HD 4000 small graphical bugs / 3570K temperature concern

[idata]

I have an HD4000 connected to a 32" LG LCD in my bedroom via HDMI. The first issue I have is if I turn my PC on without my TV on or my TV on the cable input while the computer boots up, some of my desktop icons (most of them on the right side) get re-arranged.

The second issue I have is, I use a screen saver and when the computer is on and I switch to my cable TV input then back to my PC input (HDMI) while the screensaver is running, the right side of the screen (about 3 inches of the screen) shows my desktop while the rest of the screen shows the screensaver going.

I also have computer speakers connected to the onboard audio on my motherboard (VIA chipset). When I have my speakers on and I switch away from my PC input (HDMI) I hear a sound as if you were to disconnect a device from your PC, which is probably caused by the GPU losing signal with the display.

It seems to me that these two issues are related and that if you're not on the input the PC is connected to, it doesn't send a video/audio stream/signal so if I turn my PC on not being on the PC input, there's issues. If I switch inputs then switch back, there's issues. Both cases seem to be causing issues on the right side of the screen.

Is it possible to send a constant signal/stream through the HDMI output, even if there's no connection with the display? Or at least, can you fix these graphical issues that popup? It's not a big deal by any means, but it is annoying nonetheless.

I'm using WIndows 7 64-bit and the latest Intel graphics drivers on your website.

My second question is completely unrelated, but I have concerns about my 3570k temperatures. I'm using the stock cooler and I'm not overclocked but my temperatures running x264 video encoder causes my temperatures to reach 70c in a 22c room. I'm using RealTemp to check temperatures. That seems high for a stock system, even if I'm using the stock cooler. Should I be concerned at all?

[idata]

Just thought of something that may or may not explain why my CPU temps are a little higher than some others. If you use the onchip graphics, will that increase your CPU temperature read by RealTemp/CoreTemp etc? I'm sure I'm in the minority of people who actually use the onchip HD 4000.

While running x264, I'm obviously not stressing the GPU that much, but WIndows does have some visual effects that will put some load on the GPU.

[idata]

Good move on the CPU cooler. Lot's of decent ones out there. Good case air flow really effects the retail CPU coolers. There are less than a dozen cases that have perfect air flow while there quite a few more that have decent air flow. I've worked with more great looking cases with poor air flow than I can remember.

At any rate, worth checking your air flow. Hopefully you have at least one front intake or side panel fan blowing cool air on to the system. If not, it's worth adding one. Some cases have front fans that can't get cool air to the back. The easiest fix is retrofitting a 120mm cooling fan on to the drive cages facing the system using a couple zip ties. It'll pull more cool air back and shut those retail coolers (and most any cooler) right up.

[idata]

You're correct Daniel. Fully loading the iGPU for an extended time does heat the chip up a bit more. It basically adds another hot head to the cooler plate which is already working really hard to transfer heat from the limited die-IHS surface. Also, if you've installed Lucid Virtu MVP, the iGP is a bit active and "On" by default even if you're using a video card. If you're using any of the supported software labels, you're definitely heating up that CPU. But boy is it fast.

Also wanted to clarify, it's the TIM that was used between the IHS and CPU die that really affected the cooling properties. It was poorer quality and apparently not ideal for this size of die. The processors are shrinking, but the voltage and heat are still a constant issue based on the technology. Something still has to make a solid connection in order properly cool. That's why all the current stock going out went back to the original method.

[Xenon2000]

I did my test both open air no case and inside my HAF 912. Which currently has a 120mm intake and exhaust. And is extremely open flow with all the cabling behind the motherboard. Temps were the same with the stock cooler both in and out of the case.

Personally I thought BTX formfactor was the "perfect" airflow cooling setup. But that never took off. As for ATX cases, I think the HAF 912 that I have now and many like it have excellent air flow options with top and rear venting, etc. When I am done it will have either 2x120mm or 1x 200m top and front fans. Plus a side intake fan. Either way it will have a matched intake and exhaust setup. PSU is basically isolated being bottom mounted with it's own intake and exhaust. Some prefer positive pressure and some negative pressure, based on dust and static pressure on fans, etc. I will just go 1:1 since that is how this case is mostly setup.

I should be getting my replacement 3770K today. So it will be nice to test the stock temps again.

[Xenon2000]

"That's why all the current stock going out went back to the original method."

I didn't see a quote option in the forum, so I just copy/pasted your comment.

Do you have a link to share about this? Are you saying that only an original shipment of early Ivy Bridge CPUs were released with the TIM and that they have been returned and are now using the Fluxless Solder method? If that is true, I would love to read about that. And I would love to know how to confirm from my CPU serial number which one I have.

[plee21]

TechSergeant wrote:

Also wanted to clarify, it's the TIM that was used between the IHS and CPU die that really affected the cooling properties. It was poorer quality and apparently not ideal for this size of die. The processors are shrinking, but the voltage and heat are still a constant issue based on the technology. Something still has to make a solid connection in order properly cool. That's why all the current stock going out went back to the original method.

Hi

I have no over heating issues on Ivy Bridge, nor does anyone else using the processor as designed.

Over heating is only a problem if you push the CPU higher than it is designed for, i.e. over-clocking. Over clocking always causes heat issues for any CPU hence a whole industry around the sale of water coolers and over-sized/over priced chunks of metal. Isn't that half the attraction of over-clocking, the challenge of installing elaborate cooling solutions and getting it down by another few degrees so you can up the voltage a touch more and add a few more 100MHz?

Intel never said Ivy Bridge would over clock more than Sandy Bridge.

Whatever interface material Intel uses is used for a reason.

Just for anyone else reading here, there is no heat problem with Ivy Bridge, the only issue is expectations about over-clocking gained from assumptions before anyone saw an Ivy Bridge CPU just hasn't been met.

Regards

Phil

[Xenon2000]

How did you do a quote like that? I don't see a reply with quote option.

I know you are responding to TechSergeant, but I was having heat issues with the stock cooler at stock speeds.

I agree, there should be no heat issues until you significantly overclock and overvolt the cpu. But I was hitting 95-102C with a properly seated stock cooler at stock speeds. But I get the replacement cpu today, so hopefully it was the cpu.

I also agree that Intel would have a reason for using the TIM method they use for each CPU. But I would love to know what it was in this case. If it was mostly cost and there was little heat difference all stock, like within 2C, then fine. Though people have already argued that even a small difference is no excuse to save money on the K versions.

But if there is a large thermal difference, they better have a great reason that has nothing to do with price. And again, hopefully my first 3770K was just defective.

[plee21]

Hi

Shawn wrote:

How did you do a quote like that? I don't see a reply with quote option.

Click reply to the post you want to quote then click the quote icon in the edit tool bar.

Shawn wrote:

I agree, there should be no heat issues until you significantly overclock and overvolt the cpu. But I was hitting 95-102C with a properly seated stock cooler at stock speeds. But I get the replacement cpu today, so hopefully it was the cpu.

I also agree that Intel would have a reason for using the TIM method they use for each CPU. But I would love to know what it was in this case. If it was mostly cost and there was little heat difference all stock, like within 2C, then fine. Though people have already argued that even a small difference is no excuse to save money on the K versions.

But if there is a large thermal difference, they better have a great reason that has nothing to do with price. And again, hopefully my first 3770K was just defective.

That does seem very hot, hopefully it was the CPU. As long as the ambient temperature is around 38 degrees in the case or lower than the stock heat-sink and fan should be absolutely fine. It can take several days for the TIM on the heatsink to bed in so it can drop a bit over time.

Well everything comes down to price in the end . Perhaps the CPU just can't take the solder without damage or it's safer to have none solid means of interfacing to allow movement due to heat and cooling.

Good luck with the new CPU.

Regards

Phil

[idata]

If Intel truly is shipping the current stock back, then I'm kinda pissed quite frankly. Intel should offer up a free exchange program if that's the case. I bought the first batch of 3570ks.

[Xenon2000]

phil_l wrote:

Hi

Shawn wrote:

How did you do a quote like that? I don't see a reply with quote option.

Click reply to the post you want to quote then click the quote icon in the edit tool bar.

Shawn wrote:

I agree, there should be no heat issues until you significantly overclock and overvolt the cpu. But I was hitting 95-102C with a properly seated stock cooler at stock speeds. But I get the replacement cpu today, so hopefully it was the cpu.

I also agree that Intel would have a reason for using the TIM method they use for each CPU. But I would love to know what it was in this case. If it was mostly cost and there was little heat difference all stock, like within 2C, then fine. Though people have already argued that even a small difference is no excuse to save money on the K versions.

But if there is a large thermal difference, they better have a great reason that has nothing to do with price. And again, hopefully my first 3770K was just defective.

That does seem very hot, hopefully it was the CPU. As long as the ambient temperature is around 38 degrees in the case or lower than the stock heat-sink and fan should be absolutely fine. It can take several days for the TIM on the heatsink to bed in so it can drop a bit over time.

Well everything comes down to price in the end . Perhaps the CPU just can't take the solder without damage or it's safer to have none solid means of interfacing to allow movement due to heat and cooling.

Good luck with the new CPU.

Regards

Phil

I see, it's not listed as text. It's the little icon. I am used to the text button for that. Thanks for pointing that out here.

I put my details on page 1 so you likely didn't see it. So I will repeat. My room is 25C to 27C. My temps for the CPU where both in the out of the case. And no way my case is 38C inside. But regardless, same temps both in and out of the case.

[idata]

I have no over heating issues on Ivy Bridge, nor does anyone else using the processor as designed.

Over heating is only a problem if you push the CPU higher than it is designed for, i.e. over-clocking. Over clocking always causes heat issues for any CPU hence a whole industry around the sale of water coolers and over-sized/over priced chunks of metal. Isn't that half the attraction of over-clocking, the challenge of installing elaborate cooling solutions and getting it down by another few degrees so you can up the voltage a touch more and add a few more 100MHz?

Intel never said Ivy Bridge would over clock more than Sandy Bridge.

Whatever interface material Intel uses is used for a reason.

Just for anyone else reading here, there is no heat problem with Ivy Bridge, the only issue is expectations about over-clocking gained from assumptions before anyone saw an Ivy Bridge CPU just hasn't been met.

To Phil: Begging your pardon, that is the official marketing message. Can we assume you're speaking for Intel?

And to confirm to readers, there is no mass thermal problem with Ivy Bridge. However, we aren't overclocking nor discussing overclocking. More than just a few sources have confirmed that the Core i7-3770K specifically do heat up compared to something like Sandy Bridge Core i7-2700K under normal use. Whatever the reasons for IVB design, let's just say we're all extremely happy that the 3770K is only a 77W processor. I have a new retail 3770K in route for a little more confirmation.

Moving back to Shawn, it looks like there is plenty of air flow. I always focus on the basics and double check cooler mounting, thermal paste and contact. Personally, I use Arctic Silver 5 and aftermarket coolers on all my higher end Intel processors because I like them to be quiet, if not more silent. With my current 3770K, that's a little more of a challenge so I've had to do a little more optimizing. Hopefully, you get yours figured out.

[Xenon2000]

TechSergeant wrote:

Moving back to Shawn, it looks like there is plenty of air flow. I always focus on the basics and double check cooler mounting, thermal paste and contact. Personally, I use Arctic Silver 5 and aftermarket coolers on all my higher end Intel processors because I like them to be quiet, if not more silent. With my current 3770K, that's a little more of a challenge so I've had to do a little more optimizing. Hopefully, you get yours figured out.

Got the replacement. It's definitely cooler now.

Using IBT 2.53 (Intel Burn Test) and RealTemp 3.70

Previous 3770K: Stock speeds, stock cooler, room 25c, case open

Standard: 95c to 102c within a minute or so. Stopped at about 3 mins.

Maximum: 105c within 2 mins (after 16 GB is fully loaded) and then the cpu starts to throttle. Stopped when I saw throttling.

New 3770K: Stock speeds, stock cooler, room 26c, case open

Standard: 87c to 92c, but took about 5 mins to get there. Ran for 15 mins. Mostly stayed below 90c

Maximum: 91c to 97c, but also took about 5 mins to get there. Ran for 10 mins. Mostly stayed in the 87-91 range.

So I have no idea what was the deal with my previous 3770K and stock cooler. Maybe it was a defect in the copper mating surface of the previous cooler. But these temps make me happy as I know that IBT causes a lot of heat and I won't see this normally.

I did a 5 min test using the latest 64bit Blender 3D with the standard current benchmark blend file. And it utilized all 8 thread. Total utilization was 75-80%, temps ran an average of 75c. Max peaks which didn't happen often, was 80c.

So now that I am happy about my chip. My Noctua NH-D14 is on it's way. Should be here Tuesday. After that I will do my same test at stock speeds. While I normally use (and have) Arctic Silver 5, many people are showing tests that put the Noctua compound within 1 c of the AS5. So I will likely just use the Noctua paste that it comes with.

And now to buy the rest of my case fans and hopefully I have time next week to set my slight overclock and move on with my life. And I will likely buy the extra Intel warranty, though I likely won't need it as I plan on staying around 4.2 to 4.4 Ghz. Maybe 4.5 Ghz if the temps are decent.

[Xenon2000]

Daniel wrote:

Just thought of something that may or may not explain why my CPU temps are a little higher than some others. If you use the onchip graphics, will that increase your CPU temperature read by RealTemp/CoreTemp etc? I'm sure I'm in the minority of people who actually use the onchip HD 4000.

While running x264, I'm obviously not stressing the GPU that much, but WIndows does have some visual effects that will put some load on the GPU.

Right now I am only using the HD4000 iGPU. Using AIDA64 with RealTemp 3.70, if I choose all the stress tests other than GPU, I get about 85-90c for my coolest to hottest core. If I add GPU to the test, it goes up to 90-95c as my core ranges. In a room that is currently about 27c as I turned off the AC to make sure I could listen to the fan the whole time without interruption.

So it looks like a small 5c increase for maxing the GPU on top of the CPU. Not bad I don't think. Of course I don't have the tools to do a performance per watt analysis. And this is just 1 tool I tested for the GPU on top of CPU testing. I am happy.

[plee21]

Hi

TechSergeant wrote:

To Phil: Begging your pardon, that is the official marketing message. Can we assume you're speaking for Intel?

And to confirm to readers, there is no mass thermal problem with Ivy Bridge. However, we aren't overclocking nor discussing overclocking. More than just a few sources have confirmed that the Core i7-3770K specifically do heat up compared to something like Sandy Bridge Core i7-2700K under normal use. Whatever the reasons for IVB design, let's just say we're all extremely happy that the 3770K is only a 77W processor. I have a new retail 3770K in route for a little more confirmation.

Moving back to Shawn, it looks like there is plenty of air flow. I always focus on the basics and double check cooler mounting, thermal paste and contact. Personally, I use Arctic Silver 5 and aftermarket coolers on all my higher end Intel processors because I like them to be quiet, if not more silent. With my current 3770K, that's a little more of a challenge so I've had to do a little more optimizing. Hopefully, you get yours figured out.

I don't work for Intel. I'm merely pointing out that Intel haven't sold or promoted Ivy Bridge as being better for over-clocking, or that you will see lower temperatures when monitoring the core, so there is no point complaining to Intel about it

When using the CPU with the retail cooler and at it's rated speeds, in case that doesn't exceed 38 degrees, they work absolutely fine, all with the specification Intel have published and there is no over heating issue, and to my knowledge there has been no recall, this just seems to be scare mongering. (Yes there might be individual cases where a heatsink isn't fitted correctly or some other issue causes higher temperatures, but applies to any CPU.)

Also using a burn in program to test the CPU is expected to cause high temperatures and thermal throttling, that is all be design. This has been the case with Intel CPUs for a while, simply because the retail thermal solution and overall design of the CPU is designed for typical real world use. To have a retail thermal solution and CPU die cope with these sort of stress tests that see every part of the CPU ramped up to maximum, would mean that 99.999% of people are paying lots of extra money for absolutely no benefit in their real world use, just so a few people see low temperatures when stress testing. That makes no sense for anyone, except a select few, so the whole design is around cooling for real-world use.

Thermal throttling is happening day in day out as essentially Turbo Boost is nothing more than thermal throttling but sold positively as us getting extra speed, not less, but thermal throttling is what it is. So essentially the CPU only runs at its highest maximum rated speed when it is cool and has little work to do, as soon is it is pushed, it throttles down, if it gets hotter still, it throttles down even more.

CPUs heat up, they always have. Overall the heat emitted from Ivy Bridge while doing the same or even more work than Sandy Bridge is less by 15 or 20 watts. The "hot spot" reported by the diode on each CPU is reporting higher temperatures, no surprise given everything is closer together, but is still within specification. If the CPU overheats because it simply can't move the heat fast enough from the chip to the heatsink because it is under extreme workload, thermal throttling kicks in, which it is design to do, as that is all part of the cooling solution.

Regards

Phil

[plee21]

Hi

Got the replacement. It's definitely cooler now.

Using IBT 2.53 (Intel Burn Test) and RealTemp 3.70

Previous 3770K: Stock speeds, stock cooler, room 25c, case open

Standard: 95c to 102c within a minute or so. Stopped at about 3 mins.

Maximum: 105c within 2 mins (after 16 GB is fully loaded) and then the cpu starts to throttle. Stopped when I saw throttling.

New 3770K: Stock speeds, stock cooler, room 26c, case open

Standard: 87c to 92c, but took about 5 mins to get there. Ran for 15 mins. Mostly stayed below 90c

Maximum: 91c to 97c, but also took about 5 mins to get there. Ran for 10 mins. Mostly stayed in the 87-91 range.

So I have no idea what was the deal with my previous 3770K and stock cooler. Maybe it was a defect in the copper mating surface of the previous cooler. But these temps make me happy as I know that IBT causes a lot of heat and I won't see this normally.

I did a 5 min test using the latest 64bit Blender 3D with the standard current benchmark blend file. And it utilized all 8 thread. Total utilization was 75-80%, temps ran an average of 75c. Max peaks which didn't happen often, was 80c.

So now that I am happy about my chip. My Noctua NH-D14 is on it's way. Should be here Tuesday. After that I will do my same test at stock speeds. While I normally use (and have) Arctic Silver 5, many people are showing tests that put the Noctua compound within 1 c of the AS5. So I will likely just use the Noctua paste that it comes with.

And now to buy the rest of my case fans and hopefully I have time next week to set my slight overclock and move on with my life. And I will likely buy the extra Intel warranty, though I likely won't need it as I plan on staying around 4.2 to 4.4 Ghz. Maybe 4.5 Ghz if the temps are decent.

Hardly a big difference.

Each CPU is tested and calibrated with a set of voltages it must run at, so CPUs vary. You've probably just got a CPU that was slightly better than the first and so requires a bit less voltage, so when pushed you save a few degrees, although unlikely to give you much advantage when you start overclocking due the exponential nature of power usage. The difference might even be explained by slightly better fitting of the heatsink and thermal compound this time round.

Statistically you can't draw any conclusions there was anything wrong with the first CPU.

Regards

Phil

[Xenon2000]

phil_l wrote:

Hi

Hardly a big difference.

Each CPU is tested and calibrated with a set of voltages it must run at, so CPUs vary. You've probably just got a CPU that was slightly better than the first and so requires a bit less voltage, so when pushed you save a few degrees, although unlikely to give you much advantage when you start overclocking due the exponential nature of power usage. The difference might even be explained by slightly better fitting of the heatsink and thermal compound this time round.

Statistically you can't draw any conclusions there was anything wrong with the first CPU.

Regards

Phil

Are you kidding me? Hardly a difference? I would say that 95-105c with cpu throttling, vs mostly 87-91 and never throttling; is a HUGE difference. It's the difference of being able to use my PC or not.

I have been buying CPUs since the Cyrix days. Point is, a long time. And while any CPU or component can have failures or issues, this is the first time I have returned a CPU. Ever. It very well could have been the cooler that was the issue before. Likely it was not the CPU. But I only have 1 Ivy Bridge CPU, compatible motherboard, and Intel stock cooler. So I was not in a position to determine if it was the CPU or cooler. So I had to return the bundle.

And my new Bundle (CPU/Cooler) works as I expect. I expect IBT to cause excessive heat. That is why I use it. And with my new bundle and every CPU combo I have ever used it on, it has never caused throttling on a stock speed processor. Only with overclocking and determining how far I can push a large cooler. Which helps me determine where I want t leave my overclocking.

But the point is. Regardless of what Intel or anyone else says. I expect a stock CPU and stock cooler in a cool room of 25c or lower, running at stock speeds, to be able to handle an IBT stress test or other heavy stress test, without ever throttling. And my previous bundle didn't pass. And as I said before. Very likely the cooler was defective, not the CPU. But I had to return both to get my tests to pass. And that is a huge difference to me.

Oh, and of course I can't statistically draw any conclusions. I only have stats from 2 bundles and no cross testing of stock coolers. I can't prove anything. And I am not trying to prove anything. I found this thread because I knew my bundle was not right to be throttling during my testing. So I was searching to find info on temperatures with stock cooler and stock speeds. Shockingly, that is very difficult as almost everyone posting temps is using non-stock coolers and/or overclocking. Which I will be as well, but I like a baseline first.