hi,
here result from my bottleneck pc (my friend says so, bcause i use i3-3220)
OS: Windows 7 64bit
GPU: GTX 970 4GB
Time: 08:02:16
I was reading up on this last night as well.
Rumour mill:
I wouldn’t hang ten on current 980s just yet if anyone’s so inclined.
OS: Windows 7 64bit
GPU: 2x GTX Titan
Time: 04:50
Here is the 2.72 benchmark scene tested with blender 2.73 and I am matching LordOdin tile size of 256x256.
OS: Windows 8.1
GPU: GTX 780 OC + GTX 780 OC
Tile: 256x256
Time: 2:56.31
Comparing LordOdin benchmarks times with 2x970 OC with my 2x780 OC
2.72 scene 2x970 OC: 4:08.06 and with 2x780 OC: 2:56.31
2.73 scene 2x970 OC: 4:17.55 and with 2x780 OC: 4:18:84
This timing shows my previous point that the 2.73 benchmark scene is much more Maxwell GPU friendly.
I also time 2.71 benchmark scene with blender 2.73 in case LordOdin wants to tested:
OS: Windows 8.1
GPU: 2x780 OC
Tiles: 256x256
Time: 1:25.77
Before I submit a bug report: Is anyone-preferably with a Maxwell chip- able to render the scene with branched and 128 diffuse/glossy samples? I instantly get a driver reset…
And who decided to lock the render slots during rendering? Makes noise comparison impossible unless you sit through the whole render…
GTX980/WIn7 64 7m31s btw.
Oh well theres your problem lol
Ok here is 2.71 test on 2.73
Windows 10 x64
GTX 970 OC + GTX 970 OC
01:46.74
First 770 on here, I see. My benchmarks have only been run once.
OS: Ubuntu 14.04
GPU: EVGA GTX 770 4GB
2.73 Benchmark: 11:57.71
2.72 Benchmark: 9:42.13
Are both of these ran on 2.73?
Just to clarify if you guys are going to compare older ones be sure to use 2.73
Sorry, I should have specified. Yes, both benchmarks were run with the Linux version of Blender 2.73
Hi, I add link for the 2.72 bench file to first post, please render both with files with 2.73.
User who did render sponza file allready please render the 2.72 bench and add new results.
I will try to add the results to my spreadsheet.
Windows 8.1 64bit.
GPU: GTX Titan Black (x4)
GPU1 - 6:09.07
GPU2 - 6:08.41
GPU3 - 6:18.81
GPU4 - 6:06.33
4x GPUs - 1:39.73.
Hi, new entries with a few 2.72 file results.
[ATTACH=CONFIG]356465[/ATTACH]
Very interesting, mib
@Fikey, will add your time for 2.72 file next time.
Attachments
OS: Windows 8.1
GPU: 3 x GTX Titan OC/LC
No changes to benchmark settings
Sponza: 03:11.08
2.72 file: 02:09.82
mib2berlin - At the start of the thread you mention “Benchmark use now 256x256 tiles” but I’m seeing 200x240. Am I missing something? Also, it’s always a little strange to open a benchmark and get the Blender Auto-run warning. Of course I trust you, but you might consider stripping unnecessary python in future benchmarks.
I cut this post because I was probably just cluttering up the thread with nonsense.
Thank you for your time and effort here everyone and special thanks to mib2berlin.
-Larry
Reran the 2.72 benchmark in 2.73:
GTX980 - 7:01.38
GTX980 + GTX460 - 5:40.02
Here are the old times:
GTX980 - 7:42.54
GTX980 + GTX460 - 6:27.90
Actually these times are just from the larger tile size, so I’m not seeing much change from 2.72.
I redid both the renders: (I also did a bios update as the bios of the rampage V was like 601 where current is 901 is much more stable at booting)
GTX 980 G1 x 2
sponza: 3:57.91
272bench: 3:30.7
OS: Win7 64
GPU: GTX 770 4GB Factory OC (GPU 1137/1189 Memory 7010)
2.73 benchmark rendered with 2.73 - 11:58.37
2.72 benchmark rendered with 2.73 - 10:20.04
2.72 benchmark rendered with 2.72a - 10:48.38
GPU temp peaked at 58c on all tests.
-LP
ok, here comes the 2.73 benchmark again (from previous post) and the 2.72 benchmarl added - I put it all in one so that mib2berlin has is easier with the lists (thx for maintaining these for us!)
OS: Win8.1, i7-3770k, 32GB RAM
2x780 6GB Strix (Asus - 941MHz GPU; the name says OC edition, but I am not sure if overclocking is built in or has to be turned on via GPU tweak tool - I did not touch the factory settings)
2.73 benchmark with 2.73
256x256 tile size
2x 780 6GB strix (double card)
05:14.23
1x 780 6GB strix (single card)
09:23.48
2.72 benchmark with 2.73
2x 780 6GB strix (double card)
03:30.44
1x 780 6GB strix (single card)
06:57.27
Yeah I deleted all that before I saw you quoting me in your post which didn’t appear until hours later. I so hate that blasted moderated first 10 posts rule.
Anyway, I was mainly just curious. Whether a bug (doubtful) or a boost for the 9’s being multiplied in that complex scene, it is not a problem in my little world. Thanks for the reply anyway.
-LP
Blender 2.73 / Win7 64b / AMD X6 1090T / 16GB
GTX 570 (1.2GB)
2.72 benchmark file: 14:32:43
2.73 benchmark file: 17:24:59