ViDock 2 - My experiences so far

totally true, FSX is the hog of hogs.

 

Meh, FSX is just cpu hungry get a good quad and your good to go.

GTA4 is cpu hungry too but also requires a lot of vram.

Crysis is the GPU hungry game

So they can each place the death toll on your computer depending on what it has in it. Most laptops do not have quad cores in them but those that do should handle FSX really well.

 

I dug up this article on tomshardware.com

http://www.tomshardware.com/reviews/pci-express-scaling-analysis,1572-8.html

It's quite interesting. It seems though that GeForce has more trouble running at 1x than ATI cards.

 

Any benchmarks of FSX on a tricked-out quad-core system with a weak GPU? If I remember correctly you own a system with a quad-core and a decent high-end GPU so you should be able to max FSX out on native resolution.

 

Yeah seen a quad and like a 8800 for desktop and it ran it great, also for almost everybody when they released the quad core patch performance in the game doubled if they had a quad core cpu, showing it was the cpu holding it back. If you start customizing the game with custom textures and things, it can get much more graphically demanding.

I got a free copy of FSX Deluxe from microsoft but sold it before I ever opened it because I needed the money and didnt have a computer that could play it at the time. Kinda wish I had it now. But I will most definitly have to get it when I get a HD projector or do my 3x monitor setup.

 

nice find, props. ill have to look into that a little. im hoping that only ati can hold this. get them a place in the market they own so they can get something on the green giant. plus, ati's are cooler .

 

Tomshardware had a review of the previous ViDock. They concluded that any card faster than Radeon 2600 is waste of money since you can't get any performance benefit. Yet, ViDock2 runs 3x as fast. So there's something going on in here.

randomdude, I preffer NVIDIA actually, since they always had better linux support ;-) And I have big interest in CUDA, since I do scientific computing. But other than that, ATI sounds good.

There are some big architectural differences between NVIDIA and ATI. That might be one reason for different results. Or it might be just drivers. Maybe ATI, without even knowing, just did the "right" thing for devices like ViDock. Or maybe it's just coincidence?

I hope Stenier32 can do COD4 benchmarks ;-) After what masterchef said, I'm interested to see that.

 

well ya, that makes sense. you were saying you were programer and nvidia w/ cuda is right up your ally. for me, ati does just fine and i would rather support them with my $ than the rebraded green giant. if ati could get thier drivers up to part with nvidia, they would pwn.

either way, an ati 4890 kicks an nvidia 94/300 anyday.

ps, i never really understood the whole architect thing of gpu's

 

yes
something is going on here
vidock 2 is a completely different product than the vidock 1 that was in that review
it is newer and has a much different purpose (vidock 1 was not targeted for 3d performance)

 

I grabbed the COD4 disc and dropped it in my bag. I have played through it - so I will copy the save game folder over to the borrowed laptop.

I own 3 copies of FarCry -- can't find a single one. 2 DVD and one original CD version.... I will keep looking.

 

sweet, loving it steiner. dont kill yourself though.

so, what ports are on the back of this thing? never actually figured out?
hdmi?... what else? kind of a basic question after we just had like 10 pages on the bottlenecks and theorys of this thing lol. but why not .

 

Typically it will have VGA, DVI and HDMI ports, but could also have Dual DVI.

 

Thanks Steiner ;-)

If you'll have few more minutes, there is a benchmark for Resident Evil 5 that was recently released:

http://fileshack.com/file.x/14772/Resident+Evil+5+Benchmark+Version

It might be good to try ;-) Don't know how good port is it, but that's one of the latest games.

 

thanks donald.

kk night peoples.

 

Downloading - and currently installing COD4.Odd - but some of the pics I uploaded are gone. Will try to find them and put them back into the OP.

 

The forum has been malfunctioning a bit lately - maybe that's somehow related to missing pics. I've been getting "Database Error" message quite often during last 3 days.

 

OK -- installed and played a bit with COD4.

Settings used:
1280*720
4X AA, max AF
Trilinear filtering
High on all quality settings


Laptop GPU --
Unplayable. Sucked. Fraps was reporting about 8 FPS inside buildings, walking outside from there dipped to 0, then up to about 3-4.


ViDock - 4670
Playable - but I would probably drop something down to take care of the transitions from inside buildings to outside. Inside buildings I would sit in the mid 30s, walking outside there would not be a noticiable stutter per say, just a perceived slow down to ~15 FPS, then it would climb again to high 20s.

Completely playable, but I would probably change something to make it play a little faster.

COD4 is a game that plays best about mid 20s and above. Crysis is completely playable in the teens.

 

i think that is kind of what we expected, a bigger stutter than normal because it is sending all the new stuff to the gpu at the slow bandwidth. cool. were there any random spikes down in frames? not when going from one setting to the other, but just throughout the game?

 

This stuttering is interesting I assume it might be ironed out with the more powerful vidock, unless it is simply the bandwidth limitation as mentioned above that is the sole cause.

 

No other stuttering. Smooth other than that. I am keeping the laptop around this time for a couple of days. Will try to play this weekend.

 

Thanks! Keep up the good work! Did you manage to run that RE5 benchmark?

BTW, I think it would be interesting to see if disabling AA/AF improves the performance.

 

No - thanks for the reminder. I have it on my main laptop. Will take it home and install it.

 

Resident Evil 5 Benchmark: -- All DX10 on Windows 7 - 64bit
1152x864
AA - 4X
Motion Blur - On
Shadow Detail - High
Texture detail - low
Overall quality - high

16.9 Avg FPS

+++++++++++++++++++++++++++++==

1152x864
AA - 2x
Motion blur - off
shadow - high
texture - low
overall - high
17 fps avg.

++++++++++++++++++++++++++++++++++
1280x960
AA - off
motion blur - off
shadow - med
texture - low
overall - med
26.2

Shadows make a big difference.

Let me know if you want it run at a specific resolution with settings - I will try to do so.

 

Thanks Steiner!

Hmm... that looks much slower compared to 4670 in a desktop, which gets some 44fps with 1280x720 and all settings maxed.

Maybe it's DX10. I think DX10 is much much slower. Also, did you use FRAPS with RE5 benchmark?

 

also, 1152x864 is like 70k more pixels. but still, shouldnt do that much. maybe it is dx10. who knows...

 

Thanks for all the help Steiner. Looking to pick up the ViDock3 when it comes out. +rep

 

Used DX10 - can use DX9 in the morning. I will report back.

 

Sorry - But I call BS.

I ran the benchmark again on the Lenovo with ViDock2.

1280x720
DX9 -
settings on high, but motion blur off and AA off.
Fixed benchmark:: 20.1 FPS
Variable Benchmark:: 28.2

I thought that double would be a a little odd on something without the cable as a "bottleneck" - so I ran the benchmark on my laptop.
reminder - Sager NP5797 - T9900 (3.06Ghz), 4Gb RAM, SSD, GTX280M.
This time I maxed out everything - but left the resolution at 1280x720.
This means - all on High, 8x AA, Motion Blur: On. fixed -- 44.8 FPS.
Variable:: 70.7


I think that you read something that was incomplete on a desktop getting over 40FPS with everything maxed. Benchmarks aside, the 4670 is less than half the GTX280M.

I stand by my notion that this device allows you to game with hardware not created for gaming. But your casually tossing out numbers in ignorance kind of irritates me.

 

so, i found this
http://www.dynamictube.com/youtube/...t-evil-5-hd-4670-by-gargarov=L0hCTki4C14.html

it has a 3.2 cpu which may make a difference also, but ya. more stuff for us to pick at.

and more tests from stiener! awesome! thanks.

 

Great job benching steiner.

Vidock definately allows older hardware or non gaming asociated hardware to buff up.

Will look into one of these for my old sony vaio fe21s.

 

Quick question: looking at specifications, expresscard 2.0 speeds are equal to superspeed usb 3.0 speeds... do you think that the vidock 4 would come usb 3.0 ready as well?

 

if it does, it should have a thing so that you can have 2-3 usb cables for more bandwidth. many laptops usually double up USB's on a side right next to each other.

 

with usb 3.0 it may be likely to finally power the laptops own screen.

 

i don't think you understand what "USB" stands for
its a serial bus

 

they usually double up on usbs for power, i thought...

 

universal serial bus... i know that. would u mind enlightening me on what exactly that is though. I am really trying to learn as much as i can about computers right now. im curious how everything works

thx

 

wilse, there are USB external graphic cards out there. They "suck" though. I mean they can display higher res than, say, old integrated graphic card; you don't get any 3d acceleration though. So there might be some way, who knows?

 

sure no prob
because it is a serial bus, you cannot increase bandwidth to the external device by having it plugged into more than one usb port

 

Correct...the double plug in is only to get more Amps for powering external devices.

 

So you can only use this vidock with an external monitor?

 

Yes. ....

 

Resident Evil 5 Benchmark version at 1280x800/32-bit

Obtained on a Toshiba P8800-2.66 and T7500-2.2 Dell M1330 using HD4670, HD4500, 8800M GS (highest=25FPS) using DIY ViDock solution shown here. Compared against the Resident Evil 59.5 FPS on a E5200-3.2/HD4670 at full x16 pci-e 2.0 speeds here captured from this video.

Performance drop across 1x pci-e bus

I've found some info on the web giving an idea of the loss in performance going from 16x/2.0 or 16x/1.0 pci-e to 1x/2.0 or 1x/1.0. The 1x/2.0 data gives an idea of how much faster expressport 2.0 is over 1.0. Note HD4350 and 8600GT benchmarks below are particularly good comparisons.

1/ HD4350 compared: 8x/2.0, 1x/2.0, 1x/1.0
2/ 8600GT 16x/1.0, 1x/1.0 (inc Phys-X with 9600GT)
3/ 7300GT 16x/1.0, 1x/1.0

(1) HD4350's performance summarized below shows the 2x->1x performance drop averages out to be *higher* than the 16x->2x drop, especially the 3dmark06 values. Percentage below given as compared to a 16x bus:

Code:

ATI HD4350    |  16x/1.0 |  1x/2.0 (2x/1.0) | 1x/1.0
--------------+----------+------------------+-----------------
Crysis        |  16.56   | 15.01 (-9.34%)   | 13.63 (-17.69%)
3dmark06      |   2747   |  2568 (-6.52%)   |  1911 (-30.43%)
3dmarkvantage |    649   |  631  (-2.77%)   |   540 (-16.80%)

The 1x/2.0 is the same performance you'd get from an expressport 2.0 setup.

pci-express scaling analysis for current nvidia and ATI cards

Tomshardware PCI Express Scaling Analysis gives an idea of what handicapping x1 bus would give with older graphics cards. Does anyone have access to current HD4670, HD4770, HD48x0, nv9600GT, nv 9800GT, nvGTS250 could do the same sort of testing? It would be good to know x1 and x2 performance of current ATI and Nvidia offerings to determine what is the sweet spot for overall performance.

Modding ViDock for x2 mode?

Does anyone with a ViDock want to try to pass through the 4 extra pci signals from their mini pci-e slot to see if their video card will then work at pci x2 speed (10Gbps full duplex). ICH8M datasheet here suggests this is possible.

 

How would the ViDock2 with the 4670 compare to the Gefore 8600 GT GDDR3 card in my Dell XPS 1530?

 

It should perform better since you are comparing a class 2 laptop card that is now 2 generations old to a current class 2 desktop video card.

 

Can you put a more powerful card then the 4670 in the Vidock2 and get better performance then the 4670, or will it not show any improvement over the 4670 due to the expresscard slot.

 

Yes you can but in the next versions of the Vidock.

 

Wouldn't a 4770 work just the same? The 4770 only uses maybe 2 watts more power, only difference is it uses gddr5 instead of gddr3.

 

4770 will work better then 4670 and so on it's not about the power consumption different cards will have different performances at x1.

 

Does Vidock 2 require Windows 7? If not, how would one operate in Vista that is restricted to one driver for multiple graphics adapters?

 

Vidock doesn't require windows 7 it works with xp too but with 7 you will have some features that in xp aren't available.
PS:It doesn't work with vista.