ECC RAM at 85C?

[H0R53]

I just logged onto my NAS via RDP and I noticed FB-DIMM-0 is at 85C while the rest are mid-60s. Is this an anomaly and safely dismiss this or is it something to worry about? RAM is relatively new.

[ArduinoBen]

Are the RAM sticks near some other components that may be generating lots of heat?  For example if some of the are closer to the PSU they may get some of the heat from there.

[Aw_Ginger_Snapz]

Is it DDR2 ECC FB dims? When I had my old workstation, all 8 DIMMs of ram ran at about 90 to 115 degrees during normal to heavy workloads. (I discovered this through software and hardware methods) I dismissed this as being a normal occurrence because it didn't affect the stability of my machine in any way. This did make me nervous however, so I opened up my case and placed a normal floor fan in the side panel area to make sure airflow wasn't an issue. However, this didn't fix the problem. If you have heat sinks on your DIMMs, then it could be a bad thermal pad.

[H0R53]

2 minutes ago, Aw_Ginger_Snapz said:

Is it DDR2 ECC FB dims? When I had my old workstation, all 8 DIMMs of ram ran at about 90 to 115 degrees during normal to heavy workloads. (I discovered this through software and hardware methods) I dismissed this as being a normal occurrence because it didn't affect the stability of my machine in any way. This did make me nervous however, so I opened up my case and placed a normal floor fan in the side panel area to make sure airflow wasn't an issue. However, this didn't fix the problem. If you have heat sinks on your DIMMs, then it could be a bad thermal pad.

Yes, they are DDR2 ECC FB-DIMMs, which I know can get extremely hot. So the temps I'm getting are relatively good? They do have heat spreaders. What's the highest temperature that I should accept as 'normal'? I have a fan almost directly on them and it seems to do its job fairly well. They are hot to the touch and if I put my finger on them too long it burns.

 

Bad thermal pad? I didn't realize they can go bad.

 

My NAS is loosely based off of a Dell Precision 690. Here's a screenshot of the RDP session as of writing.

[Aw_Ginger_Snapz]

13 hours ago, H0R53 said:

Yes, they are DDR2 ECC FB-DIMMs, which I know can get extremely hot. So the temps I'm getting are relatively good? They do have heat spreaders. What's the highest temperature that I should accept as 'normal'? I have a fan almost directly on them and it seems to do its job fairly well. They are hot to the touch and if I put my finger on them too long it burns.

 

Bad thermal pad? I didn't realize they can go bad.

 

My NAS is loosely based off of a Dell Precision 690. Here's a screenshot of the RDP session as of writing.

I wouldn't sweat the temps of your RAM because they enterprise grade electronics. They are meant to be virtually indestructible. Also I had a thermal pad that dried up in one of my DIMMs, making it essentially useless for transferring heat. Granted, the DIMM was more than 10 years old by the time I discovered the occurrence.

 

My old workstation was based off of an HP XW8600, which used the 5400 series of Xeons. the airflow of the case was god awful, and I bought a RAM fan specifically to remedy this problem. Like I said, my temps of 8 FB DIMMMs peaked at about 115 degrees under heavy load.

 

I wouldn't worry much about the temps especially since your using this machine just as a NAS. The worst that could happen is one of your DIMMs die, which is very unlikely. Considering how cheap DDR2 ECC is, it wouldn't be that big of a deal to get a new DIMM.

 

However, based off your screenshot, I do find it rather strange that you have a single DIMM that is much hotter than all of the others. It could be its placement in your NAS, it could be picking up heat from something else, or it could be failing. If it is the latter, do you get any memory error messages during bootup?

[H0R53]

3 minutes ago, Aw_Ginger_Snapz said:

I wouldn't sweat the temps of your RAM because they enterprise grade electronics. They are meant to be virtually indestructible. Also I had a thermal pad that dried up in one of my DIMMs, making it essentially useless for transferring heat. Granted, the DIMM was more than 10 years old by the time I discovered the occurrence.

 

My old workstation was based off of an HP XW8600, which used the 5400 series of Xeons. the airflow of the case was god awful, and I bought a RAM fan specifically to remedy this problem. Like I said, my temps of 8 FB DIMMMs peaked at about 115 degrees under heavy load.

 

I wouldn't worry much about the temps especially since your using this machine just as a NAS. The worst that could happen is one of your DIMMs die, which is very unlikely. Considering how cheap DDR2 ECC is, it wouldn't be that big of a deal to get a new DIMM.

 

However, based off your screenshot, I do find it rather strange that you have a single DIMM that is much hotter than all of the others. It could be its placement in your NAS, it could be picking up heat from something else, or it could be failing. If it is the latter, do you get any memory error messages during bootup?

The only errors are fan errors because it's a fully custom build based off a Dell board and Dells can be anal about disconnected hardware.

 

The DIMM is the top vertically, and heat rises, but there is a fan blowing directly on it if you check out the thread linked in my last reply. The hottest I see the DIMMs get is 85. I'll run Passmark and monitor temps. It's in a closet with no monitor so I use RDP.

 

The machine doesn't even need 16GB of RAM, it was only forty bucks so I got it. Unused RAM is wasted RAM.

 

Sometimes I use it for video rendering when my main workstation is occupied by another project. Even then the two Xeons don't even come close to hot, 60C at most.

 

HP machines are crap. I've had three personal rigs die on me that were HP or HP-based. Not a single machine I've had that's been a Dell based off of a Dell has ever died. The X54xx series is awesome, I have a Vostro-200 based machine with a hacked BIOS running my old X5460. Trying to sell it.

 

It's funny though, the DDR2 in my NAS came from an HP machine, so maybe that explains it.

 

More than likely the RAM is hot because it's the top of the stack, so nothing to worry about.

[wzrd]

35 minutes ago, Aw_Ginger_Snapz said:

I wouldn't sweat the temps of your RAM because they enterprise grade electronics. They are meant to be virtually indestructible. Also I had a thermal pad that dried up in one of my DIMMs, making it essentially useless for transferring heat. Granted, the DIMM was more than 10 years old by the time I discovered the occurrence.

 

My old workstation was based off of an HP XW8600, which used the 5400 series of Xeons. the airflow of the case was god awful, and I bought a RAM fan specifically to remedy this problem. Like I said, my temps of 8 FB DIMMMs peaked at about 115 degrees under heavy load.

 

I wouldn't worry much about the temps especially since your using this machine just as a NAS. The worst that could happen is one of your DIMMs die, which is very unlikely. Considering how cheap DDR2 ECC is, it wouldn't be that big of a deal to get a new DIMM.

 

However, based off your screenshot, I do find it rather strange that you have a single DIMM that is much hotter than all of the others. It could be its placement in your NAS, it could be picking up heat from something else, or it could be failing. If it is the latter, do you get any memory error messages during bootup?

@H0R53 This. The industry standard for enterprise grade electronics is 125C. Most MOSFETS, VRMs, etc., are actually designed to work at 125C or lower. DDR2 ECC dumps an insane amount of heat, I have seen far, far worse temps than that on them 

[Aw_Ginger_Snapz]

13 hours ago, H0R53 said:

The only errors are fan errors because it's a fully custom build based off a Dell board and Dells can be anal about disconnected hardware.

 

HP machines are crap. I've had three personal rigs die on me that were HP or HP-based. Not a single machine I've had that's been a Dell based off of a Dell has ever died. The X54xx series is awesome, I have a Vostro-200 based machine with a hacked BIOS running my old X5460. Trying to sell it.

 

It's funny though, the DDR2 in my NAS came from an HP machine, so maybe that explains it.

WHO ELSE LOVES PROPRIETARY HARDWARE??? I DO!!!

 

But in all seriousness, yea HP machines suck. My old workstation had two Xeons that were x5470s.

I decided to ditch it after I bought  GTX 1070 and realized how bad the bottleneck was. Unfortunately the only thing I could do was upgrade. And To be honest, I don't regret it. I just need to find something to do with the old workstation...

 

The 5400 series of chips were great because they were essentially Core 2 chips and they can be found for super cheap on ebay. But my only issue with them is that they lack many features I would have thought they should have like hyper-threading, and DDR3 support just to name a few. But their feature set is too close to their mainstream counterparts. I think that the series has aged horribly. If anyone wants a cheap gaming rig, then they are going to get a Q6600 before they get something like a Xeon E5450. I'd definitely do that because the platform is slightly cheaper, and slightly newer. I mean heck, if you get a Core 2 Quad with a high enough FSB speed, you can take advantage of much higher speed memory then any of the 5400 series chips support. I've had a Q6600 smoke my x5470s in games like GTA V simply because the memory speed was higher.

With the 5400 series chips you are limited to DDR2 ECC @ 667Mhz. On the other hand you can get a mainstream or enthusiast chip like the Q6600 (FSB of 1066 Mhz) or QX9650 (FSB of 1600) which both support DDR3, perform better, and cost a little bit less. But that's just the gaming side of things. But if you didn't want to game, then just buy the X5600 series chips. More features, more lifespan, and way more performance.
 

[H0R53]

10 minutes ago, Aw_Ginger_Snapz said:

WHO ELSE LOVES PROPRIETARY HARDWARE??? I DO!!!

 

But in all seriousness, yea HP machines suck. My old workstation had two Xeons that were x5470s.

I decided to ditch it after I bought  GTX 1070 and realized how bad the bottleneck was. Unfortunately the only thing I could do was upgrade. And To be honest, I don't regret it. I just need to find something to do with the old workstation...

 

The 5400 series of chips were great because they were essentially Core 2 chips and they can be found for super cheap on ebay. But my only issue with them is that they lack many features I would have thought they should have like hyper-threading, and DDR3 support just to name a few. But their feature set is too close to their mainstream counterparts. I think that the series has aged horribly. If anyone wants a cheap gaming rig, then they are going to get a Q6600 before they get something like a Xeon E5450. I'd definitely do that because the platform is slightly cheaper, and slightly newer. I mean heck, if you get a Core 2 Quad with a high enough FSB speed, you can take advantage of much higher speed memory then any of the 5400 series chips support. I've had a Q6600 smoke my x5470s in games like GTA V simply because the memory speed was higher.

With the 5400 series chips you are limited to DDR2 ECC @ 667Mhz. On the other hand you can get a mainstream or enthusiast chip like the Q6600 (FSB of 1066 Mhz) or QX9650 (FSB of 1600) which both support DDR3, perform better, and cost a little bit less. But that's just the gaming side of things. But if you didn't want to game, then just buy the X5600 series chips. More features, more lifespan, and way more performance.
 

Dell's fan headers are the worst. On Optiplex machines you just swap pins 1 and 3 and there are no errors, but at least on my precision I'm just using Molex and everything else as high as it can go.

 

HP is garbage, the Spectre and its relatives are the only things keeping HP alive aside from Itanium chips that HP collabs with IBM and Intel on.

 

The PCI-e 2.0 and 2.1 bottleneck is huge for newer cards. On older ones like GTX 580, not really.

 

It's funny, my Precision 690's BIOS has an option to turn HT on or off, except the LGA 771 chips don't even have it. Strange, huh?

 

The X5355s have a 1333MHz FSB. My X5460 runs its RAM at 800MHz. Also on Tyan boards with DDR3 1600 you can crazy OC the 54xx chips.

 

Thanks for the help.

[Aw_Ginger_Snapz]

14 hours ago, H0R53 said:

Dell's fan headers are the worst. On Optiplex machines you just swap pins 1 and 3 and there are no errors, but at least on my precision I'm just using Molex and everything else as high as it can go.

 

HP is garbage, the Spectre and its relatives are the only things keeping HP alive aside from Itanium chips that HP collabs with IBM and Intel on.

 

The PCI-e 2.0 and 2.1 bottleneck is huge for newer cards. On older ones like GTX 580, not really.

 

It's funny, my Precision 690's BIOS has an option to turn HT on or off, except the LGA 771 chips don't even have it. Strange, huh?

 

The X5355s have a 1333MHz FSB. My X5460 runs its RAM at 800MHz. Also on Tyan boards with DDR3 1600 you can crazy OC the 54xx chips.

 

Thanks for the help.

Lol, HP announced that they were going to discontinue making servers with Itanium after this year, and Open VMS was getting ported to x86. Rip

[H0R53]

1 hour ago, Aw_Ginger_Snapz said:

Lol, HP announced that they were going to discontinue making servers with Itanium after this year, and Open VMS was getting ported to x86. Rip

Itanium was good while it lasted. 

 

Screw HP. The only things I buy from them are replacement parts for clients, and even then I tell them to invest in another company like Dell.

[Aw_Ginger_Snapz]

13 hours ago, H0R53 said:

Itanium was good while it lasted. 

 

Screw HP. The only things I buy from them are replacement parts for clients, and even then I tell them to invest in another company like Dell.

I don't really like HP or Dell for that matter. I've had bad experiences with both. I'm to the point where if I wanted a server for example, then I'll just invest a little more in off the shelf parts from Supermicro, Asus, or Asrock. Heck if I really had a deep pocket, I would rebuild servers by Sun Microsystems. Their Sunfire and T-series lineups are dope. Those things are extremely overbuilt. But for normal desktops, I prefer just off the shelf parts.

[H0R53]

3 minutes ago, Aw_Ginger_Snapz said:

I don't really like HP or Dell for that matter. I've had bad experiences with both. I'm to the point where if I wanted a server for example, then I'll just invest a little more in off the shelf parts from Supermicro, Asus, or Asrock. Heck if I really had a deep pock, I would rebuild servers by Sun Microsystems. Their Sunfire and T-series lineups are dope. Those things are extremely overbuilt. But for normal desktops, I prefer just off the shelf parts.

My PoS machine for logging repair tickets is a Dell Optiplex I built out of replacement parts, just in a custom case. Everything else, including RAM, is factory.

 

What happened with your Dell machine?

[Aw_Ginger_Snapz]

13 hours ago, H0R53 said:

My PoS machine for logging repair tickets is a Dell Optiplex I built out of replacement parts, just in a custom case. Everything else, including RAM, is factory.

 

What happened with your Dell machine?

I can't remember the specific model of the server. I can tell you that it used the 5400 series chips (originally L5420, upgraded to a e5450), no expansion cards, 1 SSD, and 2 HDD, with 16 Gb of non buffered ECC RAM @667 Mhz. Both of the original redundant power supplies died on me while I was using low power chips. Then I upgraded the chips and got beefier official dell power supplies that were brand new to replace the broken ones. The new power supplies worked for a few months, then suddenly died on me for seemingly no reason and took the rest of the server with them. Everything was unusable, except for my storage drives. The only task that I had assigned to the bloody thing was Plone. All of my Sun, IBM Xserves (another great overbuilt server), and custom servers were absolutely fine. They were all plugged into the same UPS, and the same outlet. It was the most bizarre hardware failure I've ever seen. Since then, I'm a little scared to go back to dell. I've been just buying IBM XServes (M2 & M3 Class). They work fantastic!

[H0R53]

10 minutes ago, Aw_Ginger_Snapz said:

I can't remember the specific model of the server. I can tell you that it used the 5400 series chips (originally L5420, upgraded to a e5450), no expansion cards, 1 SSD, and 2 HDD, with 16 Gb of non buffered ECC RAM @667 Mhz. Both of the original redundant power supplies died on me while I was using low power chips. Then I upgraded the chips and got beefier official dell power supplies that were brand new to replace the broken ones. The new power supplies worked for a few months, then suddenly died on me for seemingly no reason and took the rest of the server with them. Everything was unusable, except for my storage drives. The only task that I had assigned to the bloody thing was Plone. All of my Sun, IBM Xserves (another great overbuilt server), and custom servers were absolutely fine. They were all plugged into the same UPS, and the same outlet. It was the most bizarre hardware failure I've ever seen. Since then, I'm a little scared to go back to dell. I've been just buying IBM XServes (M2 & M3 Class). They work fantastic!

The problem was the Xeons, those are underspecced for the power they are rated for. Probably pulling more power than the PSUs could handle at once.

 

Tell Linus about it. "Haunted Hardware" lol.

 

My Precision's PSU is still pulling 80% efficiency after 10 years of service, 9 of which I don't know the case. I found the Precision on the curb with a white label that said "Kitchen".

 

Give Dell another shot, get a cheapo Optiplex SFF machine.

[Aw_Ginger_Snapz]

13 hours ago, H0R53 said:

Give Dell another shot, get a cheapo Optiplex SFF machine.

Actually, my aunt dropped off a SFF Optiplex 745(?) on my doorstep, claiming that it didn't work.... The issue was that it was running Vista, had a really shitty Celery cpu (LGA 775), and a crappy HDD. I would have thrown it out, but this PC caught my eye. It had DDR3 support, and had a full length graphics card slot. So I replaced the Celery with a Q6600, threw in 4Gb of 1066 Mhz RAM, a spare HDD, and replaced the underwhelming heat sink and it works pretty good. The only issue is that I don't have a GPU to put in it, and I don't feel like spending money on a SFF card for this machine. Especially considering that I have a much more powerful machine that needs a new GPU. (it currently has a GT210 in it because the previous one stopped working). Now the Dell, is just collecting dust in my basement. I would totally use it if I had a decent low profile GPU to throw in it. Maybe a GTX 750/1050, but unless I get it for free, I don't plan on using it for much. I'd like to keep this machine as free as possible.

[H0R53]

5 minutes ago, Aw_Ginger_Snapz said:

Actually, my aunt dropped off a SFF Optiplex 745(?) on my doorstep, claiming that it didn't work.... The issue was that it was running Vista, had a really shitty Celery cpu (LGA 775), and a crappy HDD. I would have thrown it out, but this PC caught my eye. It had DDR3 support, and had a full length graphics card slot. So I replaced the Celery with a Q6600, threw in 4Gb of 1066 Mhz RAM, a spare HDD, and replaced the underwhelming heat sink and it works pretty good. The only issue is that I don't have a GPU to put in it, and I don't feel like spending money on a SFF card for this machine. Especially considering that I have a much more powerful machine that needs a new GPU. (it currently has a GT210 in it because the previous one stopped working). Now the Dell, is just collecting dust in my basement. I would totally use it if I had a decent low profile GPU to throw in it. Maybe a GTX 750/1050, but unless I get it for free, I don't plan on using it for much. I'd like to keep this machine as free as possible.

Slap in a Q9xxx, they are more powerful and the Optiplex's don't OC so the 6600 won't do very well. You can get GPUs fairly cheap off Craigslist, a trade is technically free...

[H0R53]

22 hours ago, Aw_Ginger_Snapz said:

Actually, my aunt dropped off a SFF Optiplex 745(?) on my doorstep, claiming that it didn't work.... The issue was that it was running Vista, had a really shitty Celery cpu (LGA 775), and a crappy HDD. I would have thrown it out, but this PC caught my eye. It had DDR3 support, and had a full length graphics card slot. So I replaced the Celery with a Q6600, threw in 4Gb of 1066 Mhz RAM, a spare HDD, and replaced the underwhelming heat sink and it works pretty good. The only issue is that I don't have a GPU to put in it, and I don't feel like spending money on a SFF card for this machine. Especially considering that I have a much more powerful machine that needs a new GPU. (it currently has a GT210 in it because the previous one stopped working). Now the Dell, is just collecting dust in my basement. I would totally use it if I had a decent low profile GPU to throw in it. Maybe a GTX 750/1050, but unless I get it for free, I don't plan on using it for much. I'd like to keep this machine as free as possible.

Since this is my eight core NAS we are talking about, and it literally has every single file on my PC in its storage, I went ahead and started doing some heavy archive unpacking (over 200GB worth of data to unpack using a proprietary program, not like 7Zip, WinRar, etc)

 

30 minutes into the unpack I log onto the NAS via RDP. Attached image is what I got.

 

ALL of the DIMMS are at around 80-85 degrees.

 

Is this normal? This NAS is relatively new and I never did much with it until of late, so I've never established a baseline for DIMM temperatures.

 

Right now the coolest DIMM is 64 degrees, and it's directly below the hottest DIMM, and that doesn't make much sense in the way of thermodynamics. Shouldn't it be hotter?