Intel 2600CP SDR File Hacking

[theslim1]

I'd like to centralize the discussion going on over at the "Intel Xeon E5-2670 Deal Thread" surrounding the Intel 2600CP Motherboard, and the .SDR file that controls the various PWM Fan headers on the board in relation to temperature senors and the chassis that the motherboard is housed in. Customizing this file becomes very important when housing this motherboard in a non-intel chassis and having maximum control over the fans.

I've been researching what is needed to change the PROC1_FAN and PROC2_FAN curves when housed in a non-intel chassis, using the "Slow Ramp Setting". I've identified the domain these fans are in (Domain 0), however I'm not able to figure out where the curves are defined for the processor fans.

Controlling the other system fans is easy, for instance the curves for the "Other Chassis - Slow Ramp Setting are below on lines 10585-10591,

   // Global Stepwise Curve Record
   5B             // Stepwise Curve ID
   02             // Domain max and Count [7]-Domain Max (0=no) [6:0]-Count
   1E             // 30C
   19             // 25%
   78             // 120C
   64             // 100%

Has anyone had any success in modifying this file with custom curves for the CPU Fans? Some other reference below might help;


S2600CP Thermal Margin sensors reading as not available.
Fan SDR Hacking for the Intel S1200V3RP Motherboard [updated] - Learning Bits
Intel Xeon E5-2670 Deal and Price Tracking

 

[FunkyRider83]

I have modified the SDR file for my custom build.

Here's my build post: Dual E5-2670 build in the smallest case NZXT S340

Here's some background information: The SDR file is quite complicated in a sense that there are a handful of sections you need to edit to be able to achieve controlling the FAN speed based on a different thermal sensor and curve profile. If you use the custom xxxx_RAMP profile, then all fans rely on the single thermal sensor called BB_EDGE to vary their PWM values. This sensor reacts quite slowly so in my case, having noctua fans running at 600RPM and sloooowly ramping up is not enough. I needed the CPU fans to react quickly when the CPU is loaded. After some modification, I have linked the P2_VR sensor to both CPU fans, it approximately matches the CPU temperature and reacts pretty quickly.

I have uploaded my modified SDR file for you to compare. It sets FAN speed based on my all noctua configuration.

CASE FAN 1: NF-P14S Redux PWM-1200 RPM
CASE FAN 3: NF-P14S Redux PWM-1200 RPM
CPU FAN1: NF-F12 PWM-1500 RPM
CPU FAN2: NF-F12 PWM-1500 RPM
REAR FAN: NF-P12 PWM-1300 RPM

 

[dragonme]

I have a similar issue but with a S5520HC motherboard in a Rackable enclosure.

Rackable installed 3 x 80mm fans on the back wall connected directly to power supply though a power brake out board. They are 3 pin. Therefore they spin at 100% in a pull configuration and connecting them to the motherboard chassis fan headers with a custom enclosure selected makes them spin at 100% as well, by intel design.

The Rackable S3012 chassis, which can be seen on eBay is a half depth, 3U design. The motherboard is put in backward.. i.e. the nics and pci-e are up front, and the dual l5640s have passive heat sinks. It really cools like a 1U since the 12 drive backplane box sits on top of the board and covers about half the motherboard. so active cooling of the cpus would be difficult.

since the fans are 3 pin, would the motherboard chassis fan headers even be able to regulate the fans or would I need to find 4 pin fans?

I thought of:

1: connecting 2 of the fans to the CPU headers and attempting to have CPU temp drive the fan speeds and customize the 3rd fan in SDR

2: connect all fans to the chassis headers and customize the fan speeds in SDR by changing the sensors used for regulation giving them the CPU temp sensor number or something like that

Biggest problem I am having now is trying to read the SDR file and know for sure what it is I am editing

Additionally, since the drive box sits on top of the pci-e lanes, it has room for only 1 card on a 90 riser, so trying to figure out how to add some cards internally, but since its a half depth chassis with half that room taken up by the drive box, there is not much room to play with.

What are your thoughts guys?

This chassis has lots of power for my needs with the dual hex cores, pulls a bit more watts at idle than I would like ... about 150.. with 48gb ram.. but it really is too noisy with 3x80s going full steam!!

 

[FunkyRider83]

AFAIK Intel boards can only regulate fan speed using PWM so 3-PIN fans will always run at max speed.

 

[Russ]

I have S2600cp in P4308XXMFGN, I added a 92mm PWM fan at the rear end, powered by system fan 7 4-pin header , it spins but can not see the speed thru HWinfo or IASC, is it right?

 

[FunkyRider83]

I have S2600cp in P4308XXMFGN, I added a 92mm PWM fan at the rear end, powered by system fan 7 4-pin header , it spins but can not see the speed thru HWinfo or IASC, is it right?

Whenever you change fan configration, you have to re-run FRUSDR utility in EFI shell to detect / specify active fan headers. Inactive fan headers will not report RPM.

 

[Russ]

Whenever you change fan configration, you have to re-run FRUSDR utility in EFI shell to detect / specify active fan headers. Inactive fan headers will not report RPM.

Thanks for your response!
I did re-run FRUSDR many times after adding the rear fan, SDR part runs automatically due to detected Intel chassis, did not ask which fan is on, which is off. The fan spins so should be active but can not find RPM report. any clue? Do I need to add fan 7 manually in the SDR file?

 

[FunkyRider83]

Thanks for your response!
I did re-run FRUSDR many times after adding the rear fan, SDR part runs automatically due to detected Intel chassis, did not ask which fan is on, which is off. The fan spins so should be active but can not find RPM report. any clue? Do I need to add fan 7 manually in the SDR file?

In this case (no-pun-intended) you need to manually hack the sdr description script to include an additional fan header to your case model. Take a look at master.cfg file, which stores case-specific fan configurations, it takes a bit time and patience to learn which fan header corresponds to which section in the file.

 

[Russ]

In this case (no-pun-intended) you need to manually hack the sdr description script to include an additional fan header to your case model. Take a look at master.cfg file, which stores case-specific fan configurations, it takes a bit time and patience to learn which fan header corresponds to which section in the file.

Cool, I'll try it later today, only master.cfg is enough to add the fan in?

 

[Russ]

In this case (no-pun-intended) you need to manually hack the sdr description script to include an additional fan header to your case model. Take a look at master.cfg file, which stores case-specific fan configurations, it takes a bit time and patience to learn which fan header corresponds to which section in the file.

Looks like need to modify master.cfg as follows, then modify "OTHER_REAR_FAN" in S2600CP.sdr, am I right? since the built-in OTHER_REAR_FAN is for high spin fan

//Set SDR tags for items that are standard on specific SKU *******************************
         
            // Intel(R) Server Chassis P4000M Non Redundant Fan SKU ----------------------------------------------------
            IFSET "P4000M_BASE"
                //System Fans
                SET "FAN_120_1"
                SET "FAN_120_2"
                SET "OTHER_REAR_FAN"  // for rear case fan
             
                //Processor
                SET "PROC_1"
                PROBE "PROCESSOR" 2
                FOUND "PROC_2"         ""
             
                //Front Panel Temp
                SET "FP_TEMP"
             
                //Chassis intrusion
                SET "CHASSIS_INTRU"
             
                //Front panel NMI
                SET "FP_NMI"

            ENDIF //IFSET "P4000M_BASE"

 

[Russ]

after adding "OTHER_REAR_FAN" in master.cfg, the fan is seen on HWinfo or IASC, but the rpm is not right. Does anybody know how to modify the "OTHER_REAR_FAN" portion of SDR to report the right rpm?

 

[rayofsi]

I have modified the SDR file for my custom build.

Here's my build post: Dual E5-2670 build in the smallest case NZXT S340

Here's some background information: The SDR file is quite complicated in a sense that there are a handful of sections you need to edit to be able to achieve controlling the FAN speed based on a different thermal sensor and curve profile. If you use the custom xxxx_RAMP profile, then all fans rely on the single thermal sensor called BB_EDGE to vary their PWM values. This sensor reacts quite slowly so in my case, having noctua fans running at 600RPM and sloooowly ramping up is not enough. I needed the CPU fans to react quickly when the CPU is loaded. After some modification, I have linked the P2_VR sensor to both CPU fans, it approximately matches the CPU temperature and reacts pretty quickly.

I have uploaded my modified SDR file for you to compare. It sets FAN speed based on my all noctua configuration.

CASE FAN 1: NF-P14S Redux PWM-1200 RPM
CASE FAN 3: NF-P14S Redux PWM-1200 RPM
CPU FAN1: NF-F12 PWM-1500 RPM
CPU FAN2: NF-F12 PWM-1500 RPM
REAR FAN: NF-P12 PWM-1300 RPM

Thanks for this post. Using this temporarily until i get time to figure it out. Running case fan 3 120mm fan, and rear fan rigged up to 6x 140mm fans via pwm splitter. and 2x 140mm cpu fans. dropped my full load from ~62c to 54c. not much extra noise. Fans still not 100% but certainly much better

 

[Jude]

I have modified the SDR file for my custom build.

Here's my build post: Dual E5-2670 build in the smallest case NZXT S340

Here's some background information: The SDR file is quite complicated in a sense that there are a handful of sections you need to edit to be able to achieve controlling the FAN speed based on a different thermal sensor and curve profile. If you use the custom xxxx_RAMP profile, then all fans rely on the single thermal sensor called BB_EDGE to vary their PWM values. This sensor reacts quite slowly so in my case, having noctua fans running at 600RPM and sloooowly ramping up is not enough. I needed the CPU fans to react quickly when the CPU is loaded. After some modification, I have linked the P2_VR sensor to both CPU fans, it approximately matches the CPU temperature and reacts pretty quickly.

I have uploaded my modified SDR file for you to compare. It sets FAN speed based on my all noctua configuration.

CASE FAN 1: NF-P14S Redux PWM-1200 RPM
CASE FAN 3: NF-P14S Redux PWM-1200 RPM
CPU FAN1: NF-F12 PWM-1500 RPM
CPU FAN2: NF-F12 PWM-1500 RPM
REAR FAN: NF-P12 PWM-1300 RPM

FunkyRider83 I think your build is great. It inspired me to go ahead and purchase one of the S2600CP2J package deals from Natex. I have installed it and the dual E5-2670's in a Thermaltake Core V71 case. I am using Noctua fans and have found that the fans are spinning at 100% even though I updated the bios and selected all of the existing fans and their locations. I would love to use your modified SDR file but I have two more fans than you and I am not sure which lines I need to change in the SDR file so that they match what you have already altered.

My fan setup is as follows

Case Fan 1: NF-P14s Redux PWM 1500 RPM
Case Fan 2:NF-P14s Redux PWM 1500 RPM
Case Fan 3: NF-F12 PWM-1500 RPM
Case Fan4: NF-F12 PWM-1500 RPM
CPU Fan 1: NF-F12 PWM-1500 RPM
CPU Fan 2: NF-F12 PWM-1500 RPM
Rear Fan: NF-P14s Redux PWM 1500 RPM

Any assistance would be greatly appreciated. Thanks

 

[FunkyRider83]

FunkyRider83 I think your build is great. It inspired me to go ahead and purchase one of the S2600CP2J package deals from Natex. I have installed it and the dual E5-2670's in a Thermaltake Core V71 case. I am using Noctua fans and have found that the fans are spinning at 100% even though I updated the bios and selected all of the existing fans and their locations. I would love to use your modified SDR file but I have two more fans than you and I am not sure which lines I need to change in the SDR file so that they match what you have already altered.

My fan setup is as follows

Case Fan 1: NF-P14s Redux PWM 1500 RPM
Case Fan 2:NF-P14s Redux PWM 1500 RPM
Case Fan 3: NF-F12 PWM-1500 RPM
Case Fan4: NF-F12 PWM-1500 RPM
CPU Fan 1: NF-F12 PWM-1500 RPM
CPU Fan 2: NF-F12 PWM-1500 RPM
Rear Fan: NF-P14s Redux PWM 1500 RPM

Any assistance would be greatly appreciated. Thanks

I think one of the problems of using slow fans is they tend to trigger lower limit of RPM so the motherboard kicks in to panic mode and spin all fans to 100%. Try to rise the baseline PWM of the fans in BIOS. In my case, I have to set the base PWM value from 0 to +15 to make the fan run faster than lower limit. You don't have to flash the firmware after setting this. I can't remember the exact setting in BIOS but it should be under Cooling and Acoustics or something like that. When you set the value to i.e. 10, it adds 10% to the PWM signal so slow fans can turn a bit faster.

 

[Jude]

I think one of the problems of using slow fans is they tend to trigger lower limit of RPM so the motherboard kicks in to panic mode and spin all fans to 100%. Try to rise the baseline PWM of the fans in BIOS. In my case, I have to set the base PWM value from 0 to +15 to make the fan run faster than lower limit. You don't have to flash the firmware after setting this. I can't remember the exact setting in BIOS but it should be under Cooling and Acoustics or something like that. When you set the value to i.e. 10, it adds 10% to the PWM signal so slow fans can turn a bit faster.

OK I know where that setting is I will definitely try out that adjustment. I'm away right now so I wont be able to test it till next week.

Do you think that by doing that I will not need to "hack" the SDR file?

 

[Jude]

OK I know where that setting is I will definitely try out that adjustment. I'm away right now so I wont be able to test it till next week.

Do you think that by doing that I will not need to "hack" the SDR file?

OK, that worked.

Adjusting the PWM signal by increasing it from 0 to 20 has stopped the fans from falling back into failsafe 100% speed mode.

System at idle: all fans running at 784 RPM and BB MEM VR Temp sensor reading 41 degrees C. Not silent but close enough to it.

Cheers

 

[Gnodu]

I'm certainly no expert... but give some thought to the section jkust below the header at the bottom of my post... I recently played around with this on another server type altogether-- and it had a huge impact-- as in, with a few other changes, you can "fool" it into thinking that ALL fans are running twice as fast as they really are.

<cut>
00 // Number of tach Pulses Per Revolution: Bit 0=Fan 0, Bit 1=Fan 1, etc.; 0=2 Pulses/Rev, 1=1 Pulse/Rev(***TBD***)
<cut>

Now-- I can't speak to the S2600CP, but if (as in my case) you changed the "00" to "01", suddenly instead of 5000 RPM reading 10000 pulses per minute, the reading becomes 10000 pulses equals 10000 RPM, so the PWM speed gets cut back to bring the RPM in line. At least, I THINK that's what it does... but this certainly helped me a lot with my "jet engine" as my wife was calling my S4600LT2 server, even through a closed door. (Note- I'm still working on this project- but I can only test with 2 CPUs right now, so I'm not sure it'll work correctly w all 4 CPUs installed.

When I first played with this value, I expected to see a 50% cut in fans, but after all the other logic gets applied, it was more like a 35% reduction.

WARNING!!! As far as I can tell- this means that if the baseboard TRIES to go to 100%, it may not be able to do so (fans will cap out at max RPM as defined in each fan record-- in other words-- be careful not to adjust too low!!!!! (Personally-- I add addl fans (usually 40x40x10mm directly to heatsinks as a secondary insurance policy... at least on stuff like the RAID controller and x540 chips that are prone to running hot!)

Good luck- hope this helps-- but please experiment carefully!


//====================================================================//
//
// SDR Type C0/0A Record For Fan Sensor Scaling SDR Record
//
//====================================================================//
**SEE RIGHT BELOW THIS IN SDR FILE****

 

[Gnodu]

BTW- if you are at 784RPM, I would NOT try to go much lower- that should be fairly quiet-- I'd consider a fan swap/rewire (to a lower speed version of the same fan-- usually last 1-2 characters in the fan model number) before I'd go lower than about 800 RPM!

 

[FunkyRider83]

I'm certainly no expert... but give some thought to the section jkust below the header at the bottom of my post... I recently played around with this on another server type altogether-- and it had a huge impact-- as in, with a few other changes, you can "fool" it into thinking that ALL fans are running twice as fast as they really are.

<cut>
00 // Number of tach Pulses Per Revolution: Bit 0=Fan 0, Bit 1=Fan 1, etc.; 0=2 Pulses/Rev, 1=1 Pulse/Rev(***TBD***)
<cut>

Now-- I can't speak to the S2600CP, but if (as in my case) you changed the "00" to "01", suddenly instead of 5000 RPM reading 10000 pulses per minute, the reading becomes 10000 pulses equals 10000 RPM, so the PWM speed gets cut back to bring the RPM in line. At least, I THINK that's what it does... but this certainly helped me a lot with my "jet engine" as my wife was calling my S4600LT2 server, even through a closed door. (Note- I'm still working on this project- but I can only test with 2 CPUs right now, so I'm not sure it'll work correctly w all 4 CPUs installed.

When I first played with this value, I expected to see a 50% cut in fans, but after all the other logic gets applied, it was more like a 35% reduction.

WARNING!!! As far as I can tell- this means that if the baseboard TRIES to go to 100%, it may not be able to do so (fans will cap out at max RPM as defined in each fan record-- in other words-- be careful not to adjust too low!!!!! (Personally-- I add addl fans (usually 40x40x10mm directly to heatsinks as a secondary insurance policy... at least on stuff like the RAID controller and x540 chips that are prone to running hot!)

Good luck- hope this helps-- but please experiment carefully!


//====================================================================//
//
// SDR Type C0/0A Record For Fan Sensor Scaling SDR Record
//
//====================================================================//
**SEE RIGHT BELOW THIS IN SDR FILE****

Sounds a bit dangerous to me. Better to lower the actual PWM in the fan curve than fooling motherboard RPM feedback, if somehow your server gets loaded 100% for prolonged time, you could easily damage the system by not providing sufficient airflow.

 

[Gnodu]

That's why I put the big "WARNING" on there-- but-- I learned a LOT about how the whole system worked after I discovered this "trick". My current version SDR has been edited to heck and back-- but it WAS reset to the original "00" value. (I just went and double-checked LOL!)

It also made it so I could actually think while I was testing out other configurations and changes.