DIY eGPU experiences

ok..here is my problem in doing the PCI reallocations..the 4th step..how to do it for windows 7/32??i try to type the code in the command prompt..but the bcdedit cant be open..said some error..thus i dont understand n stuck here..can anyone explain to me..how to Create a Win7/32 bootloader menuitem using the disk image..
please help me as detail as u can..because i am totally stuck..


i think the next step..i do know how to do it..i already stucked...please reply asap..pleaseeeeee!

ok..now i succeeded do the 4th step..but there was always a problem..because when i try boot the setup 1.x at boot menu..at boot menu..unfortunately..i cannot move my selection..i hope someone can help me..huhuhu..because i dunno why on earth my arrow wont do what it suppose to do..

 

Hi all. Just found today that you can "do it your self" vidock. So here is my super question. I have a MacBook Pro. Here are the specs MacBook Pro "Core 2 Duo" 2.4 15" (08) Specs (Early 2008, MB133LL/A, MacBookPro4,1, A1260, 2198) @ EveryMac.com . As it has a ExpressCard/34 slot. Can i do DIY ViDock booted to Windows 7 32bit ???

 

Since you have the express card slot then there should be no problem doing it with windows 7.

 

This is the second Macbook post we've had. I'll reply to the early one posted:

POST presumably is like bios on a PC and it does various device allocation. If it hangs then the Mac "bios" doesn't know what to do with the 9800GTX+. So then, can't have it powered on during bootup.

Then there are two additional issues.

1. PCI Allocation for the 9800GTX. Can Win7 allocate it to 32-bit space on Macbook? If not it would appear with an error 12: cannot allocate resources against it in Device Manager. To correct the error would require some way of applying a pre-OS boot PCI Configuration. eg: DIY ViDock Setup 1.x or using grub2's setpci/write_xxxx commands.

2. Configuring the NVidia card if hotplugging. This is a quirk I found with my GTX470 testing. If the card is hotplugged in DOS while running Setup 1.x, it's detected OK, I can do PCI allocation for it OK but it will hang Win7 loading. The workaround solution I found being:

(i) apply the PCI allocation (if needed), but poweroff the NVidia card. Chainload into Win7, standby, poweron NVidia card, resume. It will be detected and function OK.

(ii) OR load a "good" 8kb PCI configuration into the NVidia card in DOS prior to chainloading. Obtaining that good configuration is done by using hwdirect to save the PCI Config space of the card after doing 1 in Win7 OR saving it in DOS using pt if can boot past the bios with the NVidia card turned on. This method would require that DIY ViDock Setup 1.x works OK.

I'm hopeful NVidia can advise on how to 'configure' the NVidia card when hotplugged.

 

ok after select from the boot menu setup DIY Vidock..what should i do..it direct me to grub:>....what should i type in that to do the realocation of my memory??please..
nando..or who ever..please help me...
or did i do sumthing wrong..?

 

hershbergien Any luck making it work? Did u try installing the nvidia 9800GTX+ drivers in osx prior turning on the card? And did you try it bootcamped in windows 7? This can help with osx drivers http://aquamac.proboards.com/index.cgi?board=hack1&action=display&thread=528 . But i need to know how it is when bootcamped to win7.

 

how do i know whether i have already PCI reallocated
or not??

 

when i try to chainload..why at the last its stated that

start cmain().....

what is that..and i wait almost 15 minutes..and it seems unresponsive..

can someone tell me..

 

Sorry for the silence, I just flew home yesterday for some hard earned holidays. No worries, I brought the GTX 460+PSU over so that I keep tinkering with it. This also means I won't be able to collect any HD5750 data until I return (early August).

Yeah I'm also suspicious of that result. The DiRT 2 bench runs are not entirely reproducible because the 8 cars are A.I. controlled and vary in how well they drive between runs. It might have been a weird run (e.g. cars piled up during the lap). I'll run some more runs in triplicate to see if I can reproduce that particular result.

 

i am sorry to bother everyone..but i am a newbie..and very interested in this DIY Vidock..

my spec is..

XPS m1530...win7/32...T9300 2.5GHz..4GB RAM..nVidia 8600mGT..
try to use XFX ATI HD5770...with PE4L+EC2C connected with ATX PSU 350W..

i am very2 sorry..if all my questions are nonsense..
i want to tell that i already succeeded (i think) in doing the PCI realocation..
using the 36bit...and has reallocated 256mb PCI windows..as stated..

and now..i am confusing how to used the ViDock..as i connected the GPU to my laptop..my laptop recognised it well in the device manager..this is before any driver installation..
but after i installed the 10.6 ATI driver..the 10.6 software can't detect my HD5770..i don't understand this is prior to what??...i tried a lot using a lot of times..and i still can't figure it out..
i also used the 9.8 ATI driver software..but it also fail to detect the HD5770..but my laptop detected it in the device manager..

what should i do nando?? i am currently blur...and helpless..

 

I have my system connected to ATI 5770. The result was fantastic.

Laptop spec: Sony Vaio CW i5 520m core with 4 GB ram + 330m


I've never used any booting method but just connect it and it runs like a charm!!!

Here is score with ATI 5770 (10.6 driver)
3DMark06: 11758

Pics:




Everest report:




But I wonder how to enable x2 mode...-_-

 

@nand

I tried combing this forum for specifics on my system, and I am confused about a few things. Hopefully someone can help me!

I currently have a Dell XPS M1330 (ICH8 version)
it is a T8300 @2.4GHz with 4 GB ram and win 7-64bit
It also has the geforce M8400 GS 128MB

I am very interested in getting this external graphics solution working for my rig. As I understand it, I have some PCI express slots that are open, but my wifi is in port 2, and this needs to be moved? Is it possible to move it to another empty PCIE slot and it will still work? (I read somewhere that the ICH8M version has 6 PCIE ports, and therefore is it possible for me to run my vidock in 2x, and still have my wifi enable in another slot?)

Also, if I just run the 1.0 x1 version, how do I set the x1E workaround (i.e. - do I still need to move my wifi to another spot?)

And finally, I cant find any data for my rig's current performance to one that has the external graphics up and running - most of the ones I saw were for a PCIE 2.0 system and such, do you know about what performance gain I would get with just the 1.0 x1 version?

One final question: Is it true that with the new catalyst 10.6 or a newer NVIDIA card, I do not have to do the PCI allocation at all to use this with 4gb ram? (essentially it would just be plug in/install drivers and play?))

Any help would be much appreciated.

 

A good one to get sorted to confirm x1E is only required for ATI or older NVidia cards.

Can you also do HD5750@x1 benchmarks even if it's in August? That is by far the most common link speed and would probably just confirm my results showing a GTX4xx is the better performance choice. Eg: I got 3dmarkvant.gpu of 8k on my 2510P+GTX470@x1 or just shy of 4k for my HD5750@x1. That's with a 3-year old [email protected] cpu. New Calico test in 3dmarkvantage with the GTX470@x1 is spectacularly smooth HD5750 is jumpy. Exactly as observed in NFS Shift - the GTX470@x1 plays it great. The HD5750@x1 is unplayably jumpy. Yes.. the 2510P+GTX470 plays NFS shift smoothly.

1. ensure your HD5770 is seen in Everest attached to a pci-e port (Motherboard->Chipset->South Bridge)

2. If your HD5770 is detected but no driver loaded then it will appear as a VGA device in Device Manager->Display Adapters. Uninstall that driver.

3. re-install Catalyst 10.6 which upon detecting your HD5770 will load a driver for it.

Your experience is now linked off the first page. You have no options for either x1E or x2 mode. Only way to get noticable better performance is to get a GTX460 which you'll need this info for..

Hybrid Graphics owners

We've had two Sony owners with Hybrid Intel/NVidia graphics. One owner tested a NVidia card finding the Sony Switchable gfx driver caused some conflict and could not get it to work.

The good news is I found that these systems (eg: HP DV6-3006TX, Sony) bootup with both the IGP and the extra NVidia/ATI graphics active with the IGP being the primary. On the DV6-3006TX I just disabled the x16 port hosting the HD5670 using 'setpci -s 0:0.0 54.l=0:8' so the HD5670 was never registered on the PCI bus. This had the secondary beneficial outcome of freeing up an additional 256MB of 32-bit PCI space perfect for hosting the desktop graphics.

A Hybrid Graphics menuitem has been added to the next beta Setup 1.0e2 version to enable/disable the IGP/discrete graphics or anxious power users can refer to this for the time being.

All the implementations shown are pci-e 1.0 2.5Gbps speed. The series-5 chipsets are pci-e 2.0 compliant for the power management component but link speed is still 2.5Gbps. There are currently three variations on performance shown on the first page: x1 1.0, x1E and x2 1.0.

XPS M1330

Based on the M1330 port layout consider two options.

1. do a x2E link using your wifi and wwan ports. Means could go a NVidia GTX4xx or ATI HD5xxxx, both with great performance. Would mean having access doors to the underside of your system OR removing the plastic cover.

2. do a simpler x1 1.0 link using your expresscard slot port4. In that case I would suggest going a GTX4xx for it's superior x1 1.0 performance as shown by the GTX470 versus HD5750 benchmarking.

 

Thank you so much for the prompt reply Nando4;

I will think more about which configuration I want to do -

I do have another question, is there anyway to preserve wifi function AND have a x2 config without needing the USB WIFI? meaning, can I use the port 1 and port 2 for x2 and then move the wifi to another port?

thanks again!

@ ithildin

Where did you buy you gainward GTX 460, and how much was it - I think im gonna run a 1.0 x1 for now with the GTX 460 1gb - (maybe go to x2 in the future)
Im looking to spend around 200 for the card. let me know!

 

Thanks for comment, nando4. But I will look for ATI 5850, for I really want to play Arma 2 with a better Dx 9 performance. If I get ATI 5850, I will post the improvement from 5770. 460 OC version is really nice to have but I need more performance for Arma 2 that is hard coded only with Dx 9.

Updated: Here is RE5 Bench result:

 

If the HD5770 is not peforming well then suggest get a GTX460. A HD5850 will not give you much more than a HD5770 at x1 1.0 speed. A HD57xx+ really needs x1E to give it a bit more oomph. Worth noting that NFS Shift is a DX9 compile and runs far far better on the GTX470 than the HD5750. Pretty much all GTX460's can be overclocked to ~GTX470 levels.

At the very least would be worth getting a GTX460 from a retailer with a generous return policy to do comparative testing with Arma 2.

 

I found other portable dock companies beside Magma:


Spectrum this looks expensive...


DECA Japanese company, expensive about $350


HP Riser Card this looks nice and cheap,,, $32-ish?


HP PCI Expansion Blade (perhaps too long??) = expensive... $700ish

I really wish someone could build PCI enclosures + room for TFX/Flex ATX PSU for cheap!!!

Perhaps someone interested to buy HP riser card? I'd be glad to hear his experience

 

I've decided to run the 1.0 x1 for now with the geforce gtx460. I had a couple of questions though. Are there some brands I should stay away from due to overheating/too much power consumption? Also since I'm only running x1 do I need to get the 1 gb version for optimal performance? Or will the 768mb be comparable. Also which power supply should I use?

Finally, if I put ports 1+2 on my m1330 Dell as shared x2, can I put my current wifi pcie card into ports 3 or 4?

I'm in the u.s. And am having trouble finding some good geforce gtx460 availability. Also other than the geforce gtx460, is there a cheaper option that would still give me excellent performance on x1? My main goal is to play the new sc2 on high (or dare I say ultra) settings, as right now it only works ok on low settings

 

Okay, so I've got a HD 5750 connected with the PE4L to my Thinkpad, and it starts up fine. I can get the laptop to output through the video card to my external monitor. However, after several minutes of running, even if I'm just doing simple web browsing, the monitor just all of sudden reports that there's no signal and I'm forced to restart the laptop to do anything.

Any ideas what's wrong?

 

I've repeated some the DX9 checks and indeed that first result is not reproducible. Here are the results of the 7 subsequent runs:

DiRT 2	Samples	Min FPS	Avg FPS	DirectX
GTX 460 x1E	4771	30.32	56.68	9
GTX 460 x1E	4985	35.03	58.51	9
GTX 460 x1E	5781	44.56	65.70	9
GTX 460 x1E	4776	35.53	56.83	9
GTX 460 x1E	5094	43.64	60.36	9
GTX 460 x1E	4876	36.79	57.32	9
GTX 460 x1E	5395	35.40	63.72	9

I sat and watched every single one of these runs. Basically, every time some of the cars in front crash or the AI driver lags behind, the game has to render fewer cars in the picture and so the frame rate climbs (>60 avg fps). The median entry regarding Avg FPS is highlighted - 58.51 fps which is very similar to the DX10 result, confirming that there is no performance difference between x1 and x1E on Fermi GPUs.

 

GTX460: 768MB versus 1GB We don't have benchmarks to show if the extra vram and ROPS of the 1GB GTX460 justifies the 15% price premium over the 768MB card on our x1 1.0 setup. There's also a Sparkle 2GB GTX460 which would theoretically be the best card to get for highest x1 1.0 performance.	GTX460-768MB GTX460-1GB VRAM/bus 768MB/192-bit 1GB/256-bit ROPS 24 32 RRP US$199 US$229

ATX PSU to drive a GTX460

I expect users will overclock their GTX460s so budget for a PSU capable of delivering 225W ([email protected]) from a single rail. The GTX460 uses two 6 pin pci-e power connectors. A molex-to-6pin pci-e adapter will be supplied with the card. Means look for a PSU with an additional 6-pin pci-e power connector.

A quality, inexpensive newegg PSU matching that criteria being a US$30AR Corsair 400W rated at 12V@30A (360W).

Thank you for taking the time out to confirm this important point.

This means the GTX4xx delivers full x1 1.0 performance on Intel chipsets. ATI cards needs the x1E tweak, specific to only some systems with the right port layout. Furthermore, ATI have expressed no interest in solving their handshaking/performance issue to get full x1 1.0 performance on Intel chipsets.

The new best-bang-per-buck DIY ViDock card: The NVidia GTX460

I'll now be recommending the NVidia GTX460 as the new best bang-per-buck DIY ViDock x1 1.0 video card, taking over from the ATI HD57xx series. Leaving then only the recommended lower power 12V/75W HD5670 or 12V/80W HD5750 setups which might change in August as well when NVidia are speculated to release their 75W products: GF106-based GTS4xx, probably GTS420/440.

 

@ Nando

I will be buying the GTX 460 and that recommended power source. I have another quick question (BTW I really appreciate all your time and patience).

Will running DIY vidock setup have any permanent effect on system stability and performance when not running with the vidock? As in does it make bios changes that would mess up my computer? Also, if it does make these changes, is there a way to backup the original settings to restore it if it messes up?

Also just wanted to confirm that the PE4L would work with the GTX460 and said powersupply Thanks!

 

DIY ViDock Setup 1.x makes no permanent changes to bios. It just changes PCI configuration registers which get reset back to defaults upon doing a cold reboot.

PE4H versus PE4L for GTX4xx use

The GTX460 is smaller than my GTX470 so consider that when reading my comments below.

Yes, the PE4L+ATX PSU works perfectly well with my GTX470. I do however prefer to use the PE4H as it has a longer, wider and heavier base so can self-support the GTX470 in an upright position. That's desirable to allow airflow to cool components on either side. The mini HDMI socket is mounted in an easier to access spot too.

The PE4L on the otherhand is not firmly planted with the GTX470 installed. It would need to be mounted on say a metal ruler to give it a stable self-supporting base with airflow. I had a temporary setup placing a buffer sheet/box between the GTX470 and the ATX PSU, similar to this for it to lean against. Not the best solution for airflow. The PE4L is however better for my lightweight portable 12V/80W HD5750 setup.

If not wanting to engineer a self-supporting base for a PE4H AND if you want to keep costs as low as possible then consider a US$85 PE4H + US$10-ground shipping from Harmonic Inversion. PE4H has the other advantages of x2/x4 link connectivity with your notebook OR desktop system, the latter using a hwtools' US$25 HP4A.

 

So I read that with the Nvidia GTX cards , it runs into problems when hotplugging. Does this only happen when restarting the computer? (meaning that if I just wake up my computer from sleep with the express card still plugged in, it would work?

Also My system has 4GB of ram --> do you know if dell M1330's can do the appropriate PCI reallocation?

(Solution#4: Use an NVidia desktop card which whose driver works can use a 256MB window in > 4GB space (not: mobile doesn't allow 64-bit relocation). --does this mean that Mobile cards dont allow 64-bit relocation?

 

If you don't need Setup 1.x then it's just boot Win7, standby, poweron DIY ViDock, resume.

On the DV2000 I needed Setup 1.x to do PCI allocation AND I needed the GTX470 off during bootup to get past the bios as with it on would just start beeping. This is how I managed to get it all going:

Recipe for hotplugging GTX470 in DIY ViDock Setup 1.x

PE4H 2.1 with it's 7 or 15 second PCI Reset delay might eliminate the need for steps 1-3 AND might make the GTX470 be in a 'configured' state making step 5 irrelevant. I'll have to wait and see once I have one to test with.

Once I had the working pci.bat generated it was very easy to get the GTX470 working:

Automated Startup

My dilemma now how to 'configure' a hotplugged GTX470 which I've asked NVidia for help. That might be necessary for x2 1.0 setups as standby/resume can lose the x2 port settings. I have one workaround that I use on my systems - save a 'configured' NVidia GTX470 8kb PCI register dump and load it into the hotplugged GTX470 before chainloading. Result is I can even chainload straight into Win7 without it hanging. Seems to work but I'll need to test more before I commit that into Setup 1.0f.

Also not sure if GTX460 vbios behaves different to my GTX470.

Dell M1330 specifics

We don't have one example M1330 as yet so I can only speculate. I believe your M1330 will need PCI compaction to work with 4GB as I know a 4GB M1530 needs it, or RAM removed to work. I found NVidia mobile gpus (eg:8800GT) didn't want to be 36-bit relocated. Not a problem since the GTX470 was fine being relocated to 36-bit space on the Inspiron 1440 with Win7/32.

CO-existing Nvidia mobile and desktop drivers

I had previously worked offline with a Clevo owner and we did manage to get the 9300M+9600GT(desktop) working concurrently in Win7. Meaning then the mobile and desktop NVidia drivers can co-exist.

 

As I understand it - there was a tested M1330 with a GTX260 right here: http://forum.notebookreview.com/gam...e-out-how-make-diy-vidock-16.html#post5122549

I hope so too! - if not I guess Ill return the GTX and settle for a radeon...

I guess my main concern now is making sure that the PCI reallocation will work - otherwise this is all a lost cause!

 

May I know what is this H4PA? I couldn't find it anywhere in the hwtools website.

 

Whoops. It's a HP4A. See here.

 

Hey Nand,

Before I took the plunge and shelled out my 300 dollars, I was kinda hoping you could tell me if my system could actually accomplish the PCI reallocation (i would really like to keep my 4gb of ram as its in 2-2gb sticks and I dont want to take one out or buy a new 1Gb stick)

I attached the device manager shots below (in two images, as it got cut off)

I hope this is the information you need

thank you so much in advance

 

Dell XPS M1330 PCI config space analysis

Your Device Manager pics show TOLUD=E0000000 (3.5GB). Dell have set a high TOLUD on your M1330, exactly the same as on the Inspiron 1440 so as to maximise the 32-bit RAM available in Win7/32 and XP.

So you have 512MB of 32-bit space to accomodate PCI devices. 277MB is taken by the 8400M, 44MB by the systemboard. Other devices take a small bite too. In short, there is not a chance to find 128+64+32 required host a GTX4x0 in 32-bit space.

So you'd be counting on Win7 to work with your GTX460 allocated to 36-bit space. It certainly works fine with the PM45+X4500 equipped Inspiron 1440. Will it work with a PM965+8400M equipped M1330? Only way to get 100% confirmation is to try it.

A contigency plan *if* you had updated to the latest NVidia drivers, run Setup 1.x to allocate your GTX460 to 36-bit space and still couldn't get Win7 cooperation would be to swap a 2GB module with 1GB. The liberated 512MB of PCI config space would then provide ample room to host the GTX460 in 32-bit space. Your system would register as having 3GB of system RAM.

 

ati hd5770 1 gb successfully installed on lenovo t400 - t9800 @2.93 ghz in x1-mode via expresscard
win7 x64, 4 gb ram installed

3dmark06 (standard) = 10696
re5 benchmark (dx9) ~ 51 fps
re5 benchmark (dx10) ~ 36 fps

works like a charm on medium resolutions, no pci-allocation necessary!
on higher resolutions fps drop...

 

Thank you for posting your experience.

If want higher FPS in 3D apps/games as shown in th e HD5750@x1 versus GT470@x1 benchmarks consider RMAing the fresh HD5770 for a US$200 GTX460-768MB. This non-response by AMD requesting assistance in obtaining full x1 1.0 performance on Intel chipsets being a good reason for the RMA.

 

What the... I can't believe the answer you got on the AMD forum. It was a perfectly legitimate question regarding a basic problem in AMD cards. Saying it was rude, self-important, condescending and misinformed doesn't quite nail it...!

 

And the thread is locked!!! I just registered to object against that mod, searching where is the reply button, hell now I know where it is!!

 

the mods awnser was really crude...
I got the ati very cheap, that's why I will keep it until prices for gtx 460 go down.
maybe I will try to use x2 mode for home use.
is there already a work-around for lenovo t400?

 

Hey nand

I know you said only way to know if it will work for sure is to try it out, but is there another to yes before I shell out the cash? What prevented other systems from using the 36 bit allocation? Was it just the ati cards?

 

x1E or x2 is only way to get the HD5770 to have more bandwidth to play with. TO use either would requiring investing in a PM3N and attaching it to your turbo memory slot. On the x200 at least the turbo memory slot was hard to get to. It required lifting the keyboard to access so was not really a workable solution.

We don't have a confirmed 4GB M1330 or M1530 as yet. Closest confirmation I have is a 965PM+9800GTX M1730 showing 36-bit PCI allocation active on the expresscard slot (port4) here. So it is likely to work, but keep the contigency of running with 3GB in mind 'just in case'. Haven't heard back from Scrat9876 to see the M1730+HD5670 is working. Would be interested to know if it worked using Setup 1.x's 36-bit selective compaction, with a confirmed Device Manager screenshot showing 36-bit allocation.

So then the cheapest way to approach it would be aquiring just the hwtools gear and connecting a borrowed HD5xxx or GTX2xx/GTX4xx + PSU just as a trial run before investing in the GTX460+PSU. FYI: I've also updated my PSU recommendation to a $30AR Corsair 400W (80% certified) 360W/single rail. A great bargain PSU for the specs.

 

+1 for the Corsair PSU recommendation. My 400W Corsair is rock solid and barely audible over the GTX 460 ventilator even under strain. It's pretty cheap too.

 

But I wonder there could be any driver conflict between 330m and 460 GTX. Anyways, I ordered 460 GTX from Newegg. But they don't offer refund. If this 460 GTX fails, I will put this on eBay.

 

Which one did you order: Im looking at these two:


Newegg.com - SPARKLE SXX4601024D5SNM GeForce GTX 460 (Fermi) 1GB 256-bit GDDR5 PCI Express 2.0 x16 HDCP Ready SLI Support Video Card

and

Newegg.com - GIGABYTE GV-N460OC-1GI GeForce GTX 460 (Fermi) 1GB 256-bit GDDR5 PCI Express 2.0 x16 HDCP Ready SLI Support Video Card

And they both say full refund within 30 days.

Also @Nando
The sparkle card linked above (first link) - says it needs a 450W Powersupply minimum - can that 400W one you mentioned earlier still work? ALSO which do you recommend out of the above linked cards!
finally, do I understand correctly, that I will be trying to relocate the GTX 460 to 36bit (something that is supported by it, and I just need to try it and see if it works) - If i read correctly on the Inspiron 1440 - it looked as if the IGP was relocated to allow the external to be hosted on the PCI bus. In my case, I want to relocate the GTX 460 straight to 36-bit space and hope it hosts?

** I was looking at the earlier device manager shots I uploaded, and it looks like Port 4 ( my express card slot) is already in 36-bit space correct?

 

Yes your port4 is using memory in 32-bit and 36-bit space. That's a promising start. The IGP relocation you are referring to was how I got the HD5750 working on the Inspiron 1440 initially as ATI 10.3 driver refused to work in 36-bit space. Since Catalyst 10.6 now supports 36-bit PCI allocation I no longer need to do that anymore. NVidia's Forceware 257.71 also uses 36-bit PCI allocation on my Win7/32 setup.

If I was to buy one I'd go the US$235 MSI Cyclone GTX460 reviewed here with a 725Mhz factory overclock and very quiet non-reference fan.

550W recommendation is for a desktop system, motherboard+hdds+gfx. You're only running the gfx which requires 225W. The US$30AR 400W Corsair CX with a single 12V/30A rail (360W) will easily meet that 225W requirement. 80% certification means it's energy efficient too, saving you $$ on electricity bills. Has ithildin's seal of approval

I worked with a Clevo owner offline and got a 9300M + 9600GT working. So the mobile+desktop drivers can co-exist. The standard NVidia GT330M driver can be used on the CW as it has no hybrid gfx.

 

OK i think im gonna pull the trigger.

The newest Nvidia drivers still support the 36-bit allocation right? therefore theoretically mine should be a PNP solution?
-do you have any idea why that M1730 failed?

Im pretty excited - Im gonna take the plunge and get the P4EL + the recommended power supply above, along with the Gigabyte GTX 460

Additionally to be on the safe side I may go buy a cheap 1 gb ram module to run it in 3gb if all else fails.. Ill let you know how it goes!

I really appreciate the help

 

Ah,,,I didn't check all of 460s...I ordered this:
Newegg.com - EVGA 768-P3-1362-AR GeForce GTX 460 (Fermi) Superclocked 768MB 192-bit GDDR5 PCI Express 2.0 x16 HDCP Ready SLI Support Video Card

It is a 768 MB version with OCed clock. (Core clock is 763 MHz)


@Nando, thanks for comment. I really hope this 460 GTX with 763 MHz OCed version may work!!!

 

Hello, I am planning on purchasing the ExpressCard External Graphics x1 from Harmonic Inversion. I don't plan on using it for better graphics but for a capture card, the Blackmagic Intensity Pro. It's 1x PCE Express card. I was wondering before I purchase, if this is possible and if I will run into any problems.

Specs:
Gateway FX P-7805u
Upgraded to Core 2 Duo T9900 and 8gb of ram
Windows 7 64 bit Ultimate

Thanks

 

This is just what I need honestly. I find that adding a good video card is what adds price, weight and power consumption to a laptop.

But basically all I need is this:
ExpressCard External Graphics x16 with a video card and a PSU and I'm fine right?

 

basically those are the components you need , but your laptop must also be compatible with the configuration - check the first page of this post for some examples and methods for getting it to work.

 

Hey nando

Is there no possibility in joining port 4+5 together to make X2, has anyone tried it before? , i want to see if there is chance for an x2 link before buying a PE4L

 

just overclocked the hd5770 from stock (850/1200) to 950/1300
=> 3dmark rose from 10696 to 10955
conclusion: no real improvement!

3dmark vantage gpu score: 5884

 

No chance. I was hopeful for x2 1.0 on the tested DV4 and Inspiron 1440 trying all the permutations of port 1+3+6 and port 2+5 combos respectively. None worked. The test confirming Intel's datasheet: a x2 link must use the physical port1+2 OR 3+4 OR 5+6 OR 7+8.

x2 capability is therefore limited by what I/O port links the manufacturer set in their design for the mPCIe and expresscard slots.

If the only option you have is x1 1.0 (not x1E) with your Intel chipset then suggest go a NVidia GTX4xx for maximum performance. If need a compact 12V/75W setup then can wait a few weeks for the 75W Fermi card, likely called a GTS4xx to be released.

Yup. I too found overclocking at x1 1.0 link speed gave no real improvement. Only when I had x1E bandwidth did it overclocking show some decent FPS/3dmark returns.

 

Two questions...

1) From reading this thread it appears that you can't match desktop performance using just your ExpressCard slot. If you had an ExpressCard 2.0 slot, could you match it? Or would you need 2 ExpressCard 2.0 slots?

2) I have yet to see any laptops be released with ExpressCard 2.0...when will that start?