Overclock.net › Forums › Industry News › Hardware News › [PugetSystems] Titan X Performance: PCI-E 3.0 x8 vs x16
New Posts  All Forums:Forum Nav:

[PugetSystems] Titan X Performance: PCI-E 3.0 x8 vs x16 - Page 5

post #41 of 48
I'm curious as to why the 8x got higher results at 1080p in some tests.
post #42 of 48
Quote:
Originally Posted by Ghoxt View Post

With the near identical results, I'm surprised they did not monitor the I/O data rate on the components along the path to the GPU's. Ultimately not doing so leaves the test as inconclusive to what is, or is not, going on for the results seen.


Several Data files to be loaded into the GPU coming from the SSD would pass directly through the X99 chipset via DMI 2.0 to the CPU then out the 40 PCI lanes. I think this is more relevant to their confusing findings. they also may not have access to the Nvidia GPU monitoring tool that engineers can use to profile the GPU stack.

I'm doubtful any of these tests are fetching assets from the drives frequently enough to matter, especially If they ran the test more than once sequentially, as everything should have been preloaded already.
Quote:
Originally Posted by axiumone View Post

I think that the actual GPU used for this test isn't as important as the result. I'd expect the results to be the same for any nvidia chip.

The more powerful the GPUs the higher the frame rate and the more coherency and compositing data needs to go over PCI-E.
Quote:
Originally Posted by Master__Shake View Post

cards been out almost 2 years and now someone is doing a scaling review?

The review isn't really about the cards, but those are Pascal Titan Xes.
Quote:
Originally Posted by tpi2007 View Post

There is one fundamental test that they should have done: test the card running in PCIe 2.0 x16 mode. PCIe 2.0 x16 is not exactly the same as PCIe 3.0 x8 because the overhead of the protocol is different. The bandwidth overhead on PCIe 2.0 is 20% whereas it's only around 1.54% in PCIe 3.0. In theory, it should end up more or less the same given the bandwidth numbers they start with, but I'd like to know for sure.

On a semi-related tangent, back when AGP was king, and NVIDIA GPUs could handle stupid high AGP clocks, AGP 4x @ 100MHz was faster than AGP 8x at default 66.7MHz.
Quote:
Originally Posted by DarkIdeals View Post

People are asking for more tests because the current tests have shown MUCH larger than normal improvement with 16x/16x than we've seen in the past.

Relatively few recent tests have been with mutli-GPU setups and people still often fail to grasp that multi-GPU setups use PCI-E bandwidth completely differently from single GPUs.
Quote:
Originally Posted by Lon3Tr3k View Post

This was tested and shown to make a difference on cards earlier than the Pascal series of cards, as early as the 980 Ti.

Much earlier than that.
Quote:
Originally Posted by Lon3Tr3k View Post

The problem with the testing the GamersNexus did was that he didn't even use multiple cards for SLI, he just used a single card in a x16 and then a x8 slot and concluded there was no difference. rolleyes.gif The difference is only there when using SLI.

As anyone with a clue would expect.

Single GPU means none of frame compositing that is needed for multi-GPU rendering modes and none of the VRAM asset coherency requirements...4x lanes would be enough for most all single GPU cases, even if 8x/8x cripples dual card performance.
Quote:
Originally Posted by xxpenguinxx View Post

I'm curious as to why the 8x got higher results at 1080p in some tests.

I suspect that the larger chunks of data that are moved with 16x would be responsible for more overhead and more IMC/PCI-E controller contention and/or latency....which could outweigh the bandwidth needs of coherency in some apps, especially at more CPU limited settings.

Just a semi-educated guess though.
Primary
(15 items)
 
Secondary
(13 items)
 
In progress
(10 items)
 
CPUMotherboardGraphicsRAM
5820K @ 4.2/3.5GHz core/uncore, 1.175/1.15v Gigabyte X99 SOC Champion (F22n) Gigabyte AORUS GTX 1080 Ti (F3P) @ 2025/1485, 1... 4x4GiB Crucial @ 2667, 12-12-12-28-T1, 1.34v 
Hard DriveHard DriveHard DriveCooling
Plextor M6e 128GB (fw 1.06) M.2 (PCI-E 2.0 2x) 2x Crucial M4 256GB 4x WD Scorpio Black 500GB Noctua NH-D15 
OSMonitorKeyboardPower
Windows 7 Professional x64 SP1 BenQ BL3200PT Filco Majestouch Tenkeyless (MX Brown) Corsair RM1000x 
CaseMouseAudio
Fractal Design Define R4 Logitech G402 Realtek ALC1150 + M-Audio AV40 
CPUMotherboardGraphicsRAM
X5670 @ 4.4/3.2GHz core/uncore, 1.36 vcore, 1.2... Gigabyte X58A-UD5 r2.0 w/FF3mod10 BIOS Sapphire Fury Nitro OC+ @ 1053/500, 1.225vGPU/1... 2x Samsung MV-3V4G3D/US @ 2000, 10-11-11-30-T1,... 
RAMHard DriveHard DriveHard Drive
1x Crucial BLT4G3D1608ET3LX0 @ 2000, 10-11-11-3... OCZ (Toshiba) Trion 150 120GB Hyundai Sapphire 120GB 3x Hitachi Deskstar 7k1000.C 1TB 
CoolingOSPowerCase
Noctua NH-D14 Windows 7 Pro x64 SP1 Antec TP-750 Fractal Design R5 
Audio
ASUS Xonar DS 
CPUMotherboardGraphicsRAM
i7-6800K @ 4.3/3.5GHz core/uncore, 1.36/1.2v ASRock X99 OC Formula (P3.10) GTX 780 (temporary) 4x4GiB Crucial DDR4-2400 @ 11-13-12-28-T2, 1.33v 
Hard DriveHard DriveCoolingOS
Intel 600p 256GB NVMe 2x HGST Travelstar 7k1000 1TB Corsair H55 (temporary) Windows Server 2016 Datacenter 
PowerCase
Seasonic SS-860XP2 Corsair Carbide Air 540 
  hide details  
Reply
Primary
(15 items)
 
Secondary
(13 items)
 
In progress
(10 items)
 
CPUMotherboardGraphicsRAM
5820K @ 4.2/3.5GHz core/uncore, 1.175/1.15v Gigabyte X99 SOC Champion (F22n) Gigabyte AORUS GTX 1080 Ti (F3P) @ 2025/1485, 1... 4x4GiB Crucial @ 2667, 12-12-12-28-T1, 1.34v 
Hard DriveHard DriveHard DriveCooling
Plextor M6e 128GB (fw 1.06) M.2 (PCI-E 2.0 2x) 2x Crucial M4 256GB 4x WD Scorpio Black 500GB Noctua NH-D15 
OSMonitorKeyboardPower
Windows 7 Professional x64 SP1 BenQ BL3200PT Filco Majestouch Tenkeyless (MX Brown) Corsair RM1000x 
CaseMouseAudio
Fractal Design Define R4 Logitech G402 Realtek ALC1150 + M-Audio AV40 
CPUMotherboardGraphicsRAM
X5670 @ 4.4/3.2GHz core/uncore, 1.36 vcore, 1.2... Gigabyte X58A-UD5 r2.0 w/FF3mod10 BIOS Sapphire Fury Nitro OC+ @ 1053/500, 1.225vGPU/1... 2x Samsung MV-3V4G3D/US @ 2000, 10-11-11-30-T1,... 
RAMHard DriveHard DriveHard Drive
1x Crucial BLT4G3D1608ET3LX0 @ 2000, 10-11-11-3... OCZ (Toshiba) Trion 150 120GB Hyundai Sapphire 120GB 3x Hitachi Deskstar 7k1000.C 1TB 
CoolingOSPowerCase
Noctua NH-D14 Windows 7 Pro x64 SP1 Antec TP-750 Fractal Design R5 
Audio
ASUS Xonar DS 
CPUMotherboardGraphicsRAM
i7-6800K @ 4.3/3.5GHz core/uncore, 1.36/1.2v ASRock X99 OC Formula (P3.10) GTX 780 (temporary) 4x4GiB Crucial DDR4-2400 @ 11-13-12-28-T2, 1.33v 
Hard DriveHard DriveCoolingOS
Intel 600p 256GB NVMe 2x HGST Travelstar 7k1000 1TB Corsair H55 (temporary) Windows Server 2016 Datacenter 
PowerCase
Seasonic SS-860XP2 Corsair Carbide Air 540 
  hide details  
Reply
post #43 of 48
Quote:
Originally Posted by DarkIdeals View Post

Really? Did you even read the results? They showed up to a 30% difference in some cases! Yes the average was less, but you were seeing 5%, 8%, 15% etc.. in several different style tests including games, workstation software, benchmarks etc.. thats FAR from "no difference in realistic scenarios". People are asking for more tests because the current tests have shown MUCH larger than normal improvement with 16x/16x than we've seen in the past.

Yes. I did. Day one.
Did you? And did you assess that benching Heaven @ low settings in 3x 4K surround is the realistic scenario? Moreover needing more than 80 fps? For what? Your 30 or 60 Hz panel?
Or are you investing in the future for those $5K 120Hz 4K panels that your Titan X card doesn't even have the interface to drive above 60Hz?

For realistic gaming scenarios - at least the few tested in the OP - the differences are there, but 5% here and there where you are already super fast is diminishing returns. Really diminishing returns. In many cases is easily dismissable as non-relative in RL. Like endlessly beaching about whether a Ferrari 488 is faster or a McLaren 650S when in fact both are so fast that the difference is undetectable outside a track or a drag-strip: scenarios that are NOT RL for the vast majority of people that will OWN the thing, and that would be flabbergasted driving either. The "forum-warriors" that won't even get in the drivers seat of a 600~700HP supercar? Please...keep it real.

And obviously you did not read all tests or understood that the lower overhead in slower than 16x/16x configurations actually benefit the cards and those post higher scores than the "all-out" 16x/16x, or you did and you choose to disregard that fact and just bundle all those tests as one that caters to your predetermined narrative.

But as I ironically referred to, you are just "wowed" by them using workstation software...which you don't really understand, nor they report it to be consistently faster with A vs. B scenario, yet you have to use for your arguments...meh, by my post so far you should have realized my opinion about your comprehension, and the lack of my sympathy for your ironic move quoting all these people that disagree with your predetermined & fact-immune narrative "to teach them", while you just embarrass yourself.

Is PCIe bandwidth important? Sure.
Much like FP64/DP calcs etc etc.
You can invent all kinds of metrics & setup scenarios to illustrate that, but claiming RL benefits is a totally different ballpark.

By the time PCIe 3.0 8x will be irrelevant, PCIe 4.0 will be out, we won't be as limited by current panels and DisplayPort generations (probably will be USB-C / Thunderbolt anyways) and your 40x lane i7-E will be obsolete by much cheaper parts - much like your Titan X or 1080Ti...till then, keep it real, call a Spade a Spade, and judge epeen for what it is.
FooBox
(20 items)
 
 
Camera Gear
(10 items)
 
CPUMotherboardGraphicsRAM
Xeon E5-1650 V3 3.5GHz Dell C610 Quadro K4200 4GB 32GB PC-2133 DDR4 
Hard DriveCoolingOSMonitor
Crucial M550 256GB Dell Windows 7 Professional Dell U2412M 
MonitorKeyboardPowerCase
Dell U2412M Dell KB-212-B Dell 850W 80+ Gold T5810 
Mouse
Logitech G700 
CPUGraphicsGraphicsGraphics
Canon EOS 6D  EF 24-70mm f/2.8L II USM  EF 16-35mm f/4L IS USM EF 70-200mm f/4L IS USM 
GraphicsOptical DriveOptical DriveOptical Drive
EF 35mm f/2 IS USM Manfrotto 190XPROB Tripod Manfrotto 685B NeoTec Monopod Manfrotto 410 Junior Geared Head 
Optical DriveCase
Manfrotto 496 Ball Head Lowepro Mini Trekker AW 
  hide details  
Reply
FooBox
(20 items)
 
 
Camera Gear
(10 items)
 
CPUMotherboardGraphicsRAM
Xeon E5-1650 V3 3.5GHz Dell C610 Quadro K4200 4GB 32GB PC-2133 DDR4 
Hard DriveCoolingOSMonitor
Crucial M550 256GB Dell Windows 7 Professional Dell U2412M 
MonitorKeyboardPowerCase
Dell U2412M Dell KB-212-B Dell 850W 80+ Gold T5810 
Mouse
Logitech G700 
CPUGraphicsGraphicsGraphics
Canon EOS 6D  EF 24-70mm f/2.8L II USM  EF 16-35mm f/4L IS USM EF 70-200mm f/4L IS USM 
GraphicsOptical DriveOptical DriveOptical Drive
EF 35mm f/2 IS USM Manfrotto 190XPROB Tripod Manfrotto 685B NeoTec Monopod Manfrotto 410 Junior Geared Head 
Optical DriveCase
Manfrotto 496 Ball Head Lowepro Mini Trekker AW 
  hide details  
Reply
post #44 of 48
Quote:
Originally Posted by pcfoo View Post

Yes. I did. Day one.
Did you? And did you assess that benching Heaven @ low settings in 3x 4K surround is the realistic scenario? Moreover needing more than 80 fps? For what? Your 30 or 60 Hz panel?
Or are you investing in the future for those $5K 120Hz 4K panels that your Titan X card doesn't even have the interface to drive above 60Hz?

For realistic gaming scenarios - at least the few tested in the OP - the differences are there, but 5% here and there where you are already super fast is diminishing returns. Really diminishing returns. In many cases is easily dismissable as non-relative in RL. Like endlessly beaching about whether a Ferrari 488 is faster or a McLaren 650S when in fact both are so fast that the difference is undetectable outside a track or a drag-strip: scenarios that are NOT RL for the vast majority of people that will OWN the thing, and that would be flabbergasted driving either. The "forum-warriors" that won't even get in the drivers seat of a 600~700HP supercar? Please...keep it real.

And obviously you did not read all tests or understood that the lower overhead in slower than 16x/16x configurations actually benefit the cards and those post higher scores than the "all-out" 16x/16x, or you did and you choose to disregard that fact and just bundle all those tests as one that caters to your predetermined narrative.

But as I ironically referred to, you are just "wowed" by them using workstation software...which you don't really understand, nor they report it to be consistently faster with A vs. B scenario, yet you have to use for your arguments...meh, by my post so far you should have realized my opinion about your comprehension, and the lack of my sympathy for your ironic move quoting all these people that disagree with your predetermined & fact-immune narrative "to teach them", while you just embarrass yourself.

Is PCIe bandwidth important? Sure.
Much like FP64/DP calcs etc etc.
You can invent all kinds of metrics & setup scenarios to illustrate that, but claiming RL benefits is a totally different ballpark.

By the time PCIe 3.0 8x will be irrelevant, PCIe 4.0 will be out, we won't be as limited by current panels and DisplayPort generations (probably will be USB-C / Thunderbolt anyways) and your 40x lane i7-E will be obsolete by much cheaper parts - much like your Titan X or 1080Ti...till then, keep it real, call a Spade a Spade, and judge epeen for what it is.

Riiight... except that there is a noticeable performance differences in certain games, yet there isn't in others. For example, there clearly is a difference in games such as the Witcher 3, Rise of the Tomb Raider, Project Cars, and the newer Battlefield titles, as well as others. The difference in these in more like ~10-15%, which actually is pretty noticeable. Personally, I play many of these types of games where there is a difference, which is a much more accurate "real world" representation than Heaven. I don't know about you, but I don't actually spend a lot of time running the Heaven benchmark just to run it, let alone play it.

It all comes down to the types of games that you plan on playing, but the trend is only likely to continue towards favoring x16/x16.
Edited by Lon3Tr3k - 9/22/16 at 9:05am
    
CPUMotherboardGraphicsRAM
i7-6850k @ 4.5 GHz Rampage V Edition 10 EVGA 1080 FTW 32 GB G.Skill Trident Z 3200 MHz CL14 
Hard DriveHard DriveHard DriveOptical Drive
M4 Crucial 256GB Samsung 840 Pro 512GB Samsung 850 PRO 1TB LG Super Multi Blue 
CoolingCoolingCoolingCooling
Alphacool NexXxos XT45 420mm Radiator Alphacool NexXxos UT60 280mm Radiator XSPC Raystorm CPU Block XSPC Dual Bay Res/Pump 
OSMonitorMonitorKeyboard
Windows 7 Pro 27" 1440p Samsung S27B970D 23" 1080p 120hz Samsung S23A700D Corsair K70 Lux RGB 
PowerCaseMouseAudio
CoolerMaster V1000 NZXT Switch 810 Razer Deathadder Chroma Klipsch Promedia 2.1 
Audio
V-Moda Crossfade M-100 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i7-6850k @ 4.5 GHz Rampage V Edition 10 EVGA 1080 FTW 32 GB G.Skill Trident Z 3200 MHz CL14 
Hard DriveHard DriveHard DriveOptical Drive
M4 Crucial 256GB Samsung 840 Pro 512GB Samsung 850 PRO 1TB LG Super Multi Blue 
CoolingCoolingCoolingCooling
Alphacool NexXxos XT45 420mm Radiator Alphacool NexXxos UT60 280mm Radiator XSPC Raystorm CPU Block XSPC Dual Bay Res/Pump 
OSMonitorMonitorKeyboard
Windows 7 Pro 27" 1440p Samsung S27B970D 23" 1080p 120hz Samsung S23A700D Corsair K70 Lux RGB 
PowerCaseMouseAudio
CoolerMaster V1000 NZXT Switch 810 Razer Deathadder Chroma Klipsch Promedia 2.1 
Audio
V-Moda Crossfade M-100 
  hide details  
Reply
post #45 of 48
Except it is Q3 2016...now, after YEARS of having the same narrative, at last and there are some games that with extremely fast Pascal cards and the Titan X do better with x16/x16...in some cases 15% (Witcher), in others much less or no difference (like Project Cars you've mentioned), mostly in resolutions above 1440p where there aren't really high Hz panels to qualitatively tell a difference. It is just benchmarks.

In all these scenarios, the fps attainable in 8x/8x is very very playable, and the minimum FPS is most of the times the same. The newer HB bridge does more than 16x/16x does.
You might be suffering @ 90 fps AVG @ Witcher 3 vs. 110 fps AVG, both scenarios @ same MIN fps, but I doubt most people in RL will be able to tell a difference, especially on a 60Hz panel.

Same with Tomb Raider...the engine has issues so you might get 200+ fps with the 16x16x and the HB, but you will still get deeps @ 40-ish FPS in any configuration. Being 10% faster @ 44 fps MIN vs 41 fps MIN, is NOT a real gain, much like maxing out at 240+ fps vs. 195+ fps is not a RL gain, cause YOU WON'T FEEL IT..you will definitely notice the 4x FPS deeps on either case, not the above 120+ fps peaks.

Yes, there are spikes where the 16x/16x does show gains, but it is on paper. Unless ofc your are gaming 144Hz @ 4K which probably you don't.

Long story short, if you go SLI, you can perfectly do it on s115x (aka 8x/8x) and still see most of the benefits. It is splitting hair to say otherwise, and many scenarios where a higher-clocked s115x CPU will do better than a contemporary s2011-x, and more than bridge the difference.
FooBox
(20 items)
 
 
Camera Gear
(10 items)
 
CPUMotherboardGraphicsRAM
Xeon E5-1650 V3 3.5GHz Dell C610 Quadro K4200 4GB 32GB PC-2133 DDR4 
Hard DriveCoolingOSMonitor
Crucial M550 256GB Dell Windows 7 Professional Dell U2412M 
MonitorKeyboardPowerCase
Dell U2412M Dell KB-212-B Dell 850W 80+ Gold T5810 
Mouse
Logitech G700 
CPUGraphicsGraphicsGraphics
Canon EOS 6D  EF 24-70mm f/2.8L II USM  EF 16-35mm f/4L IS USM EF 70-200mm f/4L IS USM 
GraphicsOptical DriveOptical DriveOptical Drive
EF 35mm f/2 IS USM Manfrotto 190XPROB Tripod Manfrotto 685B NeoTec Monopod Manfrotto 410 Junior Geared Head 
Optical DriveCase
Manfrotto 496 Ball Head Lowepro Mini Trekker AW 
  hide details  
Reply
FooBox
(20 items)
 
 
Camera Gear
(10 items)
 
CPUMotherboardGraphicsRAM
Xeon E5-1650 V3 3.5GHz Dell C610 Quadro K4200 4GB 32GB PC-2133 DDR4 
Hard DriveCoolingOSMonitor
Crucial M550 256GB Dell Windows 7 Professional Dell U2412M 
MonitorKeyboardPowerCase
Dell U2412M Dell KB-212-B Dell 850W 80+ Gold T5810 
Mouse
Logitech G700 
CPUGraphicsGraphicsGraphics
Canon EOS 6D  EF 24-70mm f/2.8L II USM  EF 16-35mm f/4L IS USM EF 70-200mm f/4L IS USM 
GraphicsOptical DriveOptical DriveOptical Drive
EF 35mm f/2 IS USM Manfrotto 190XPROB Tripod Manfrotto 685B NeoTec Monopod Manfrotto 410 Junior Geared Head 
Optical DriveCase
Manfrotto 496 Ball Head Lowepro Mini Trekker AW 
  hide details  
Reply
post #46 of 48
While I agree to some degree on the point about minimum frames being close on both, I also think that this is probably misleading in the sense that looking at the 97 percentile frame minimum would be a lot better representation of the felt drops in frames. It's too bad that no one has done this sort of comparison yet, that I know of, but I would imagine that with the average framerate being much higher in some cases, the higher the 97 percentile of frames should be as well.

Yes, there are only a few games that really benefit so far from x16/x16, so is it worth upgrading a system purely for this? No, and I don't think anyone is suggesting that. But what it comes down to is this: if building a new system, or upgrading CPU/motherboard anyway, it's should definitely be worth considering going for 40 lanes over 28 lanes for the minimal cost difference, if you ever plan on multi-gpu down the road. Especially for someone who doesn't plan on upgrading every year.
    
CPUMotherboardGraphicsRAM
i7-6850k @ 4.5 GHz Rampage V Edition 10 EVGA 1080 FTW 32 GB G.Skill Trident Z 3200 MHz CL14 
Hard DriveHard DriveHard DriveOptical Drive
M4 Crucial 256GB Samsung 840 Pro 512GB Samsung 850 PRO 1TB LG Super Multi Blue 
CoolingCoolingCoolingCooling
Alphacool NexXxos XT45 420mm Radiator Alphacool NexXxos UT60 280mm Radiator XSPC Raystorm CPU Block XSPC Dual Bay Res/Pump 
OSMonitorMonitorKeyboard
Windows 7 Pro 27" 1440p Samsung S27B970D 23" 1080p 120hz Samsung S23A700D Corsair K70 Lux RGB 
PowerCaseMouseAudio
CoolerMaster V1000 NZXT Switch 810 Razer Deathadder Chroma Klipsch Promedia 2.1 
Audio
V-Moda Crossfade M-100 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i7-6850k @ 4.5 GHz Rampage V Edition 10 EVGA 1080 FTW 32 GB G.Skill Trident Z 3200 MHz CL14 
Hard DriveHard DriveHard DriveOptical Drive
M4 Crucial 256GB Samsung 840 Pro 512GB Samsung 850 PRO 1TB LG Super Multi Blue 
CoolingCoolingCoolingCooling
Alphacool NexXxos XT45 420mm Radiator Alphacool NexXxos UT60 280mm Radiator XSPC Raystorm CPU Block XSPC Dual Bay Res/Pump 
OSMonitorMonitorKeyboard
Windows 7 Pro 27" 1440p Samsung S27B970D 23" 1080p 120hz Samsung S23A700D Corsair K70 Lux RGB 
PowerCaseMouseAudio
CoolerMaster V1000 NZXT Switch 810 Razer Deathadder Chroma Klipsch Promedia 2.1 
Audio
V-Moda Crossfade M-100 
  hide details  
Reply
post #47 of 48
Has anyone addressed that "CPU bottlenecks" might be present in those SLI test results?
Logix
(23 items)
 
ProjectZero V3.0
(11 items)
 
To get list...
(1 item)
CPUMotherboardGraphicsGraphics
4820k @ 4.5Ghz Asrock X79 Extreme6 NVIDIA GeForce GTX 780 w/ EK waterblock XXL NVIDIA GeForce GTX 780 w/ EK waterblock XXL 
RAMHard DriveHard DriveHard Drive
G.Skill Ares 16Gb 9-11-10-28-2N 2133Mhz Hitachi 1Tb Seagate 1Tb Seagate 4Tb 
Hard DriveOptical DriveCoolingOS
Samsung EVO 250Gb Liteon Bluray Custom W/C Windows 7 Ultimate 64-bit 
MonitorMonitorKeyboardPower
LG IP237 Dell U2713HM Corsair K90 OCZ ZX 1250w 
CaseMouseMouse PadAudio
Corsair 900D Corsair M90 Razer Vespula Auzentech X-Meridian 7.1 2G 
AudioAudioOther
Roccat Kave XTD Digital Swans M10s 2.1 Speakers Razer Orbweaver 
CPUMotherboardGraphicsRAM
Intel Core i7 2700K Asus P8Z68 Pro GEN 3 Gigabyte GTX 570 OC Edition G.Skill 2x4GB 2133Mhz 11-11-11-30-2N 
Hard DriveCoolingOSMonitor
Seagate 1Tb Noctua D14 Windows 7 Home Premium 64-bit Samsung 50" LCD TV 
PowerCaseAudio
Corsair HX850 Aerocool X-Predator Kenwood AMP w/ Custom Home Theatre 
  hide details  
Reply
Logix
(23 items)
 
ProjectZero V3.0
(11 items)
 
To get list...
(1 item)
CPUMotherboardGraphicsGraphics
4820k @ 4.5Ghz Asrock X79 Extreme6 NVIDIA GeForce GTX 780 w/ EK waterblock XXL NVIDIA GeForce GTX 780 w/ EK waterblock XXL 
RAMHard DriveHard DriveHard Drive
G.Skill Ares 16Gb 9-11-10-28-2N 2133Mhz Hitachi 1Tb Seagate 1Tb Seagate 4Tb 
Hard DriveOptical DriveCoolingOS
Samsung EVO 250Gb Liteon Bluray Custom W/C Windows 7 Ultimate 64-bit 
MonitorMonitorKeyboardPower
LG IP237 Dell U2713HM Corsair K90 OCZ ZX 1250w 
CaseMouseMouse PadAudio
Corsair 900D Corsair M90 Razer Vespula Auzentech X-Meridian 7.1 2G 
AudioAudioOther
Roccat Kave XTD Digital Swans M10s 2.1 Speakers Razer Orbweaver 
CPUMotherboardGraphicsRAM
Intel Core i7 2700K Asus P8Z68 Pro GEN 3 Gigabyte GTX 570 OC Edition G.Skill 2x4GB 2133Mhz 11-11-11-30-2N 
Hard DriveCoolingOSMonitor
Seagate 1Tb Noctua D14 Windows 7 Home Premium 64-bit Samsung 50" LCD TV 
PowerCaseAudio
Corsair HX850 Aerocool X-Predator Kenwood AMP w/ Custom Home Theatre 
  hide details  
Reply
post #48 of 48
I noticed framerate degredation with PCI-E 3.0 @ 8x/8x relative to 16x/16x years ago with only 3x 1080p monitors and only 2x 780 Tis, let alone 3x 1440p and 2x 980 Tis, let alone 3x 4k and 3x 1080 Tis (AKA Titan X Pascal).

It makes me cringe everytime I read ppl say the difference is only like 2% (unnoticeable) all these years
Quote:
Originally Posted by ProjectZero View Post

Has anyone addressed that "CPU bottlenecks" might be present in those SLI test results?
I highly doubt it unless there's a coding problem. They're not doing CFD (Computational Fluid Dynamics) for a new part of Ferari's Formula 1 car.
Edited by spin5000 - 9/30/16 at 1:38am
PC
(11 items)
 
  
CPUMotherboardGraphicsRAM
i7-5930K @ 4.5/3.3 Ghz ASRock Fatal1ty X99X Killer 2x (SLI) GTX 980 Ti @ 1500/7750 16GB (4x4GB) DDR4 2400MHz 11-11-11-24-1T 
Hard DriveCoolingOSMonitor
Sam 840 Pro 256GB, WD Vlcrptr 1TB Swiftech H320 (push/pull) Windows 8.1 64-bit 3x Dell S2716DG 
PowerCaseAudio
Antec HCP-1200 Cooler Master HAF 932 Auzentech X-Fi Forte 7.1 
  hide details  
Reply
PC
(11 items)
 
  
CPUMotherboardGraphicsRAM
i7-5930K @ 4.5/3.3 Ghz ASRock Fatal1ty X99X Killer 2x (SLI) GTX 980 Ti @ 1500/7750 16GB (4x4GB) DDR4 2400MHz 11-11-11-24-1T 
Hard DriveCoolingOSMonitor
Sam 840 Pro 256GB, WD Vlcrptr 1TB Swiftech H320 (push/pull) Windows 8.1 64-bit 3x Dell S2716DG 
PowerCaseAudio
Antec HCP-1200 Cooler Master HAF 932 Auzentech X-Fi Forte 7.1 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [PugetSystems] Titan X Performance: PCI-E 3.0 x8 vs x16