Haswell Temp Readings

[Revolution 11]

4,2 Ghz Idle: 30C

4.2Ghz Load:

Hyper 212 Evo with 2 fans push/pull, Arctic Silver paste I7 4770k, G1 Sniper 5. Ambient room temp: 23C

Vcore 1.150. Anything above 1.2V gets really hot and the 212 can't keep up. I was going for 4.4Ghz, but I couldn't get stable under 1.2V

Please don't use Arctic Silver. It was good maybe 10 years ago.

That said, wow Haswell is hot. Would a better air cooler help or is the only solution is to delid?

 

[Kenmitch]

Nobody else brought it up so I thought I would at least.

CPU temp when viewed in uEFI/BIOS is not the idle temp as your core(s) are somewhat under load....Extent of load? Just no energy savings enabled I think.

 

[ShintaiDK]

Nobody else brought it up so I thought I would at least.

CPU temp when viewed in uEFI/BIOS is not the idle temp as your core(s) are somewhat under load....Extent of load? Just no energy savings enabled I think.

Some BIOS temperature measurements still uses an external sensor under the socket and adds a best guess offset to it.

 

[willomz]

212 Evo is a fine cooler for most systems but it can't handle any significant amount of heat:

A Phanteks PH-TC12DX is quite a modest air cooler and could bring temps down by 20 degrees.

 

[Kenmitch]

Some BIOS temperature measurements still uses an external sensor under the socket and adds a best guess offset to it.

If that is the case them the core temps would even be greater than shown.

I figure they are showing package temp if anything. Of course I could be wrong.

 

[willomz]

If that is the case them the core temps would even be greater than shown.

Well depends if the guess is too high or too low. You can be wrong either way.

 

[Redoitall]

4,2 Ghz Idle: 30C

4.2Ghz Load:

Hyper 212 Evo with 2 fans push/pull, Arctic Silver paste I7 4770k, G1 Sniper 5. Ambient room temp: 23C

Vcore 1.150. Anything above 1.2V gets really hot and the 212 can't keep up. I was going for 4.4Ghz, but I couldn't get stable under 1.2V

Those temperatures remind me of my pre delided 3770 K. I think that as long as you do not run your chips too close to the tj max you should not worry. however do worry about voltages these chips are very voltage sensitive. You could easily blow it with one run of cinebench. I tell you from experience I did it with one of my 3770k delided the temps were all good but went to 1.525 @ 5.00 ghz and my chip literally just wont work unless I have it on higher voltages. It will also wont run on offset any longer

 

[aigomorla]

So we are not talking desktop chips? Single socket Xeons like this?
http://ark.intel.com/products/75057/Intel-Xeon-Processor-E3-1280-v3-8M-Cache-3_60-GHz

I can assure you that you can find Haswell Xeons in blades soon, if not already.
http://www.redorbit.com/news/techno...t-density-server-solutions-with-coming-intel/

Cooling is not a problem either. Servers already contain jet fans that can run up to 15000rpm.

The Xeon 1200 V3 chips will carry either no onboard graphics, the GT1 chip or the GT2 chip . The GT2 chip that some of the Xeon 1200 V3 chips will carry is reportedly just as powerful as the Intel HD Graphics 4600 that we'll see on the consumer Haswell chips. The units with an integrated graphics chip will have the model name ending in a 5, rather than a 0, for those that don't carry integrated graphics.

Server line is getting the real haswell.
You desktop line got a fubard laptop version.

And ONTOP:

However, ChinaDIY recently got samples of the upcoming Haswell Xeon E3-1200 V3 processors and when equipping the new chips on a Z87 retail motherboard, the chip didn’t fit in the socket.

Meaning entirely new board.. because...

The reason for this lies in the alignment of the side notches that don’t fit in with the Z87 LGA 1150 motherboards.

which implies... its a totally different CPU if pins arent the same.

This could be a possible hint at Intel preparing a different socket interface for their Xeon E3-1200 V3 line up which would further be compatible with Broadwell line of Server CPUs later in 2014.

Yup... YAY~ were back on the poorman's board system... and the rich man board system again...
Only this time the poor man's cpu is NOT the same as the rich man's CPU. :\

So its no longer i7 vs Xeon.. but LGA20XX l7 vs LGA1150 i7 vs Xeon i7....
i7's were all suposed to be i7!!! not anymore.... It no longer applies anymore.

Read more: http://wccftech.com/intel-haswell-xe...#ixzz2VNTjMpw5


Dont inflat haswell.. its a flomp by intel every way i see it on the desktop line.

 

[ShintaiDK]

Server line is getting the real haswell.
You desktop line got a fubard laptop version.

And ONTOP:



Meaning entirely new board.. because...


which implies... its a totally different CPU if pins arent the same.


Dude... dont inflat haswell... Its not special... and this is coming from an intel fan.

Its the exact same chips. And the socket is the same. You can also see compability at quite alot of mobo makers.

You own link even states:

Update Via Computerbase: Apparently ChinaDIY were mistaken. At least in the pictures shown it is not a “Haswell” sample.The number of pins is not 1156 and 1150th Moreover, the figures “CM80616003180AG” on the heatspreader on an older ”Clarkdale” chip (Core i3-530) out. That this does not fit into the new socket LGA1150 is therefore hardly surprising.


Read more: http://wccftech.com/intel-haswell-x...compatible-z87-lga-1150-socket/#ixzz2VNTjMpw5

 

[aigomorla]

Its the exact same chips. And the socket is the same. You can also see compability at quite alot of mobo makers.

You own link even states:

i just googled what was the fastest... as the V3 series are still wraped in major amounts of NDA.

My friend who have the chips are saying the same thing.
The HASWELL we are all waiting for ... is not HASWELL... its more Broadcomm on a completely different board and setup.
This haswell is a laptop bin made for desktops.
Thats why the IHS doesnt even seat properly.... because LAPTOP grade processors have no IHS to begin with.

Intel is trying this model out and seeing how it goes.... this is why real servers arent seeing these chips.
And yes i do mean the bigger chips.. the ones which are suposed to replace LGA2011.

 

[allenk09]

212 Evo is a fine cooler for most systems but it can't handle any significant amount of heat:

A Phanteks PH-TC12DX is quite a modest air cooler and could bring temps down by 20 degrees.

Yeah I'm looking at a new cooler. This is just too much heat. I'd like to get a stable 4.4 overclock below 90C and with the Hyper 212 it's just not going to happen

 

[WhoBeDaPlaya]

Interesting experiments :

[Memory Speed / Voltage]

Used the same chip, 4.4GHz @ 1.2V (fixed, not adaptive), Antec 920, IC Diamond, Crucial Ballistix Elite 1866 9-9-9-27-2T 1.5V
All runs were using the XMP timings (fixed at 9-9-9-27-2T) - only the speed and voltage were changed. Verified to be error free in memtest86+ 5 RC1.
Highest core temperature during a 25-pass IBT run is reported, averaged across 2 runs.

1333 - 1.4V : 75C
1333 - 1.5V : 75C
1866 - 1.4V : 79C
1866 - 1.5V : 79C

Conclusion : Memory voltage doesn't seem to make a difference, but speed definitely does.


[BCLK/multi]

Jack sh*t. Can't seem to get the Z87-GD65 to take the 1.25 BCLK ratio/strap.

 

[allenk09]

Please don't use Arctic Silver. It was good maybe 10 years ago.

That said, wow Haswell is hot. Would a better air cooler help or is the only solution is to delid?

I repasted with the stock Hyper 212 Cooler Master paste and lost 4C on load, and a few idle. Arctic Silver I would say, is that bad. Wow.

BOOM BABY. 6.8Ghz on air!

Just kidding, CPU-Z was showing some weird clocks when updating frequencies.

 

[Ares202]

I repasted with the stock Hyper 212 Cooler Master paste and lost 4C on load, and a few idle. Arctic Silver I would say, is that bad. Wow.

BOOM BABY. 6.8Ghz on air!

Just kidding, CPU-Z was showing some weird clocks when updating frequencies.

Did you need to raise any other voltages apart from vcore to get to 4.2ghz?

 

[allenk09]

Did you need to raise any other voltages apart from vcore to get to 4.2ghz?

Increased both chipset voltages by like .015. Ram is 1.53 (1.5 stock) Completely stable @ 4.2Ghz with 1.21V core.

I can't get it any higher on this cooler. It sits at 91C on linpack. So 4.2 is where I'm staying for now.

 

[Ares202]

Increased both chipset voltages by like .015. Ram is 1.53 (1.5 stock) Completely stable @ 4.2Ghz with 1.21V core.

I can't get it any higher on this cooler. It sits at 91C on linpack. So 4.2 is where I'm staying for now.

My I5-4670K isnt getting above 55-58C at stock setting running Prime95, so I'm hoping for a decent overclock, but i'm not getting my hopes up too much reading some early reports.

Can you cook eggs at 91C?

 

[Jamil20]

My I5-4670K isnt getting above 55-58C at stock setting running Prime95, so I'm hoping for a decent overclock, but i'm not getting my hopes up too much reading some early reports.

Can you cook eggs at 91C?

What cooler are you running?

 

[allenk09]

My I5-4670K isnt getting above 55-58C at stock setting running Prime95, so I'm hoping for a decent overclock, but i'm not getting my hopes up too much reading some early reports.

Can you cook eggs at 91C?

Luckily, playing BF3 it chills down at 63C. Only the burn tests really cook it.

 

[Ares202]

What cooler are you running?

Only a Corsair H50 (with dual fans) and MX-2 thermal paste, nothing is lapped.

Room temp is 23-24c

Luckily, playing BF3 it chills down at 63C. Only the burn tests really cook it.

Not so bad then

 

[Ares202]

double post

 

[BallaTheFeared]

Luckily, playing BF3 it chills down at 63C. Only the burn tests really cook it.

So why not bump it up for your actual use case instead of testing based on a theoretical max your chip will never be exposed to?

I was under the impression that Haswell had really good throttle, so I don't think there is much harm in going for a bit more "gaming" and just avoiding stability tests that use AVX.

 

[Ares202]

So why not bump it up for your actual use case instead of testing based on a theoretical max your chip will never be exposed to?

I was under the impression that Haswell had really good throttle, so I don't think there is much harm in going for a bit more "gaming" and just avoiding stability tests that use AVX.

+1

You should try for higher, and if your not comfortable with that, drop it down a little

 

[JimmiG]

Just got my Haswell build completed today (minus GPU).

My temps @ stock are about 54C full load in Prime95 8 threads (turbo to 3.7 GHz). The temperature in the room is about 26C (Sweden is experiencing a rare weather phenomena known as "summer"). I'm using the Phanteks PH-TC14PE cooler in a Fractal Design XL R2 case.

For me, Aida64, RealTemp and HWMonitor all show the same values. Speedfan is completely off at 13C idle or something.
Not as bad as I expected, but I guess the temp quickly goes out of control once you start overclocking.

 

[allenk09]

Just got my Haswell build completed today (minus GPU).

My temps @ stock are about 54C full load in Prime95 8 threads (turbo to 3.7 GHz). The temperature in the room is about 26C (Sweden is experiencing a rare weather phenomena known as "summer"). I'm using the Phanteks PH-TC14PE cooler in a Fractal Design XL R2 case.

For me, Aida64, RealTemp and HWMonitor all show the same values. Speedfan is completely off at 13C idle or something.
Not as bad as I expected, but I guess the temp quickly goes out of control once you start overclocking.

Prime95 makes my 4770k yawn and sit at 64C on full load. Linpack/IntelBurnTest makes it cry.

Try IntelBurnTest on maximum.

 

[Kingduqc]

Prime95 makes my 4770k yawn and sit at 64C on full load. Linpack/IntelBurnTest makes it cry.

Try IntelBurnTest on maximum.

I don't get why people use those test, There is literally no workload on the planet that will get even 10c close to those so what's the point? You could run bf3, F@H, video encoding, image rendering, 100 chrome tabs and running cine bench and it will still be cooler then those tests.