Overclock.net › Forums › Industry News › Hardware News › [Toms] Micro-Stuttering And GPU Scaling In CrossFire And SLI
New Posts  All Forums:Forum Nav:

[Toms] Micro-Stuttering And GPU Scaling In CrossFire And SLI - Page 14

post #131 of 177
Quote:
Originally Posted by JofaMang View Post
Totally. I had CF4870s (sapphire 512s) and was going crazy with the MS, so much so that I ended up taking one out, but the CF6870s (sapphires as well) on an almost identical platform didn't bother me at all.

I am down to one 6870 due to a power surge in my building that wiped out my comp, am waiting for 7000 to reinvest in a CF setup, but otherwise wouldn't hesitate to go for CF6870s again.
While I had CF HD4890s with no problems at all, with two screens too.
    
CPUMotherboardGraphicsRAM
Intel Core i5 3570k @ 4.5Ghz ASRock Z77 Pro3 Powercolor Radeon HD7950 3GB @ 1150/1350 4x4GB G.Skill Ares 2000Mhz CL9 
Hard DriveHard DriveHard DriveHard Drive
Samsung 840 250GB Western Digital Black 1TB WD1002FAEX Seagate Barracuda 3TB ST3000DM001 Samsung Spinpoint EcoGreen 2TB 
Optical DriveCoolingCoolingCooling
Pioneer DVR-220LBKS Noctua NH-D14 Scythe Gentle Typhoon 1850rpm Corsair AF140 Quiet Edition 
CoolingOSMonitorMonitor
Arcitc Cooling Acclero Twin Turbo II Arch Linux x86-64, amdgpu BenQ G2220HD BenQ G2020HD 
KeyboardPowerCaseMouse
Ducky Shine III Year of the Snake, Cherry Blue Silverstone Strider Plus 600w CoolerMaster CM690 II Black and White SteelSeries Sensei Professional 
Mouse PadAudioOther
Artisan Hien Mid Japan Black Large ASUS Xonar DX NZXT Sentry Mesh 30w Fan Controller 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Intel Core i5 3570k @ 4.5Ghz ASRock Z77 Pro3 Powercolor Radeon HD7950 3GB @ 1150/1350 4x4GB G.Skill Ares 2000Mhz CL9 
Hard DriveHard DriveHard DriveHard Drive
Samsung 840 250GB Western Digital Black 1TB WD1002FAEX Seagate Barracuda 3TB ST3000DM001 Samsung Spinpoint EcoGreen 2TB 
Optical DriveCoolingCoolingCooling
Pioneer DVR-220LBKS Noctua NH-D14 Scythe Gentle Typhoon 1850rpm Corsair AF140 Quiet Edition 
CoolingOSMonitorMonitor
Arcitc Cooling Acclero Twin Turbo II Arch Linux x86-64, amdgpu BenQ G2220HD BenQ G2020HD 
KeyboardPowerCaseMouse
Ducky Shine III Year of the Snake, Cherry Blue Silverstone Strider Plus 600w CoolerMaster CM690 II Black and White SteelSeries Sensei Professional 
Mouse PadAudioOther
Artisan Hien Mid Japan Black Large ASUS Xonar DX NZXT Sentry Mesh 30w Fan Controller 
  hide details  
Reply
post #132 of 177
Quote:
Originally Posted by Brutuz View Post
While I had CF HD4890s with no problems at all, with two screens too.
A buddy with CF 4890s didnt have any issues at all, it may have been a VRAM issue on my end (being as the cards were only 512mb models). Its also no secret that the 4890s were a substantial improvement on the 4870s in all ways.
    
CPUMotherboardGraphicsRAM
Phenom II x4 955 @ 3.8ghz Gigabyte 890GPA-UD3H Sapphire 6870 4gb G. Skill 1600mhz CL8 
Hard DriveOptical DriveOSMonitor
2TB WD Black LG DVD 22x Burner Windows 7 Ultimate 64 BenQ 22" 1920x1080 5ms 
PowerCase
Antec Earthwatts 650w CM Elite 370 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Phenom II x4 955 @ 3.8ghz Gigabyte 890GPA-UD3H Sapphire 6870 4gb G. Skill 1600mhz CL8 
Hard DriveOptical DriveOSMonitor
2TB WD Black LG DVD 22x Burner Windows 7 Ultimate 64 BenQ 22" 1920x1080 5ms 
PowerCase
Antec Earthwatts 650w CM Elite 370 
  hide details  
Reply
post #133 of 177
did my eyes just fooled me??

or my sight just gets worse...
Anti-Wuttz!!
(13 items)
 
  
CPUMotherboardGraphicsRAM
AMD A6-3650 @ 4ghz Asrock A75Pro4-M AMD HD6530D OCZ Flex 
Hard DrivePowerCase
160gb WD Cavar Blue Bench Disk HEC Rapter Open Air 
  hide details  
Reply
Anti-Wuttz!!
(13 items)
 
  
CPUMotherboardGraphicsRAM
AMD A6-3650 @ 4ghz Asrock A75Pro4-M AMD HD6530D OCZ Flex 
Hard DrivePowerCase
160gb WD Cavar Blue Bench Disk HEC Rapter Open Air 
  hide details  
Reply
post #134 of 177
Microstutter is a myth created by "The Man" to keep us down.
The War Machine
(15 items)
 
  
CPUMotherboardGraphicsRAM
i5 4790k Gigabyte Z97X Gaming 5 MSI 980 ti 6G 16GB DDR3 
Hard DriveCoolingOSMonitor
Intel SSD 480GB Noctua NH-D15 Windows 10 64 bit 39" 1920x1080 
PowerCase
EVGA SuperNOVA G2 650 Fractal Design Define R5 
  hide details  
Reply
The War Machine
(15 items)
 
  
CPUMotherboardGraphicsRAM
i5 4790k Gigabyte Z97X Gaming 5 MSI 980 ti 6G 16GB DDR3 
Hard DriveCoolingOSMonitor
Intel SSD 480GB Noctua NH-D15 Windows 10 64 bit 39" 1920x1080 
PowerCase
EVGA SuperNOVA G2 650 Fractal Design Define R5 
  hide details  
Reply
post #135 of 177
My first SLI rig was with a MSI K8N NEO 4 Mobo running an Athlon 64 4000+ @ 3GHz. I ran 8800GT SLI on that board. Back then Micro Studder was more visible. On the 750i it was less of an issue, be it that of driver enhancements or just higher performance. From my experience SLI setups require a lot of driver testing and trying different driver settings to smooth things out, and most of the time its possible to eliminate visible micro stutter, if you work at it.

If your getting "hitching" or what not its a bad driver install or the drivers are not mature enough yet for that particular game. Running Win XP 32bit doesn't help either, in fact a 32bit OS for SLI is a no no.

While there may still be some micro studder now a days, its no where near what it used to be, even still, a single powerful card will most definitely appear smoother if you don't put the time into it to smooth things out.

Edit: Just wanted to add this. Just because your using SLI or Crossfire doesn't mean you can max out the frame buffer and expect things to be silky smooth. Doing so can actually make things worse than it would if using a single card.
Edited by Drobomb - 8/25/11 at 9:06pm
Still pushing
(13 items)
 
  
CPUMotherboardGraphicsRAM
E8400 E0 @ 4.32GHz 1.36v EVGA 750i FTW GTX470 765/1530/1848 (unlocked 465) OCZ Blade 9200LV @ 1080MHz 5-5-5-18-33 
OSMonitorPowerCase
XP 32bit /Win 7 64bit /Arch Linux Samsung PX2370 PCP&Cooling 610w Silencer piece of cardboard 
  hide details  
Reply
Still pushing
(13 items)
 
  
CPUMotherboardGraphicsRAM
E8400 E0 @ 4.32GHz 1.36v EVGA 750i FTW GTX470 765/1530/1848 (unlocked 465) OCZ Blade 9200LV @ 1080MHz 5-5-5-18-33 
OSMonitorPowerCase
XP 32bit /Win 7 64bit /Arch Linux Samsung PX2370 PCP&Cooling 610w Silencer piece of cardboard 
  hide details  
Reply
post #136 of 177
Quote:
Originally Posted by adjas View Post
read? just base your opinion on the reactions of fellow OCN'ers.
Lol unfortunately if you do that, you wont learn anything.

Sent from my HTC Incredible S
Bernard
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 950 @ 3.71 Ghz | 1.17v Rampage III Extreme EVGA GTX 580 Corsair Dominators GTX 6gb 1925MHz 7-7-7-20 
Hard DriveOptical DriveOSMonitor
64Gb C300 SATA 6 Gb/s, 2 x 320 Gb F4's RAID 0 ASUS 24x -_- Windows 7 Home(Who needs ultimate?!) ASUS VH226H 21.5 Inch 1080p 
PowerCaseMouseMouse Pad
Corsair HX1000 + Custom Sleeving NZXT Tempest Evo Razer Naga NONE 
  hide details  
Reply
Bernard
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 950 @ 3.71 Ghz | 1.17v Rampage III Extreme EVGA GTX 580 Corsair Dominators GTX 6gb 1925MHz 7-7-7-20 
Hard DriveOptical DriveOSMonitor
64Gb C300 SATA 6 Gb/s, 2 x 320 Gb F4's RAID 0 ASUS 24x -_- Windows 7 Home(Who needs ultimate?!) ASUS VH226H 21.5 Inch 1080p 
PowerCaseMouseMouse Pad
Corsair HX1000 + Custom Sleeving NZXT Tempest Evo Razer Naga NONE 
  hide details  
Reply
post #137 of 177
Thread Starter 
haha, love how this thread has been voted down to 2 star just because some are butt-hurt with their dual-card heater.


microstutter is defined as large varying frame render time. even a slight change in FPS could mean a single frame has taken twice longer to render than most other frames. in another words, looking at FPS graph are meaningless. Tom's article is trying to say that dual-GPU doesn't really produce smooth enough experience simply because one frame could take 1ns while the next could take 1s. of course if both frames take 0.5s would be a smoother experience. but both would be recorded as 2 FPS.




Brettjv's post he linked is a well written OCn version of same explanation:
http://www.overclock.net/nvidia/9901...on-thread.html
post #138 of 177
Quote:
Originally Posted by brettjv View Post
Probably because 4870's were out back in the days when ATI drivers were actually attempting to distribute the frametimes evenly

Some folks may remember that there was a time, actually a little over 1 year ago, when ATI put out a crossfire enhancement package in between driver sets (and it was incorporated into the next driver package), that like magically improved crossfire scaling by like 10-15% overnight. This was before 6xxx-series, at the time, we were still rocking 5850/5870's.

I've been suspicious ever since that time that the way they accomplished this was by re-coding the drivers to abandon attempting to evenly space the output of frames over time on multi-gpu setups ... and I've mentioned it a number of times on OCN.

I believe that they calculated that, since reviewers don't review microstutter, they only review FPS, that they'd sell more multi-gpu setups by throwing caution to the wind and letting the cards fall into whatever pattern of outputting frames that 'came natural'.

Which means, basically, one card never WAITS for the other

I'd be interested to see if you still think their current XFire enhancements are causing the micro-stutter to be so bad.

As long as minimum frames never go below 60, you usually can't notice the stuttering very much.(see frame times above 20ms)

Here's an article that explains what you're talking about. Ignore the Nvidia charts as you'll see at the end of the article that we can't measure frame times with Nvidia until they release the API for 3rd party apps since they have a frame metering device that interferes with frame times.




http://techreport.com/articles.x/21516
Edited by PoopaScoopa - 9/11/11 at 9:31am
post #139 of 177
Quote:
Originally Posted by razaice View Post
Microstutter is a myth created by "The Man" to keep us down.
agreed, don't really believe in this article. since SLI 460 is almost close to 580 and vendors want consumers to pay more and go for the 580. Propaganda = profits.
    
CPUGraphicsRAMHard Drive
i5-2410M @ 2.9ghz Nvidia GT 540M 1GB DDR3 4GB DDR3 1333mhz 500GB 5400rpm 
OSMonitorPowerCase
Win 7 Home 64 1366x768 gloss 90w power brick Acer 
  hide details  
Reply
    
CPUGraphicsRAMHard Drive
i5-2410M @ 2.9ghz Nvidia GT 540M 1GB DDR3 4GB DDR3 1333mhz 500GB 5400rpm 
OSMonitorPowerCase
Win 7 Home 64 1366x768 gloss 90w power brick Acer 
  hide details  
Reply
post #140 of 177
Quote:
Originally Posted by robwadeson View Post
agreed, don't really believe in this article. since SLI 460 is almost close to 580 and vendors want consumers to pay more and go for the 580. Propaganda = profits.
Congrats, you seized an opportunity to expose the secret evil plots of the man all while beating a dead horse. For two birds with one stone I say well done sir.
Retail Edge Build
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 2600k 4.2 Ghz GIGABYTE GA-P67X-UD3-B3 SLI MSI 560 4GB (2x2) G Skill 1600 1.5v cas 8 
Hard DriveOptical DriveOSMonitor
500GB WD RE4 ASUS DRW-24B1ST Windows 7 Pro 64bit ASUS 26" + LG 23" 
KeyboardPowerCaseMouse
Saitek Cyborg Seasonic X-series 650W Gold CM HAF 922 Steelseries Xai 
Mouse Pad
X Trac pads pro 
  hide details  
Reply
Retail Edge Build
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 2600k 4.2 Ghz GIGABYTE GA-P67X-UD3-B3 SLI MSI 560 4GB (2x2) G Skill 1600 1.5v cas 8 
Hard DriveOptical DriveOSMonitor
500GB WD RE4 ASUS DRW-24B1ST Windows 7 Pro 64bit ASUS 26" + LG 23" 
KeyboardPowerCaseMouse
Saitek Cyborg Seasonic X-series 650W Gold CM HAF 922 Steelseries Xai 
Mouse Pad
X Trac pads pro 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [Toms] Micro-Stuttering And GPU Scaling In CrossFire And SLI