I've bought a new nVidia Quadro NVS-290 (PCI-E x1) video card.
It can successfully fit in any of the x8 slots.
I've tried it in the n°5 as well as in n°1 slots.
The card is well-seen by Windows 2008 R2 (driver installed and no error in the device manager).
But when connecting a monitor on any of the VGA cable (there are 2 on the NVS 290), either before boot or after, I don't get a signal and the monitor always turns into powersaving mode...
I've seen an option in the Unified Server Configurator where I can disable the on-board Matrox W200 video card.
Is it the solution ?
But I'm a bit afraid to do that, because if the nVidia still doesn't work, I won't have the onboard-card any more either (so, how to re-enable blindly the default card ?) !
What's your opinion ?
Thanks in advance.
Solved! Go to Solution.
This graphics board I used was a Zotac GT218- ION graphics card (512MB DDR3 memory, PhysX by NVIDIA, Open GL 3.2, Dual-Link DVI, 64-bit ) PCIe x1. You don't have to hacksaw off anything because the PCIe x1 physically doesn't fit into the motherboard, it doesn't even reach the motherboard. You need a riser/jumper (PCIe x1) to jumper the 2 connectors. After installing, follow the instructions from skylarking.
I really like the enhanced graphics now.
theflash, 'hack job' is dismissive of the research i did to perform this 'quality modification' :emotion-5:
If a company wants to stay successful in business, they must both support and react to their customers needs when developing new products. I guess we will see what if anything Dell has learned about simple issues like modern integrated graphics capabilities, physical x16 PCIe slots on the m/b along with bios options to disable on board graphics and allow insertion of x16 PCIe graphics cards across their entire new range of PE servers when they are finally released.
But Dell isn't alone in being passive to their customers needs as i also contacted a few graphics card manufacturers and could not get past the script techs to ask simple questions of the hardware designers. To have these companies recognize that a low power 8x PCIe graphics card is rather simple to produce and may hit a niche market adding something to their bottom line is like relieving oneself in the wind.
Now based on the fact that better graphics on servers is an often asked question on this forum, at least in the short time i have been here, there does seem to be a niche market. And having discussed this issue with Dell customer service and technical support i hope they take it on board, though i have my reservations.
But i do indeed expect warranty and support with any systems or product i have purchased.
For the T610, Dell provided a bios disable feature for integrated graphics, so it's hardly 'unrealistic' to expect them to support their customer and take the small extra step to provide a part number to a x1 PCIe graphics card that should work with this server. But they would not do that, nor could they state what consequences or lost feature would occur should i install a PCIe graphics card (no crash screen capture via idrac, etc). I was left out in the cold and had to resolve the issue on my own. This is very poor support indeed.
And judging by my other experience with Dell support, your correct to say it is 'ridiculous' to expect Dell could cope with providing some real technical information for what is a very simple modification to a x16 PCIe graphics card, let alone the even simpler requests i have made.
After all, i had such issues trying to get plastic hdd blanking plates for the front of my T610 because i had to removed 4 of my 15K 300GB SAS drives which i wont/cant have installed as they make the system fans spin like crazy. Seems Dell just can't understand that the sound of a jet is 'ridiculous' but they will not fix it, as 'its a server'. Never mind that many use these pedestal servers in offices. As such, the end result is considerable funds tied up with 4 unused SAS drives sitting in my cupboard because of Dells inability to react to their customers needs and release new firmware fixing this noise/cooling issue. Yes ridiculous indeed.
And FYI, modifying an expensive motherboard by cutting the PCI slot means i take full responsibility and loose warranty for the motherboard. Whereas modifying an old unused and unloved graphics card was a no brainer and Dell is still responsible for T610 warranty and system support.
Along with the design of their new PE servers, Dell also needs some further work on its support organization.
Yablargo, you'll have a tougher time as you can't disable your on board graphics in BIOS. Have you tried an ATI based PCIe card?
I'm not saying it is unrealistic or ridiculous for Dell to support after-market or add-in graphics (indeed I wish they would) ... what I said was unrealistic and ridiculous was that Dell not only support but provide documentation and instruction on modifying the system or add-in hardware as you have described. It's like expecting Ford to provide documentation and support for adding a NOS booster to a Mustang - it's not their responsibility.
I'm not saying they don't need to work on a great many things, but let's keep perspective. How many legitimate uses for a server require higher-end graphics that what come onboard? I'm not one to define "legitimate", but most servers are set up and run quietly [no pun intended, I agree they SHOULD run quietly) in the background, being access very infrequently - especially directly.
This isn't to say that they can't be "repurposed" to do something other than typical server functions, but as that was not their design, using it for something other than what it was designed for is not Dell's responsibility to make it work, but the user's. By calling it a hack job, I didn't intend to belittle what real research and effort went into it, but let's call it what it is - a hack job (I have a Barracuda firewall box running Windows 7 as a PC for my kids - it's a hack job). There's nothing wrong with it - in fact, I admire this sort of thing - but we can't expect the manufacturers of these devices to support these types of things by providing documentation on doing so.
We do agree on much and thanks for the smile your unintended pun gave me :emotion-1:
I do like your vehicle analogy but what i can add to it is that vehicle manufacturers do provide a wealth of detailed technical information to aid their customers with some aspects of their modifications. Obviously they don't warranty any customer made mods but it's not uncommon for them to recognize a market niche and take such mods on board releasing new products incorporating them (though vehicle manufacturers work at a slower than glacial pace).
My gripe is that Dell and others just shuts the door on any request for technical information and leave you on your own and out in the cold.
One must remember that PCs are in theory an open architecture but manufacturers continue to try to put that genie back in the bottle. Standards which cost $$ and an NDA agreement just to view them, modification of standards to make system components incompatible between vendors, signed boot loaders under the guise of added security, DMCA laws that make repurposing legally questionable. It's getting worse and the march continues with the end game being that even simple mods or repurposing will be impossible. We'll all be forced to hand over more money and buy another system or be made criminals for hacking.
In the area of your speciality, i am amazed that vendor specific protocols seem to exist which results in vendor specific firmware for raid cards and hdd forcing people to compromise support if able to use non vendor hdd. Such RAID specific protocols should have been standardized so that any hdd can be purchased from any vendor and plugged into another vendors raid card in your system and work with all error codes and conditions being correctly handled by the standard ptotocol. In Dells case, it was only after a large backlash from customers that they released firmware to allow non certified drives on our dell raid controllers but no standardization has taken place. Standardization is not rocket science but part standardization is a business strategy that denies all of us the best feature sets with the only aim being to lock us in. It's abhorrent and something which should be resisted in the most strongest way.
Back to graphics, the number of legitimate users of a server requiring better graphics, is not relevant. As a legitimate owner of a server, Dell provided me with no help for even a part number for a PCIe x1 graphics card, or an extract from a PCIe spec or any other technical info so your correct to say that Dell needs to work on a great many things. But to expect a technical doc that specifies how to mod a card, even a simple mod, is unrealistic and ridiculous :emotion-7:
About the RAID ... there is standardization of settings but not of configurations. Standardization exists in the available settings on the disks and controllers, but each of the dozens of settings are configurable ... things such as timeout, retries, error recovery, timings, cache settings, etc. that are all configured at the factory based on what it is going to be used for. Consumer drives, for example, generally have no redundancy, so their timings and error recovery settings must be more lenient ... they must be much tighter when used in a high performance RAID controller. Each OEM (not just Dell) tweaks these settings (just like the auto analogy car makers all have similar engines, they just tweak the injection, orientation, capacity to what they feel is a superior configuration) to be a better performing system (cutting edge settings) or more reliable system (safer settings) ... and one could argue these points either way. With these tweaks to the controller though, drives must be tuned to respond/interact with the controller in the way that the controller expects for it to work properly. Retail controllers are usually configured with "average", "generic", unconfigured, and/or versatile settings, allowing most retail drives (with similar generic/unconfigured drive settings) to work ok with them. I'd be kidding myself to think that OEM's are only thinking of the customer when configuring their controllers and drives - one only needs look at the prices that OEM's charge for their drives to see that they are taking advantage of the situation, even if the situation itself is valid ... but there are real differences in the settings and using the wrong drives can have a negative affect on performance/reliability. The price the OEM charges for their drives (often 20-40% higher than the same drive from reputable resellers) is what really irks me about this.
"My gripe is that Dell and others just shuts the door on any request for technical information and leave you on your own and out in the cold."
This goes for Dell employees too ... support staff (including levels 2 and 3) do not have access to the types of documentation and specs we sometimes look for/need. Not sure what the information block is coming from engineering.
Having dealt with this issue on a T410 server, I would like to point out the following; the T410 and T610 is not designed to use add on video cards, and they the systems are not designed to disapate more than 15 watts of heat for non-dell cards (25 watts for Dell storage cards). It is all in the manuals.
I am aware of the heat limitation Dell has placed on the PCIe slots which they have (poorly) documented in the Hardware Owners Manual and (more clearly) in the Technical Guide.
In the case of the T610, the machine has 5 PCIe slots, 2 of which are x8 and 3 of which are x4 slots. Table 3-1 of the HOM states the priority order of expansion cards, how many can be installed and what power they can consume. The table footnote states that a maximum of 2 cards whose power exceeds 15W can be installed. Above the table itself there is the statement that no more than 2 of the five slots can can have a power consumption greater than 15W (25W max). The HOM needs to be corrected as its poorly written and easily misinterpreted. The TG states in section 11 that the system supports 25W max from the first two slots but only 15W max from the remaining 3 slots due to thermal limitations not power limitations.
As for the PCIe specification, this defines 2 classes of slots, normal and graphics (though can't remember the correct terms) which can feed 25W max and 75W max respectively. So any limitation Dell has placed on the system is an issue of heat dissipation and not one of power consumption per sei (just as stated in the TG).
So, it then seems that we can have a total of 2x25W + 3x15W = 95W of heat produced before the machine will have any issues with non force cooled server cards (as they all are). If you put fan cooled non server cards in the system, then you will also be able to get 25W for each slot => 125W in total (but it's for you to ensure the heat will be effectively removed from the case, Dell doesn't guarantee it)
Why any power differential would be documented at 25W for Dell Storage and 15W for non Dell storage cards is also a mystery. Has Dell coded something in the BIOS to differentiate between Dell and non Dell cards and takes action accordingly - I think (hope) not. It's just lame documentation in the HOM that has the (un?)intended consequence of pushing purchase of their 'certified' storage controller cards and HDD.
Finally, the PCIe spec defines the initial power up process for cards. If you have a graphics card with no PEG connectors inserted, this initial power draw limit is 25W from the slot. Then once system reset is released, the system provides the slot_power_limit to the card and if higher than the initial power draw, the card can then draw to this new power limit (normally 75W for a graphics slot). In the case of non graphics slots, the slot_power_limit is always going to be 25W. So as long as you use low power cards, you will be able to very safely have your directx11 and enjoy it.
Where the PCIe spec is not clear is how a graphics card will apportion operational power draw between the PCIe slots and PEG connectors should the slot_power_limit remain at 25W (rather than the 75W of a normal graphics slot). Since there is no guidance here, different high power cards may not behave as expected and have issues under heavy use when they can't make up for the shortage (75W required - 25W available = 50W shortage) from the PEG connectors.And the graphics card manufacturers will not divulge any information.
So if you want better graphics and Dell wont help, take a hacksaw and modify a low power graphics card and feel assured that all should be OK afteer doing your homework.
If you had specific issues with your T410, then do tell.
Very interesting, but besides the point. The technical specifications clearly says you can't do it. Why, I have no idea ! Clearly, if you have to take a hacksaw to to a plugin card to make it fix, the card was not designed for the slot.
ALoubert, the original poster had an issue with getting his nVidia Quadro NVS-290 to work so we helped to resolve his problem. Foveator also had an issue and he seemed delighted with the resultant video on his 20" monitor after adding a graphics card. Others weren't so lucky and had more difficulty since Dell, by choice, did not provide a bios which allows the owner to disable on board video. In these instances, some got their video to work, some not.
These above issues had nothing to do with mechanical fit.
Like others, i could have used a x1 PCIe graphics card in the T610 in which case it would fit in any of the x8 or x4 slots. But i found reuse of an existing x16 card i had lying around to be interesting project and at $0 cost. As some may also be interested in how to modify x16 cards, i felt sharing would be in good spirit. Many would not contemplate such a mod due to their lack of understanding of the technology or the loss of (card) warranty that results. If modding is not your thing, that's fine - don't do it, use a x1 PCIe graphics card.
But the issue with heat and power (which i have mentioned it in other posts) is a very minor one and easily solved. People just need to be aware of the technical and warranty issues before they start such a venture. Trying to discourage them from using discreet graphics cards because of a misleading statement in Dell documents adds nothing to the discussion. As for Dell documents being clear, maybe you should read the appropriate parts of the HOM and TG again. Hardly the be all and end all of clarity.
And so you understand the heat dissipation limits, server cards do not have fans on them for reliability reasons. These cards rely on the system fans for their cooling airflow. In servers, these system fans can be optioned as redundant fans. Obviously Dell has decided that for the T610, slots 3, 4 & 5 can't get as much air flow as is need for the PCIe spec slot limit of 25W, so they downgraded the slot power limit to 15W per slot.
The issue about Dell and non Dell cards is just rubbish and should be given the contempt it deserves.
If everyone listened to what Dell recommends, we would all be forced to use Dell certified storage cards and certified drives and pay a huge premium for it - or move to another vendor that didn't have such lock-in. Luckily many many owners were disgusted at such limitations and to Dells credit they listened and released new firmware for some Dell cards allowing non certified drives to be used. Unfortunately they didn't fix the issue for all Dell storage cards.
I don't know about you, but i despise forced lock in and such designed limitations.
The real problem is Dells designed limitation (x8 slot and missing bios options) and seemingly indifferent attitude to their customers. If they really cared, the would have resolved the issue by releasing updated system bios that allowed all these old and crippled servers to be repurposed via an expansion graphics card update (and sold some x1 or newly designed x8 PCIe cards along the way - a lost goodwill opportunity).
If people want to blindly follow what's written, that's their choice but i would hope people want to see for themselves. And it's information that can help in the latter.