frame 76
Average Tri/Sec: -21175906, Tri/Frame: -649693
Recorded/Played Tris ratio: -1.41
!TimeDemo Run 2 Finished.
Play Time: 61.07s, Average FPS: 32.75
Min FPS: 18.82 at frame 1048, Max FPS: 68.84 at frame 101
Average Tri/Sec: -21286948, Tri/Frame: -650041
Recorded/Played Tris ratio: -1.41
!TimeDemo Run 3 Finished.
Play Time: 61.03s, Average FPS: 32.77
Min FPS: 18.82 at frame 1048, Max FPS: 68.84 at frame 101
Average Tri/Sec: -21278246, Tri/Frame: -649336
Recorded/Played Tris ratio: -1.41
TimeDemo Play Ended, (4 Runs Performed)
Min - 18.82, Max 61-68 fps, Avg - 32-33 fps Note: seems timedemo 1 had an anamoly again, as its showing the average fps as being higher than the last 3 tests, but i'm going to chalk that up to a glitch and disregard the 37fps avg.
Seems the "sweet-spot" for gtx260 SLI in Crysis is 1680*1050 High, 4xMSAA, 16xAF - overclocked or stock clocks, i'd say this is the playable limit and best balance of IQ/FPS i'm sure that higher min/max fps could be reached with AA or AF disabled, but I can't be bothered and its kinda a meaningless fluff test anyways for higher end hardware - again, no signs of microstuttering, frame bottom outs, or otherwise throughout any of these tests.
3DMARK Vantage - 1680*1050, 4xMSAA, 16xAF, Extreme Quality, 1:2 Post Processing scale - I couldn't do a "Extreme" benchmark, as my screen doesn't support 1920*1200, so you get an "Extreme settings" with 1680*1050 instead.
I take 3DMARK with a grain of salt anyways, but I just thought i'd post it seeing as EVGA does include it in the software package.
GPU Test 1: 28.69 FPS
GPU Test 2: 22.89 FPS
CPU Test 1: 1034.53 PLANS/S
CPU Test 2: 115.93 STEPS/S
Feature Test 1: 987.87 Gtexels/S
Feature Test 2: 9.83 Gpixels/S
Feature Test 3 44.44 FPS
Feature Test 4: 22.87 FPS
Feature Test 5: 33.19 FPS
Feature Test 6: 62.92 FPS
Yeah I don't think you'll experience much micro-stutter at 1680x1050, it would need to be a games setting and resolution where the single card would struggle to produce consistenly above 30fps where it would becoming recordable, and likely only when that would be below 20fps that it would be noticeable, because you'd need that point where the second card was unable to render that 1 frame in it's expected time frame portion of the 60fps workload.
--------------- You need a license to buy a gun, but they'll sell anyone a stamp (or internet account) - REDGREEN. GA to SK HD Freedom: 45 5F E1 04 22 CA 29 C4 93 3F 95 05 2B 79 2A B2
I think this is the last GPU upgrade i'll be doing until earliest next spring ^^
THG finally posted a 4870 review with DDR5 talk
Also in today's newsreel, Nvidia will have dx10.1 support by Q1 2009 they claim. Guess my prediction was right, i'd expect to see a similar newsreel stating Windows Vista becoming the mainstream desktop OS by Q4 2008 as well
Yea, except that AMD's software doesn't support Quadfire in the 4800 series.
I mean sure, I coulda gone with crossfire 4870s - but by the time I bought an x48 board it would've been more expensive in the longrun for the exact same performance
What incredibly unimpressive results. The MIN and AVG FPS are pathetic, and these were to be the make of cards that could run Crysis singly.
While you did a service by providing these results and I thank you for your hard work, it just demonstrates how stagnant and unresponsive NVidia has become.
A fast GeForce 8 can run any game except Crysis pretty much any way you want, the 9's only confirmed that claim. What are these X's supposed to do?
Well, honestly - going from 8800 GT SLI to gtx260 SLI I get almost 4times the framerate at the same resolution/IQ settings
On a comparison, I could run 8800 GT SLI in 1680*1050 16xAF, 2xAA - for most levels up to Assault harbour I was doing okay at min 12-max 40 fps.
But on the gtx260s, I just finished playing through assault harbour on 1680*1050, 16xAF 4xAA and it was liquid smooth and no hitching ****.
The gtx260s don't bottom out like the 8800 GTs in SLI did - I don't get hitches or lags when panning the screen over large areas like I did on my 8800s in Crysis at times.
I'd say overall, its a much better experience - benchmarks never really tell the whole story; but its the only standardized test I can give you short of frapsing hours of gameplay and finding someone to host it
I'll update this over time whenever something new comes out like Stalker: CS - I might fraps some CoD4 with 16QAA/TRAA and see how that taxes it. If anything right now I really feel like I lack a selection of games that will really tax these GPUs other than Crysis
Message edited by ovaltineplease on 07-04-2008 at 03:44:06 AM
No, i'm saying that AMD hasn't gotten 4 single cards to work properly let alone scale on their 790 FX platform
I'm quite hopeful for 4870x2 quadfire on Intel x48 though; i'm eager to see how that scales because its supposed to be upwards of 180% scaling for a single GPU unlike 130- 140% for Crossfire 4870s - so maybe the Quadfire 4870x2 will have some nice scaling as well
Yeah, it looks like the 4800 X2s are going to kick ass. I sure hope so, AMD could use the boost, and Nvidia could use a little kick to get them moving again. Still, looks like you've got some very nice cards on your hands. That's a pretty cool overclock - stock heatsinks and fans, right?
