Need to cool down some Pentium Ds

[Keysplayr]

Originally posted by: Markfw900
Keys, since you were not the poster I didn;t see if buried in the thread. I re-read the entire thread and found this:

Funny thing is, I tried to get them to let me buy AMD parts but the programmers rallied against me. LOL. Seems they like to code on Intel rigs exclusively.

So, I APOLOGISE AND TAKE BACK THE FANBOY REFERENCE !

I also found a comment you made that said "we all know AMD runs cooler" or something like that, so again, I will try to be more productive in this thread. You might want to point out to the the fanboys who rallied against you NOW the errors of their ways.

I thank you for that Mark. Believe me, I tried to convince them with perfect logic. But they would not hear me. And they have the final word on what they code on. Even so, the PD830's are really rather quick for what it's being used for. It blows away Dual 3.06 Xeons by a surprisingly large margin. So, it's no biggie that we use Intel. It's just fine.

 

[Keysplayr]

Originally posted by: Duvie

Originally posted by: keysplayr2003

Originally posted by: Markfw900
keysplayr2003, fine, you live with your power hogs. Our company has a data center in Corona CA, and the city doesn't have enough power for it (no sh!t !), so we have 8 semi rigs with generators powering the data center idenfinitely, until the city can fix its power problems. Right now we only use Wintel servers, and they calculated that we could save enough power, just by eliminating the Xeon servers, and going with the new Opteron servers to get rid of 2, maybe 3 of those semi's. At todays gas prices, we could pay for the hardware. I don;t have the details, but we are talking hundreds of servers, not counting the mainframes.

Edit, and you idle at 46c ? I can do that with both core @full load when the amient gets down to 70f. Load at 65 ? I can't get that high as far as I have tried to crank the voltage ! Yea, you won't need a heater this winter....You can be nice an cosy with your P-D's fanboy.

Hmmm. Sounds to me like I must have hit a nerve Mark. Apologies. I'm just trying to call out all of the over the top BS coming out of some peoples yaps around here. I'm no fanboy despite your mudslinging slander, as I mentioned I even tried convincing them to go AMD for the lower power consumption and cooler setups. Oh, you must have missed that part intentionally or wiped it from your mind on the fly as you were reading it. These programmers insisted on Intel rigs. And I stand here before you to say what I have found in using them, and get called a fanboy by you Mark. I stand here and tell you that your multiple 6000 rpm fan reference and multiple hard drives not letting enought juice to the CPU to "heat it up properly" was BS. And you still have the retort to call me a fanboy.

I'm gonna have to ask why your so offended, and why you chose to call me a fanboy.

Waiting for a credible reply in light of the information I have given you.


Duvie: 65C is the absolute max I have seen the temp after hours and hours of full bore crunching. Both cores at 100% for overnighter jobs and still cannot get over 65C.

It varies but only in lower temps. I see 64C, 63C and 62C at full blown crunching, but never exceeds 65C. Room temps are usually 72F. Electronically Climate controlled.
And if dust gets built up on the rigs fans, would you not occaisonally maintain them just as equally if it were an AMD rig? or would you leave the dust cake up on your fans ordinarily? If your the enthusiast you say you are, you would be in your case almost every day checking it for such things.

Actaully when I had my northwood with more case fans meant more dust and I had to religiously clean it about every 3-4 weeks as temps would climb...I run about 2 less case fans now which produces less dust and I clean t about every 6-8 weeks but it doesn't get even anywhere close to as dirty and the temp usually dont raise but 1-2c......

I am asking you case I have seen offices and the computers by no means get a thorough cleanninglike I give them even within a 3-4 month schedule...

All I am saying is putting it that close to the potential limit and require potentially "enthusiast" level cleaning frequently, seems a bit much to ask for the lowest grade dual core running stock out of the box...Imagine your same setup with a 200-400mhz increase of the 830 or 840....could you keep it cool???

I understand what you mean. But, it will not require enthusiast level cleaning. We have had Dual Xeon systems running for these programmers for better than a year. When I open the cases, sure there is dust, but nothing to worry about. And by the way, These 4 machines are using the 830's.

 

[Keysplayr]

Originally posted by: TGS

Originally posted by: keysplayr2003


Lithan: Huge amounts of data are processed on these rigs. A few thousand 256MB files at a time. 500GB drives just came out so we will be using those in the next workstations.

On a sidenote here, how exactly are 4 drives processing upwards of half a terabyte worth of IO at a time? 4 ATA drives to be more precise. Granted, it could save the data out to the disks and crunch portions in and out of memory, but I really doubt that 4 drives are moving anywhere close to a quarter to half a terabyte at once.

Now if the backend was a SAN backend. I could see a multi-bussed, high spindle count array doing a great deal of legwork to that sort of IO demand. I just can't see any ATA devices coming anywhere, in the realm of reason, near those numbers. The amount of memory for those arrays to process realtime would be outrageous.

Let me clear this up for you. Not a few thousand files at "the very same time" but consecutively. One after the other after the other and on and on. Non-stop. The data that is collected using special algorithm's on each 256MB file is enormous and requires a lot of hard drive space. I never said the 4 drives were moving a quarter to half a terabyte at once. I did not make it clear and you assumed. Does it make more sense this way?

 

[Duvie]

Originally posted by: keysplayr2003

Originally posted by: Duvie

Originally posted by: keysplayr2003

Originally posted by: Markfw900
keysplayr2003, fine, you live with your power hogs. Our company has a data center in Corona CA, and the city doesn't have enough power for it (no sh!t !), so we have 8 semi rigs with generators powering the data center idenfinitely, until the city can fix its power problems. Right now we only use Wintel servers, and they calculated that we could save enough power, just by eliminating the Xeon servers, and going with the new Opteron servers to get rid of 2, maybe 3 of those semi's. At todays gas prices, we could pay for the hardware. I don;t have the details, but we are talking hundreds of servers, not counting the mainframes.

Edit, and you idle at 46c ? I can do that with both core @full load when the amient gets down to 70f. Load at 65 ? I can't get that high as far as I have tried to crank the voltage ! Yea, you won't need a heater this winter....You can be nice an cosy with your P-D's fanboy.

Hmmm. Sounds to me like I must have hit a nerve Mark. Apologies. I'm just trying to call out all of the over the top BS coming out of some peoples yaps around here. I'm no fanboy despite your mudslinging slander, as I mentioned I even tried convincing them to go AMD for the lower power consumption and cooler setups. Oh, you must have missed that part intentionally or wiped it from your mind on the fly as you were reading it. These programmers insisted on Intel rigs. And I stand here before you to say what I have found in using them, and get called a fanboy by you Mark. I stand here and tell you that your multiple 6000 rpm fan reference and multiple hard drives not letting enought juice to the CPU to "heat it up properly" was BS. And you still have the retort to call me a fanboy.

I'm gonna have to ask why your so offended, and why you chose to call me a fanboy.

Waiting for a credible reply in light of the information I have given you.


Duvie: 65C is the absolute max I have seen the temp after hours and hours of full bore crunching. Both cores at 100% for overnighter jobs and still cannot get over 65C.

It varies but only in lower temps. I see 64C, 63C and 62C at full blown crunching, but never exceeds 65C. Room temps are usually 72F. Electronically Climate controlled.
And if dust gets built up on the rigs fans, would you not occaisonally maintain them just as equally if it were an AMD rig? or would you leave the dust cake up on your fans ordinarily? If your the enthusiast you say you are, you would be in your case almost every day checking it for such things.

Actaully when I had my northwood with more case fans meant more dust and I had to religiously clean it about every 3-4 weeks as temps would climb...I run about 2 less case fans now which produces less dust and I clean t about every 6-8 weeks but it doesn't get even anywhere close to as dirty and the temp usually dont raise but 1-2c......

I am asking you case I have seen offices and the computers by no means get a thorough cleanninglike I give them even within a 3-4 month schedule...

All I am saying is putting it that close to the potential limit and require potentially "enthusiast" level cleaning frequently, seems a bit much to ask for the lowest grade dual core running stock out of the box...Imagine your same setup with a 200-400mhz increase of the 830 or 840....could you keep it cool???

I understand what you mean. But, it will not require enthusiast level cleaning. We have had Dual Xeon systems running for these programmers for better than a year. When I open the cases, sure there is dust, but nothing to worry about. And by the way, These 4 machines are using the 830's.

Where those Xeons within 5c of possible clock thorttling and then detrimental to performance??? I doubt it...

Yeah I saw the 830 after the comment was made....so almost the bottom!!!!

Again it doesn't take much dust on a fan blade to impede 10% air flow and when you have 5c to work with that is not much....


I guess I am just a bit disappointed ppl are happy with a stock product being only about 6-7% away from overheating to the point the system will downclock itself in a act of self preservation at expense of performance the owner paid for....

Granted this has not happened yet but it is sad no less the headroom is soooo low...I onmly came close to that edge after Ocing 1.1ghz and .08v....Maybe INtel needs to do a better job of determing the TDP so that designers can build more effective cooling....

 

[AkumaX]

you think its a good or bad thing that the 820D doesn't come with speedstep/throttling?

 

[Keysplayr]

Originally posted by: Duvie

Originally posted by: keysplayr2003

Originally posted by: Duvie

Originally posted by: keysplayr2003

Originally posted by: Markfw900
keysplayr2003, fine, you live with your power hogs. Our company has a data center in Corona CA, and the city doesn't have enough power for it (no sh!t !), so we have 8 semi rigs with generators powering the data center idenfinitely, until the city can fix its power problems. Right now we only use Wintel servers, and they calculated that we could save enough power, just by eliminating the Xeon servers, and going with the new Opteron servers to get rid of 2, maybe 3 of those semi's. At todays gas prices, we could pay for the hardware. I don;t have the details, but we are talking hundreds of servers, not counting the mainframes.

Edit, and you idle at 46c ? I can do that with both core @full load when the amient gets down to 70f. Load at 65 ? I can't get that high as far as I have tried to crank the voltage ! Yea, you won't need a heater this winter....You can be nice an cosy with your P-D's fanboy.

Hmmm. Sounds to me like I must have hit a nerve Mark. Apologies. I'm just trying to call out all of the over the top BS coming out of some peoples yaps around here. I'm no fanboy despite your mudslinging slander, as I mentioned I even tried convincing them to go AMD for the lower power consumption and cooler setups. Oh, you must have missed that part intentionally or wiped it from your mind on the fly as you were reading it. These programmers insisted on Intel rigs. And I stand here before you to say what I have found in using them, and get called a fanboy by you Mark. I stand here and tell you that your multiple 6000 rpm fan reference and multiple hard drives not letting enought juice to the CPU to "heat it up properly" was BS. And you still have the retort to call me a fanboy.

I'm gonna have to ask why your so offended, and why you chose to call me a fanboy.

Waiting for a credible reply in light of the information I have given you.


Duvie: 65C is the absolute max I have seen the temp after hours and hours of full bore crunching. Both cores at 100% for overnighter jobs and still cannot get over 65C.

It varies but only in lower temps. I see 64C, 63C and 62C at full blown crunching, but never exceeds 65C. Room temps are usually 72F. Electronically Climate controlled.
And if dust gets built up on the rigs fans, would you not occaisonally maintain them just as equally if it were an AMD rig? or would you leave the dust cake up on your fans ordinarily? If your the enthusiast you say you are, you would be in your case almost every day checking it for such things.

Actaully when I had my northwood with more case fans meant more dust and I had to religiously clean it about every 3-4 weeks as temps would climb...I run about 2 less case fans now which produces less dust and I clean t about every 6-8 weeks but it doesn't get even anywhere close to as dirty and the temp usually dont raise but 1-2c......

I am asking you case I have seen offices and the computers by no means get a thorough cleanninglike I give them even within a 3-4 month schedule...

All I am saying is putting it that close to the potential limit and require potentially "enthusiast" level cleaning frequently, seems a bit much to ask for the lowest grade dual core running stock out of the box...Imagine your same setup with a 200-400mhz increase of the 830 or 840....could you keep it cool???

I understand what you mean. But, it will not require enthusiast level cleaning. We have had Dual Xeon systems running for these programmers for better than a year. When I open the cases, sure there is dust, but nothing to worry about. And by the way, These 4 machines are using the 830's.

Where those Xeons within 5c of possible clock thorttling and then detrimental to performance??? I doubt it...

Yeah I saw the 830 after the comment was made....so almost the bottom!!!!

Again it doesn't take much dust on a fan blade to impede 10% air flow and when you have 5c to work with that is not much....


I guess I am just a bit disappointed ppl are happy with a stock product being only about 6-7% away from overheating to the point the system will downclock itself in a act of self preservation at expense of performance the owner paid for....

Granted this has not happened yet but it is sad no less the headroom is soooo low...I onmly came close to that edge after Ocing 1.1ghz and .08v....Maybe INtel needs to do a better job of determing the TDP so that designers can build more effective cooling....

I agree that it is sad, for overclockers. Not sad for those who will sit down and use the computer the way it was intended to be used and not even know what overclocking is.

This 5C "window" does not matter. It only matters if the thermal limit is reached. I could happily use these systems for 5 years straight at 65C and not worry about throttling. If and when the event occurs where these systems throttle, it will take only a side panel removal and a quick blast of air to remedy.

Now, I cant say the same thing for the PD840 or PD840HT because they will logically run a bit warmer than these 830's. I just wanted to say that they really do get the job done. Nice CPU's albeit no where near the best they could've been.

 

[Duvie]

Originally posted by: AkumaX
you think its a good or bad thing that the 820D doesn't come with speedstep/throttling?

I think it is a good thing...However it is good for catching that catastrophic fan failure, that 1 in a 1000 event that may kill your cpu within its product use...However to have it happen becuase of deficient cooling that is bundled with the cpus by the manufacturers is sad...

Sad for all not just ocers...

 

[Lithan]

Originally posted by: keysplayr2003

Originally posted by: Lithan

Originally posted by: keysplayr2003

Originally posted by: Markfw900

Originally posted by: BGuardian75
There is much misinfo concerning intel on this obviously pro AmD forum. The OP posted temps that are way out of the ballpark for the PD which indicates something is wrong with the way mwave built these machines. They should idle mid 40s on stock and reaching 65c or so on load which to us AMD lovers appears insane but those machines shouldn't be locking up. I would look at cleaning up the wiring and using some higher rpm fans and cpu coolers with actual thermal grease.

If the HSF is OEM, and uses a thermal pad, Intel says that should be enough, but this is part of the problem ITS NOT. Thats Intels fault. And it is also a fact they DO run very hot and DO quite often throttle due to that. Now that locking up ?? I don't know they should throttle, but all of the points made are valid. You should have to have 3 or 4 6000 rpm fans to cool the case. It is also possible that the 5 hard drives are too much for the PSU's in addition to the large amount of current the PD's take.

You actually just said that out loud didn't you... That the 5 hard drives suck up so much juice that the PD isn't getting enough juice to heat it up? 3 or 4 6000 rpm fans to cool the case? What are you smoking? I stated all the cooling I have in these machines and it's meager compared to what you say is required. So what now? Do I have 4 rare machines that no one else has? Am I cooling with water? TEC's? Phase Change? No. Stock Intel HSF with thermal pad, a single 120mm fan @900rpm and an 80mm fan running default in the rear. You can even count the 2 PSU fans on the Enermax. Although they spin too slowly to even hear them. Prescotts/Smithfields run relatively hot compared to A64's/X2's.
So?

Lithan: Huge amounts of data are processed on these rigs. A few thousand 256MB files at a time. 500GB drives just came out so we will be using those in the next workstations.

Why not use independant drive servers? I've never heard of someone using workstations as their own storage on anywhere near that scale.

Ahh, well different companies would each have there own way of doing things correct?
They have their reasons, and each workstation needs to be standalone for specific reasons and is the way the software is being developed by the programmers.

These systems will ultimately end up in different locations running medical devices that crunch "big" data. Hence the requirement of huge "MOBILE" storage because for the most part will not always have access to LAN's. Don't try to suggest a better way of doing things (although it is appreciated) because this has been discussed numerous times at a hundred meetings. The way it is, is the best way it can be done for the current application.

I read that and all I hear is "some manager who had no clue what he was doing insisted on it". Maybe I've worked tech too long.

 

[Pabster]

Unfortunately you'd be surprised how many companies you might work for where the mere mention of anything other than Intel not only raises eyebrows but will get you fired.

 

[Keysplayr]

Originally posted by: Lithan

Originally posted by: keysplayr2003


Why not use independant drive servers? I've never heard of someone using workstations as their own storage on anywhere near that scale.

Ahh, well different companies would each have there own way of doing things correct?
They have their reasons, and each workstation needs to be standalone for specific reasons and is the way the software is being developed by the programmers.

These systems will ultimately end up in different locations running medical devices that crunch "big" data. Hence the requirement of huge "MOBILE" storage because for the most part will not always have access to LAN's. Don't try to suggest a better way of doing things (although it is appreciated) because this has been discussed numerous times at a hundred meetings. The way it is, is the best way it can be done for the current application.

I read that and all I hear is "some manager who had no clue what he was doing insisted on it". Maybe I've worked tech too long.[/quote]

You hit the nail right on the head.

 

[Bingo13]

840EE with Gigabyte G-Power PDU21 at 3.6. 45c idle, 57c load. Not much more than my FX57.

 

[Pabster]

Originally posted by: Bingo13
840EE with Gigabyte G-Power PDU21 at 3.6. 45c idle, 57c load. Not much more than my FX57.

Nice!

 

[TGS]

Originally posted by: keysplayr2003

Originally posted by: TGS

Originally posted by: keysplayr2003


Lithan: Huge amounts of data are processed on these rigs. A few thousand 256MB files at a time. 500GB drives just came out so we will be using those in the next workstations.

On a sidenote here, how exactly are 4 drives processing upwards of half a terabyte worth of IO at a time? 4 ATA drives to be more precise. Granted, it could save the data out to the disks and crunch portions in and out of memory, but I really doubt that 4 drives are moving anywhere close to a quarter to half a terabyte at once.

Now if the backend was a SAN backend. I could see a multi-bussed, high spindle count array doing a great deal of legwork to that sort of IO demand. I just can't see any ATA devices coming anywhere, in the realm of reason, near those numbers. The amount of memory for those arrays to process realtime would be outrageous.

Let me clear this up for you. Not a few thousand files at "the very same time" but consecutively. One after the other after the other and on and on. Non-stop. The data that is collected using special algorithm's on each 256MB file is enormous and requires a lot of hard drive space. I never said the 4 drives were moving a quarter to half a terabyte at once. I did not make it clear and you assumed. Does it make more sense this way?

again

A few thousand 256MB files at a time.

I asked for clarification on your poor wording. I fully understand concurrent versus consecutive operations. Using the wording "At a time" implies parallel operations. I assumed due to the non-standardized methodology and limited scalability in the way the application is designed, that it is a stove piped project. Seeing how I outlined how it logically should work by parsing portions into memory to work on, and not all at once, I really fail to see how I assumed anything. Frankly, it looks like I was spot on in the assessment of how the application moves the data.

 

[tiap]

Originally posted by: Bingo13
840EE with Gigabyte G-Power PDU21 at 3.6. 45c idle, 57c load. Not much more than my FX57.

Same temps here with a PD830 in an Antec P180 Gigabyte 955Royal stock Intel heatsink with pad
Stable and doesn't throttle.

Originally posted by: aka1nas
They were for a client who is doing heavy video encoding and of course they wanted to go Prescott, so I talked them into a dual-Core setup as it sort of made sense and was a minor change to the specs my coworker had made up.

I will take a look at that Swiftech Cooler and the XP120. The case, as mentioned above, is a Coolermaster Centurion with a 120mm exhaust and an 80mm front intake.

This machine is used for vids and tmpge uses dual core, works great, you don't want to crash an hour into an encode.
Don't worry about actual temps too much, it's Intels chip, so go by their specs, not amd's. Ive been using one of the first Prescott 2.8 800fsb cpu running at 65C load without probs or throttling.
I do highly recommend the P180 and a possibly a better cpu cooler if you want.

 

[Pabster]

ThrottleWatch is an easy way to check for throttling.

But beware, on the D series, it only monitors the first core. If the second core is throttling, you won't be able to tell with TW.

 

[aka1nas]

Intel's Spec for the PD830 is max temperature of 69.8C, so these machines are going over on load. Thankfully, the new heatsinks wil show up soon and we can get those temps down.

 

[Duvie]

Originally posted by: Pabster
ThrottleWatch is an easy way to check for throttling.

But beware, on the D series, it only monitors the first core. If the second core is throttling, you won't be able to tell with TW.

rumor has it open 2 cpu_zs and one of the cores or if not both will have the multiplier change dynamically...

 

[aka1nas]

I can confirm that 2 CPU Z instances were showing this when I was running stress tests on the Pentium D machines. They only dropped 1 multiplier though(from 15x to 14x). The 2 cores would switch independently of each other.

 

[Pabster]

Interesting.

 

[Bingo13]

Originally posted by: Pabster

Originally posted by: Bingo13
840EE with Gigabyte G-Power PDU21 at 3.6. 45c idle, 57c load. Not much more than my FX57.

Nice!

I probably should mention I have two 7800GTX cards in the board also, although the Pentiums run "hot" compared to their AMD counter parts I have had less issues with this setup than my FX57 when it comes to fsb/memory sensitivity due to voltage increases. I am running the 840EE at stock voltages across the board at 3.6, it will do 4.0 with some juice but the load temps will hit 68c or better at times.