|
Maximilien wrote: all fans works, so there is power to all the components, including the graphic cards. Maybe. As I pointed out in the Lounge, some PCIE graphics cards need the extra "top" power connected. It may be a matter of split power rails rather than paralleled. (I haven't got one to play with, so I can't confirm this.)
Asking Gigabyte sounds a real good idea.
Cheers,
Peter
Software rusts. Simon Stephenson, ca 1994.
|
|
|
|
|
Peter_in_2780 wrote: some PCIE graphics cards need the extra "top" power connected. It may be a matter of split power rails rather than paralleled. (I haven't got one to play with, so I can't confirm this.)
Well, there is only one power plug on each card, I plug in the dedicated pci-e plug from the power supply, each one with a different cable.
I don't know what a split power rail is ... translation or image ?
Watched code never compiles.
|
|
|
|
|
According to the ATX spec you can only put a relatively limited number of amps on a single part of the 12v power generation hardware inside the PSU (called rails). Early high wattage 12V centric PSUs followed this restriction and split 12V into two or more rails, with the potential result that if you tried to pull too much 12V from a certain subset of the plugs it would fail because you maxed out one rail even though you were well short of the total limit. The fun is that the rail structure was rarely (if ever) documented; so unless your particular PSU was reviewed and dismembered by a EE on a site like Jonny Guru/[^] you'd have no way of knowing what the rails were except by trial and error. Most (all?) new PSUs simply disregard the rail amperage restrictions and put all 40, 60, 80, etc amps on a single rail to make it easier for users. (I don't know if they put any sort of current limiting hardware on a per cable level; avoiding yanking currents high enough to melt wires down a single cable was part of the reason behind splitting rails).
You can fiddle around with different PCIe plugs on different cables; but with a new PSU it's unlikely to be an issue. Especially since split rail designs still should be able to run several PCIe plugs/rail and your 6850's only use a single plug each.
Did you ever see history portrayed as an old man with a wise brow and pulseless heart, waging all things in the balance of reason?
Is not rather the genius of history like an eternal, imploring maiden, full of fire, with a burning heart and flaming soul, humanly warm and humanly beautiful?
--Zachris Topelius
|
|
|
|
|
You could also check the BIOS version. Version 1.A have "Improved VGA card compatibility" and I did run into problems a few year back when I upgraded a videocard to a Radeon 4580 on an Intel motherboard with a Q35 chipset. I could see the motherboard splash screen but it will freeze sortly after
http://www.msi.com/product/mb/P67A-G43--B3-.html#?div=BIOS[^]
Also check your pci-e connector as Peter said. Each GPU should have a 6 pins connector at the back. They are necessary for the card to work properly. And these connnector should come from the PSU, not the twin molex to pci-e connector bundled with the card. These adaptors are for those who don't have enough pci-e connector on their PSU (read older).
Best of luck.
Vince
Remember the dead, fight for the living
|
|
|
|
|
I will (re)check the bios tonight (but I think I've already done that)
Vincebl wrote: Also check your pci-e connector as Peter said. Each GPU should have a 6 pins connector at the back. They are necessary for the card to work properly. And these connnector should come from the PSU, not the twin molex to pci-e connector bundled with the card. These adaptors are for those who don't have enough pci-e connector on their PSU (read older).
Everything is connected properly.
Thanks.
Watched code never compiles.
|
|
|
|
|
At least one of the card work (I only brought one to work).
I put it in my colleague computer (I have a crappy dell with no slots) and boom it worked and booted properly (booted up to Win7)
At of today, lunch time:
- New System is fully working with old graphic card.
- New Graphic card is fully working (at least one of the 2) on different computer (ASUS Motherboard)
- New System is not working with the New graphic card.
Next step :
Check to see if I upgraded the Motherboard BIOS with all patches from the manufacturer.
Thanks again.
Watched code never compiles.
|
|
|
|
|
I bought a GTX580, and it needed both the additional power plugs connected.
Also, i thought i had enough power with my 600W PSU, but found that the individual rails in the PSU did not supply the sufficient juice, and had to by a larger one. Although 600W was theoretically enough, the individual rail loads could not accommodate it.
so 2 things;
1) make sure you connect all the power plugs
2) make sure the PSU can supply sufficient current on the 12V rails that you are using.
|
|
|
|
|
While I was sleeping other members have provided some good information. The 'split rails' I mentioned is the converse case to Dan's (although the reasoning is the same). My split was load-side, not supply-side; that is, different sections of the card get their power from different connectors. Bear in mind that all connectors have a limited current capacity, and graphics cards draw a LOT of current, easily enough to overload a single connector. If you simply use multiple connectors in parallel, there's no guarantee that the current will be shared evenly, so you can still have a connector overloaded. (You often see multiple pins on the same connector paralleled. This works better because the pin-to-pin variations are less, so the sharing is likely to be more even.)
From your other post (3rd in this thread), it looks like you have addressed this issue anyway. It's probably only of academic interest now, but you could experiment in your one-card setup to see if my theory applies in this case. (Pull the 'top' power plug and see if it still works.)
Hope this makes sense - early morning, caffeine deficiency...
Cheers,
Peter
Software rusts. Simon Stephenson, ca 1994.
|
|
|
|
|
(I'm not the patient man)
When to the store this morning, got another motherboard, a gigabyte MB (p67a ud4 b3) to go with the gigabyte graphic cards that I have.
All seems to be working fine, I can at least boot with one of the 2 graphic card, will make sure it works with one, and in a couple of days, will put it the second GPU.
The old MSI motherboard will get shipped back to newegg next week (will do all the paper work from work)
Thanks.
Watched code never compiles.
|
|
|
|
|
I have an old Linksys Wireless-G Router and A new Linksys Wireless-N Router. Is it possible to set up my old G router as an Access Point with the N router without having a Cat5 connected between the two(2). I would like to place the G router in a different room, without running any cables.
|
|
|
|
|
|
Hi, I plan to purchase iphone4. I heard that iphone4 could not send or receive bluetooth from the other phone (htc, nokia...). Is it true? If it is true, are there any possible way to make iphone4 could send bluetooth to any other phone legally (without a jail break)?
any iphone4 user please help to comment.
Thank in advance,
|
|
|
|
|
|
You will need a dongle.
see here[^]
------------------------------------
I will never again mention that I was the poster of the One Millionth Lounge Post, nor that it was complete drivel. Dalek Dave
CCC Link[ ^]
Trolls[ ^]
|
|
|
|
|
My current workstation is starting to sing it’s slow song of death, at least I’m getting a forewarning.
Our company buys from Dell and I’ve narrowed the choices down to:
Optiplex 390 with i5 2400 with 4GB RAM and a 512MB AMD Radeon HD 6350
Or
Optiplex 780 Core 2 Duo E5800 4GB RAM and a 256MB NVIDIA GeForce 9300
Anybody have any experiences with either system or recommendations?
Is there another system from Dell I could be looking at?
It was broke, so I fixed it.
|
|
|
|
|
The former. The two generations newer i5-2400 stomps the 2 generations older E5800 silly. Anandtech bench comparison[^] The 8600 is the closest fit I could find and somewhat faster than the 5800 on account of having 3x the cache, and an extra 133mhz of clock speed.
I'm not sure what the 6350's specs are to make a comparison; but it's 2 major (4 total) generations newer than the 9300 and both have model numbers indicating that they're relatively low end parts.
3x12=36
2x12=24
1x12=12
0x12=18
|
|
|
|
|
|
i5 is definitely better, not sure why you're even comparing those.... but if at all possible, core i7 is a GREAT processor!
Also, what OS are you operating? If you're intending on going to Win7, be aware its a RAM hog.
|
|
|
|
|
Thanks Albert, it's my workstaion at work that's dying so I have to spec one that's reasonably priced. I saw the 390 with the i5 and it was essentially the same price as the 780 so wondered if there was something special about the Duo2.
Thanks for the info, I appreciate it. My 5
It was broke, so I fixed it.
|
|
|
|
|
When you say it's a RAM hog do you mean that it uses a lot of RAM for core functionality, or that it uses as much RAM as you can give it to work faster instead of just having spare RAM sitting there unused? (ie SuperFetch[^])
cheers,
Chris Maunder
The Code Project | Co-founder
Microsoft C++ MVP
|
|
|
|
|
That's a good question... I think the answer is somewhere in the middle. Win7 does work a lot faster than XP for a lot of things (the boot time is a lot better too), but it uses a lot of memory just sitting there (thanks in part to the flashier user interface).
Now with that said, I wouldn't put WinXP on anything new, but in comparison, Linux can definitely run smoother on more modest configurations (and you have the option of using a flashy interface like Unity/Gnome3 or running on something simple yet fast like KDE/Gnome2).
Oh... and the reason I had mentioned the RAM to the OP is because I wouldn't buy a new desktop meant for Win7 with 4GB, I'd get at least 6GB. My personal desktop has 9GB, runs smooth. My laptop has 4GB but I run Ubuntu on that. My work computers, well... they're a diverse bunch .
|
|
|
|
|
I have one computer that I can't modify in any way... Industrial thing that can't be changed.
Two displays can be plugged to the same VGA port...
My problem is that when two displays are there the graphics card has not enough power to show the image... and the external display looks nice, but the original one becomes blurry...
Is there any device that could allow me to remove that effect? plugging two displays to the same port without affecting the graphic card?
Thank you in advance...
|
|
|
|
|
|
Hello everyone! I am new to minifilter drivers and I have thought to design one that just appends a non-existent file to a specified folder, that is, let's say for the folder \test\, when I open this folder with explorer I would like to see its content plus a non-existent file (whatever it's name is).
I know that I should intercept IRP_MJ_DIRECTORY_CONTROL, and append the FILE_BOTH_DIR_INFORMATION structure corresponding to the file I want to "add", but I don't know how to do this . One easy solution would be to create a IRP response with the mentioned structure in the pre-operation and let the minifilter drivers bellow mine do the rest, but I am not sure this is possible nor how to do this. (Another possibility would be to append this information in the post-operation)
Any help is appreciated, thanks in advance!!!
Santi
|
|
|
|
|
Member 8027003 wrote: I know that I should intercept IRP_MJ_DIRECTORY_CONTROL, and append the
FILE_BOTH_DIR_INFORMATION structure corresponding to the file I want to "add",
but I don't know how to do this
I dont know much about minifilter drivers for file system stuff, but:
When you register as a minifilter driver your IRP handler gets the Irps before the go down the chain. (There are numerous ways of doing this, async, sync, with or without completeion routines. Oneys book is good for the details on this). You want a synch with completion routine handling of the Irp.
So you handle the IRP_MJ_DIRECTORY_CONTROL Irp in your Device control func you registered at DriverEntry.
Now it gets into the unknown for me. I would imagine you allocate some new memry which is big enough for the original Irp buffer (can be in various places depending how the IOCTL is formed: METHOD_BUFFERED METHOD_NEITHER etc) and the additional data.
You then copy the old Irp data over, add the new data at the end. And point the Irp at this new buffer.
You send it down.
In your completion routine you point the Irp to the original buffer. Delete the one you allocated, and complete the Irp so it goes back to the originator.
Give it a go. If it BSODs you know it isnt good,
==============================
Nothing to say.
|
|
|
|