Sign in with
Sign up | Sign in
Your question

New Nas CPU - i5 Sandybridge, ivybridge, or i3 ivy?

Last response: in CPUs
Share
August 5, 2012 6:38:03 PM

Going to upgrade my NAS with a bit more storage using ZFS which is RAM and CPU intensive, especially once de-dedupe is more common and turned on in future freenas builds.

Fry's had an i5-2400 for $118 which got me thinking quad core but the TDP was high, can't decide if too high.

I have no need to overclock but may undervolt/underclock a bit to keep it nice and cool in an ITX case. Do I need to go with an unlocked processor to really have the last bit of control like a 3570K?

What's the real world difference between a 3450 and a 2400 in terms of heat in a small case?

August 5, 2012 6:56:39 PM

This link will help you understund why i say no to ivy cpu's

http://giatrakis.wix.com/john-giatrakis#!home/mainPage

in terms of heat in a small case as you said is NO way anyway...

Find a motherboard with onboard grapchic controller and deactivate it from the bios when you done installing to reduce the heat even more.
m
0
l
a c 116 à CPUs
August 5, 2012 7:49:48 PM

mstang1988 said:
What's the real world difference between a 3450 and a 2400 in terms of heat in a small case?

The 3450 (75W) has lower TDP which means less heat (~20W less) for the HSF and VRM to dissipate. You could also aim for the 3470S/3450S (65W) to shave yet another 10W or 3470T (35W) to shave 40W more. 30-60W less in what would otherwise be a ~130W SFF system should make a fairly noticeable difference in case temperatures if it has somewhat limited airflow.

As far as undervolting goes, this depends on the motherboard.
m
0
l
Related resources
August 5, 2012 8:01:31 PM

InvalidError said:
The 3450 (75W) has lower TDP which means less heat (~20W less) for the HSF and VRM to dissipate. You could also aim for the 3470S/3450S (65W) to shave yet another 10W or 3470T (35W) to shave 40W more. 30-60W less in what would otherwise be a ~130W SFF system should make a fairly noticeable difference in case temperatures if it has somewhat limited airflow.

As far as undervolting goes, this depends on the motherboard.


There is no way to cool down these cpu's what part of:

(Link Removed by moderator)

You don't understund?

In a NAS system there is no room for massive cpu coolers anyway, 2 NF-P12 (2 x 1,08 Watts) are more than enough if he deactivates the onboard graphics card he can connect it to his pc and working it as a storage from there anyway.
m
0
l
a c 116 à CPUs
August 5, 2012 8:15:58 PM

Giatrakis said:
You don't understund?

You are the one who does not understand that Ivy Bridge is perfectly fine. Millions of those CPUs are out there and the world has not gone through thermonuclear meltdown. Hundreds of people on these forums bought one and are perfectly happy with their IB as well.

So please just quit heat-trolling.
m
0
l
a b à CPUs
August 5, 2012 8:22:43 PM

InvalidError said:
You are the one who does not understand that Ivy Bridge is perfectly fine. Millions of those CPUs are out there and the world has not gone through thermonuclear meltdown. Hundreds of people on these forums bought one and are perfectly happy with their IB as well.

So please just quit heat-trolling.


You have to give him 10/10 for trolling though he even went away for a week and made his own website!
m
0
l
August 5, 2012 9:00:08 PM

I used more close to the earth examples "the road to cpu cooler", "highways" but my english are not very best and i know it.

I asked Pinhedd's help that is always more accurate than me and toled me he will rewrite some parts...

What else can i do?
m
0
l
a c 190 à CPUs
August 5, 2012 9:15:09 PM

Heat is heat, temperature is temperature, running hot doesn't mean it made more heat. Temperature is a function of heat generated, cooling performance, and ambient temperature so while ivy may seem to run hotter that would be because of worse cooling performance the total heat generated is less so the air temperature immediately around it will be higher but left in a sealed box it will heat up the air slower than a sandy bridge CPU.


Also, your page has some issues. Capacitors don't cease to function properly above their rated temperature, they just age faster. 85C capacitors will still work at 105C they will just age about 4x faster(aging rate approximately doubles for every 10C above the rating), also the caps around the CPU aren't getting anywhere near 90C when the CPU is at 90C, the fins on the heatsink are likely only around 50C since there is a temperature gradient in the heatsink that progress from the hotpoint(the base) to the cooler points(the tips of the fins) the air around the heatsink is no where near the temperature of the CPU itself so that piece is a chunk of FUD.


Also, thermal camera pictures with no scale shown to tell you how hot each color is are meaningless, cameras change their scale depending on the hottest item in the picture to provide you with contrast. Dark red is the lowest and dark blue is the highest in each picture but i have no idea if red is 20 C and blue is 30 C or if red is 0 C and blue is 200 C without a scale.



As for the actual topic at hand, reducing the TDP of the CPU used means that you don't need as much airflow to keep temperatures at a rational level, undervolting will further reduce it. The Phenom IIs could get a signficant drop in TDP from undervolting(they were set much higher than needed), I am not sure about how much of an effect it has on ivy bridge but reducing voltage will always reduce heat output so it is worth a shot.
m
0
l
a c 79 à CPUs
August 5, 2012 9:28:31 PM

Giatrakis said:
I used more close to the earth examples "the road to cpu cooler", "highways" but my english are not very best and i know it.

I asked Pinhedd's help that is always more accurate than me and toled me he will rewrite some parts...

What else can i do?


learn some basic physics please, specificaly around temperature Vs Heat energy. I have my 3570k idling at 30-33, maxing at 60-62 @ 4.3Ghz on cheap air.

in a small case the number of air changes per minute can be a lot higher, its air changes that gets rid of the heat build up, so it can run happily in a small box.
m
0
l
a b à CPUs
August 5, 2012 9:33:03 PM

it is amazing how many people do not understand heat-temperature relationship. Nice post Hunter315.
-Bruce
m
0
l
August 5, 2012 9:40:41 PM

Thanks for the debate guys! My question was more academical. Intel typically rates a class of CPU's at a given TDP and they will often perform under depending on where they rank in their class, at least that's my observation. I was interested in real world observations of these CPU's and their characteristics.

The guy that hates IB can leave the discussion.

m
0
l
August 5, 2012 10:43:33 PM

13thmonkey said:
learn some basic physics please, specificaly around temperature Vs Heat energy. I have my 3570k idling at 30-33, maxing at 60-62 @ 4.3Ghz on cheap air.

in a small case the number of air changes per minute can be a lot higher, its air changes that gets rid of the heat build up, so it can run happily in a small box.


Run a prime 95 torture ONLY 4 THREADS, 6000 bytes (6Mbytes) minimal fft, half the ram of your system ussage and post me a screen shot.

I need only system, cpu temprature from speedfan. But i did warn you i know something you don't ok?

(Link Removed by moderator)

even indigo 2 extream faild to these cpu's....
m
0
l
a c 109 à CPUs
August 5, 2012 10:48:15 PM

Giatrakis said:
Run a prime 95 torture ONLY 4 THREADS, 6000 bytes (6Mbytes) minimal fft, half the ram of your system ussage and post me a screen shot.

I need only system, cpu temprature from speedfan. But i did warn you i know something you don't ok?


And your speedfan shot says 'Intel core 2 quad' which is not ivy bridge by far....
m
0
l
August 5, 2012 10:49:58 PM

Yes it was 105 watts processor and i asked HALF of that processor was doing... were is the problem?

I mean he wrote "I have my 3570k idling at 30-33, maxing at 60-62 @ 4.3Ghz on cheap air." There is no problem? you bet his 79 watts cpu can't handle 4 threads with his cache full and half the ram? I know it.

Ps I'm still trying to find a cooling solution for ivy's cpus when i will get finaly Succeed i will post in the web page for sure trust me, what to post the failures? Amuffin what do you think about this one?
m
0
l
a c 79 à CPUs
August 6, 2012 6:24:52 AM

Giatrakis said:
Yes it was 105 watts processor and i asked HALF of that processor was doing... were is the problem?

I mean he wrote "I have my 3570k idling at 30-33, maxing at 60-62 @ 4.3Ghz on cheap air." There is no problem? you bet his 79 watts cpu can't handle 4 threads with his cache full and half the ram? I know it.

Ps I'm still trying to find a cooling solution for ivy's cpus when i will get finaly Succeed i will post in the web page for sure trust me, what to post the failures? Amuffin what do you think about this one?


will respond tonight. Although I usually use IBT/linpack as its hotter. Any particular version of prime 95?

It was video encoding all day yesterday should I wait for a few days before it is cool enough to use again?
m
0
l
August 6, 2012 8:37:23 AM

13thmonkey said:
will respond tonight. Although I usually use IBT/linpack as its hotter. Any particular version of prime 95?

It was video encoding all day yesterday should I wait for a few days before it is cool enough to use again?


You can follow the steps 1 2 3:

http://mersenne.org/freesoft/

I don't know how many days you must wait untill your cpu to cool down is usually getting 1-2 minutes...

Do you want me a link for the speed fan?

http://www.almico.com/speedfan.php

You know how to press the printscreen right?

m
0
l
a c 79 à CPUs
August 6, 2012 5:43:52 PM

you see the 2.67 version of p95 uses additional functionality on the ivb chips, making them run hotter, 2.66 and earlier run cooler. I didn't want you to claim it was invalid for me using the wrong version.
m
0
l
August 6, 2012 5:48:06 PM

13thmonkey said:
you see the 2.67 version of p95 uses additional functionality on the ivb chips, making them run hotter, 2.66 and earlier run cooler. I didn't want you to claim it was invalid for me using the wrong version.


I don't claim something I allready don't know and the fact is your cpu can't handle 4 threads with your cache full and the half ram of the system without getting 90 c to the cores (-15 c from the die temprature) and way above 105 to the system use any prime95 you like the same result over and over again and you can't do nothing to stop it.

I know why and i post the link for you the thing is when you finally realize it.

PS i tell you what try a little you will see it then stop it no way to make 2 hours with these tempratures... i mean really really i search any possible solution my self beacouse this cpu is no go for me
m
0
l
a c 79 à CPUs
August 6, 2012 5:58:16 PM

What clocks do you want it run at? stock or OC? Will do it in a few hours, how long do you want it run for?
m
0
l
August 6, 2012 6:01:46 PM

Even in stock it makes no diffrence read this carefull and tell me what part is not clear to you

(Link Removed by moderator)
m
0
l
a c 79 à CPUs
August 6, 2012 6:05:12 PM

I know what I know, and I know that ivb works fine. I also know that you don't know all that you think you know.
m
0
l
August 6, 2012 6:10:10 PM

Then run a test with any prime you like and post me a screen shot.

prime 95 torture 4 threads ONLY minimal/maximal fft all the cpu cache 8192 (8MBytes) + half your system ram.

m
0
l
a c 116 à CPUs
August 6, 2012 6:13:23 PM

13thmonkey said:
What clocks do you want it run at? stock or OC? Will do it in a few hours, how long do you want it run for?

He's the resident thermal troll, keeps repeating that Ivy Bridge CPUs run hot enough to initiate nuclear fusion; I doubt you will manage to make him shut up no matter how much proof you collect and present. There are already plenty of people running IBs well under 80C full-load, often less than 70C but he keeps saying it is impossible to run IB at less than 90C.
m
0
l
August 6, 2012 6:16:07 PM

Yes i know you let him run it in stock speed and show us..
m
0
l
a c 79 à CPUs
August 6, 2012 6:18:20 PM

I know, worth 30mins on prime to see what rubbish he comes up with. All day encoding 100% usage, + SWTOR (its rubbish) and nothing over 60C.

In a small case there should be no problem, as the airflow will be that much better. It might also help the OP make a call with some actual data, unless he's given up due to the back and forth.
m
0
l
August 6, 2012 6:19:42 PM

prime 95 torture 4 threads ONLY minimal/maximal fft all the cpu cache 8192 (8MBytes) + half your system ram.

speedfan and i will post this link in my web page with special 10nx to 13thmonkey that did the test for me.

task manager memory usage, cores , cpuz

stock speed

PS forget the link i will post the screen shot to my web page directly we are trolling enythig else except the real test for hours...
5 mins will be enough come on... Nikor did it fasterrrr
m
0
l
a c 79 à CPUs
August 6, 2012 7:57:47 PM

peaked at 66, fan at 70%, most of the time just above 60 on 1 core, 8GB custom using large and small FFT's. @4.3Ghz, because I don't want to fiddle with bios.
m
0
l
August 6, 2012 8:06:43 PM

i can't see the system temprature, The 8192 in the fft is 720K are you jocking? in the speed fan i see only 2 hds what happen to the system? all this time for this? and you like to put in in my wabe page!!!

lets go prive ok ?



in my image you see clear 4096K fft with 270k fft is 45 50 c to this processor and my image you can zoom and you see the 400 watts load.. BTW the cpu had 54 c in the bios... and tyou have 60 wiuth 270 kbytes ussage ? i USE 4 Mbytes pal are you blind?
m
0
l
a c 79 à CPUs
August 6, 2012 8:23:20 PM

expert fan shows package temp at 41, mobo at 34, core temps are 58,62,61 and 62. Task manager shows 10.4GB in use. The fft size bears no relation to ram usage. 177W at the wall. GTX470 idling.

Whats not to understand?
m
0
l
August 6, 2012 8:27:10 PM

YOU have 270kb fft i toled you minimal 8196 maximal 8196 the prime 95 will show 8196K is the same ?

ABOVR is is 4 Mbytes fft what 270 kbytes? is notthing to me a 270 fft no more 45 to the cores..


YOY are an Addict in Tom's Hardware do a torture test with minimal fft = 8196 maximal fft = 8196 4 threads only only 10 minites and need full image enough crap 2 days now...
m
0
l
a c 79 à CPUs
August 6, 2012 8:29:22 PM

and not in gibberish that is? you said minimal fft. if you look up and down its done large and small ffts. pick one i'll do it again, pick wisely.
m
0
l
August 6, 2012 8:35:08 PM



are you having problem with prime95 ? 8 kbytes and 270 kbytes is the same with 4096 kbytes ? is 4 Mbytes fft are you blind?

and addict in tom's hardware 2 days now can't run a custome prime95 test,....
m
0
l
a c 79 à CPUs
August 6, 2012 8:37:30 PM

being insulting on purpose? be back in 30mins.

and you instructions as to what settings were rubbish.
m
0
l
August 6, 2012 8:39:51 PM

i will ask deeply sorry for any word i said but you post me an image with 270 kb fft

i wrote prime 95 torture 4 threads ONLY minimal/maximal fft all the cpu cache 8192 (8MBytes) + half your system ram.

5 minutes no more.... and speed fan dosen't giving the system is printing the gpu temprature 2 times...

270 kb ftt in the above machine is NOTHING 45 c and you have 60 c and you are ok?

NIkorr in my web page got right he did it in secs....
m
0
l
a c 79 à CPUs
August 6, 2012 8:51:05 PM

i have 2 gpus, not sli, on later
m
0
l
August 6, 2012 8:51:49 PM

i have 36 c to the system and 54 c to the cpu and is NOT running only prime...

13thmonkey if you want an image of yours in my wabe page and special 10nx written do it straight and i wand FULL image to be clicable and zoom... use [ img ] not [ url ]

USE CPUID harware monitor i need to see the system is your motherboard temprarture btw...

PS do you want me to give you my msn accound i have camera in english i'm far more better...

the fft length was 8 Kbytes above btw.... i mean.... don't do 30 minutes again 5 minutes you are dead with the tempraters that you giving...
m
0
l
a c 79 à CPUs
August 6, 2012 9:43:52 PM

Giatrakis said:
i have 36 c to the system and 54 c to the cpu and is NOT running only prime...

13thmonkey if you want an image of yours in my wabe page and special 10nx written do it straight and i wand FULL image to be clicable and zoom... use [ img ] not [ url ]

USE CPUID harware monitor i need to see the system is your motherboard temprarture btw...

PS do you want me to give you my msn accound i have camera in english i'm far more better...

the fft length was 8 Kbytes above btw.... i mean.... don't do 30 minutes again 5 minutes you are dead with the tempraters that you giving...


hosting images
m
0
l
August 6, 2012 9:45:54 PM

i cant see noting again how come the image i gave above is so clear and you can see enything? you don't have to it 30 minutes with the numbers that we are playing you get high fast 5 minutes are enough maybe you need to change the DPI in the letters to be bigger....

i cant zoom in that image anyway can we go prive? in there i can give my email to send it to me to see it.
m
0
l
a c 79 à CPUs
August 6, 2012 9:49:00 PM

click the second image
m
0
l
a c 109 à CPUs
August 6, 2012 9:49:58 PM

Click on the image.
m
0
l
August 6, 2012 9:50:53 PM

ok i got it and i zoomed you have safe tempratures lets go prive... i need more infos we finished from here...
m
0
l
a c 79 à CPUs
August 6, 2012 9:52:58 PM

59-63-62-57 max @4.3 @8k @ 8GB ram ran for 5mins, stable temps after first 10-15s
m
0
l
a c 79 à CPUs
August 6, 2012 9:53:37 PM

we'll have the discussion here. Thats how things are discussed and the truth comes to light.
m
0
l
a c 79 à CPUs
August 6, 2012 9:53:43 PM

Anyway, i'm going to get some sleep.
m
0
l
August 6, 2012 9:56:06 PM

read the prive i will put anything to the web page our nick your name whatever you like.

I need a photo from a thermometer, photos of your room the cooling system thermal paste was used we must be accurate and boulletprooff don't sleep now... I have a JOB and stopped for you trust me.
m
0
l
a c 79 à CPUs
August 6, 2012 10:07:43 PM

xigmatek achillies, icdiamond, about 20-21C in room. All in P180B, with 2x120mm on front (1 in 5-1/4 bays), 2x120mm at rear/top, positive pressure (probably).

No aircon in house, filters on fans, can run more tomorrow.
m
0
l
August 6, 2012 10:14:17 PM

we must do the test in 26-28 do you have a photo? *** the public nobady cares for a thread anything will be uploaded to the web page
m
0
l
a c 190 à CPUs
August 6, 2012 10:23:52 PM

You don't need to do the test in 26-28, an increase in the ambient temperature of 5 C will raise the load temperatures by 5 C assuming identical cooling and fan speed, what is actually going to happen in this scenario though is his fan is going to kick up another 5-10% and the temperatures are only going to increase by about 3 C.

There is nothing mythical about what is going on here, please read up on thermal systems, wikipedia is a good place to start
http://en.wikipedia.org/wiki/Heat_transfer

As newton's law of cooling puts it very simply,
"The rate of heat loss of a body is proportional to the temperature difference between the body and its surroundings."

If you increase the temperature of the room you will have the same increase in temperature of heated objects within the room if their cooling is the same at both ambient temperatures because that will maintain the temperature difference between the object and its surroundings, but computers are not simple hot objects, they have fan curves so unless you run the fan at 100% at 21 C and 28 C you aren't going to get a flat temperature increase, it is going to be less than the change in ambient. Since he has 30% of his fan speed remaining you aren't going to see a massive increase in temperature like you keep thinking he should based off testing a totally different and unrelated processor, now quit derailing threads and spreading FUD about Ivy Bridge.
m
0
l
August 6, 2012 10:33:58 PM

Assuming Giatrakis is NOT trolling which i think he ISN'T...I believe he just got a bad CPU...not even bad more like one in a thousand that runs considerably hotter than other 3570Ks

Could this be a possibility?
m
0
l
August 6, 2012 10:36:49 PM

I gave my photo, the photos of the room, thermometer in the incomings, I can't give all the things to public without these any data are unssuable do you belive the 20 c? but when i give 32 c like this:



nobody can tell is a fake... 32 c

to 13thmonkey i did send everyting we goanna do our tests together... and will be 100% accurate the data and booletproof just because someone said i have 20 in my room you will beliave him?

if he can't come to me with camera online and thermometers in the incomings live any data are unsuable to me.

So if the 20 c were true do the calculations again for 32 ambient. I HAVE 423 watts heat load without 2 open windows i can't managed no ***.. The air condition is uselles with such a load anyway....

When someone starts saying with this prime95 i have hotter cpu what you expect? Well and mine is hotter with new prime 95 but this means is more efficient and in the end nobody can be sure that he is trying really hard to deliver relaible data to the public..........


Ps plain thermometers not digital crap... The plain thermometer is using larger attached surface to the ambient and is always more accurate.
m
0
l
!