Dell Workstation Owners Club

[lexluthermiester]

An update on success or failure of this approach will be forthcoming! Thanks to all!

Nice! Looking forward to that. Kinda curious about how it's setup. If you would feel ok with it, take photos and show us how you have it setup.

 

[DMFCAVU]

Nice! Looking forward to that. Kinda curious about how it's setup. If you would feel ok with it, take photos and show us how you have it setup.

Photos, no problem at all. I spent a good bit of time last night finishing up the wiring, getting the massive wire looms tucked away, running the temp sensors and ATX Passthrough. Since it was late, I decided against a "test" since knowing if it was not 100% successful, I would be up all night diagnosing the problem(s). So tonight it will get plugged in and the "button" pushed!

 

[DMFCAVU]

OK, a bit of progress. Not there yet, but so far, so good.

This was the first startup after piecing it back together, really in bench test mode. Did not want to pack everything up, hit the button only to discover there's a problem requiring disassembly.

It's running as these photos were taken. This was the second attempt to start, the first try resulted in no video, and a diagnostic code with lights AND sound that got my attention. Shut it down, started researching codes, only to learn there was no diagnostic code listed for 5, 2, 3/3 flashes ending with 3 flashes and beeps. Hmmm. Time to scratch the head a bit. Went through everything, making sure things were secure, plugged where they should be, made certain the monitors were switched on. Fired up again, then the Dell logo shows up and it goes right to pre-boot testing - and that's when I remembered removing the HDD carrier with the built-in mini fan had triggered the pre-boot test previously. Sure enough, not getting a response from the fan reports it failing. Next step is disable that fan in the BIOS. Also got the same beep/flash codes, so the issue the first attempt was the monitors were not on, and thus I did not see the pre-boot test running.

First two pics show the two fans added to pull more air from the front of the case, hopefully getting more heat off the GPUs. If I can avoid adding cooling to those, all the better. With the dual cards and each one taking up two slots, I'm not even certain they can be cooled. The fact they've cost me 2 PCIe slots is annoying enough.



Next two photos - the cooler rests on top of the 5820. It is surprisingly quiet with the three 120mm fans, The side shot really is reason to celebrate. It's running with the side panel off. Looks like we finally beat the intrusion switch, alarm, PITA. Not 100% out of the woods yet. Uncertain of the impact of multiple changes, the ATX pass-through was not connected, although everything else on the cooler is functioning, pump and fan status, temp probes functioning. The temp probes are all reporting 2 degrees Celsius lower than HWINFO64 (26 vs 28). Makes sense, the water block probes are stuck to the top and 2 sides. HWINFO 64 is picking up its temps internally. You may also notice the 18" cable plugged into the Power Distribution Board - That socket was previously unused and intended for a 2 CPU configuration. I wanted access to additional 12 volt power, so purchased a cable, and have already confirmed there's 12 volts in each of the PCIe connectors - the 6 and 6+2 pin. Another win. Once I get a cable made with PCIe 6 or 8 female on one end and SATA power15 pin female on the other, the rear fans and their Noctua controller will be installed. Another option I'm considering for those fans is using an independent, variable power source, that way I can run them up or silence them at will. Believe me, those fans are load and they will move a lot of air. 2000-10,000 RPM, and a 1 amp load. With two of them, this is why I did not want to use a MB header for power. Not sure if I'll tackle the rest tonight, may be tomorrow. Adios!



Last two shots are more of the side panel MIA. A good look at the NVIDIA Quadro RTX 5000's consuming a LOT of real estate. The PCIe card support is not installed since the next step is inserting the ATX Passthrough in the loop to test the theory. Stand by!

 

[lexluthermiester]

OK, a bit of progress. Not there yet, but so far, so good.

This was the first startup after piecing it back together, really in bench test mode. Did not want to pack everything up, hit the button only to discover there's a problem requiring disassembly.

It's running as these photos were taken. This was the second attempt to start, the first try resulted in no video, and a diagnostic code with lights AND sound that got my attention. Shut it down, started researching codes, only to learn there was no diagnostic code listed for 5, 2, 3/3 flashes ending with 3 flashes and beeps. Hmmm. Time to scratch the head a bit. Went through everything, making sure things were secure, plugged where they should be, made certain the monitors were switched on. Fired up again, then the Dell logo shows up and it goes right to pre-boot testing - and that's when I remembered removing the HDD carrier with the built-in mini fan had triggered the pre-boot test previously. Sure enough, not getting a response from the fan reports it failing. Next step is disable that fan in the BIOS. Also got the same beep/flash codes, so the issue the first attempt was the monitors were not on, and thus I did not see the pre-boot test running.

First two pics show the two fans added to pull more air from the front of the case, hopefully getting more heat off the GPUs. If I can avoid adding cooling to those, all the better. With the dual cards and each one taking up two slots, I'm not even certain they can be cooled. The fact they've cost me 2 PCIe slots is annoying enough.

View attachment 366227View attachment 366228

Next two photos - the cooler rests on top of the 5820. It is surprisingly quiet with the three 120mm fans, The side shot really is reason to celebrate. It's running with the side panel off. Looks like we finally beat the intrusion switch, alarm, PITA. Not 100% out of the woods yet. Uncertain of the impact of multiple changes, the ATX pass-through was not connected, although everything else on the cooler is functioning, pump and fan status, temp probes functioning. The temp probes are all reporting 2 degrees Celsius lower than HWINFO64 (26 vs 28). Makes sense, the water block probes are stuck to the top and 2 sides. HWINFO 64 is picking up its temps internally. You may also notice the 18" cable plugged into the Power Distribution Board - That socket was previously unused and intended for a 2 CPU configuration. I wanted access to additional 12 volt power, so purchased a cable, and have already confirmed there's 12 volts in each of the PCIe connectors - the 6 and 6+2 pin. Another win. Once I get a cable made with PCIe 6 or 8 female on one end and SATA power15 pin female on the other, the rear fans and their Noctua controller will be installed. Another option I'm considering for those fans is using an independent, variable power source, that way I can run them up or silence them at will. Believe me, those fans are load and they will move a lot of air. 2000-10,000 RPM, and a 1 amp load. With two of them, this is why I did not want to use a MB header for power. Not sure if I'll tackle the rest tonight, may be tomorrow. Adios!

View attachment 366229View attachment 366230View attachment 366231

Last two shots are more of the side panel MIA. A good look at the NVIDIA Quadro RTX 5000's consuming a LOT of real estate. The PCIe card support is not installed since the next step is inserting the ATX Passthrough in the loop to test the theory. Stand by!

View attachment 366232View attachment 366233

That's impressive! Nicely done. You got some cable management to do, but otherwise it looks very good so far. My one question: Purple fluid?

 

[DMFCAVU]

That's impressive! Nicely done. You got some cable management to do, but otherwise it looks very good so far. My one question: Purple fluid?

What can I say? A Prince fan??? One thing you be certain of, there will NOT be any freakin' lights on fans, the case, zero, zip, nada!

What can I say? A Prince fan??? One thing you be certain of, there will NOT be any freakin' lights on fans, the case, zero, zip, nada!

Yep, not nearly done with the cables. Still have the fan hub to mount and cables to route. Also need to put some edge protection around the wiring passthrough openings. Maybe a grommet around the hoses. Probably going to reroute the temp probe leads (the white sleeved wires) to exit where the ATX Passthrough is (the red sleeved wires). Fingers crossed this passthrough kill switch works!

 

[lexluthermiester]

What can I say? A Prince fan???

Fair enough. Was only curious.

 

[uco73]

OK, a bit of progress. Not there yet, but so far, so good.

What did you do to make your computer work with the case page open? On my computer, when I open the case page, the computer immediately shuts down.

 

[LoveLiveHonoka]

For anyone with the Dell T5810 with WINBOND 25Q128FVSG BIOS chip that needs Turbo Boost Unlock modified BIOS, I made one here. Flash with CH341a programmer. Disable C-States in BIOS after flash.
Flashing process is very easy, you can follow the YouTube tutorial uploaded by Miyconst. I did not need to remove the CPU, I only removed PCIe devices to make the BIOS chip more accessible, and had PSU plugged in. For people with Macronix International MX25L12839F BIOS chip, use Miyconst's BIOS.

A follow up to this, I made a different one which should be more stable using the payne driver. Has a -70mV core and -50mV cache undervolt. There is no need to disable C-States with this one. Flash with CH341a.
(Remember: this BIOS is for motherboards with the WINBOND 25Q128FVSG chip).

Dropbox

www.dropbox.com

SHA256 = 152b98611534e62b1a3079945065262cccadc0f7869fb09a66219a08a4dfe125

 

[DMFCAVU]

What did you do to make your computer work with the case page open? On my computer, when I open the case page, the computer immediately shuts down.

Yep, had that annoying problem too Not any more. Assuming you know the location and method of operation for the intrusion switch, it looks like it's a NC switch, so when the side panel is in place, there's a metal bracket on the side panel that depresses the switch when it's in place, "opening" the circuit. As long as the panel is in place, the circuit remains open and the system continues to run.. The action of removing the side panel closes the circuit, triggering the intrusion alarm and accompanying shutdown. My project for the cooling system had 2 objectives, first, get the cooler operating correctly and 2, silence/render the freakin' intrusion function inoperative. Success on both points.

Killing the intrusion system required removal of the switch and wiring, disconnecting it from the header located on the "Front Panel" devices daughter board, under the ODD/Flex Drives on my 5820. That said, i did not test the functionality of the system with the switch and wiring removed, as I had plans to repurpose the Intrusion circuit. At this step of the process though, all I did was connect one end of a passthrough circuit to a connector and inserted it in the header. The circuit would be "Open" under normal operations, and would close only when another device closed its circuit.

If the intrusion circuit works as I understand it to function, removal/unplugging the connector alone (leaving the switch in place) should silence the intrusion function and the system will continue to run or start up with the case open.

Good luck.

Dennis

 

[DMFCAVU]

Fair enough. Was only curious.

LexLutherMiester and interested parties,

We have a winner, and congrats and thanks to SiggiTheHun for the smokin' hot tip. I was skeptical, but willing to try anything. Several folks said no, that won't work, including tech support at Koolance and Dell. Works exactly as it should, ran a real work test, set the threshold temps low, then intentionally ran the system under load. Triggered the alarm when the CPU hit the preset threshold, fan went to max speed, pump is always at 10/10 unless set lower. After a few times of the system running the fans at max and causing the temps to drop I had to get creative - decided to shut the cooling system down and wait for temps to climb into the 40's. Worked like a charm. Switched system on, as soon as it reported the higher temp, alarm went off, fans and pump switched to max and within seconds the system shut down. So a win-win and I owe Siggi a beer! Thanks for the help.

Dennis

 

[SiggiTheHun]

Good on you! Dennis - and thanks for the kind words.
Saw your avatar (TF Harley?) and just thought of a Tether switch.

..tho mulling over the impact of the forced stop to the ongoing computing process..

 

[DMFCAVU]

Good on you! Dennis - and thanks for the kind words.
Saw your avatar (TF Harley?) and just thought of a Tether switch.

..tho mulling over the impact of the forced stop to the ongoing computing process..

Siggi, good eye on the Top Fuel Harley. Yep that’s me enjoying the best 6 seconds of life. That shot was the US Nationals, 2018, around the 1000’ mark, 215+, still pulling hard. Notice the mist behind the bike? Nitro leaking from a loose fuel line. Cost me fuel pressure = lost horsepower = only a 6.41 pass, and a ticket to eliminations at the Big Go. If anyone’s interested, I may post a shot of John Force dropping in our pit to give the bike a look - and proceeding to determine we were nuts for riding these bikes. If you didn’t know, we would suck 2.5 gallons of nitro in 6 seconds. In 2018/2019, nitro in drums was $40/gallon.

Dennis The Nitro Menace!

Good on you! Dennis - and thanks for the kind words.
Saw your avatar (TF Harley?) and just thought of a Tether switch.

..tho mulling over the impact of the forced stop to the ongoing computing process..

Siggi - valid point regarding the computing process. For me, I would rather restart processing a map than torch an expensive CPU, and I was seeing some crazy temps before the cooling system.

Again, thanks for the tip. You pushed the ball over the goal line.

 

[lexluthermiester]

LexLutherMiester and interested parties,

We have a winner, and congrats and thanks to SiggiTheHun for the smokin' hot tip. I was skeptical, but willing to try anything. Several folks said no, that won't work, including tech support at Koolance and Dell. Works exactly as it should, ran a real work test, set the threshold temps low, then intentionally ran the system under load. Triggered the alarm when the CPU hit the preset threshold, fan went to max speed, pump is always at 10/10 unless set lower. After a few times of the system running the fans at max and causing the temps to drop I had to get creative - decided to shut the cooling system down and wait for temps to climb into the 40's. Worked like a charm. Switched system on, as soon as it reported the higher temp, alarm went off, fans and pump switched to max and within seconds the system shut down. So a win-win and I owe Siggi a beer! Thanks for the help.

Dennis

Thinking I missed some context, but hey, your happy with the results and that's what counts at the end of the day! Glad we could help and it's working well for you!

Very impressive rig by the way! The first time I saw a photo of one of those units it intrigued me.

 

[uco73]

Good luck

Thx man alot.

 

[SiggiTheHun]

What did you do to make your computer work with the case page open? On my computer, when I open the case page, the computer immediately shuts down.

Either unlink/cut the red cable - 3rd (from top) pin on the I/O-boards top connctor - but keep the 1st and 2nd pin connected with each other (black cables).

Or remove the latch from the sidepanel and clamp some piece of plastic or so of adequate size to push down the switch in the switch´s latch opening instead.

Siggi, good eye on the Top Fuel Harley. Yep that’s me enjoying the best 6 seconds of life. That shot was the US Nationals, 2018, around the 1000’ mark, 215+, still pulling hard. Notice the mist behind the bike? Nitro leaking from a loose fuel line. Cost me fuel pressure = lost horsepower = only a 6.41 pass, and a ticket to eliminations at the Big Go. If anyone’s interested, I may post a shot of John Force dropping in our pit to give the bike a look - and proceeding to determine we were nuts for riding these bikes. If you didn’t know, we would suck 2.5 gallons of nitro in 6 seconds. In 2018/2019, nitro in drums was $40/gallon.

Dennis The Nitro Menace!

Way to go!
My sincere respect, appreciation and admiration!

You, Sir, definitely merit that recent Elvis´ witness´ prosaic verdict!

 

[latulippen901]

Recently bought my wife a t5810, unfortunately the ram config is probably holding it back some, it has 40gb of ram, a 32gb stick and 2x4gb sticks. I did get a xeon e5 1660v3 for it but will getting a quad channel set of 8gb ram be worth it?

 

[lexluthermiester]

the ram config is probably holding it back some, it has 40gb of ram, a 32gb stick and 2x4gb sticks.

Yup, that'll do it. That system should have 8 DIMM slots. I would suggest finding matching sticks of the 4GB sticks, 2 more would give a total of 16GB, or you could find 3 more of the 32GB sticks for a total of 128GB. The 2x4GB would be the most cost effective. Keep the 32GB and the 2 4GB sticks as spares. Lots of possibilities. If you're going to match up, make sure you get the same or close to the same model numbers of the sticks you have. Mismatched RAM can cause issues.

I did get a xeon e5 1660v3

Nice. Good CPU.

but will getting a quad channel set of 8gb ram be worth it?

Yes, that would be a solid config!

 

[sk111]

The 1660v3 is great. I recently upgraded my Dell t5810 with it. But to get most of it you need to overclock it with Throttlestop.

 

[Astebreed]

Hello guys! First post of mine here.
Got a T5810 with ram and 1630v3 recently, purchased 2667v4 from taobao for like $25, planning to got an a770 and learnt about the resizable bar things. Found a confirmation of t5810 on the rebar modding project https://github.com/xCuri0/ReBarUEFI
Luckily I purchased the external EEPROM flasher before, so will give it a try then.
Good luck for myself then

 

[sk111]

Hello guys! First post of mine here.
Got a T5810 with ram and 1630v3 recently, purchased 2667v4 from taobao for like $25, planning to got an a770 and learnt about the resizable bar things. Found a confirmation of t5810 on the rebar modding project https://github.com/xCuri0/ReBarUEFI
Luckily I purchased the external EEPROM flasher before, so will give it a try then.
Good luck for myself then

Yes I have done it but I found for my gpu rx 6650xt was stuttering so I removed it. Also heard the same from users of other platforms and gpus the same thing. It is a hit and miss. May not worth the effort.

 

[Astebreed]

Yes I have done it but I found for my gpu rx 6650xt was stuttering so I removed it. Also heard the same from users of other platforms and gpus the same thing. It is a hit and miss. May not worth the effort.

Thanks for such valuable information.
I think it still worth a try since ARC will lose pretty much performance without it. Afterall I'm feeling more confident since there is a successful precedent. It should at least work as usual if I failed since I have the flasher right?

 

[sk111]

Thanks for such valuable information.
I think it still worth a try since ARC will lose pretty much performance without it. Afterall I'm feeling more confident since there is a successful precedent. It should at least work as usual if I failed since I have the flasher right?

Yes if you get a backup of your bios and keep it safe you can always revert if it goes wrong. I have done it many times.

 

[Astebreed]

Yes if you get a backup of your bios and keep it safe you can always revert if it goes wrong. I have done it many times.

Just tried getting backup of it but reported "IC not respond" Choose to give up for now

 

[Astebreed]

Yes if you get a backup of your bios and keep it safe you can always revert if it goes wrong. I have done it many times.

Hi sir, just successfully modded since I don't know the PSU should be connected . Also, my laptop cannot afford such power current (it's obviously that the programmer can be recongnized without clipping the BIOS chip, also the power LED on programmer will be not as same bright as it should be) so that I borrowed another from my roommate. Downloading 3d mark to see how the performance will change now

 

[sk111]

Hi sir, just successfully modded since I don't know the PSU should be connected . Also, my laptop cannot afford such power current (it's obviously that the programmer can be recongnized without clipping the BIOS chip, also the power LED on programmer will be not as same bright as it should be) so that I borrowed another from my roommate. Downloading 3d mark to see how the performance will change now

Well for me you dont need the psu connected in both of my T5810s. Just need to plug the external programmer to a usb3 port as it has more voltage and you are good. Seems though you got it working .
One of the standards to check is cpu-z and the cinebench... ops thought you are checking processor. Yes 3dmark is good.