As most people know, Intel's "Ivy Bridge" is supposed to be the successor of the previous "Sandy Bridge" cpu architecture. A lot of people who like to overclock their cpu's have noted that they have problems with heat even though IB is supposed to be cooler than SB considering that it has a smaller cpu die and runs on less voltage. The issue lies beneath the IHS (the heat spreader). Intel has used cheap thermal paste that must be replaced beneath the IHS, or else you will continue you have heat problems. Here is a video (not by me) that goes through the steps and shows the results of replacing the cheap silicon paste that Intel used:
If anyone has worked on PS3's, its basically the same process as when you are cleaning and replacing the thermal paste under the IHS on the GPU.
EDIT: if you go to the actual video on youtube, the person who made the video has time skips setup for different parts of the video (ex. Temps before XX:XX, Removing IHS XX:XX, etc..)
Such a stupid way for Intel to cut corners. It reminds me of my c2d where I had to lap the IHS with sandpaper to get a decent temperature because it was concave as fuck.
On November 23 2012 02:46 HotCookies wrote: very helpful thanks for mentioning it.
No problem bro. Anything to help out.
On November 23 2012 03:07 Aando wrote: Such a stupid way for Intel to cut corners. It reminds me of my c2d where I had to lap the IHS with sandpaper to get a decent temperature because it was concave as fuck.
Yea I feel you. Its the same issue with 360's and ps3's. Companies want to cut corners and save money, but they aren't realizing that the customer is suffering. I wonder if doing this could fix issues with some AMD cpu's as well although I highly doubt it.
On November 23 2012 03:16 a176 wrote: this has been known for a ltitle while. but this specific video, 90c, really? i cant believe that.
I thought that the cold obsession has been already dealt with, even on this particular forum. 90 degrees for a modern Intel CPU is completely normal. I have an i5 in my laptop and i goes to 90-95 degrees everytime something mildly interesting happens on the computer.
It is not agaist any specification to run at this temperature. All the current Intel processors have integrated safeguards that shut the CPU off the milisecond the temperature actually reaches a remotely dangerous. Mind you, the CPU is not made of meat, but silicon.
On November 23 2012 03:16 a176 wrote: this has been known for a ltitle while. but this specific video, 90c, really? i cant believe that.
I thought that the cold obsession has been already dealt with, even on this particular forum. 90 degrees for a modern Intel CPU is completely normal. I have an i5 in my laptop and i goes to 90-95 degrees everytime something mildly interesting happens on the computer.
It is not agaist any specification to run at this temperature. All the current Intel processors have integrated safeguards that shut the CPU off the milisecond the temperature actually reaches a remotely dangerous. Mind you, the CPU is not made of meat, but silicon.
The fact that its a desktop cpu WITH an aftermarket cooler and still runs a 90c is silly though. Im running a Pentium Dual Core right now that I oc'd from 1.6ghz to 3.2ghz (a 200% oc) running at 1.325v and it doesnt get that hot (on stock cooling). highest it gets is 70c and thats under p95 stress test. Sorry, but laptop cpu's generally run hotter anyway because of the terrible cooling. A desktop cpu shouldnt be above 80c unless under extreme conditions.
On November 23 2012 03:16 a176 wrote: this has been known for a ltitle while. but this specific video, 90c, really? i cant believe that.
I thought that the cold obsession has been already dealt with, even on this particular forum. 90 degrees for a modern Intel CPU is completely normal. I have an i5 in my laptop and i goes to 90-95 degrees everytime something mildly interesting happens on the computer.
It is not agaist any specification to run at this temperature. All the current Intel processors have integrated safeguards that shut the CPU off the milisecond the temperature actually reaches a remotely dangerous. Mind you, the CPU is not made of meat, but silicon.
i can imagine 80-90 using the stock low profile intel cooler under extreme load. but definitely not for a tower cooler. and especially one in open air!
On November 23 2012 03:16 a176 wrote: this has been known for a ltitle while. but this specific video, 90c, really? i cant believe that.
I thought that the cold obsession has been already dealt with, even on this particular forum. 90 degrees for a modern Intel CPU is completely normal. I have an i5 in my laptop and i goes to 90-95 degrees everytime something mildly interesting happens on the computer.
It is not agaist any specification to run at this temperature. All the current Intel processors have integrated safeguards that shut the CPU off the milisecond the temperature actually reaches a remotely dangerous. Mind you, the CPU is not made of meat, but silicon.
Well, Tj. max for an Ivy is 105c and it's lower for a Sandy (100c for the mobiles if I remember correctly and something like 95-98 for the desktops ) so if you have temps at 95c whenever something "mildly interesting" happens I'd definitely take a look at that.
I am surprised this is not common practice for overclockers. Before I went mac we used to replace the paste on everything including video cards. I do imagine nvidia got pretty good at that after Fermi tho^^
On November 23 2012 07:19 ToKoreaWithLove wrote: I am surprised this is not common practice for overclockers. Before I went mac we used to replace the paste on everything including video cards. I do imagine nvidia got pretty good at that after Fermi tho^^
Well, first of all it's a bit risky and can damage the CPU if you're a bit clumsy. And secondly the gains aren't really that high. With a decent tower cooler you can do a good OC and still very low volt, most CPUs doesn't go much higher anyway. Mine runs fine in 4.2ghz and 1.0v for example. Very cool, never over 70C in stresstests. And I run it without a fan... The temperature differences between sandy bridge and ivy is only critical if you want to go over 4.5ghz on the 50% of the CPUs that can (the volt increases skyrockets the temps on ivy).
On November 23 2012 02:10 LgNKami wrote: As most people know, Intel's "Ivy Bridge" is supposed to be the successor of the previous "Sandy Bridge" cpu architecture. A lot of people who like to overclock their cpu's have noted that they have problems with heat even though IB is supposed to be cooler than SB considering that it has a smaller cpu die and runs on less voltage. The issue lies beneath the IHS (the heat spreader). Intel has used cheap thermal paste that must be replaced beneath the IHS, or else you will continue you have heat problems. Here is a video (not by me) that goes through the steps and shows the results of replacing the cheap silicon paste that Intel used:
If anyone has worked on PS3's, its basically the same process as when you are cleaning and replacing the thermal paste under the IHS on the GPU.
EDIT: if you go to the actual video on youtube, the person who made the video has time skips setup for different parts of the video (ex. Temps before XX:XX, Removing IHS XX:XX, etc..)
is this still an issue with the newer batches? it was some sort of an issue when ib first came out but the discussions about it somewhat died down so i thought the issue was resolved. some forums that i go to generally just went for lower overclocks since it is still slightly better than sandy chips.
That vid is pretty jank he does a bunch of things i wouldn't recommend while taking off the spreader, also you don't really need to put the spreader back on, assuming you have the right type of cooler just putting it directly on the cpu die works just fine and avoids the spreader which mobile cpu platforms been doing for ages and amd gpu's no longer use head spreaders. At worst you use a shim essentially what the heat spreader is but you can get one out of pure cooper instead of what the ihs is, nickel plated copper, or you just have to run a small mod on your mobo, but fuck it you're ripping the ihs off your expensive cpu why not go the extra mile. It's intels fault anyways for not soldering the cpu to the ihs
On November 23 2012 11:34 semantics wrote: That vid is pretty jank he does a bunch of things i wouldn't recommend while taking off the spreader, also you don't really need to put the spreader back on, assuming you have the right type of cooler just putting it directly on the cpu die works just fine and avoids the spreader which mobile cpu platforms been doing for ages and amd gpu's no longer use head spreaders. At worst you use a shim essentially what the heat spreader is but you can get one out of pure cooper instead of what the ihs is, nickel plated copper, or you just have to run a small mod on your mobo, but fuck it you're ripping the ihs off your expensive cpu why not go the extra mile. It's intels fault anyways for not soldering the cpu to the ihs
I 100% agree. The video is very sloppily done and he did a lot of things that I do not agree with, but it gives people an idea about what to do with their stuff.
There's a thread that goes in depth describing the issue, personally I wouldn't recommend replacing the thermal paste before you're sure you're having an issue with overheating, like i said personally I haven't had any issues at all.
here's a quote from an intel rep from the newegg forum
Sorry I didnt get back on this before. The reason for the change to TIM (Thermal Interface Material) was a cost move. This move had to do with our overall processor line up and keeping the costs as low as reasonable without impacting performance. For the great majority of our processors this move doesn't have any impact at all on performance.
However when it does come to overclocking as some of you have pointed out you are not getting as high of clock speed when overclocking as you did from the 2nd generation Intel® Core™ processors. Most reports in the enthusiast space found that the with the Intel Core i5-2500K about 80% of them would top off at 4.5GHz unless you got into extreme cooling. Most people are finding with the Intel Core i5-3570K are finding that you can reach 4.2GHz to 4.3GHz on the 3rd generation Intel Core i5-3570K without changing the voltage at all. Since it is the voltage that is really leading to the heating issues this means that you are getting a processor that is running at a reasonable temperature without turning to higher end cooling.
Due to IPC (Instructions Per Cycle) improvements on the 3rd generation Intel Core processors you are getting right around a 6% improvement over the 2nd generation in overall performane. So at 4.3GHz you are getting performance that is right at the level that you would from the 2nd generation processors.
In the end as reported on Tomshardware, Anandtech and HardOCP the performance of the 3rd generation Intel Core processors when overclocking is basically a "wash" with the 2nd generation processors when overclocking.
What the future may hold with our processors I don't know but I will say without question that the enthusiast space is important to us and we are going to continue to try to make the best processors that we can.
On November 23 2012 03:16 a176 wrote: this has been known for a ltitle while. but this specific video, 90c, really? i cant believe that.
I thought that the cold obsession has been already dealt with, even on this particular forum. 90 degrees for a modern Intel CPU is completely normal. I have an i5 in my laptop and i goes to 90-95 degrees everytime something mildly interesting happens on the computer.
It is not agaist any specification to run at this temperature. All the current Intel processors have integrated safeguards that shut the CPU off the milisecond the temperature actually reaches a remotely dangerous. Mind you, the CPU is not made of meat, but silicon.
The fact that its a desktop cpu WITH an aftermarket cooler and still runs a 90c is silly though. Im running a Pentium Dual Core right now that I oc'd from 1.6ghz to 3.2ghz (a 200% oc) running at 1.325v and it doesnt get that hot (on stock cooling). highest it gets is 70c and thats under p95 stress test. Sorry, but laptop cpu's generally run hotter anyway because of the terrible cooling. A desktop cpu shouldnt be above 80c unless under extreme conditions.
Yes, it is silly, in the sense that it is probably a fault somewhere in the cooling and could be made better. The question is, is there a point in doing it, when the CPU can run just fine on this temperature?
On November 23 2012 03:16 a176 wrote: this has been known for a ltitle while. but this specific video, 90c, really? i cant believe that.
I thought that the cold obsession has been already dealt with, even on this particular forum. 90 degrees for a modern Intel CPU is completely normal. I have an i5 in my laptop and i goes to 90-95 degrees everytime something mildly interesting happens on the computer.
It is not agaist any specification to run at this temperature. All the current Intel processors have integrated safeguards that shut the CPU off the milisecond the temperature actually reaches a remotely dangerous. Mind you, the CPU is not made of meat, but silicon.
Well, Tj. max for an Ivy is 105c and it's lower for a Sandy (100c for the mobiles if I remember correctly and something like 95-98 for the desktops ) so if you have temps at 95c whenever something "mildly interesting" happens I'd definitely take a look at that.
But this is exactly it - Tj for a desktop Ivy really is 105. That means that 90 is full 15 degrees below that. Can you please point me to any source of Tj being 95 for mobile Sandies? I always thought it is flat 100.
Anyway, I have a pre-sandy CPU, which has 105 so I am golden
Intel CPUs before Ivy Bridge used to throttle at a temperature far below the shutdown temperature, usually in the 80-85C range. So the CPU doesn't run "fine" at 90C - it runs very slowly. That may have changed with Ivy Bridge, but I'd be surprised.
The engineering logic behind this is that a CPU running at higher temperature needs more voltage to run at the same frequency. By implementing a throttle temperature, CPU manufacturers can increase the rated frequency and/or lower the stock voltage while maintaining stability. Turbo boost complicates that a little, but the principle remains.