Overclock.net › Forums › Industry News › Hardware News › [PCper] Frame Rating Dissected: Full Details on Capture-based Graphics Performance Testing
New Posts  All Forums:Forum Nav:

[PCper] Frame Rating Dissected: Full Details on Capture-based Graphics Performance Testing - Page 12

post #111 of 420
applaud.gif Good job Nivdea, your slowly winning me over, but I cant afford you.
post #112 of 420
Quote:
Originally Posted by Kuivamaa View Post

"What the overlay does is it draws a single column of colour on each frame as it leaves the game engine". This is a quite crucial part of this testing method. Does anybody actually know when this happens? Before even drivers take effect ,just before hitting the screen, or somewhere in between?

...

If columns do survive the pipeline unaffected this could be moot of course. Still curious on that though.

It puts the color bars on at the same place where Fraps records the frame time, at the Present call. Not sure what the pipeline could do to affect the color bars themselves, although I guess it could mess with the timing somehow. Seems like it would be pretty easy to check if it did - although I think PCPer mentioned that the color overlay portion was the one thing that wasn't open-sourced.
post #113 of 420
Quote:
Originally Posted by JCPUser View Post

You keep saying that frame capture only tells part of the story, but since it is basically acting as the monitor it would seem to tell the whole story, IMO. Could you elaborate on that point? Thanks.

However, I do understand though if you are saving your repsonse for your review site... tongue.gif

It's easy to understand.

If you're locked with a 60 Hz monitor/tool every single extra frame per refresh displayed is throwing outdated info to the screen. Having a faster graphics solution only adds visual artifacts and more outdated info, hence worse. Somehow it turns to be better with these new methods.

If reviewers are trying to measure the user experience they're failing big time. My last hope is [H] where Brent actually cares about user experience.
post #114 of 420
Quote:
Originally Posted by Forceman View Post

It puts the color bars on at the same place where Fraps records the frame time, at the Present call. Not sure what the pipeline could do to affect the color bars themselves, although I guess it could mess with the timing somehow. Seems like it would be pretty easy to check if it did - although I think PCPer mentioned that the color overlay portion was the one thing that wasn't open-sourced.

Yeah, Ryan Smith did explain that in detail on article and comments. This could span from being totally legit (nvidia knowing their strengths and exhibiting them for the world to see) to totally biased (a line of code stretching/loosening things at a point if it doesn't detect nvidia hardware) etc. If the measured delta wasn't that big on pcper tables, I wouldn't have even raised an eyebrow-always found SLi to be more consistent. But those numbers make cfx to seem like worthless which it isn't. They are most certainly into something good here, but there are loose ends still.
Mastodon Ryzen
(12 items)
 
HP Z220
(8 items)
 
 
CPUMotherboardGraphicsRAM
R7 1800X Asus Crosshair VI Hero Sapphire RX Vega 64 reference Gskill TridentZ 
Hard DriveOptical DriveCoolingOS
Pny SSD 240GB Crucial MX100 CM Nepton 280L Win 10 
MonitorPowerCaseMouse
Acer Predator XG270HU Freesync XFX 750W Pro HAF XM Logitech G502 
CPUMotherboardGraphicsCooling
i7 3770 HP Quadro K2000 HP 
OSPowerCaseMouse
Win 7  HP 400W HP CMT RAT 7 
  hide details  
Reply
Mastodon Ryzen
(12 items)
 
HP Z220
(8 items)
 
 
CPUMotherboardGraphicsRAM
R7 1800X Asus Crosshair VI Hero Sapphire RX Vega 64 reference Gskill TridentZ 
Hard DriveOptical DriveCoolingOS
Pny SSD 240GB Crucial MX100 CM Nepton 280L Win 10 
MonitorPowerCaseMouse
Acer Predator XG270HU Freesync XFX 750W Pro HAF XM Logitech G502 
CPUMotherboardGraphicsCooling
i7 3770 HP Quadro K2000 HP 
OSPowerCaseMouse
Win 7  HP 400W HP CMT RAT 7 
  hide details  
Reply
post #115 of 420
Quote:
Originally Posted by Kuivamaa View Post

Yeah, Ryan Smith did explain that in detail on article and comments. This could span from being totally legit (nvidia knowing their strengths and exhibiting them for the world to see) to totally biased (a line of code stretching/loosening things at a point if it doesn't detect nvidia hardware) etc. If the measured delta wasn't that big on pcper tables, I wouldn't have even raised an eyebrow-always found SLi to be more consistent. But those numbers make cfx to seem like worthless which it isn't. They are most certainly into something good here, but there are loose ends still.

The TechReport numbers for CFX didn't look as significantly different as the ones PCPer was showing (TR had about a 50% increase in BF3, for example, while PCPer was showing close to 0%), although that could just be because TR showed an average FPS number and PCPer only showed a graph.

As for the color bars thing, I'm sure there is a way Nvidia could manipulate it to mess with the numbers, but I really doubt they'd go to all that trouble for something that would almost certainly be discovered. There is a lot of attention on this now, from a lot of websites, so you'd think any weird anomalies like that would get noticed. Plus AMD already acknowledged that they have issues, and that they prioritized CorssFire performance for latency over frame times, so it doesn't sound like these results are really unexpected to them.
post #116 of 420
Quote:
Originally Posted by Forceman View Post


As for the color bars thing, I'm sure there is a way Nvidia could manipulate it to mess with the numbers, but I really doubt they'd go to all that trouble for something that would almost certainly be discovered. There is a lot of attention on this now, from a lot of websites, so you'd think any weird anomalies like that would get noticed. Plus AMD already acknowledged that they have issues, and that they prioritized CorssFire performance for latency over frame times, so it doesn't sound like these results are really unexpected to them.

Yeah,that's my guess as well. This isn't 2003. I feel the key to this whole issue is the very nature of those short/runt frames. What they exactly are, what part on the whole latency issue do they play etc. And another thing-someone should actually check this little device on nvidia cards and a variety of drivers from the past 12-18 months or something. That should tell us when exactly nvidia started to focus on this issue.
Edited by Kuivamaa - 3/27/13 at 6:15pm
Mastodon Ryzen
(12 items)
 
HP Z220
(8 items)
 
 
CPUMotherboardGraphicsRAM
R7 1800X Asus Crosshair VI Hero Sapphire RX Vega 64 reference Gskill TridentZ 
Hard DriveOptical DriveCoolingOS
Pny SSD 240GB Crucial MX100 CM Nepton 280L Win 10 
MonitorPowerCaseMouse
Acer Predator XG270HU Freesync XFX 750W Pro HAF XM Logitech G502 
CPUMotherboardGraphicsCooling
i7 3770 HP Quadro K2000 HP 
OSPowerCaseMouse
Win 7  HP 400W HP CMT RAT 7 
  hide details  
Reply
Mastodon Ryzen
(12 items)
 
HP Z220
(8 items)
 
 
CPUMotherboardGraphicsRAM
R7 1800X Asus Crosshair VI Hero Sapphire RX Vega 64 reference Gskill TridentZ 
Hard DriveOptical DriveCoolingOS
Pny SSD 240GB Crucial MX100 CM Nepton 280L Win 10 
MonitorPowerCaseMouse
Acer Predator XG270HU Freesync XFX 750W Pro HAF XM Logitech G502 
CPUMotherboardGraphicsCooling
i7 3770 HP Quadro K2000 HP 
OSPowerCaseMouse
Win 7  HP 400W HP CMT RAT 7 
  hide details  
Reply
post #117 of 420
AMD aknowledged that their CFX cards suffer from frame latency issues - which is true. They have not commented on what these guys are accusing them of, which is throwing in fake frames.

I have yet to try out a CFX config but i doubt thousands of people would not have noticed that their new $300-$400 video card did absolutely nothing for them. Frankly i think there is a flaw somewhere in their testing software but we shall see.

If by some chance this turns out to be true, and AMD have been tricking their customers - enthusiasts who find 1 is not enough so go out and buy more- then that would probably the worst thing they have ever done in this industry. It's a very serious accusation which is why i'm not jumping to point fingers just yet, i can't accept that they would do something so absolutely stupid.
Bender
(18 items)
 
  
CPUMotherboardGraphicsGraphics
Core i5-2500k @ 4.6Ghz Gigabyte Z68X-UD3H-B3 Sapphire AMD R9 290X Tri-X (1050/1300mhz)  EVGA GeForce GTX 1080 (How do you OC?) 
RAMHard DriveHard DriveOS
8GB 1600Mhz OCZ Vertex 3 240GB SSD OCZ Vertex 3 240GB SSD windows 10 64bit 
MonitorKeyboardPowerCase
Asus MG279Q Logitech G510  Corsair TX750 CM II 690 Advanced 
MouseMouse PadAudioAudio
CM Sentinel Advance  some big corsair one beyerdynamic DT 770 Pro 250 Ohm SoundBlaster Z Soundcard 
  hide details  
Reply
Bender
(18 items)
 
  
CPUMotherboardGraphicsGraphics
Core i5-2500k @ 4.6Ghz Gigabyte Z68X-UD3H-B3 Sapphire AMD R9 290X Tri-X (1050/1300mhz)  EVGA GeForce GTX 1080 (How do you OC?) 
RAMHard DriveHard DriveOS
8GB 1600Mhz OCZ Vertex 3 240GB SSD OCZ Vertex 3 240GB SSD windows 10 64bit 
MonitorKeyboardPowerCase
Asus MG279Q Logitech G510  Corsair TX750 CM II 690 Advanced 
MouseMouse PadAudioAudio
CM Sentinel Advance  some big corsair one beyerdynamic DT 770 Pro 250 Ohm SoundBlaster Z Soundcard 
  hide details  
Reply
post #118 of 420
Quote:
Originally Posted by specopsFI View Post

This is the day when GPU testing was forever changed.

You think so? No what this is telling you is the same thing some us have known for quite some time. Who is paying the bills? That is who gets the favored review. Its been back and forth, one side is just as bad as the other green and red.

At the end of the day though if you plunked down $400 or $600 on equipment, and failed to follow the golden rule (BUYER BEWARE) articles like this will grate your nerves.

At the same time you have to wonder how camp red does so well on valley bench mark. Then camp green just creams them in top 30 3d mark 13 bench mark.
http://www.overclock.net/t/872945/top-30-3d-mark-13-fire-strike-scores
MyCleanPC
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 4770k ASUS MAXIMUS VI EVGA master blaster Corsair Vengence  
Hard DriveOptical DriveOSMonitor
OCZ SSD raid0 samsung Win 7 Samsung 
PowerCase
Enermax rev 1050 Stacker 832 
  hide details  
Reply
MyCleanPC
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 4770k ASUS MAXIMUS VI EVGA master blaster Corsair Vengence  
Hard DriveOptical DriveOSMonitor
OCZ SSD raid0 samsung Win 7 Samsung 
PowerCase
Enermax rev 1050 Stacker 832 
  hide details  
Reply
post #119 of 420
Quote:
Originally Posted by th3illusiveman View Post

AMD aknowledged that their CFX cards suffer from frame latency issues - which is true. They have not commented on what these guys are accusing them of, which is throwing in fake frames.

I have yet to try out a CFX config but i doubt thousands of people would not have noticed that their new $300-$400 video card did absolutely nothing for them. Frankly i think there is a flaw somewhere in their testing software but we shall see.

If by some chance this turns out to be true, and AMD have been tricking their customers - enthusiasts who find 1 is not enough so go out and buy more- then that would probably the worst thing they have ever done in this industry. It's a very serious accusation which is why i'm not jumping to point fingers just yet, i can't accept that they would do something so absolutely stupid.
I believe them, you aren't going to notice the missing performance unless you have something of the actual expected performance to compare it to. The placebo effect is also quite relevant here. That said, I don't think AMD is intentionally faking anything, they just aren't timing delivery and confirmation right.
Edited by TranquilTempest - 3/27/13 at 7:24pm
1
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 930 @ 3.6 Gigabyte x58a ud3r Gigabyte GV-R6870C-1GD Mushkin Redline 
Hard DriveOptical DriveMonitorKeyboard
2x 1TB Spinpoint samsung dvd burner Samsung p2370 + Mitsubishi Diamond Pro 930SB WASD keyboards v1 semi custom w/ cherry browns 
PowerCaseMouse
Antec CP-850 Antec P183 CM Storm Spawn 
  hide details  
Reply
1
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 930 @ 3.6 Gigabyte x58a ud3r Gigabyte GV-R6870C-1GD Mushkin Redline 
Hard DriveOptical DriveMonitorKeyboard
2x 1TB Spinpoint samsung dvd burner Samsung p2370 + Mitsubishi Diamond Pro 930SB WASD keyboards v1 semi custom w/ cherry browns 
PowerCaseMouse
Antec CP-850 Antec P183 CM Storm Spawn 
  hide details  
Reply
post #120 of 420
I'm surprised it took this long to consider variance in addition to central tendency. Statistics 101.
    
CPUMotherboardGraphicsRAM
Intel Core i7 920 SABERTOOTH X58 Gigabyte GTX 970 Corsair  
Hard DriveHard DriveHard DriveCooling
WD 320GB WD 1TB Hitachi 1TB CM Hyper 212 Plus 
OSMonitorMonitorPower
Windows 10 Pro 24" Sceptre 20" Asus Corsair TX750 
Case
Antec 1200 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Intel Core i7 920 SABERTOOTH X58 Gigabyte GTX 970 Corsair  
Hard DriveHard DriveHard DriveCooling
WD 320GB WD 1TB Hitachi 1TB CM Hyper 212 Plus 
OSMonitorMonitorPower
Windows 10 Pro 24" Sceptre 20" Asus Corsair TX750 
Case
Antec 1200 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [PCper] Frame Rating Dissected: Full Details on Capture-based Graphics Performance Testing