• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Unsolvable Dell XPS 430 XFX Radeon HD 6670 problem

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

misternumberone

Member
Joined
Jan 17, 2012
Location
Texas
(I put this in the motherboard section because it concerns a number of parts and I wasn't sure what the most major one would be)
I am absolutely stumped by this. I'm not going to list what I have tried because a list of what I have not tried would be more appropriate, and I don't know what I haven't tried!
I had a system as follows:
Dell Dimension 9100 aka XPS 400(Intel 945P Express chipset on MoBo)
Intel Pentium D 960(3.6 Ghz, 4MB L2 cache, dual core)
4GB DDR2
DVDRW drive paired with CDRW drive in IDE as master/slave
Dell Card Reader(Internal USB 2)
SB live! 24-bit 5.1 sound card(PCI)
2-port USB 3 card(PCIe x1)
5-port USB 2 card(PCI)
XFX AMD Radeon HD 6670 1GB GDDR5(PCI-e x16)
TP-LINK 300MBps Wireless-N card(PCI)
Green 500 Watt PSU(EXTREMELY BUDGET PSU- GREEN IS ACTUALLY THE BRAND)
A 120mm case fan and a 140mm CPU fan and a 3-pipe heatsink
Intel 320 series 80GB SSD(SATA II)
Seagate Barracuda 7200 RPM 160GB HDD(SATA II)
Windows 7 Ultimate 64-bit Service Pack 1

this system always worked fine for me in every way, except for one thing;
the HD 6670 always ran at PCIe x1, instead of x16, detracting from its performance. This never bothered me too much, as it worked okay as it was, but I was confused as to why it did that. After I was unable to fix it or gain an idea of why it did that, I just let it be and ignored it. Also, there are 3 ever-so-slightly bulgier than normal capacitors in the corner of my 9100/400 board. I know from experience how troubling worn-out caps can be, but these are not bulgy enough to seem remotely problematic and I have never had any trouble from them.

Recently, I decided to upgrade to a better PC. I bought the following in good condition and either new or used and certified working:

Intel Core 2 Quad Extreme QX9650(3 Ghz, 12MB L2 Cache, Quad Core)
Dell XPS 430 Motherboard(with Intel x48 chipset on)
Dell XPS 430 Case (with front panel/power button and HDD activity indicator light and fans very similar to those in the 9100/400)
DVD RW Drive (SATA II)
Dell XPS 430/420 Card Reader(internal USB 2)
4 sticks of 2 GB each DDR3-1333 RAM (8GB total)
XFX Pro 550 Watt Full Wired PSU(definitely NOT a cheapo PSU)

and I took the following from my old rig to put in it:
the USB 3 card
the USB 2 card
the wireless card
and the Radeon HD 6670.
the Heatsink



If it had worked, then you wouldn't be reading this.



I did numerous tests and- I'm sorry, did I say numerous? I meant ABOUT ONE HUNDRED AND SIXTY tests, and I discovered that no matter what I put where and what extra hardware that I have in my extensive collection of parts I used and what other PCs I tested things in, everything works great in every combination I can possibly imagine except for one: If I put the HD 6670 in the XPS 430 MoBo.


What happens is this: After I plug the 6670 in and power the board, it turns on by itself, with the fans on, and the CPU fan on the ridiculously loud highest setting(if I have the fans plugged in)and has the MoBo) and power button lights (2 blue) on and turns on the HDD activity light on for a fraction of a second(if I have any H1Ds or the lights plugged in). It does not do anything else. One second later, it turns itself off and back on twice. It then does not give any feedback, make any noise from the speaker, or display anything on the screen, caps lock num lock and scroll lock do not do anything on a keyboard if plugged in(nor does any other keyboard input). It just sits there with the lights on and the CPU fan on high. There is a fan on the HD 6670; that comes on, but only the CPU fan goes on high. The heatsink on CPU is definitely good enough; the original XPS 430 would use the same one. It must be turned off by cutting power or connecting the two wires the power button connects for a few seconds. It can then be turned back on by pressing the power button, with the same results except that it does not turn itself off and back on twice. It only does that when it is first supplied power. It only gets to this if it has ram and a CPU in it and both the 4pin cpu and 24pin board connectors are in. Otherwise it gives normal errors for no ram and/or no CPU. I even tried it with the Pentium Ds I have, which use the same lga775 socket and have a FSB speed compatiable with the x48 chipset (800Mhz), though I did not expect them to work because they are not supposed to work on that chipset or BIOS. (speaking of BIOS, there is only one version of the BIOS from dell for that M/B, so no BIOS update) It acted like they were not there. As a warning: DO NOT EVER PUT A CPU IN A BOARD UNLESS THE FRONT SIDE BUS SPEED OF THE CPU IS SUPPORTED BY THE CHIPSET, ESPECIALLY IF THE FSB OF THE CPU EXCEEDS THAT SUPPORTED BY THE CHIPSET!!! I am glad that I didn't have to learn the hard way on that one.

Anyway, I can't possibly figure out in the remotest what could be the problem. As I said before, I have tried every combination and tested everything I can in other PCs(and that one itself without the HD 6670, but a different GPU, in) and it's just that board with that GPU. I just don't understand why they don't get along! If you have any ideas, please tell me! If you want pics, tell me and I'll post!

I think that it might be possible that, through some crazy Dell OEM madness, the board won't power more than a certain amount through a PCIe slot, no matter what the limit of the PSU is? Because the original Dell XPS 430 graphics card took a PSU connector, and so did not get all of its power through the PCIe slot. My Radeon HD 6670 gets all of its power through the PCIe slot. Do graphics cards that have a PSU connector get all the power that a PCIe slot is SUPPOSED to deliver, and then some with the PSU connector, or do they get the bulk of their power through the PSU connector and just use the PCIe slot mainly for communication with the board?

EDIT: The USB 3.0 card takes an optional floppy-drive-style power supply connector. Not that it matters.
EDIT: What happens is when the board does this this is exactly how it says it doesn't recognize a CPU.
EDIT: The board does not work with two graphics cards in it, even if it "likes" both of them. However it does work with the single graphics card in the 8x slot as long as it "likes" it.
EDIT: I was going to try to compare the board to the extremely similar XPS 420 and Precision T3400, but when I got a T3400 board this is what happened: http://www.overclockers.com/forums/showthread.php?t=711274

(I copied this from my post on another forum because I didn't want to recompose and type the whole thing again)
 
Last edited:
:welcome:

I think that it might be possible that, through some crazy Dell OEM madness, the board won't power more than a certain amount through a PCIe slot, no matter what the limit of the PSU is? Because the original Dell XPS 430 graphics card took a PSU connector, and so did not get all of its power through the PCIe slot. My Radeon HD 6670 gets all of its power through the PCIe slot. Do graphics cards that have a PSU connector get all the power that a PCIe slot is SUPPOSED to deliver, and then some with the PSU connector, or do they get the bulk of their power through the PSU connector and just use the PCIe slot mainly for communication with the board?
Something like this is my guess as well. Any board from Dell or any other OEM has a very high probability of sucking. It's a bummer, but I would find a new motherboard from someone like Asus or Gigabyte.
 
This post is now extremely outdated, and the PC is now VERY fixed and VERY different, But I still have the problem with the mobo not "liking" certain seemingly random Graphics Cards.
 
The PCI-E thing is likely Catalyst's fault.

Reminds me of back in 2003. Reminds me of a Catalyst control bug where after installing Catalyst for my Radeon 9000 Pro on my Asus A7N266-VM, it would always default to AGP 1X!

But it worked fine if I manually moved the AGP slider in the Catalyst control panel to 4X and rebooted.
 
I once had a system that worked great as long as you didn't actually install it into its case. If you did that it would constantly power on and then off. The issue was that for some reason that case with that motherboard in that setup constantly shorted out. No matter how many times I pulled it out and put it back in it would short out. I simply purchased another case to fix my problem.

I don't know how much more time you want to put into trying to figure this mystery out but to me it sounds like a short somewhere if it only happens when you put a video card in. I'd say try pulling everything out of the case and booting it that way, if it works then you have at least figured out the problem. If it doesn't then at least you know the case isn't the issue.

Good luck. Thats a nice processor even now days. :)
 
Yeah this post isnt very relevant; the OP is very old before I did all my testing. But, It's DEFINITELY NOT a short lol I have, after all these months, managed to try every combination of everything I have everywhere under every condition I have with a "Disliked" card in one of my two Dell XPS 430 mobos with same exact results every time. I use a Radeon HD 4890 OC edition and I wastrying for crossfire when I realized that it doesn't like having 2 GPUs in it. Even if the GPUs are two that it normally likes when they are by themselves.
 
Yeah this post isnt very relevant; the OP is very old before I did all my testing. But, It's DEFINITELY NOT a short lol I have, after all these months, managed to try every combination of everything I have everywhere under every condition I have with a "Disliked" card in one of my two Dell XPS 430 mobos with same exact results every time. I use a Radeon HD 4890 OC edition and I wastrying for crossfire when I realized that it doesn't like having 2 GPUs in it. Even if the GPUs are two that it normally likes when they are by themselves.

Pay attention Mate!!

I had the same bloody problem with my 4870x2...

If you have a look on that blue plastic support, you can see that there's a litte space for the side of the pcb of the gc... well on some of those, that pulls the pcb on the pci x16 support and it miss the contact with the pins inside... try to fit the card, without this blue support and keep a finger on the pcb, in order to bend it a little bit up or down, and make some test... I think it will work...
 
Back